Robin Tolmach Lakoff - Context Counts. Papers On Language, Gender, and Power (2017)

i
Context Counts
ii
iii
Context Counts
Papers on Language, Gender, and Power
Robin Tolmach Lakoff
Edited by Laurel A. Sutton
1
iv
1
Oxford University Press is a department of the University of Oxford. It furthers
the University’s objective of excellence in research, scholarship, and education
by publishing worldwide. Oxford is a registered trade mark of Oxford University
Press in the UK and certain other countries.
Published in the United States of America by Oxford University Press

198 Madison Avenue, New York, NY 10016, United States of America.
© Oxford University Press 2017
All rights reserved. No part of this publication may be reproduced, stored in

a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by license, or under terms agreed with the appropriate reproduction
rights organization. Inquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above.
You must not circulate this work in any other form

and you must impose this same condition on any acquirer.
Library of Congress Cataloging-in-Publication Data

Names: Lakoff, Robin Tolmach, author. | Sutton, Laurel A., editor.
Title: Context counts : papers on language, gender, and power /
Robin Talmoch Lakoff; edited by Laurel Sutton.
Description: New York, NY : Oxford University Press, [2017] |
Includes bibliographical references and index.
Identifiers: LCCN 2016024895| ISBN 9780195119893 (pbk. : alk. paper) |
ISBN 9780195119886 (cloth : alk. paper)
Subjects: LCSH: Language and languages—Gender—History—20th century. |
Gender studies—History—20th century. | Feminism—History—20th century.
Classification: LCC P120.S48 L323 2017 | DDC 306.44—dc23
LC record available at https://lccn.loc.gov/2016024895
9 8 7 6 5 4 3 2 1
Paperback Printed by WebCom, Inc., Canada
Hardback printed by Bridgeport National Bindery, Inc., United States of America
v
Illa ipsa loquitor

vi
vii
CONTENTS
Acknowledgments ix
Contributors xi
Introduction 1
By Laurel A. Sutton
1. Language in context (1972) 7
Introduction by Sally McConnell-Ginet
2. The logic of politeness; or, Minding your P’s and Q’s (1973) 37
Introduction by Sachiko Ide
3. Excerpts from two 1974 papers: Pluralism in linguistics;
Linguistic theory and the real world 57
Introduction by Birch Moonwomon
4. You say what you are: Acceptability and gender-related
language (1977) 85
Introduction by Mary Bucholtz
5. Stylistic strategies within a grammar of style (1979) 101
Introduction by Deborah Tannen
6. When talk is not cheap: Psychotherapy as conversation (1979) 137
Introduction by Joan Swann
7. Some of my favorite writers are literate: The mingling of oral
and literate strategies in written communication (1982) 151
Introduction by Jenny Cook-Gumperz
8. Persuasive discourse and ordinary conversation, with examples
from advertising (1982) 183
Introduction by Janet S. Shibamoto-Smith
9. Doubletalk: Sexism in tech talk (1983) 209
Introduction by Susan M. Ervin-Tripp
10. My life in court (1986) 225
Introduction by Susan Blackwell
viii
11. The way we were; or, The real actual truth about generative
semantics: A memoir (1989) 241
Introduction by Georgia Green
12. Review essay: Women and disability (1989) 299
Introduction by Suzette Haden Elgin
13. Pragmatics and the law: Speech act theory confronts
the First Amendment (1992) 315
Introduction by Susan C. Herring
14. The rhetoric of reproduction (1992) 335
Introduction by Laurel A. Sutton
15. True confessions? Pragmatic competence and criminal
confession (1996) 353
Introduction by Linda Coleman
16. Afterword 371
by Robin Lakoff
Index 385
[ viii ] Contents
ix
ACKNOWLED GM EN TS
This book would not have been possible without the expert assistance of
Julia Bernd, who provided eagle-eyed editing, insightful comments, and
tireless research, all of which helped shape this book into a volume worthy
of its author.
Thanks also to Jocelyn Ahlers for creating Chapter 3 through her wise
choices, and for giving me the final push to finish this volume.
This book’s contributors have shown more patience than I thought pos-
sible; I offer thanks and apologies in equal measure. I will always mourn the
fact that the amazing Suzette Haden Elgin passed away before she could
see this book in print.
And finally, thanks to Robin Lakoff for being Robin: she was, and is, the
feminist linguist we need—and sometimes even deserve.
x
xi
CONTRIB U TORS
Susan Blackwell, Department of Language, Literature and Communication,

Vrije Universiteit Amsterdam
Mary Bucholtz, Department of Linguistics, University of California, Santa
Barbara
Linda Coleman, Department of English, University of Maryland
Jenny Cook-Gumperz, Department of Education, University of California,
Santa Barbara
Suzette Haden Elgin, Associate Professor Emeritus (Retired), Department
of Linguistics, San Diego State University; Director, Ozark Center for
Language Studies
Susan M. Ervin-Tripp, Department of Psychology, University of California,
Berkeley, Emerita
Georgia Green, Department of Linguistics, University of Illinois, Emerita
Susan C. Herring, Professor of Information Science and Linguistics, School
of Informatics and Computing, Indiana University
Sachiko Ide, Department of English, Japan Women’s University, Retired
Sally McConnell-Ginet, Department of Linguistics, Cornell University,
Professor Emerita
Birch Moonwomon, Department of English, Sonoma State University
Janet S. Shibamoto-Smith, Department of Anthropology, University of
California, Davis, Professor Emerita
xii
Laurel A. Sutton, Department of Linguistics, University of California,

Berkeley/Sutton Strategy
Joan Swann, Emeritus Chair of English Language, The Open University UK
Deborah Tannen, Department of Linguistics, Georgetown University
[ xii ] Contributors
1
Introduction
BY L AUREL A . SUT TON
T his book did not exist, so it became necessary to invent it.

Or edit it, anyway. The papers collected in this volume represent
over twenty years of groundbreaking research by Robin Lakoff; they origi-
nally appeared in diverse journals, conference proceedings, and volumes
devoted to specialized topics in linguistics. Just try to find them in your
university library.
Language and woman’s place, Lakoff’s seminal 1975 book, remains her
most-cited work, turning up in almost every paper written about gender
and language, and often positioned as Lakoff’s first and last word on the
subject. The importance of Language and woman’s place in the field of socio-
linguistics (and indeed feminism) cannot be overestimated, an impact dealt
with eloquently and at length by Bucholtz and Hall (1995). And yet I think
the long shadow of her book has obscured Lakoff’s subsequent work on
language and gender, language and law, and language and politics, a point
discussed by Bucholtz (2004) in her introduction to the newly annotated
edition of Language and woman’s place. Think of this book as a halogen lamp
in that shadow.
It was not easy to choose from Lakoff’s full list of publications. I wanted
to put together a book that served many needs: a retrospective, a reader,
a history, a reference, and a guide to Lakoff’s theoretical views. While the
papers presented here cover many topics, from hardcore transformational
grammar to advertising devices to judicial speech to anti-abortion propa-
ganda, they all rely on sound linguistic analysis combined with Lakoff’s
keen insight into the use, misuse, and abuse of language. Underlying all
2
her work is Lakoff’s understanding of the many ways in which power and
social relations are expressed in everyday utterances; and it is this under-
standing, always expressed so clearly, that keeps these essays fresh and rel-
evant. Naturally, the examples and “current” linguistic theory discussed in
the oldest papers included here (1970s) are dated, but even these provide
an insider’s view into the field and the feeling of the times.
All of the contributors—Lakoff’s peers, noted linguists in their own
right—gave generously of their time and energy to provide introductions
to each chapter, framing Lakoff’s work in a historical and personal context.
The introductions span a broad range of perspectives, from mini-research
papers to deeply personal anecdotes. They are truly the icing on the cake.
Those who know Lakoff only as a feminist may be pleasantly surprised
by the diversity of subjects covered in this volume; those who know her
only as the author of Language and woman’s place will now have an account
of her linguistic research and writing from that time until the late 1990s.
And for those young scholars just beginning to think about language,
I hope this book provides a rich resource of intelligent commentary and
analysis to which they will return again and again.
REFERENCES
Bucholtz, Mary. 2004. Introduction. Language and woman’s place: Text and

commentaries, ed. by Mary Bucholtz, 3–14. Oxford: Oxford University Press.
Bucholtz, Mary, and Kira Hall. 1995. Introduction: Twenty years after Language and
woman’s place. Gender articulated: Language and the socially constructed self,
ed. by Mary Bucholtz and Kira Hall, 1–22. New York and London: Routledge.
Lakoff, Robin. 1975. Language and woman’s place. New York: Harper and Row.
[ 2 ] Context Counts
3
Introduction to “Language in context”

BY SALLY MCCONNELL-G INET
W hen “Language in context” first appeared in 1972, it broke much new

ground, turning linguists’ attention to the rich and subtle resources
languages offer their speakers for articulating and managing their social
relations with one another. Lakoff was certainly not the first to observe,
for example, that English modals might express different degrees of polite-
ness, and that their value in a particular utterance is very heavily depen-
dent on extralinguistic features of the context in which that utterance is
made. Traditional descriptive linguists had made such observations. She
was, however, one of the very first linguists trained in the generative gram-
mar tradition to suggest that such phenomena merit linguistic explanation
and that their analysis might be especially crucial for illuminating cross-
linguistic comparisons. The word “explanation” is key here. Lakoff aims
in this paper to go beyond taxonomy and description to formulate some
general explanatory principles. Her comparisons of English and Japanese
are designed to show the potential universal applicability of some of these
principles and to explore some of the ways in which their instantiation dif-
fers cross-linguistically and cross-culturally. This emphasis on the search
for general explanatory notions was very much in keeping with the tenets
of generative grammar, the framework within which she had begun her
own linguistic work.
At the time this paper first appeared, theories of semantic competence
were just beginning to be developed in generative linguistics. There was
already a debate about the nature of the semantics–syntax interface. The
4
“generative semantics” camp (with which Lakoff was identified) held that
semantic phenomena “drive” the grammar. That is, semantic representa-
tions are in some sense basic (and, for most of these and other linguists,
universal), and it is language-particular grammars that yield actual surface
forms “from” semantic input. “Interpretive semantics,” on the other hand,
took syntactic representations of some kind as input to the semantics, with
semantics feeding on syntax rather than vice versa. But perhaps the critical
difference was that the generative semanticists took semantic representa-
tions to be fundamentally syntactic, governed by exactly the same prin-
ciples that operate generally in syntax; interpretive semantics left open
the possibility that semantic representations might be quite different from
syntactic.
Now the question of how best to describe and analyze the dependence
of interpretation on context—a major issue that “Language in context”
raises—was not really being systematically addressed by either semantics
camp. Even less attention was being paid to the social ramifications of lan-
guage use. Although a few linguists had begun to explore proposals from
the philosophy of language about the heavily contextual character of lin-
guistic communication (especially those inspired by Grice 1968), the field of
linguistic pragmatics did not yet exist. Sociolinguistics was also in its very
early stages, and virtually all the work available (e.g., Labov 1972) empha-
sized socially conditioned variation of the sort that distinguishes dialects
and plays a role in language change. Ideas about what is now sometimes
called “communicative competence” were still in their infancy; “interac-
tional sociolinguistics” and “discourse analysis” were not yet part of lin-
guists’ vocabularies. Lakoff’s article, published in the prestigious and
widely read journal Language, was an important spur to subsequent work
on questions of the importance of context to the understanding of natural
language utterances and their social effects.
In rereading “Language in context,” I was surprised to find that it did
not actually address in much detail the question of how investigations of
contextual matters were to be integrated with the rest of linguistic inquiry.
As I remembered, there is the claim that “traditional transformational
grammar” (which was accompanied by no semantics, much less any prag-
matics) fails when confronted with language in context. The idea is that
the “applicability” of grammatical rules must be conditioned by contextual
phenomena because a rule is considered “inapplicable” in contexts where
its application would produce any kind of oddness at all. Contextual mat-
ters as well as other semantic phenomena are seen as essential inputs to
the grammar, the task of which Lakoff assumes (without argument) is to
predict the acceptability or appropriateness (and perhaps even the social
5
efficacy) of utterances—that is, sentences uttered in particular contexts.

“Grammaticality” is no longer a notion that applies to sentences as such but
is reserved for sentences together with social contexts, which include inter-
locutors and their sociolinguistically relevant properties (including their
attitudes) and relations. But Lakoff never really considers the possibility
that social appraisal of utterances might be explained outside grammar
proper—that social and syntactic deviance might be very different kinds of
phenomena. Or, if not extragrammatical, social and other contextual prop-
erties of utterances might be assigned to syntactically well-formed senten-
tial structures without figuring at all in syntactic derivations.
In the end, however, this generative semantics/pragmatics stance does
not really matter very much. What I found especially interesting in reread-
ing the paper is how little Lakoff’s theoretical position (against which
I have often argued, and which frames the paper as a main point) actually
affects her discussion. Whether we pack contextual factors into the syntax
or not, we still need somewhere to say something about what particular
sentential structures indicate about the preferred contexts of their use and
what effect their utterance has on contexts. What she shows quite convinc-
ingly is that relative social status, age, sex, and other aspects of social iden-
tities and relations have linguistic underpinnings not only in languages
like Japanese, which have a whole array of forms apparently specialized
for social interactive purposes, but even in languages like English, which
at first glance might appear socially neutral. Although English speakers do
not express deference or superiority or consideration in the same ways nor
under exactly the same circumstances that Japanese speakers do, Lakoff
shows that they must attend to such factors at least implicitly in their lin-
guistic performance. And they have a rich tacit knowledge of the sociolin-
guistic implications that discriminate among alternative ways of conveying
what is roughly “the same” message.
“Language in context” set the stage for “Language and woman’s place”
(Lakoff 1973), first published the following year. Lakoff is already noting
that interlocutors’ gender is sociolinguistically important in English as
well as in Japanese, and she is already hypothesizing, for instance, that
tag questions might be an important component of the sociolinguistic
construction of gender in English (of course, she doesn’t put it in quite
these terms, which were not really available then). And she already forges a
strong link between matters of gender and questions of male privilege and
female subordination. Although her feminist voice was to strengthen in
later work, we can hear it beginning to speak in this early article. Her use
of generic masculines, which continues in “Language and woman’s place,”
may jar some contemporary feminist readers. She writes, for example, that
L A N G UAG E I N C O N T E X T [ 5 ]
6
the use of certain particles in Japanese provides “implicit personal infor-

mation about the speaker, about his sex and status, relative to that of the
addressee” (p. 23). A sentence like “the speaker of Japanese must make his
(or her) sex explicit in most conversations” (n. 5) suggests, however, that
she was chafing a bit against what was then the overwhelmingly standard
pronominal usage.
In short, “Language and context” is still very much worth reading. It
is, of course, the product of its time, and in some ways it seems dated.
But it makes some important general points about the social power of
language that can stand even if the theoretical trappings and the detailed
claims about how English works (and, to a lesser extent, Japanese) might
be problematic.
“Language in context” pushed me and many others who read it as gradu-
ate students to think more seriously about the social implications of lan-
guage, and especially to pay attention to the nitty-gritty detail of socially
significant linguistic choices. In the quarter-century since this paper first
appeared, the study of language in context has become an increasingly
important area of linguistic inquiry. Robin Lakoff’s voice was one of the
first and most forceful to address topics in this area, and its articulate
energy was all the more remarkable in 1972, given the dismissive attitude
of most mainstream linguists toward such work.
Thank you, Robin, for daring to speak out as a linguist on matters that
the leading (mainly male) research linguists had dismissed as outside the
province of linguistics. And thank you especially for inspiring others of us
to try to follow your example.
REFERENCES
Grice, H. P. 1968. The William James Lectures. Published with other material
in H.P. Grice (1989), Studies in the way of words. Cambridge, MA: Harvard
University Press.
Labov, William. 1972. Sociolinguistic patterns. Philadelphia: University of
Pennsylvania Press.
Lakoff, Robin. 1973. Language and woman’s place. Language in Society 2.45–79.
7
CHAPTER 1
Language in context (1972)
Traditional transformational grammar attempts to define the conditions on the applica-

bility of grammatical rules on the basis of superficial syntactic environment alone. This
paper discusses a number of examples in several languages that show that such a goal
is unattainable—that, in order to predict correctly the applicability of many rules, one
must be able to refer to assumptions about the social context of an utterance, as well as
to other implicit assumptions made by the participants in a discourse.
W hen studying exotic languages, the speaker of English often runs

into odd facts. As if the syntactic, lexical, and morphological pecu-
liarities with which other people’s languages are unfortunately replete were
not enough to confound the English speaker, he encounters still odder
details: things which, as far as he can see, have no analogs in English at
all. It is certainly bad enough to encounter case languages, or languages
with complex and synthetic tense systems, or absolute constructions, or
six words for “snow”; but at least these are analogous to things that occur
in English. But what about certain still stranger phenomena? How does
the native speaker ever learn these weird distinctions? How can he ever
remember to make them, in the course of ordinary conversation? Doesn’t
he inevitably (though accidentally) offend everyone he encounters, or
incessantly stamp himself as a boob?
This paper originally appeared in Language 48:907–27. Reprinted here with permis-
sion of the publisher.
I should like to thank the following people, who have served as informants or made
valuable suggestions regarding the Japanese data: Chisato Kitagawa, Tazuko Uyeno,
and Kazuhiko Yoshida. I should also like to thank George Lakoff for much helpful dis-
cussion. All errors and misinterpretations are, of course, my own responsibility.
8
The problems I am referring to will of course be immediately recogniz-

able to anyone who has done any reading about almost any language that
is not English—that is, I should think, any linguist. I refer to phenomena
such as the following:
(i) Particles, like doch in German, or ge in Classical Greek, or zo in

Japanese. How do you know when to use them? And how do you
know when not to? Are they inserted in sentences randomly? Since
these particles do not add to the “information content” conveyed
by the sentence, but rather relate this information content to the
feelings the speaker has about it, or else suggest the feelings of the
speaker toward the situation of the speech act, it is sometimes rather
cavalierly stated that they are “meaningless.” If this were really true,
it would of course be impossible to misuse them. But we all know
that there is nothing easier for the non-native speaker.
(ii) Honorifics. Asian languages, Japanese in particular, are infamous
for containing these. Using them in the wrong situation will, one
is assured, result in instantaneous ostracism. But how do you know
when the situation is wrong? The non-native speaker apparently
never sorts it out. Can the native speaker (who is linguistically
naive) be expected to do any better?
(iii) Many languages have endings on verbs, or special forms related to
the verbal system, that are used to suggest that the speaker himself
doesn’t take responsibility for a reported claim, or that he does—
that he is hesitant about a claim he is making or confident of its
veracity. How can a speaker keep track of these mysterious concepts?
Are speakers of other languages conceivably that much smarter than
we are? Then why don’t they have a man on the moon?
The purpose of this paper is to explore these questions. I will not really
attempt to answer the question, “How do they do it?”—we don’t know how
people do even the simplest and most obvious linguistic operations. But
what I will show is that these phenomena also occur in English. It is often
not superficially obvious that we are dealing, in English, with phenomena
analogous to politeness or hesitance markers in other languages; there
are often no special separate readily identifiable morphological devices.
Rather, these distinctions are expressed by forms used elsewhere for other
purposes. Therefore it is easy to imagine that they are not present at all.
But I hope to show that the reverse is true; and further, that if the pres-
ence and uses of these forms are recognized, several of the most difficult
9
problems confronting such diverse areas as theoretical linguistics and the

teaching of second languages will be solvable. Thus I am in effect making
two claims, the first of theoretical, the other of practical, interest:
(a) Contextually linked linguistic phenomena are probably identifiable,

to one extent or another, in all the languages of the world. But one
language may have special markers for some or many of these possi-
bilities, while another language may utilize forms it uses elsewhere for
other purposes. One language may require that these markers be pres-
ent, while another may consider them optional, or to be used only in
case special classification is desired, or for special stylistic effects. (As
we shall see later in this paper, Japanese is apparently a representative
of the first class of languages, English of the second. Hence, as many
speakers of Japanese have said to me, English sounds “harsh” or “impo-
lite” to them; while to the speaker of English, Japanese often gives the
effect of being unbelievably subtle, making inordinately many unnec-
essary distinctions.) But we should ask, not only whether a language is
one type or the other or a mixture of both, but also whether this fact
about a language is related to any other facts, deep or superficial, about
its structure. Since questions of this sort have not been studied in any
disciplined way heretofore, nothing is known at present. The answers,
if ever found, would be of interest in studies of the lexicon, the forms
of logical structure, the identification of linguistically relevant types of
presupposition, and many other areas with which linguistic theorists
are at present concerned.
(b) If one is to teach second-language use successfully—so that a non-
native speaker can use the language he is learning in a way reminiscent
of a native speaker, rather than a robot—then the situations in which
forms of this type are usable in a given language must be identified. It
is obviously useless to try to list or pinpoint the superficial syntactic
configurations where they are correctly used; examples will be given
later in the paper that illustrate the problem. We must then identify
the means by which the second language makes these distinctions, and
pair the two, although in terms of superficial syntax, the two languages
will appear to have little in common.
We may distinguish certain aspects of context from others. Some have

universal linguistic relevance; others may be linguistically relevant under
certain conversational situations but not others, or for certain cultures but
not others; and still others may never be linguistically relevant as far as
L anguage in context [ 9 ]

10
we know. So it is normally true in all languages and all situations that one
must somehow make clear the type of speech act involved: are you asking
a question, making a statement, or giving an order? Ambiguities in this
regard are generally not tolerated.1 Some languages require that you know
more about the speech situation than this. English sometimes requires
overt notice as to whether the speaker believes a past-time event is rele-
vant to the present, by the use of the perfect tense rather than the preterit.
Other languages require that there be overt expression of the identity of
speaker and/or addressee: What are their respective social positions? And,
related to this, of course, what are their respective ages and sex? English
only sometimes requires that these be recognized overtly; other languages,
such as Japanese, require it much more often. But it is hard to think of a
language that requires one special overt marker if the speaker has blue eyes,
and a different one if the speaker has brown eyes. This is contextual infor-
mation, as real and available to the speakers of a language for the purpose
of making distinctions as are differentiation of age and sex; yet only the
latter two often occur as linguistically significant contextual information.
In any case, I trust that, by the end of this discussion, it will be per-
fectly clear that there are areas of linguistic competence that cannot be
described in any theory that does not allow an integration of information
about the context in which the discourse takes place—sometimes erro-
neously referred to as “realworld” as opposed to “linguistically relevant”
situation—and the purely linguistically relevant information the sentence
seeks to convey: superficial syntax, choice of lexical items, and semantics
aside from contextually relevant meaning elements.
I shall try to substantiate some of the claims I have been making by
looking at examples.
We all know, or at least know of, languages that employ honorifics
as essential elements in sentences. Sometimes they occur with personal
names, and in these cases it is fairly easy to see what is going on: one usu-
ally assumes that the speaker either actually is lower in status than the
addressee, or is speaking as if he were. In the latter case, which is perhaps
the more usual in conversational situations, it is assumed that this linguis-
tic abasement occurs for reasons of politeness. But an important question
1. Gordon & G. Lakoff (1971) discuss a number of interesting cases where, if one looks
only at superficial syntactic configurations, apparent ambiguities of this type do in fact
exist: e.g., It’s stuffy in here, most normally a declarative statement, may, under specific,
contextually determined conditions, be interpretable as an imperative, equivalent to
Please open the window. As they show, this does not indicate that such sentences really
are ambiguous between the two interpretations: it indicates rather that context must
play a role in the interpretation of sentences.
11
is usually glossed over: Why is it polite for the speaker to suggest that the
addressee surpasses him in status? In some languages we find honorifics
related to non-human items, to show that the speaker considers them of
importance in one way or another. How is this related to any notion of
politeness, which is a concept involving behavior between human beings?
Another problem is that many languages apparently have two kinds of
honorifics. One is the kind I have just mentioned. But going hand in hand
with this is the use of forms that humble or debase the speaker himself,
or things connected with him. Translated into English, this often has ludi-
crous results, e.g., “Honorable Mr. Snarf have some of my humble apple
pie?” This sort of translation is ludicrous for several reasons, but perhaps
principally because, by translating the honorific and dis-honorific, if I may
use that term, with overt adjectives, the sense of the sentence has been
palpably altered. In the original language, the sense of superiority or infe-
riority conveyed by the honorifics is presupposed, or implicit. The use of
adjectives like honorable and humble makes these concepts explicit. So
what had been a tacit suggestion, in effect, is now made overt. The English
translations do not, I think, allow the monolingual speaker of English to
get any sense of how a speaker of Japanese feels when he is addressed
with -san. But I believe there are locutions in English whose force comes
close to that of the true honorific, because the differentiation in status
they establish is implicit rather than overt. These forms are also used for
the sake of politeness (as adjectives like humble and honorable never are).
I said earlier that these contextually linked forms had not been recog-
nized in English partly because the forms utilized for this purpose had
other, more obvious uses. English modals are a case in point. Certain uses
of the modal must are parallel to the use in other languages of special
honorific forms:
(1) You must have some of this cake.

(2) You should have some of this cake.
(3) You may have some of this cake.
Let us assume, for the purpose of analyzing these sentences, a special social
situation: a party, at which the hostess is offering the guests a cake that
she baked herself or at least selected herself, and which she therefore takes
responsibility for. In such a social context, (1) is the most polite of these
forms, approaching in its range of appropriateness that of a true honorific
in languages that have such forms. Further, although in theory (2) should
be more “polite” than (1), in actual use it is not: in the situation established
above, the use of (2) would be rude, while (1) would be polite. And (3),

12
which might at first seem the most polite form, actually is the least. Why
is this?
Finding the answer lies partly in determining what constitutes polite-
ness, and of course, its opposite, rudeness. If we can define these notions,
then the uses of these modals will be seen to be governed by the same
assumptions of politeness as govern the use of honorifics; once the prin-
ciple is understood, it can be transferred from language to language. What
we are dealing with here is something extralinguistic—the way in which
individuals relate to one another—that directly affects the use of lan-
guage. We must understand something about non-linguistic social inter-
action before we can see the generalization that is in effect regulating the
use of sentences like (1)–(3), along with the use of affixes like -san and o- in
Japanese.
It is obvious, of course, that what passes for politeness in one culture
will appear to a member of another culture as slavishness or boorishness.
We are all familiar with examples of this. Then how can we talk about
universal conditions governing the use of honorifics and other politeness
markers? I think we can assume that there is a universal definition of
what constitutes linguistic politeness: part of this involves the speaker’s
acting as though his status were lower than that of the addressee. What
may differ from language to language, or culture to culture—or from
subculture to subculture within a language—is the question of when it
is polite to be polite, to what extent, and how it is shown in terms of
superficial linguistic behavior.2 Although a speaker may know the univer-
sal definition of politeness, he may apply it at the wrong time or in the
wrong way if he attempts to transfer the uses of his own language directly
into another; hence the ludicrousness that results from taking a polite
concept implicit in one language and making it explicit. If, in a given lan-
guage, one’s own possessions are customarily followed by a marker of
humility (a situation which perhaps can be symbolized by (4) below), it
does not follow that (5), in which what is implicit in the marker in (4)
is made explicit, is a reasonable English translation of (4). In fact, as has
2. So, for example, if an officer in the Army (a subculture with special status-related
rules) gives a command to a private, he will not normally preface his command with
please. Although in most English-speaking groups the use of please prefaced to an
imperative is a mark of politeness, to use please in this situation will be interpretable as
sarcastic. Again, in some cultures it is considered polite to refuse an invitation several
times before one is conventionally “prevailed upon” to accept: if a speaker from such a
culture finds himself in one where it is considered polite to accept invitations at once
with thanks, confusion and worse will inevitably ensue, with each party impressing the
other as unbelievably boorish or stupid.
13
already been noted, the effect of such bogus translations is generally

laughable, and rightly so:
(4) Have some of this cake—yecch.

(5) Have some of this revolting cake.
My claim is that a sentence like (1) is a much closer translation of (4) than

(5) is, although (5) stays closer to the superficial syntax of the original lan-
guage. The task of the translator then is compounded: he must translate
contextual and societal concepts—contexts that are, strictly speaking,
extralinguistic—in addition to merely translating words and ideas and
endings.
Let me try to be more specific in identifying (1), but not (2) or (3), as an
honorific form in an extended sense of the term. (I will define “honorific”
as a form used to convey the idea that the speaker is being polite to the
hearer.) At first it seems contradictory to say that a sentence containing
must is more polite than one using should or may. Going by the ordinary
uses of the modals, must imposes an obligation, while should merely gives
advice that may be disregarded, and may allows someone to do something
he already wanted to do. Surely it should be more polite to give someone
advice, or to let someone do as he wishes, than to impose an unavoidable
obligation upon him.
Normally this is true, but under special conditions the reverse is the
case, and this is the situation in (1)–(3). If we want to understand why
these modals work as they do here, we must ask: Under what realworld
conditions is it appropriate to use each of these modals? So, for example,
if the use of must expresses the imposition upon its superficial subject of
an obligation (whether by the speaker or by someone else, with the speaker
merely reporting the fact), under what conditions in the real world is it
necessary to impose an obligation? The answer is simple: it must be the
case that the person on whom the obligation rests would not do what he
is instructed unless he were obliged to do it. That is, the assumption is
that performing the act is distasteful, requiring coercion of the superficial
subject.
Now in a normal situation it is not polite to coerce anyone, since,
among other things, such action reminds him that you are his superior in
power. Thus must is normally used for politeness only when it is a second-
hand report that an obligation is imposed, on the addressee or on a third
person, by someone other than the speaker. In this situation, the speaker
is not using must as a means of coercion through his greater power or pres-
tige; but he is so doing when must reflects the speaker’s own imposition

14
of an obligation. In a sentence like (1), the most natural assumption is

that the speaker himself is imposing an obligation on the hearer. Then
why is (1) a polite offer? Why does one not take umbrage when such a
sentence is spoken to one, as a dinner guest, by one’s hostess? We seem to
be faced with an utterance that is, in a special sense, “ambiguous.” This is,
of course, no normal type of ambiguity, since it cannot be disambiguated
by linguistic context or by paraphrase. Rather, the addressee, hearing a
sentence like (1), disambiguates it in terms of the social situation in which
he is exposed to it.
Let me be more precise. Suppose you overhear the sentence Visiting rela-
tives can be a nuisance in isolation. You have no way of knowing whether the
speaker is talking about relatives who visit, or the act of visiting one’s rela-
tives. But if the hearer has also heard prior discourse, and if, for example,
this discourse was concerned with a discussion of the properties of rela-
tives, and when relatives were a nuisance, the hearer is able to disambigu-
ate the sentence by linguistic means.
Now we know that the modal must is actually an amalgam of several mean-
ings, all related but differentiable. (I will confine my discussion to the root sense
of must for obvious reasons.) As suggested above, these related meanings are:
(a) The speaker is higher in rank than the superficial subject of must, in
sent. (1) identical with the addressee. As such the former can impose
an obligation on the latter.
(b) The thing the addressee is told to do is distasteful to him: he must be
compelled to do it against his will.
(c) Something untoward will happen to the addressee if he does not carry
out the instruction.3
Any of these assumptions might be primary in a given instance. In non-

polite situations, normally (a) is paramount in sentences like (6), and (c) in
cases like (7); it seems to depend on context.
3. It seems reasonable to believe that, of the three assumptions comprising the
meaning of must, (a) and (c) are first-order presuppositions, and (b) second-order. The
reason for making this claim is that (a) and (c) can be questioned, as is typical of first-
order presuppositions, while (b) cannot, as the following examples show. In reply to,
e.g., You must take out the garbage!, the respondent might retort, under the appropriate
circumstances, with You can’t make me! or Who’s gonna make me? (which contradict (a),
and are equivalent to “You don’t have the authority”), or with So what if I don’t? (which
contradicts (c), and is equivalent to “If I don’t do it, I won’t suffer”). But he cannot reply
with *I want to anyhow! (which would be a contradiction of (b) and equivalent to “I am
not being made to do it against my will.”).
15
(6) You must clean the latrine, Private Zotz: this is the Army, and I’m
your sergeant.
(7) You must take this medicine, Mr. President, or you will never get
over making those awkward gestures.
Theoretically, then, a sentence like (1) should be triply ambiguous, and two

of the ambiguities should be rude. In fact, if taken out of context, such
a sentence would be just as mysterious to the hearer as Visiting relatives
can be a nuisance. But just as with the latter, (1) is swiftly disambiguated if
one is aware of the context. For (1), it is extralinguistic context: one knows
one is being addressed by the hostess proffering her cake, and one accord-
ingly decides on meaning (b). (Of course, if (1) were spoken by a member of
the Mafia whose wife had baked the cake, the range of possible choices of
meaning might be wider.)
Why are should and may less polite in this context? In the case of these
modals, we are making rather different assumptions about the willingness of
the subject to perform the act, and it is here, I think, that the non-politeness
lies. With should, there is normally no assumption that the action is to be per-
formed against the subject’s will: the speaker is making a suggestion to the
addressee to do something that might not have occurred to him, but there is
no hint that he would be averse to it, or would have to be compelled to do it.
In fact, the use of should indicates that the speaker is not in a position to use
duress to secure compliance: he can suggest but not coerce. In non-polite use,
then, should is more polite than must since the speaker is not suggesting his
status is such that he can coerce the addressee. But this implies that he need
not coerce the addressee, and for this reason the “humbling” force of must
is absent. But should by itself is not really a politeness marker: it does not
humble the speaker, but merely makes him the equal of the addressee. So the
use of should in the dinner party situation is not particularly polite: in fact,
it is rather rude, since the hostess is suggesting that it would be better for
the addressee if he had some cake—that is, that the cake is too good to miss.
From this assumption, the implication follows that the hostess’ offering is a
good thing—contrary, as we have seen, to the rules of politeness. As a further
example of this, consider what happens if the hostess should overtly make the
same suggestion. The same sense of impropriety ensues from (8) as from (2):
(8) Have some of this delicious cake.
But if another guest is offering the cake, both (2) and (8) are perfectly
appropriate and usual, since the guest is not praising his own property. This
shows that implicit and explicit assumptions—in this case, of the value of

16
one’s own possessions—work the same way in determining appropriate-

ness, and both work the same way as honorifics in other languages:
(9) Have some of this ‘o-cake’.4

(10) Have some of my friend’s ‘o-cake’.
(11) You should have some of her cake.
(12) Have some of her delicious cake.
Finally, it is now easy to see why may in (3) is not a polite form: in fact,
its use makes two assumptions, both of which are counter to the conven-
tions of politeness: (a) that the person who is able to grant permission (by
the use of may) is superior to the person seeking it; (b) that the person
seeking permission not only is not averse to doing the act indicated, but
wishes to do it. Then the further assumption is that, as far as the person
receiving permission by sentence (3) is concerned, having the cake is a good
thing. As with should, this is counter to the usage of politeness.
These examples show several things. First, there are uses of the modals
that reflect politeness, in terms of relative status of speaker and hearer,
and implicit desirability of the act in question. In this respect these modal
uses are parallel to the use of honorifics in other languages. Second, in
order to tell how a modal is being used, and whether certain responses to
it are (linguistically) appropriate, one must be aware of many extralinguis-
tic, social factors. Just as, in speaking other languages, one must be aware
of the social status of the other participants in a conversation in order
to carry on the conversation acceptably, so one must at least some of the
time in English, a language usually said not to require overt distinctions of
this sort.
There are many other examples of politeness conventions explicitly
realized in English. One is the use of imperatives, a task fraught with
4. According to Tazuko Uyeno, although not every Japanese noun may receive the
o-honorific prefix, those that can behave as suggested in the text. E.g., the word taku
“house” will take the prefix o- when it refers to the home of someone other than the
speaker and will occur without o- when the speaker’s own house is being referred
to. The same informant points out an interesting difference in polite usage between
Japanese and English, also relevant at this point: I have noted above that in English the
modal must, ordinarily not a polite form, may be interpreted as polite in specific social
contexts where one is able to “ignore” certain aspects of the meaning of must. But in
Japanese, this is not the case: I cannot use the word-for-word equivalent of “You must
have some cake” as a polite utterance equivalent to its English translation. It would, in
fact, be interpreted as rude under the circumstances. One must rather say something
like, “Please have some cake as a favor to me.” Thus it is not necessarily true that one
can “ignore” the same aspects of meaning in two languages.
17
perils for one who does not understand the application of levels of polite-
ness in English. For example, consider the following ways of giving an
order. When can each be used appropriately? What happens if the wrong
one is used?
(13) Come in, won’t you?

(14) Please come in.
(15) Come in.
(16) Come in, will you?
(17) Get the hell in here.
It would seem clear that these sentences are ranked in an order of descend-
ing politeness. To use (17), your status must be higher than that of the
addressee; moreover, you must be in such a situation that you don’t even
care to maintain the conventional pretense that you are addressing him
as an equal. That is, (17) deliberately asserts the superiority of the speaker
over the addressee, and as such is rude in a situation in which it is not nor-
mal to make this assertion.5 By contrast, (15) merely implies this assump-
tion of superiority: it assumes compliance, and hence suggests that the
speaker has the right to expect this compliance, and that the speaker
therefore outranks the addressee; but it does these things much more
covertly than (17). But, though not normally a rude form, it is still not
really a polite one. Again, however, we must make an exception for one
case, analogous to the one made in the first set of cases with modals: if the
5. This claim ignores the “jovial” use of sentences like (17) as used between close
friends, almost invariably male. Other examples are: Get your ass in here, Harry! The
party’s started! and What makes you think you can go by my house without coming in, you
asshole? It seems that, between close male friends in some American subcultures at any
rate, the purpose of such otherwise unpardonably rude exclamations is to say, “We’re
on such good terms that we don’t have to go by the rules.” This linguistic impropri-
ety occurs in relationships of the same degree of closeness as those which allow their
members, for example, to invite themselves over to each other’s houses—otherwise a
non-linguistic breach of propriety of similar magnitude. This illustrates again the par-
allelism of linguistic and non-linguistic concepts of politeness. These examples show,
incidentally, that English, like Japanese, makes sex distinctions in the types of sen-
tences possible. While a woman in most American subcultures would never use the
above sentences, she might use the following to much the same effect, but lacking the
obscenities: Go ahead, have some more cake, Ethel—you’re so fat, who’ll notice if you get
fatter? Between very close friends, such a remark might be taken as an acceptable joke,
but under any other conditions it is an unpardonable insult. There are other expres-
sions confined to the feminine vocabulary: in particular epithets like “gracious!” or
“dear, dear.” So English is again not so very unlike Japanese, except that the speaker
of English can refrain from these usages altogether, but the speaker of Japanese must
make his (or her) sex explicit in most conversations.

18
addressee is at the speaker’s door and is a friend, (15) is much more nor-
mal than (13)–(14) as an invitation to enter. The first two, in fact, do not
seem polite in this context: they give the impression of forced hospitality.
Here again we seem to be depending on a more complex notion of polite-
ness: both (13) and (14), like (2), imply that the addressee has the choice of
complying or not—that his status is sufficiently high with respect to the
speaker that he can obey or not as he sees fit—while (15), like (1), seems at
first to suggest that the addressee has no choice, that his status is so low
that he is obliged to obey. Yet both are relatively polite in this sort of social
context. The reason in the case of (16) is parallel to that in (1): the speaker
is implying here (by convention: he doesn’t really make this assumption, of
course; it would be bizarre if he did) that the addressee doesn’t really want
to come in, that he will enter only under duress. Since (13)–(14) do not
allow this assumption, they are less polite. So again the two definitions of
politeness—status vs. desirability of the speaker’s offering—are at odds,
and again the latter seems stronger. When the speaker is not really offer-
ing something of his own, the status assumption becomes paramount,
and (13)–(14) become more polite than (15). This is the ease in a doctor’s
office, for example, where the receptionist is more likely to use (13)–(14).
Again, (13)–(14) are likely to be used for “forced” politeness—e.g., when
inviting an encyclopedia salesman in, under duress. I am not sure why
this is so. But it is also true that a superior may address an obvious infe-
rior (for example, in the Army) by (15), with no sense of sarcasm, i.e., no
sense that he is being inappropriately polite. But if an officer addresses a
private with (13), he is necessarily being sarcastic. There is no possibility
of sarcasm, however, in the use of humbling forms of politeness, such as
are found in (1) and (15). This is reminiscent of a fact that has been known
for some time about presupposition in general: a first-order presuppo-
sition may be negated or questioned, under some conditions; a second-
order presupposition cannot be. This suggests that the type of politeness
involved in a usage like (15) or (1) is more complex in derivation than is the
simple status-equalizing case in (13) or (2). In fact, it is probably true that
the humbling type allows the status type to be deduced from it (if what
I have is no good, one can deduce that I don’t outrank you, in this respect
anyway), so that the humbling type of politeness is one level deeper than
the status type.
There are other assumptions, made in normal conversation, that are not
tied to concepts of politeness. These, too, show up in non-obvious ways in
the superficial structure. Some types which have been discussed by Grice
([1967] 1975), as well as by Gordon & G. Lakoff, are rules of conversation.
19
In a normal conversation, the participants will make the following assump-

tions, among others, about the discourse:6
Rule I. What is being communicated is true.

Rule II. It is necessary to state what is being said: it is not known to
other participants, or utterly obvious. Further, everything neces-
sary for the hearer to understand the communication is present.
Rule III. Therefore, in the case of statements, the speaker assumes
that the hearer will believe what he says (due to Rule I).
Rule IV. With questions, the speaker assumes that he will get a reply.
Rule V. With orders, he assumes that the command will be obeyed.
All these assume, in addition, that the status of speaker and hearer is
appropriate with respect to each other. (Of course, there are special situa-
tions in which all these are violated: lies, “small talk,” tall stories, riddles of
certain types, and requests as opposed to commands. But in general these
conditions define an appropriate conversational situation.)
But sometimes, even in ordinary conversational situations, some
of these rules are violated. This is analogous to violating a rule of gram-
mar: normally we should expect anomaly, lack of communication, etc. When
this is done baldly, e.g., by small children or by the insane, we do in fact
notice that “something is missing”; the conversation does not seem right.
But in ordinary discourse among normal individuals we can often discern
violations of these rules and others, and yet the total effect is not aberrant.
One way in which apparent contradictions are reconciled is by the use of
particles like well, why, golly, and really. Although these are often defined
in pedagogical grammars as “meaningless” elements, it seems evident that
6. These implicit rules show up overtly in certain locutions. Cf. the following:
{ } believe
John is a Communist, and if you don’t *obey
*answer
me, ask Fred.
{ }
*believe
Get out of here, and if you don’t obey me, I’ll sock you.
*answer
{ } *believe
I ask you whether John left, and if you don’t *obey
answer
me, I’ll be furious.
These examples show that, with each type of speech act—declaring, ordering, and
asking—an “appropriate” type of response is associated, and that this association
shows up linguistically in superficial structures.

20
they have real, specific meanings, and therefore can be inappropriately used.
It is therefore within the sphere of linguistics to define their appropriate
usage. Moreover, this appropriateness of usage seems at least sometimes
to involve the notion of “violation of a normal rule of conversation.” These
particles serve as warnings to participants in the discourse that one or more
of these rules is about to be, or has been, violated. When this warning is
given, it is apparently legal to violate the rule—that is, of course, only the
specific rule for which the warning was given. Otherwise confusion results.
I have shown elsewhere (R. Lakoff 1973) that the English particles well and
why function in this way. Well serves notice that something is left out of the
utterance that the hearer would need in order to understand the sentence—
something, normally, that he can supply, or that the speaker promises to sup-
ply himself shortly. That is, well marks a violation of the second part of Rule II.
Why indicates that the speaker is surprised at what the addressee has said: it
suggests that perhaps the prior speaker has violated Rule I, in the case of a
statement, or II, in the case of a question. Other analogous cases in English
involve special syntactic configurations rather than particles. Consider sen-
tences like the following:
(18) Leave, won’t you?

(19) Leave!
(20) John left, didn’t he?
(21) John left.
In the even- numbered examples above, we have tag- forms, one for a
command—as discussed earlier—and one for a statement. It is worth ask-
ing whether these superficially similar structures have any deeper similar-
ity: whether the reasons for applying tag-formation to imperatives are related
to the reasons for applying this rule to declarative statements. It is more or
less traditional in transformational literature to suggest that the two types
of tags have little in common aside from superficial similarities of forma-
tion. But there are reasons for supposing that, in fact, there are real semantic
reasons for this apparent superficial coincidence. This is a rather satisfying
hypothesis, if it can be substantiated: it would suggest that these two bizarre
and highly English-specific formations have a common function, so that two
mysteries may be reduced to one.
With reference to Rules III and V above, one way in which a tag question
like (20) is distinguished from an ordinary statement like (21) is that the
speaker really is asking less of the hearer. A speaker can demand belief from
someone else only on condition that he himself fully believes the claim he
is making. But the function of the tag is to suggest that the speaker, rather
21
than demanding agreement or acquiescence from the hearer (as is true in a

normal statement), is merely asking for agreement, leaving open the possi-
bility that he won’t get it. So a tag-question is really intermediate between a
statement and a question: a statement assumes that the addressee will agree,
and a question leaves the response of the addressee up to him, but a tag-
question implies that, while the speaker expects a certain sort of response,
the hearer may not provide it.7 Hence its statement-plus-question superficial
form is quite logical. The effect of the tag, then, is to soften the declaration
from an expression of certainty, demanding belief, to an expression of likeli-
hood, merely requesting it—suggesting that Rule III may be ignored.8
How do these facts lead to the conclusion that tag-questions and imper-
atives function in a parallel fashion, and that this function involves the
weakening or ignoring of normal rules of conversation? It is clear how (20)
7. As is well known, English has at least two intonation patterns associated with
tag-statements (or, as they are more commonly called, tag-questions). One, rising, is
closer to a question, as is predictable from the intonation pattern; this expresses less
certainty on the speaker’s part, and less hope of acquiescence by the addressee. The
other, falling, is nearer to a statement, expressing near-certainty, with just the mer-
est possibility left open that the addressee will fail to agree. The second type is often
found as a kind of gesture of conventional politeness, meaning something like, “I have
enough information to know I’m right, but I’m just letting you have your say, in order
to be polite.” It is interesting that some verbs of thinking, in the 1sg. present, have the
same ambiguity resolved by the same difference in intonation pattern, and both types
of locutions are used for similar purposes (cf. fn. 8 below). There is a third type of tag-
question, used when the speaker definitely knows something is true, based on personal
observation, and merely wishes to elicit a response from the addressee. This has the
particle sure inserted, as in It sure is cold in Ann Arbor, isn’t it?, vs. It’s cold in Ann Arbor,
isn’t it? The latter sentence might be used if the speaker had merely read reports that
the average temperature in Ann Arbor was 19˚. He could not, under these conditions
and if he had never been in Ann Arbor, use the former sentence. In Japanese, accord-
ing to Uyeno (1972), the particle ne expresses both the senses of the second sentence,
while its longer form nee corresponds to the first sentence.
8. In fn. 7 I alluded briefly to the uses of verbs of thinking. My point is that verbs such
as guess, suppose, believe, and sometimes think, when used in the 1sg. present, do not
describe acts of cogitation: rather, they are means of softening a declarative statement.
Consider the following sentences:
( a) I say that Fritz is a Zoroastrian.

(b) Fritz is a Zoroastrian.
(c) I guess Fritz is a Zoroastrian.
(d) Fritz is a Zoroastrian, isn’t he?
If sentences like (a) and (b) (cf. Ross 1970) express certainty on the part of the speaker
through the (overt or covert) performative verb of declaration, then (c) expresses the
speaker’s feeling that the event described in the complement of the verb of thinking
is a probability rather than a certainty. As (b) corresponds to (a), it is my contention
(made for other reasons in R. Lakoff 1969) that (d) corresponds to (c): in fact, they are
closely synonymous in many of their uses, just like (a) and (b). As pointed out in fn. 7,
the same disambiguation by intonation exists for both.

22
operates in this way, as a “softened” version of (21). We already know that

Rule V says that an order normally is given only if the giver can assume it
will be followed; and this is true of an order like (19), as well as one like (15).
But it is not true of the corresponding tag-imperatives (13) and (18). These
sentences allow the addressee the option of obeying or not, as tag-state-
ments like (20) allow the addressee the option of believing (or agreeing)
or not. So both tag-types have the same function: to give the addressee an
escape from what is normally an ironclad rule. As noted above, the parti-
cles well and why, appended to English sentences, have a similar effect: that
of showing that certain of the rules of conversation are about to be vio-
lated. Then English has at least two means for indicating this relaxation
of rules: the presence of particles, and the use of special transformational
rules for this semantic purpose and no other—in this case, of course, tag-
formation. It is not known at present whether there are languages that
are held to only one option or the other; what is known is that very few
languages other than English (actually, none I have ever heard of) utilize
such tag-formation rules. Then it should not be strange to find a language
expressing analogous functions by the use of particles. A particularly inter-
esting example is Japanese.
As with well and why in English, the use of any of the numerous particles
in Japanese is governed by the extralinguistic context: the status of the
participants (involving, among other relevant information, their sexes),
the formality of the situation, and so on. Among them is a pair of par-
ticles (ne and yo) whose function apparently is to indicate interference with
the normal rules of conversation. In fact, the only way in which one can
find a generalization about the uses of these particles is to look at them in
this way.9
Both ne and yo may be appended to any of the three sentence-
types: declarative, interrogative, and imperative. When analyzed superfi-
cially, the effect of each seems different for each different sentence-type;
but when we bear in mind the issues dealt with above, certain generaliza-
tions fall into place. Let us look at some examples. For the convenience
of readers who, like myself, are not fluent in Japanese, I have attempted
to give symbolic rather than real Japanese examples: I have used English
sentences of the appropriate types, with the Japanese particle added in its
normal place at the right.
9. All examples here are from Uyeno, who discusses these cases, with many more
examples as well as other extremely interesting particle uses in Japanese, in her dis-
sertation. Kazuhiko Yoshida and Chisato Kitagawa provided enlightening discussion
and further examples of these constructions.
23
Ne may be appended to declaratives, imperatives, and interrogatives:
(22) John is here ne. “John is here, isn’t he?” (a declarative, but
without the normal declarative demand for the hearer’s belief)
(23) Come here ne. “Come here, won’t you?” (an order, but without the
normal imperative demand for the addressee’s obedience)
(24) Is John here ne? “I wonder if John is here …?” (a question, but
without the normal interrogative demand for the addressee’s
response)
In all three cases, as the interpretations indicate, a normally obligatory rule of

conversation is relaxed: the particle ne is a signal to the addressee that he may
choose to observe the implication (one of Rules III–V) or not, as he decides.
The use of ne in Japanese (comparable to the use of tag-forms in two of the
three English types) allows the ground-rules to be suspended, as it were.
As is true of most particles, the use of ne is not completely free. Its use
is restricted to informal situations: conversation in small groups, and collo-
quial writing. (This is true of many English particles as well.) The reason for
this seems to be that the use of these particles provides implicit personal
information about the speaker—about his sex and status, relative to that
of the addressee. On the other hand, part of the idea of “formality” seems
to lie in giving as little personal information as possible, confining the dis-
course solely to the information one wishes to convey. In formal social situ-
ations, for example, the speaker does not inquire about the health of the
addressee (unless he is a doctor, in which case it is relevant to the discourse
itself), while he typically does in less formal dialog. When particles are
used in formal prose, they are ones like indeed, which give implicit informa-
tion about the relationship of the various elements in the discourse to one
another, and do not involve the speaker’s relationship toward the hearer or
his feelings toward the information he is conveying.
Aside from this general condition on the use of particles, ne is also subject
to other conditions, based on the social situation. The speaker must be aware
of the relative status of himself and his addressee in order to know whether
ne is usable in a discourse—i.e., to know whether he can offer his addressee
the right to suspend the relevant rule. Again, this can be ascertained by look-
ing at how the particle functions in discourse: roughly, it can be used in situa-
tions corresponding to those in which a speaker of English can ask, “What do
you think?” Three conditions determine when such locutions are acceptable:
(i) The status of the addressee should be somewhat higher than that of
the speaker, since offering a choice is an act of deference. (This may

24
be true even if the participants are in fact social equals, as a “hum-

bling” gesture of politeness on the part of the speaker.)
(ii) The status of the addressee cannot be very much higher than that of
the speaker, since if it is, the speaker doesn’t have the right to offer a
choice.
(iii) The status of the addressee cannot be lower than that of the speaker,
since then he would not have the right to make a choice. So we see
that both the function and the conditions on the use of ne are tied to
assumptions made by speaker and hearer about the context—social
and linguistic—in which the utterance takes place.
A similar situation can be shown to pertain in the case of yo, the other
particle mentioned above, which may be appended to declaratives, impera-
tives, and interrogatives:
(25) John is here yo. “I tell you John is here, (and you’d better believe
it).” (a declarative in which the speaker explicitly demands the
addressee’s observance of Rule III)
(26) Come here yo. “I’m telling you to come here, (and you’d better
obey).”10 (an order in which the speaker explicitly demands the
addressee’s observance of Rule V)
(27) Is John here yo? “What do you mean, is John here?” “Are you
asking me, ‘Is John here?’ ”
This last is possible only as an incredulous echo-question based on a prior

question of the addressee’s, “Is John here?” It is therefore a rhetorical ques-
tion, expecting either a positive or negative answer. The effect is: “How can
you ask such a question, when it’s so obvious what the answer must be?”
The speaker, in effect, asks why the hearer wants to have Rule IV obeyed.
10. Kazuhiko Yoshida points out that, though this sentence may be used by both men
and women, the effect is different. The translation given here is the sense it would have
when spoken by a man. If spoken by a woman, it would mean something like, “I really
hope you will come here. Please don’t forget.” A strong command has been replaced
by an earnest request. The effect of yo here in women’s speech seems to be something
like an attempt to express the idea that the speaker wishes she had the status to insist
on the observance of Rule V. The use of yo by a speaker of much lower status than
the addressee (which, in conventional Japanese society, presumably automatically
includes all women) is in a sense contradictory for reasons to be discussed below. The
contradiction is resolved by using yo to indicate a strong request, rather than a strong
injunction that cannot be disobeyed. This is still another example of how non-linguis-
tic context (such as the sex of participants in a discourse) affects the interpretation of
sentences, and therefore must be considered part of the linguistic information avail-
able to a speaker.
25
As is evident from the foregoing, the behavior of yo with questions is

more complex and much harder to understand, in terms of our tentative
generalization, than its use with either of the other two sentence-types.
But we may make a start toward analyzing it as follows. First, assume
that the addressee of (27), A, was the speaker of the immediately preced-
ing discourse—in this case the question, “Is John here?” Then, of course,
A, in asking this normal question, is implicitly making the assumption
that the addressee B, the potential speaker of (27), will follow Rule IV. B
is of course aware that observance of Rule IV is expected of him, but the
question is such that he cannot imagine why A asked the question—i.e.,
why A expects Rule IV to be followed. So what B is doing in effect, by
using sentence (27), is to make Rule IV explicit by calling it into ques-
tion. Sentence (27), then, means something like, “I don’t see why I have
to answer this question, ‘Is John here?’ ”; or, perhaps, “Make it explicit
to me why I should be expected to reply.” Where the statement and the
command make explicit the fact that they anticipate the addressee’s com-
pliance with the rule, the question followed by yo comments on the fact
that the speaker of the yo-question himself has been expected to com-
ply with a rule that he does not, in the present instance, see the reason
for. For some speakers, a positive reply is what is obviously anticipated;
for others, a negative one. It would not be surprising if some speakers
might be able to use (27) in both cases. (Intonation will differ depending
on which interpretation is intended.) There are several close parallels in
English:
(28) A: Is Agnew a liberal?

B: What do you mean, Is Agnew a liberal?
(29) A: Is Ted smoking a reefer?
B: Are you asking me whether Ted is smoking a reefer?
In both these cases, depending on context, B’s response may be construed

as being equivalent to “Of course!” or to “Of course not!” But, as explained
above, this diversity of interpretation is not contradictory, once it can be
seen that both replies reflect the speaker’s questioning the need for the
act of interrogation, or, more precisely, the need for B to follow Rule IV.
Then all three cases where the rule of conversation is insisted upon are
realized superficially in Japanese by the use of yo, and in English by the
explicit presence in the superficial structure of the performative verb, nor-
mally left implicit (cf. Ross 1970). Thus we see that, to express this notion
of insistence on observance, English employs a variation of a transforma-
tional rule (i.e., a normally obligatory rule, performative deletion, is in this

26
situation inapplicable), just as it employed a specific transformational rule,

tag-question formation, to indicate the relaxation of the observance of the
same rule of conversation. For both, Japanese employs particles. And we
see again that, although the two languages differ greatly in the grammati-
cal means by which they express this notion, both can express it relatively
unambiguously. Further, the languages express the idea in syntactically
parallel fashion for the three types of speech acts (though English does
not with the equivalent of ne-questions). This shows again that we cannot
stop our analysis at the point of superficial structure, or at the point of
logical structure, in fact: we must ask in every case what the extralinguistic
context of a sentence is, what purpose it is used for; only on that basis can
we establish whether or not sentences in two languages are parallel. And
it should be clear that a theory that does not allow the interrelationship
of linguistic and extralinguistic context cannot tell us what is held in com-
mon by yo and ne; by ne and tag-questions; by yo and the overt presence of
the performative; or, finally, by tag-questions and tag-imperatives on the
one hand, and explicit declarative, imperative, and interrogative performa-
tives on the other. This is a large chunk of linguistic material for a theory
to ignore.
In Japanese, yo is apparently much more normal for male than for
female speakers, and this is particularly true of yo with questions: all
my informants, one of them female, agree that a woman would never or
rarely use a sentence like (27). Given the conventional status of women
in the Japanese culture, it is easy to see why a Japanese woman would
never use yo. In its non-interrogative use, its purpose is to demand
compliance from the addressee. To be able to do so, the speaker must
outrank the hearer and must, in addition, be willing to make this rela-
tionship obvious. A Japanese woman would not be nearly as likely to
do this as would a man. In interrogatives, not only does the speaker do
this, but he also questions the right of the addressee (the speaker of
the prior question) to expect the rule of conversation to be adhered to,
which amounts to a still more overt declaration of higher status on the
part of the speaker.
Why should one need to make a demand of this type explicit, when
it normally is understood by the addressee anyway? One reason for the
use of yo or its equivalent occurs when the speaker has some reason to
fear that the rule in question may not in fact be obeyed by the addressee.
Although one might expect the social situations in which yo is usable to
be the reverse of those in which ne is possible, this is not quite true. We
might, for instance, assume that the higher someone is in status, the more
appropriate it might be to explicitly demand compliance to the rules. But
27
if one is sufficiently superior, he has no reason at all to fear that his injunc-
tions, explicit or implicit, will be disobeyed. Therefore, yo is most apt to be
used where the speaker is somewhat superior to the hearer, so that he has
the right to make demands, but not so much higher that he has no need to
make them. Obviously, yo cannot be used by someone of lower social status
than the addressee.11
Having given evidence that English speakers are capable of making dis-
tinctions of the first two types alluded to at the beginning of this paper, let
us now examine the third. What about the use of “dubitatives” and their
opposites, as endings on verbs or particles, to express uncertainty or cer-
tainty on the speaker’s part? I have already given examples of “dubitatives”
in English: the use of I guess or of tags, as has been shown, is essentially
dubitative in function; it is a sign that the speaker is not altogether pre-
pared to stand by his assertion, in the sense that he does not have complete
confidence in what he is asserting, since he does not—he cannot, as we
11. An apparent problem for this analysis (or, so far as I can see, any analysis) is the
fact that yo and ne may occur in sentences like this:
(a) Kore-wa anata-no hon da wa yo ne.

(b) This is your book yo ne.
If yo demands compliance with the rules of conversation, and ne allows relaxation of

the rules, is not such a sentence contradictory? As explained by Uyeno, however, the
effect of such utterances is to express the speaker’s insistence that the addressee acqui-
esce: i.e., it appears that yo modifies ne. Thus (a) has as its closest English equivalent a
sentence like
(c) This is your book, isn’t it?
This is equivalent to something like, “I guess this is your book—I certainly hope you’ll
agree.” Such a sentence might be used in circumstances like this: suppose that the
speaker of (c) has previously borrowed the book in question from the addressee. The
addressee has throughout the transaction, behaved as though the book were his to
lend. But now a third person accosts the speaker of (c), demanding the book back,
as if it had always been his. The speaker of (c)—partly because he knows or likes the
addressee better than the third person and therefore trusts him more, partly because it
is to his advantage for the book to belong to the addressee—still feels fairly confident
that the addressee really owns the book. But he is not as sure as formerly, and needs
confirmation. His use of (c) is equivalent to saying, first, “I say this is your book, and
I hope you believe it” (i.e., “This is your book ne.”) Then he adds, “I really hope you
can go along with this hypothesis; you’d better (for my sake) agree to this”—where yo
modifies and strengthens the hope of the speaker that the addressee will be able to
acquiesce: i.e., “I’m giving you a chance to relax Rule III, but I hope you don’t take it.”
I am not sure that this is precisely correct; but in any event the effect is not contra-
dictory, and it does seem as though yo modifies ne rather than the utterance itself,
particularly as an utterance like (a) is more apt to be used by women than is a normal
yo-sentence.

28
have seen—demand the addressee’s belief as he ordinarily would. The best

he can do is to ask for it. As an example, if I say
(30) John is in Antarctica
and it later turns out that (30) is not the case, my addressee may later say,
“You were wrong about (30).” If he does, I have no recourse but to agree,
providing his evidence is incontestable. But if instead I say
(31) I guess John is in Antarctica
under the same conditions, and later the addressee says, “You were wrong,”
then I have the option of replying, “No, I only said I thought (30) might be
the case.” That is, I can claim I was not really making that assertion. Thus
verbs such as guess in the 1st person singular, like tags, function as subjunc-
tives do in languages like Latin:
}
(32) Marcus Publium interfecit quod
uxorem suam corrupisset. “Marcus killed Publius
(33) Marcus Publium interfecit quod because he seduced his wife.”
uxorem suam corrupit.
Here the presence of the subjunctive in (32) indicates that the speaker
is not prepared to take responsibility for the claim that the alleged rea-
son is in fact the real reason for an action. With the indicative, as in (33),
the speaker implicitly takes responsibility. We have no natural means of
expressing this difference in English: we must resort to paraphrase. But we
do have analogous devices, illustrated by (31), usable under other gram-
matical conditions. If we were teaching English to a speaker of Latin, we
might want to exemplify this use of guess (which I do not believe is found in
Latin in this sense) by suggesting parallels with sentences like (32), rather
than by resorting to elaborate circumlocutions, which, as we have seen,
don’t really give the same idea.
English has other devices to express the speaker’s acceptance or denial
of responsibility for something in an utterance. Like the honorifics dis-
cussed earlier, these are not generally recognized as dubitatives or “cer-
taintives” (if I may coin that term), because they are not characteristically
obligatory morphemes, and because they function in only a limited subset
of sentence-types. It is not surprising, in view of our earlier findings, to
see that modals perform these functions along with many others. With
verbs of perception, the modal can displays certain semantic properties not
29
derivable from any normal definition of can. As first noted by Boyd and
Thorne (1969), under certain conditions sentences containing can appear
to be synonymous to sentences without it:
(34) I can understand French perfectly.

(35) I understand French perfectly.
But in some contexts where this should be true, particularly when the verb
is non-1st-person present, we find that although the denotative content of
the sentence pairs remains the same, one member often contains implica-
tions that are lacking in the other; e.g.,
(36) That acid-head John hears voices telling him he is Spiro Agnew,
so don’t play golf with him.
(37) That acid-head John can hear voices telling him he is Spiro
Agnew, so don’t play golf with him.
In order for (37) to be acceptable, the speaker would have to be making

the assumption that the voices were real, rather than hallucinations. Then
the effect of can in sentences such as these (again, a very restricted subset)
is to indicate doubt in the speaker’s mind as to the reality of what he is
describing—the effect of dubitative morphemes in many languages.
With these two sentences, contrast a situation in which the speaker
might normally agree that the phenomena were real which the subject of
the sentence was sensing. Then the presence of can is at least as normal as
its absence:
(38) Mrs. Snickfritz has eyes like a hawk: she can spot dust on your
carpet even if you just vacuumed.
(39) Mrs. Snickfritz has eyes like a hawk: she spots dust on your
carpet even if you just vacuumed.
There is a distinction in meaning between these two sentences, but it is not

the same as the one found in the first pair: in these, in which the first part
of the sentence establishes the reality of the dust Mrs. Snickfritz sees, the
sentence with can seems to be used merely as evidence of her superlative
ability, while the sentence without can is less an expression of approval or
astonishment than a suggestion that Mrs. Snickfritz, because of her punc-
tiliousness, is a pain in the neck.
Elsewhere in the modal system we find a device for expressing the oppo-
site of the dubitative, namely the speaker’s certainty that an event will take

30
place. This phenomenon has been referred to as “will-deletion,” though per-

haps, as I have argued elsewhere (R. Lakoff 1970), a better name is “will-
insertion.” In any case, in sentences referring to future events, the absence
of will indicates that the speaker has reason to be sure that the event will
occur, whether because it is scheduled or because he has control over it.
(Many things about this phenomenon are still unclear; there are numerous
puzzling cases and apparent counterexamples to the generalization just
given, but we can assume it is an accurate enough generalization to be used
in the present discussion.) In the previous example, the presence of the
modal can acted as a dubitative marker; here, it is the absence of will that
acts as a certaintive. Alternatively, we might view the presence of will as a
dubitative, making the speaker appear less certain about the occurrence of
an event in the future than he might be. Whichever way one looks at it, the
facts are relatively clear, as in:
(40) John dies at dawn.

(41) John will die at dawn.
In (40), the executioner is speaking; he controls John’s destiny, and has

himself arranged for John’s death. In (41), although the executioner
could say this sentence, it might also be John’s doctor speaking—
though he could not say (40), even if he were familiar with the course
of John’s disease and could be fairly sure when death would occur. He
does not (presumably) have a hand in it. However these facts are to be
interpreted, I think the use of the modals in sentences (36)–(41) can
be viewed as parallel to that of dubitatives and similar forms in other
languages.
Finally, there are still other related facts involving modals and their
paraphrases, noted by Larkin (1969). He points out that there is, for many
speakers at least, a difference in the appropriate conditions under which
these sentences can be used:
(42) My girl must be home by midnight.

(43) My girl has to be home by midnight.
By using (42), the speaker takes responsibility for the obligation. But (43)
is neutral; he may merely be reporting an obligation he does not necessarily
approve of. Compare:
(44) *My girl must be home by midnight—I think it’s idiotic.

(45) My girl has to be home by midnight—I think it’s idiotic.
31
In this case, the speaker is not taking or refusing responsibility for the fac-
tual content of the sentence, as he was in the other cases. Here the truth of
the modal notion itself is at issue—whether there really is a true “obliga-
tion” involved. I am not sure whether dubitatives in other languages can
affect or cast doubt on modality, as these can.
There are examples parallel to Larkin’s with other modals; this is not an
isolated fact about must/have to, as the previous examples were isolated
cases with can or will. This fact suggests a pervasive property throughout
the modal system. The existence in English of periphrastic modal forms
may not be due wholly to the fact that modals are syntactically defective;
there is a real need for the periphrastic forms at a semantic level as well.
Compare the following:
(46) John will shoot the basilisk.

(47) John is to shoot the basilisk.
(48) Bill may have a cookie.
(49) Bill is allowed to have a cookie.
In the first set, will is the root sense will of command: “I order that …”
Thus (46) is a direct order, for which the speaker is assuming responsibility.
In (47), he is still transmitting an order, but it may have originated with
someone else; it may not be an order he goes along with. For most speakers
of American English, (48) expresses the direct giving of permission by the
speaker. But (49) may be used to report someone else’s giving of permis-
sion. Thus, will/be to and may/be allowed to are parallel to must/have to.
I have, then, given examples of phenomena in English and other lan-
guages that bear out certain contentions:
(a) Honorifics, particles relating speaker and discourse, and dubitatives

(with their relatives) are not confined to those exotic languages that
have special exclusive markers for them. They are found in English;
but the forms used to indicate their presence are used in other ways
in other sentence types, so that they are not readily identifiable. This
indicates that languages have many and arcane ways of expressing con-
cepts; we should not assume a language cannot make a distinction just
because it has no exclusive form by which to make it.
(b) In order to assign the correct distributions to the forms under dis-
cussion, it is essential to take extralinguistic contextual factors into
account: respective status of speaker and addressee, the type of social
situation in which they find themselves, the real-world knowledge or
beliefs a speaker brings to a discourse, his lack of desire to commit

32
himself on a position, etc. We cannot hope to describe or explain large

segments of any given language by recourse only to factors which play
a role in the superficial syntax: we must take account of other levels of
language, which traditional transformational grammar expressly pre-
vents us from doing.12
REFERENCES
Boyd, Julian, and James Thorne. 1969. The deep grammar of modal verbs. Journal of
Linguistics 5.57–74.
Gordon, David, and George Lakoff. 1971. Conversational postulates. In Papers
from the seventh regional meeting of the Chicago Linguistic Society (CLS 7),
Chicago, 63–84.
Grice, H. Paul. (1967) 1975. Logic and conversation. Syntax and semantics 3: Speech
acts, ed. by Peter Cole and Jerry L. Morgan, 41–58. New York: Academic Press.
Lakoff, Robin. 1969. Syntactic arguments for negative transportation. In Papers from
the fifth regional meeting of the Chicago Linguistics Society (CLS 5), Chicago,
140–47.
Lakoff, Robin. 1970. Tense and its relation to participants. Language 46.838–49.
Lakoff, Robin. 1973. Questionable answers and answerable questions. In Issues in
linguistics: Papers in honor of Henry and Renee Kahane, ed. by Braj B. Kachru
et al., 453–67. Urbana: University of Illinois Press.
Larkin, Don. 1969. Some notes on English modals. University of Michigan Phonetics
Lab Notes 4.314.
Ross, John Robert. 1970. On declarative sentences. Readings in English
transformational grammar, ed. by Roderick A. Jacobs and Peter S. Rosenbaum,
222–72. Waltham, MA: Blaisdell.
Uyeno, Tazuko. 1972. A study of Japanese modality: a performative analysis of
sentence particles. Doctoral dissertation, University of Michigan.
12. As should be apparent to anyone familiar with other than purely transforma-
tional linguistic tradition, the notion that contextual factors, social and otherwise,
must be taken into account in determining the acceptability and interpretation of sen-
tences is scarcely new. It has been anticipated by a veritable Who’s who of linguistics
and anthropology: Jespersen, Sapir, Malinowski, Firth, Nida, Pike, Hymes, Friedrich,
Tyler, and many others. But the idea has not merely been forgotten by transforma-
tional grammar; rather, it has been explicitly rejected. Therefore, to bring up facts such
as these within the framework of recent linguistic discussion is to do more than merely
restate an old platitude. I hope that by discussing new facts, and expatiating on their
theoretical implications, I have shown that contextual factors cannot be avoided by the
linguist of any theoretical view, if he is to deal honestly and accurately with the facts
of language.
33
Introduction to “The logic

of politeness; or, Minding your
P’s and Q’s”
BY SACHIKO IDE
D eborah Tannen once said to me, “Robin was the one who inspired the
whole field” of the study of women’s language. Indeed, many women
were awakened by her seminal article “Language and woman’s place” (1973)
and were inspired both to pursue the topic of women’s language and, just
as importantly, to pursue careers as professionals in a field that had been
traditionally dominated by men and male-oriented scholarship. Needless
to say, I am one of the many women who took inspiration and strength
from Robin’s work to find a place of my own in the field.
Until I encountered Robin Lakoff’s writing, I had thought that to be a
scholar was to view the world from the ivory tower. It was when I was read-
ing a passage in “Language and woman’s place” that I realized that I had
been struggling hard to gear my mind to a male way of thinking. In that
article, she argued that to be both a woman and a scholar meant that she
had to be culturally bilingual. I could not agree more! In short, she taught
me that a woman can and should be a scholar in her own way without hav-
ing to become bilingual. She showed it herself in her thinking and writing.
Second, Robin has contributed greatly to the study of linguistic polite-
ness. Among the theories on linguistic politeness that were developed
during the 1970s and early 1980s, the theory outlined in this paper still
stands strong today as an illuminating paradigm for research. Her theory
34
can stand unchallenged by criticisms coming from the perspectives of

non-Western disciplines. After the publication of “Language and wom-
en’s place,” Brown and Levinson (1978) proposed the universal principle
of language use according to politeness; some years later, Leech (1983)
proposed his politeness principles. These theories of linguistic politeness
do not successfully incorporate politeness phenomena in societies such
as those in Japan, where honorifics form the core of linguistic politeness.
My work “Formal forms and discernment: Neglected aspects of linguis-
tic politeness” (1989) is an attempt to offer an alternative framework for
a universal theory of linguistic politeness, incorporating a non-Western
perspective.
The rules of linguistic politeness outlined in Lakoff’s paper deal with the
aspect of formality, which is crucial in analyzing politeness that uses hon-
orifics. This is where her theory seems to me strongest, though the point
is very briefly argued. In the early 1970s, when most American linguists
were attracted to the Chomskyan paradigm, she remained firmly grounded
in the real world and questioned the most preliminary assumptions about
the grammaticality of a sentence in context. She was asking, seriously and
straightforwardly, what people were doing with words, and trying to figure
out what language is in the real-world context. More than twenty years
after the publication of this work, we realize how deeply Robin understood
the phenomena of language use in context. Her theory is not culturally
biased; she has insight into universal phenomena of language in context as
a matter of human interaction. This is the real value of her work.
While many contemporary linguists have dogmatically followed the
tenets of structuralism, Robin has expanded her approach to include the
psychology of interactants and the social institutions that contextualize
and give meaning to interaction in a disciplined and theoretically sound
way. Her approach has illuminated many of the intricacies of language use
in social and cultural context. As I review the impressive body of work that
Robin has produced over the years for both academic and popular audi-
ences, what strikes me most is the sheer originality of her approach and her
down-to-earth attitude toward linguistic and language-use problems. She
has bravely challenged the difficulties in what Chomsky has called Orwell’s
problem (the explanation of the political/social/institutional systems that
instill beliefs in our lives) as compared to Plato’s problem (the discovery
of a human cognitive system). Chomsky (1981) tackled Plato’s problem at
length, but as a linguist he had little to offer regarding Orwell’s. It is Robin
who has been most successful in providing linguistic solutions to both
Plato’s and Orwell’s problems. In her writing, I can read how her thinking
flows from her sense of justice. Her attention is devoted to real people’s
35
lives, not arguments in the ivory tower. This attitude helps me understand
what I can do as a scholar of a humane discipline.
REFERENCES
Brown, Penelope, and Stephen Levinson. 1987. Politeness: Some universals in language

usage. Cambridge, UK: Cambridge University Press.
Chomsky, Noam. 1981. Knowledge of language: Its nature, origin, and use.
New York: Praeger.
Ide, Sachiko. 1989. Formal forms and discernment: Neglected aspects of linguistic
politeness. Multilingua 8(2–3).223–48.
Lakoff, Robin. 1973. Language and woman’s place. Language in Society 2.45–80.
Leech, Geoffrey. 1983. Principles of pragmatics. London: Longman.
T H E L O G I C OF P OL I T E N E S S [ 35 ]
36
37
CHAPTER 2
The logic of politeness; or, Minding

your P’s and Q’s (1973)
W e who come from the tradition of transformational grammar seem

to have spent an inordinate amount of our youth tripping over the
lumps in our rugs that contain the insoluble problems we have consigned
to that location. Among the more vicious of those lumps is the difficulty of
partial or hierarchical grammaticality: given the sort of syntactic theory
proposed in Aspects of the theory of syntax (Chomsky 1965), for instance,
one had every right to expect that a sentence would be marked as either
good or bad (*); there was no reason, if syntactic structure was the sole
decisive factor, for uncertainties and judgments of “Maybe good, if you
assume… .” But linguists who tried to be honest found as they proceeded
in their work that more and more of the sentences they dealt with did need
special markings: “Good if you assume… .,” “Good if you want the other
guy to think… .,” “Good if you don’t like… .,” and so on. That is, we needed
to worry about the context in which utterances were uttered, both linguistic
and non-linguistic; only by appeal to context could we account for the unac-
ceptability under some conditions of sentences which under other condi-
tions were unexceptionable.
We found fairly early on that assumptions speakers made about the real
world figured in judgments as to whether a particular sentence could be
This paper originally appeared in Papers from the Ninth Regional Meeting of the Chicago
Linguistic Society (CLS 9), edited by Claudia Corum, T. Cedric Smith-Stark, and Ann
Weiser, 292–305. Chicago: Chicago Linguistic Society, 1973. Reprinted here with per-
mission of the publisher.
38
used under particular circumstances; that most often, we could think in

terms of culture-wide assumptions that most speakers could be assumed
to be working under. This would include, very likely, both the philosophical
notion of logical presupposition and the majority of what have been called
pragmatic presuppositions. So, to recall one famous example, if I say (1a),
my hearer will accept the sentence only in case (1b) is assumed by him to
be the case.
(1) a. The present king of France is bald.

b. There exists at present a king of France.
And perhaps other cases are related, where although truth-values are not
explicitly involved in the test, the assumption we are talking about (the
pragmatic presupposition) will be held by virtually anyone. So we might say
that sentence (2a) is acceptable only in case (2b) is pragmatically presup-
posed by the normal addressee:
(2) a. John has lived in Paris.

b. John is still alive.
(That is, assumption (2b) permits us to use the perfect tense in (2a).)
But as we ventured out into this new way of looking at sentences, we
found more and more that a sentence which was perfectly acceptable
under one specific set of conditions might be bad under another, both
quite conceivable in the real world. I have (1969) given examples such as
those in (3): each of these sentences is good, within its particular set of
contexts:
(3) a. Who wants any beans?

b. Who wants some beans?
(The difference being, as has been pointed out, that in (a) the speaker
either has no idea whether or not beans are wanted or else assumes they
are not not wanted; but in (b) he must assume that there will be a posi-
tive response; he may, of course, prove to have been wrong, but this is the
assumption he is working under as he utters (b).)
Still more complex cases are those in which the sentence reflects the
speaker’s attitude toward his social context: more specifically, his assump-
tions about (1) the people he is communicating with: their feelings about
him, their rank relative to his; (2) the real-world situation in which he is
39
communicating: how crucial is the information he seeks to convey? does

he seek to convey information? how formal is the situation of the speech
act?; and (3) his decisions, based on (1) and (2) as to the effect he wishes
to achieve via his communicative act: does he want to reinforce the sta-
tus distinctions between himself (Sp) and the addressee (A)? To obliterate
them? Or doesn’t it matter to him? Does he want to impress A as impor-
tant? Serious? Witty? Snobbish? Does he want to change the real-world
situation, or not? No words he actually speaks may be directly applicable to
these questions; all may be raised, and answered, in a discussion, say, about
life on Mars; but the questions may nevertheless be answered unambigu-
ously, and they will have been answered, we may see upon inspection of
the dialogue, by linguistic means as much as any other: words and their
constructions are the medium of communication. So the communication
of ideas (which some might consider sociological rather than linguistic)
is effected by linguistic means. It is my contention in this paper that, if
one causes something to happen by linguistic means, whether purposely
or not, one is using a linguistic device; and it is within the domain of lin-
guistics that these questions should be explored and answered, with help,
one hopes, from anthropologists and sociologists who have been studying
these questions for years, and whose studies, we hope to suggest, may be
furthered by the use of linguistic techniques of analysis, as much as ours
may by theirs.
What I am saying, then, is that the pragmatic content of a speech act
should be taken into account in determining its acceptability just as its syn-
tactic material generally has been, and its semantic material recently has
been. We can say that, e.g., (4a)
(4) a. John threw out the garbage.
is a grammatical utterance, and (4b)
(4) b. *John threw out it.
is not, on syntactic grounds alone: the conditions on the rule of particle

movement have been violated in (4b), where they are met in (4a). Or we
might say that semantically (5a) is a better sentence than (5b), which is
unacceptable on the grounds that it violates certain semantic principles.
(5) a. The crowd dispersed.

b. *The aardvark dispersed.
T he logic of politeness [ 39 ]

40
And so we might want to say that a sentence like (6a) is a good sentence
on pragmatic (and, of course, other) grounds: it violates no assumptions
about real-world interaction; but (6b) is a very odd sentence, and purely for
pragmatic reasons.
(6) a. You can take your methodology and shove it.

b. *Can you take your methodology and shove it?
Very briefly, and glossing over exactly what is going on here, we can say
that the question form (as in (b)) is polite and the declarative form of
this sentence (as in (a)) impolite, but the communicative content of
this utterance is unalterably impolite. So if we match a polite construc-
tion with an impolite meaning, either irony or out-and-out aberrancy
will occur, on pragmatic grounds alone. A sentence like (6b) is perfectly
constructed syntactically and violates no imaginable semantic con-
straints (as would a sentence like (7), which is bad in a very different way
from (6b)):
(7) *That rock can take its methodology and shove it.
Another test of syntactic rules that has proved very useful is that of
ambiguity. Syntactic ambiguity is enthroned in the literature, proving the
need for underlying structures and transformational rules to disambiguate
utterances like (8) on purely syntactic grounds:
(8) They don’t know how good meat tastes.
Semantically there is no confusion. I have discussed cases of semantic

ambiguity, where only recourse to the relationship between semantic con-
cepts in the sentence could indicate whether and in (9) was to be thought
of as symmetric or asymmetric.
(9) The police came into the room and everyone swallowed his cigaret.
But we might also speak of ambiguities resolvable only by recourse to an

examination of the relationship between the participants in a dialogue and
their situation in the real world; this we might call pragmatic ambiguity,
and its occurrence, in a sentence like (10), would indicate to us that we
needed to incorporate pragmatically sensitive rules into our grammar:
(10) Please shut the window.
41
That is, (10) might be a truly subservient utterance in case its speaker
was really subordinate to the addressee, or equal to him and not a close
acquaintance. In this case please might mean “I’m asking you to do this
as a favor to me, since I can’t constrain you to do it.” But suppose Sp is
superior to A. In this case, the use of please is conventional. The speaker
really means something like “I’m asking you to do this, but I really have the
power to force you, I’m just acting like a nice guy.” The difference is clear
to the addressee: in the first case, he can refuse; in the second, he’d better
not, without good reason. So a response like (11) is valid as a refusal in case
(1) but not in case (2):
(11) Oh, it’s so hot in here!
There is one more possibility for an ambiguous reading of (10): suppose Sp

and A have been intimate friends for years, are on nickname terms, and
have therefore not used polite forms with each other for some time. Then
the potential speaker of (10) discovers that the addressee has been up to
some dastardly games (it need not concern us which) behind his back and
becomes furious; as far as he’s concerned, the friendship is over. Now he
and his former companion (who is, perhaps, still unaware of the situation)
find themselves together, and the speaker utters (10), with or without icy
intonation, depending on his subtlety. In this case, if A is at all acute, he
will note from the use of please alone that Sp is not kindly disposed toward
him; that there has been a change for the worse in the relationship. And he
may reply, “Hey, what’s the matter?,” referring not to the semantic content
of (10), but to its pragmatic implications. So the use of (10) may signal any
of at least three different kinds of relationships between Sp and A, and
A will respond quite differently in each case. I would call this true ambigu-
ity, of a pragmatic kind, and I would say that this is another indication
that the pragmatic component is as much a part of the linguist’s respon-
sibility as is any other part of grammar. As examples like (6) indicate, too,
pragmatics interacts with syntax and semantics and cannot be considered
apart: we need to know when the syntactic rule of question formation is
applicable, and pragmatic considerations are among those we must take
into consideration. So we must somehow extend our concept of global rules
to cover cases like these—where the conditions for the applicability of a
syntactic rule include pragmatic factors like the effect the speaker wishes
his utterance to have on the addressee.
Just as we invoke syntactic rules to determine whether a sentence is
to be considered syntactically well-or ill-formed, and in what way it is ill-
formed if it is, and to what extent, so we should like to have some kind

42
of pragmatic rules, dictating whether an utterance is pragmatically well-

formed or not, and the extent to which it deviates if it does. And just as
there are many syntactic rules that are used to generate a sentence in its
entirety, and its ill-formedness becomes progressively greater the more
rules are violated, the same can be said in the pragmatic sphere: we can
identify several types of rules, and the possible violations thereof. And just
as the applicability of a rule to a given syntactic structure may differ dia-
lectally, so may the applicability of pragmatic rules. I should like to look at
some cases, rather informally stated, but there is no reason why such rules
couldn’t, in the future, be made as rigorous as the syntactic rules in the
transformational literature (and hopefully, a lot less ad hoc).
First, we can return to the three areas of pragmatic behavior we referred
to earlier: the speaker’s assumptions about his relations with his addressee,
his real-world situation as he speaks, and the extent to which he wishes
to change either or both, or to reinforce them. We will find that two basic
rules are involved, sometimes coinciding in their effects and reinforcing
each other, more often in apparent conflict, in which case one or the other,
depending on circumstances, will supersede. Let me call these the Rules of
Pragmatic Competence:
RULES OF PRAGMATIC COMPETENCE

1. Be clear.
2. Be polite.
That is, if one seeks to communicate a message directly, if one’s princi-

pal aim in speaking is communication, one will attempt to be clear, so that
there is no mistaking one’s intention. If the speaker’s principal aim is to
navigate somehow or other among the respective statuses of the partici-
pants in the discourse, indicating where each stands in the speaker’s esti-
mate, his aim will be less the achievement of clarity than an expression of
politeness, as its opposite. Sometimes, as we shall see, clarity is politeness;
but often, one must choose between Scylla and Charybdis.
We are lucky in our work in that the rules of clarity have been formu-
lated; not fully satisfactorily, perhaps, but certainly in a valuable outline, in
Grice’s ([1967] 1975) work on the rules of conversation. It is evident that
they function as rules to the speaker to divulge the denotative content of
his speech act as clearly and with as little confusion as possible.
RULES OF CONVERSATION
1. Quantity: Be as informative as required
Be no more informative than required
43
2. Quality: Say only what you believe to be true

3. Relevance: Be relevant
4. Manner: Be perspicuous
Don’t be ambiguous
Don’t be obscure
Be succinct
I overlook here the problem (which is a problem for Grice or anyone)

of how these qualities are determined: How much is too much? What is
relevant? When is a statement obscure? These, after all, are the issues that
divide us: I make a conversational contribution that I judge to be neces-
sary, true, relevant, and perspicuous; you hear it and judge it unnecessary,
untrue, irrelevant, and obscure; and I will be puzzled if you abruptly termi-
nate the conversation. So these notions must be defined more rigorously in
terms of the separate worlds of speaker and addressee.
But a more serious difficulty is this: the rules of conversation are appar-
ently more honored in the breach than in the observance. It should be
clear to anyone looking at these rules that a normal, interesting conver-
sation violates these rules at every turn: it is the insipid or stiffly formal
conversation that hews to them. Very often the violations are signaled in
our conversation: by expressions like By the way (= “I’m violating manner,
for a reason”); As you know (= “I’m violating quantity, for a reason”); and
even without signals, we notice that violations of the rules of conver-
sation are not perceived as nonconversations, as violations of syntactic
rules are perceived as nonsentences. Rather, speakers seem to conspire,
using a kind of principle of sanity: “I assume you’re sane, unless proven
otherwise, and will therefore assume that everything you do in a conver-
sation is done for a reason: a violation of one rule will be seen as giving
precedence to another rule, or system of rules.” It seems to be the case
that, when Clarity conflicts with Politeness, in most cases (but not, as we
shall see, all) Politeness supersedes: it is considered more important in a
conversation to avoid offense than to achieve clarity. This makes sense,
since in most informal conversations, actual communication of impor-
tant ideas is secondary to merely reaffirming and strengthening relation-
ships. It is, in fact, in precisely those conversations where the content
communicated is more important than the actual act of talking that the
rules of conversation are strictly enforced. So devices like irony, exag-
geration, joking, ambiguity, and other devious conversational ploys are
normal for an informal conversational style, but not in business conver-
sations, or academic lectures, where the Rules of Conversation tend to be
in effect.

44
(Of course, this is an overgeneralization: even in the most formal sit-

uations, informal uses crop up: it’s rare to find an unmixed style, purely
formal or purely informal. There are many reasons for this, chief among
them the fact that very seldom indeed is a speech act designed purely to
impart factual information: one often seeks at the same time to impart a
favorable feeling about the factual information, best achieved by making
one’s addressee think well of one, notably through the use of the Rules
of Politeness. And similarly, even in informal conversation we sometimes
want to get down to brass tacks, however informally or apologetically; and
then we will resort to the Rules of Conversation to expedite this.)
Then what are the rules of politeness, and how are they related to the
Rules of Conversation? I shall list these rules informally below, and give
some examples.
RULES OF POLITENESS
1. Don’t impose
2. Give options
3. Make A feel good—be friendly
Now sometimes two or more of these rules may be in effect together, rein-
forcing each other; just as often, we must make a choice—are we in an R1
or an R3 situation?—and one will cancel the other out. One, that is, may
supersede the other. But how do we tell which of the rules we are enforcing
at a given time? Let us give some examples of how these rules operate:
1. Rule 1: Don’t impose. This can also be taken as meaning, remain aloof,
don’t intrude into “other people’s business.” If something, linguistically or
otherwise, is nonfree goods, in Goffman’s sense, this rule cautions us to
steer clear of it, or in any event to ask permission before indulging in it. So
we request permission to examine someone else’s possessions; and simi-
larly, if we are about to ask a question that is personal, we must normally
ask permission before we do it:
(12) May I ask how much you paid for that vase, Mr. Hoving?
But not in a case where the reply is not construable as nonfree goods:
(13) *May I ask how much is 1 + 1?
(Unless something deeper is implicit in the questioning.) (Of course, the

request for permission is more or less conventional, since you are asking
the question at the very moment you are asking permission to ask it; but
45
it’s the thought that counts. You appear to give the addressee an out (see
Rule 2), even though actually you don’t.)
Also governed by Rule 1 are certain linguistic devices: passives and imper-
sonal expressions, for instance, tend to create a sense of distance between
speaker and utterance, or speaker and addressee. Hence sentences contain-
ing these forms tend to be interpreted as polite, like the other forms of
Rule 1-governed behavior. The proper butler says (14a), not (14b).
(14) a. Dinner is served.

b. Would you like to eat?
And academic authors, as is well known and often bemoaned, tend toward
passive and impersonal sentences in their work, as well as the authorial
we, which, like the polite vous in French (and analogous cases in many
other languages), creates a distance between himself (in the vous case,
the addressee) and the other people involved in the communication. (The
authorial we is thus parallel to the vous of egalitarian nonsolidarity, as dis-
cussed by Brown and Gilman (1960); but the imperial we is parallel to the
vous of superior status.)
Related to the prohibition on nonfree goods under R1 conditions, we
find the use of technical terms to avoid mentioning unmentionables, like
sex, elimination, or economic difficulties (in our culture; other cultures
may have, of course, different unmentionables). This is the practice of
bureaucratese, medical and legal terminology, and general stuffiness. It is
to be distinguished from euphemism, which is an R2-related use. Technical
terminology seeks to divorce the subject from its emotional impact: “We’re
talking about IT all right, but it doesn’t have its usual connotations because
we’re divorcing it from all emotional content.” So you say copulation, or def-
ecation, or disadvantaged, if you have to say any of them.
Rule 2 operates sometimes along with R1, sometimes in cases where R1
would be inappropriate. R2 says, “Let A make his own decisions—leave
his options open for him.” This may seem the same as remaining aloof,
but actually it only sometimes is. So certain particles may be used to give
the addressee an option about how he is to react: some of these particles
(hedges) have been discussed by George Lakoff (1972). Some hedges also
have the effect of suggesting that the speaker feels only a weak emotional
commitment toward what he’s discussing: that is, they reflect the speaker’s
feelings about the sentence. So the use of such hedges violates R1: they are
not in place in truly formal discourse.
(15) Nixon is sort of conservative.

46
A sentence like (15) could not be used, say, in a New York Times editorial, a
format which invokes R1 almost exclusively. But it might be used in the inter-
ests of politeness: under some circumstances a speaker whose true opinion
is representable as (16) may utter (15) to avoid social unpleasantness.
(16) Nixon is an arch-conservative.
We might mention other types of sentences that allow the addressee as

much freedom as possible in making up his mind: for instance, sentences
like those of (17):
(17) a. I guess it’s time to leave.

b. It’s time to leave, isn’t it?
That is, quite apart from their basic functions, such sentences may also
function as politeness devices in accordance with R2. Obviously, both (17a)
and (17b) can be used when the speaker genuinely is uncertain of what he is
asserting. But often, too, they are used where the speaker speaks with full
confidence: he knows what he’s talking about, but does not wish to assert
himself at the risk of offending the addressee. Such sentences, under these
conditions, mean something roughly equivalent to “I say this to you, but
you’re under no compunction to believe it: I’m not trying to buffalo you.”
Whether for the sake of politeness or because the speaker really doesn’t
know the answer himself, such sentences leave the final decision as to the
truth of the sentence up to the addressee.
Just as technical terms for unmentionables are R1 devices, euphemisms
are in the realm of R2: they retain the presumption that the topic under
discussion is forbidden, but they seek to dispel the unpleasant effect by sug-
gesting that A need not interpret what is being said as THAT. (Of course,
he obviously does, so this is again conventional; but he is at least appar-
ently being given the chance of opting out, pretending that the unmention-
able topic has not been broached, and this is what makes euphemism an R2
device.) So we talk, under R2 conditions, about making it, or doing number
two, or being hard up (in the economic sense). People holding professional
discussions, where R1 holds sway generally, will resort to technical terms
rather than euphemisms, while at polite cocktail parties, if one must talk
about the thing, one uses euphemisms, as one also uses other R2 devices in
these situations; contrast the sentences of (18), the R1 cases with those (19),
and imagine, first, an anthropologist at an anthropological meeting using
each of (18) and then a society matron at an elegant party saying (19).
47
(18) (a) When the natives of Whango-whango want to
{
copulate
*do it }
they…
(b)
{
Defecation
*Making number 2 }
is generally expedited by the use of
large banana leaves, or old copies of the New York Daily News.
(c) Many of the residents of the ghetto are
{
underprivileged.
*hard up. }
(19) (a) I hear that the butler found Freddy and Marion
{
making it
?copulating }
in the pantry.
(b) Excuse me, I have to {

go to the little girls’ room.
*defecate. }
(c) Harry sold his daughter into white slavery because he was so
{
hard up.
*unprivileged. }
The point is that we respond differently emotionally to R1 and R2
devices, and hence the careful speaker will tailor his device to his purpose.
Rules 1 and 2 may be applicable together, as we have seen: avoidance of
nonfree goods may be interpreted both as a means of not imposing and
as a way of letting the addressee have his freedom. But Rule 1 and Rule 3
seem to be mutually contradictory: if they coexist in the same conversa-
tion, we must assume that, for any of various extra-linguistic reasons, the
participants are, really or conventionally, shifting their relationships with
each other. Rule 3 is the rule of politeness that seems the least “hypocriti-
cal,” although it, too, is very often used conventionally when there is no real
friendship felt. This is the rule producing a sense of camaraderie between
speaker and addressee. The ultimate effect is to make the addressee feel
good: that is, it produces a sense of equality between Sp and A, and (provid-
ing Sp is actually equal to or better than A) this makes A feel good. (But of
course, if Sp really is of lower rank than A, his invoking R3 will be seen as
“taking liberties,” and will result in the termination of the conversation on
an unsatisfactory basis.) Now it is true that R1 and R2 also are designed to
“make A feel good.” In fact, one might try to generalize and say that this
was the purpose of all the rules of politeness. But they all do it in differ-
ent ways, and R3 does it by making A feel wanted, feel like a friend. Rule 3
is the rule that produces tu in appropriate situations, in those languages that
use tu, when it is used to express solidarity. Here too, rather than the last

48
name + title used in true R1 situations, we find nicknames, or at least first

names (cf. Brown & Ford 1964); we find the sorts of particles that express
how Sp feels about what he’s talking about: this makes A a more active
participant: expressions like like, y’know, I mean. When rhetoricians warn
against these words they do it because their denotative information func-
tion is nil: they are out of place in an R1 situation. But utterances, as we
have said already, are not uttered merely to get information across; and we
must acknowledge that these meaningless particles have, in truth, a deep
meaning: they say, “R3 is in effect.” Here, too, in R3 we find the giving of
compliments, out of place in R1 situations as impositions; and we find the
use of simple forms of unmentionable words. (Brown and Ford (1964) point
out that, the more nicknames one has for a person, the more likely it is that
you have discussed intimate things with him: that is, both are parts of the R3
situation, and one follows from the other.) It has been noted that R1 and R3
are incompatible; R1 and R2 are sometimes incompatible, sometimes coex-
istent; and similarly with R2 and R3; sometimes, as with the particles, they
also serve to provide leeway for the addressee: since the speaker says like,
y’know, and so on, he means that what he’s saying is just his own feeling; so
the use of such expressions may also be construed as giving options. But the
use of nicknames cuts off options, and the use of tabooed words does too.
In a situation where we would expect R3 and get R1, the effect is a breach
of politeness, rather than a free choice between that and a (polite) refusal
to impose. So to refer back to an earlier example, if I say (10) where I have
previously been saying (20),
(20) Shut the window.
the addressee’s assumption will be that we are no longer in a state of cama-

raderie; he will have been made to feel bad, a violation of the rules of polite-
ness, rather than merely feeling he’s been left his options or has not been
imposed on.
Both interpretations should theoretically be possible; the fact that only
one is, at least in our culture, shows that R3 takes precedence over the
other rules when it is applicable. When it is not, of course, one must resort
to the others. But it seems that in middle-class American society, R3 is
gaining ground continually at the expense of R1, while in more stratified
societies, R1 seems to be given more play.
Similarly, a conversational implicature (cf. Gordon & G. Lakoff (1971))
like (21), used to mean (20), may or may not be a polite way of saying (20),
depending on the situation:
(21) It’s cold in here.
49
It may be construed as polite under R2: the addressee is, at least conven-

tionally, given the option as to how he shall interpret the sentence—as a
way of making conversation, or as an order—and may choose to respond
appropriately in either direction. But suppose the speaker of (21) is in a
superior position to the addressee. Now, if he uses (21), since he is not in
an R3 situation and not therefore merely “making small talk” about the
weather, (21) must necessarily be construed as an order, equivalent to (20).
But the fact that Sp has phrased it as a statement seems to suggest, “You
must interpret my every wish as your command, you are so far beneath
me,” and thus violates several of the rules of politeness at a blow.
One thing I would like to note briefly in passing: the rules of politeness
function for speech and actions alike. A polite action is such because it is in
accord with the dictates of one or more of Rules 1, 2, 3, as is a polite utter-
ance. So covering my mouth when I cough is polite because it prevents me
from imposing my own personal excreta on someone else (quite apart from
germs); and standing aside as someone enters a door I am in front of is
polite because it leaves him his options, that is, his freedom of movement.
This suggests that the rules of language and the rules for other types of
cooperative human transactions are all parts of the same system; it is futile
to set linguistic behavior apart from other forms of human behavior.
Now let us return to the question of the relationship between the rules
of politeness and the rules of conversation. We have noted that the rules of
conversation are in effect in non-R3 situations: that is, R1 situations, cases
of formality.
We can look at the rules of conversation as subcases of Rule 1: their pur-
pose is to get the message communicated in the shortest time with the
least difficulty: that is, to avoid imposition on the addressee (by wasting
his time with meandering or trivia, or confusing him and making him look
bad). The fact that the rules of politeness are in conflict with the rules of
conversation precisely in R3 situations suggests that this is so, and in fact if
we do interpret the rules of conversation as one kind of rule of politeness,
specifically an R1 type, we will have achieved an interesting generalization
about the way in which the rules of politeness take precedence over one
another, and the circumstances under which each is applicable. And this
will help us answer our original question: how it is that violations of the
rules of conversation save the rules of politeness. Actually, they save only
R2 and R3 politeness, and now we understand how and why this is so.
Now one objection to this formulation of the rules of politeness is that
what is polite for me may be rude for you. I am claiming here that these rules
are universal. But clearly customs vary. Are these statements contradic-
tory? I think not. What I think happens, in case two cultures differ in their

50
interpretation of the politeness of an action or an utterance, is that they

have the same three rules, but different orders of precedence for these rules.
An example: It is said that it is polite in Chinese society to belch after
a meal (if you are not the one responsible for the cooking). But this is not
polite in our society. In our society, R1 takes precedence: one must not
impose one’s internal workings on someone else. But in Chinese society R3
takes precedence: show appreciation, make the other guy feel good.
Another example: I was brought up to believe that financial questions,
like those involving sex or habits of elimination, were nonfree goods. So
one didn’t ask an acquaintance (until a very late stage in the relationship)
how much money he made, what his father did, how much any of his pos-
sessions cost. But other people apparently are perfectly free to bound into
one’s house and ask the cost of everything, how deeply in debt you are, etc.
One interpretation of this behavior is that it is plain boorish: they know
the rules but don’t care to apply them. But there is another, more gener-
ous, way. (Generosity is in accord with Rules 2 and 3.) Suppose we inter-
pret what I was taught as an affirmation of R1. But asking about people’s
possessions, showing interest in their welfare, may be interpreted as R3
behavior—making the other guy feel important to you, making him feel
like a friend. So we may have here another case where the order of prece-
dence of two rules differs dialectally. Just as someone who uses a syntacti-
cally aberrant sentence may not know the rules of English, or may have a
somewhat different set of rules from you, so someone who seems to be vio-
lating the rules of politeness may, indeed, not know the rules or be ignoring
them; but may just as well have the same rules as you, differently ordered.
One final set of cases: I have mentioned the various ways of hedging
performative utterances (the sentences of (17)), ways of weakening the
declarative force of verbs of saying, and I have suggested that sometimes
this is done in the interest of politeness; the opposite also occurs: we find
outright insistence on the force of the speech act, as in the sentences of
(22). Why do these exist?
(22) a. I’m telling you that Fred is a ratfink.

b. Young man, I’m asking you where I can find the Chairman.
c. For the last time, I’m telling you to take the chewing gum
out of your mouth.
Now these are incontrovertibly rude, in that they impose on the addressee
and destroy his options (he must recognize the speech act for what it is,
and must respond appropriately). Also, since they insist on the address-
ee’s appropriate response, they treat him not as a friend, but as an express
51
unequal. So they violate Rules 2 and 3. But they are unambiguously clear,
too: they make it very plain what the speaker intends. So in this case, the
R1–R3, or clarity-politeness conflict, is resolved in favor of R1, where ordi-
narily it is resolved in favor of R3. Is this an exception? We must first ask
when these sentences may be used. Generally they are used in desperation—
when an ordinary speech act has previously been ignored. Now in this situ-
ation, politeness may be waived—the addressee, by ignoring the speaker
in the first place, has forfeited his rights. And further, the speaker has to
get his message across, and by now has good reason to believe only the
most forceful speech act will accomplish this. So these sentences are not
true exceptions, but rather show that our rules are in general correct: they
enable us to cope with apparent exceptions, and to show why they exist.
In conclusion, then, I hope to have shown the following:
1. That we follow pragmatic rules in speaking, just as we follow semantic

and syntactic rules, and all must be a part of our linguistic rules.
2. That there are rules of politeness and rules of clarity (conversation), the
latter a subcase of the former: rules of conversation are a subtype of R1.
3. That the rules of politeness may differ dialectally in applicability, but
their basic form remains the same universally.
4. That these are not merely linguistic, but applicable to all cooperative
human transactions.
REFERENCES
Brown, Roger, and Marguerite Ford. 1964. Address in American English. Language in
culture and society: A reader in linguistics and anthropology, ed. by Dell Hymes,
234–44. New York: Harper and Row.
Brown, Roger, and Albert Gilman. 1960. The pronouns of power and solidarity.
Style in language, ed. by Thomas A. Sebeok, 253–76. Cambridge, MA &
New York: MIT Press & John Wiley and Sons.
Chomsky, Noam. 1965. Aspects of the theory of syntax. Cambridge, MA: MIT Press.
Gordon, David, and George Lakoff. 1971. Conversational postulates. Papers from the
seventh regional meeting of the Chicago Linguistic Society (CLS 7). Chicago, 63–84.
Lakoff, George. 1972. Hedges, fuzzy logic, and multiple meaning criteria. Papers from
the eighth regional meeting of the Chicago Linguistic Society (CLS 8). Chicago,
183–228.
Lakoff, Robin. 1969. Some reasons why there can’t be any some-any rule. Language
55.608–19.

52
53
Introduction to excerpts
from two 1974 papers
Pluralism in linguistics; Linguistic
theory and the real world
BY BIRCH MOONWOMON
I n the early 1970s Lakoff was writing critically about the newly established,
but already solid, generative syntax from the center of its academic cul-
ture. Long before there was an acknowledged need for an “extension” to the
standard theory, when the nonstandard challenge to the Aspects (Chomsky
1965) model was barely being developed, Lakoff was articulating criticism
of what did not even have to be called “standard,” since there was little
with which to contrast it. She was writing as a member of the generative
semanticist group. Something of particular interest in understanding the
development of her scholarship and the independence of her stance is that
she was interested in a socio-syntax, and quite willing to show this, among
a group of colleagues who would talk about supposed semantic universals,
but not social particulars.
Two shortcomings of the Aspects model especially get Lakoff’s atten-
tion. Criticism of these shows up in each of the articles reprinted
here: the privileging of formalism that (still) plagues generative gram-
mar; and the failure of the theoretical construct of autonomous syntax.
Lakoff acknowledges the importance of the profound transformation in
linguistic analysis that generative syntacticians’ work has meant, but
54
insists that as the theory became stabilized it developed a dogmatic

rigidity that has not served it well. For one thing, formalism itself has
been overemphasized. Problems which themselves can not be formulated
within the mathematical-logic-wanna-be notation of generative syntax
are declared uninteresting. In the terms of the technical lexicon of math
they are uninteresting, because they cannot be attended to with the avail-
able machinery; but this does not make such problems unimportant or
irrelevant. Lakoff points out that transformational grammar (TG), devel-
oped as philosophy or information science that rejects anything that can-
not be expressed in rules, and isolated from study of language learning
or acquisition, has produced an elegant system—but one that is increas-
ingly sterile because it is asocial. Lakoff challenges TG to look at social
and behavioral studies of language use before formalizing rules. She is
making what was at the time a very unpopular challenge to TG’s reliance
on the theoretician’s intuition.
A second shortcoming of early generative grammar, and another
one not yet corrected, is the theoretical insistence on the autonomy of
analytical levels, in particular the syntactic level. Lakoff asks what was
a truly heretical question: Is there a need for a syntactic level at deep
structure and so a need for autonomous syntax as an analytical terrain?
She claims there is no clear evidence of such a level or need. She points
out that semantic explanations for exceptions to syntactic rules bother-
somely show up in analyses. Using the art she has been trained in so well
at MIT, she is able to assert that autonomous syntax fails as a construct.
It must have seemed in really bad taste at the time to call for the dis-
carding of elegance and the abandonment of the notion of autonomous
levels.
Lakoff’s work here is not, of course, all tearing down. On the contrary,
she is concerned to get on with theoretical work that is not sterile or aso-
cial. She asks questions that tie competence to performance, including the
encompassing question: Why do several ways of saying the same thing
exist? And Lakoff asks large performance-focused questions: Why is the
usability of variants (e.g., variants for making requests) confined to a cer-
tain range of contexts?
Lakoff challenges theoretical linguistics to address matters such as
communicative affect, paralinguistic phenomena, the sociolinguistics of
language use by minorities and women, and the general association of
speech style with way of life. Her enumeration of these matters indicates
Lakoff’s interest, in the early 1970s, in both pragmatics and variation-
ist sociolinguistics—and applied linguistics as well. Asserting that lin-
guistic theory and applied linguistics have common interests, she insists
55
that the theorist needs to know what deviations from an ideal com-
petence model are found in actual language use. Sociolinguistic infor-
mation affects linguistic form. Performance facts are not random, and
selection among variants performs specific functions. Lakoff proposes
that there be pragmatic rules with accompanying syntax and semantics
rules, all fuzzy.
The promotion of fuzzy rules, evoking Zadeh’s (1965) fuzzy categories
concept, is one of several things that dates these articles, more than do the
concern with overemphasis on formalism and dependence on autonomous
levels in grammatical analysis. One is Lakoff’s faith in presumably cog-
nitively based universal information categories for the encoding of prag-
matically relevant information. It seems now strangely naive to say that it
would not be a great burden to represent all the information that a speaker
has, that this would reduce to a few things like whether information is new
or old, the uniqueness of a subject, the goodness or badness of the topic.
She is depending on a universal grammar model in terms of which it is
proper to ask questions such as how many categories from the inventory of
universal ones are pertinent in a given language’s encoding. With pragmat-
ics, Lakoff hopes that a universal underlying structure can be described in
a non-ad hoc and non-English-dominated way.
Another now-debunked idea is that women’s language is the language
of the powerless. This concept was given voice by Lakoff more than by any
other 1970s student of women’s language use, and it had her in trouble
from the start. Her remarks about women’s language anticipate her posi-
tion in Language and woman’s place (1975). Here and there it is not clear
whether she has in mind simply a stereotype of women’s use of language
and—not the same thing—an accompanying assessment of that stereo-
type as representing powerless speech or language as actually used by
women and an accompanying devaluing of that. If the latter, one would
have to know by what kinds of women and in what kinds of speech situ-
ations such language is used. Whether Lakoff has in mind what (some)
women actually do, or what they are thought to do, it is the negative eval-
uation of such behavior that is a problem. Lakoff clearly does understand
that women’s ways of doing and being are liable to be undervalued. She
notes that the people who have been turned off to formalism in syntax
have been largely women who, because only formalism is valued, have
been told their talents are not needed. A quarter of a century later it turns
out not that women do not do syntax just like men—increasingly, women
do—but that work in sociolinguistics, discourse analysis, pragmatics, and
applied linguistics is dominated by women. Lakoff as a pragmaticist has
a history of passing through the deep woods of early generative theory
I N T R O D U C T I O N T O E X C E R P T S F R O M T W O 1 9 74 PA P E R S [ 55 ]
56
and, even while there, imagining and beginning to create a performance-

responsive linguistics.
REFERENCES

Lakoff, Robin. 1975. Language and woman’s place. New York: Harper & Row.
Zadeh, Lotfi A. 1965. Fuzzy sets. Information and Control. 8:338–53.
57
CHAPTER 3
Excerpts from two 1974 papers:

Pluralism in linguistics; Linguistic
theory and the real world
PLURALISM IN LINGUISTICS
In the days when the white engineers were disputing the attributes of the feeder
system that was to be, one of them came to Enzian of Bleicheröde and said, “We
cannot agree on the chamber pressure. Our calculations show that a working
pressure of 40 atü would be the most desirable. But all the data we know of are
grouped around a value of only some 10 atü.”
“Then clearly,” replied the Nguarorerue, “you must listen to the data.”
“But that would not be the most perfect or efficient value,” protested the
German.
“Proud man,” said the Nguarorerue, “what are these data, if not direct revela-
tion? Where have they come from if not from the Rocket which is to be? How
do you presume to compare a number you have only derived on paper with a
[Because these papers contain extensive overlap, material original to each has been
collated here. Deletions are indicated with ellipses, and the footnotes renumbered. The
references for each paper have been combined and edited, and appear at the end of the
chapter.]
This paper first appeared in Berkeley Studies in Syntax and Semantics (1974), vol. 1, XIV
1–29. Berkeley: Department of Linguistics, University of California. Research underly-
ing this paper was partially supported by the NSF under grant Number NSF GS-38476.
58
number that is the Rocket’s own? Avoid pride, and design to some compromise
value.”
Thomas Pynchon, Gravity’s Rainbow ([1973] 1987) p. 314–315
Over the past dozen or so years, we have witnessed a profound transfor-

mation in linguistics: in what were the methods and goals of linguistic
analysis. Today I want to suggest that some of our early missionary zeal
may have gone too far, that there are areas of research that have been
deliberately slighted in the last decade, to the detriment of the field as
a whole.
It is beyond question that transformational theory has been of immense
benefit to our understanding of language. Chomsky’s basic insights are
striking in their beauty and clarity, especially if we can project ourselves
back in time to pre-1957 and see what the other choices were at that time for
the analysis of syntactic phenomena. It was not at all surprising that trans-
formational grammar, within a few years, spread like Christianity, making
converts who preached the new dogma with fiery zeal. And it is beyond
question that the success of transformational grammar—both in terms
of its ability to account for the sentences of English, and of its increasing
influence inside and beyond linguistics proper—can be attributed in large
measure to the formalisms Chomsky and his followers developed for repre-
senting sentence structure: they seemed to be perfectly clear and precise,
fully motivated, and capable of expressing in a few symbols generalizations
that the most perceptive linguists had felt intuitively to exist, but couldn’t
talk about for lack of a succinct vocabulary. And therefore it was only nat-
ural that, among transformational grammarians themselves, emphasis
soon was placed on formalism above almost everything else, until it was
divorced rather completely from linguistic description, and one could lis-
ten to discussions at MIT for hours on end without being given the slight-
est clue that, underneath it all, the description of languages was what was
at issue. People worried about possible underlying and derived constitu-
ent structures; notational conventions; possible ways of writing rules and
ordering them; the existence and the nature of the transformational cycle,
far removed from those directly linguistic questions which people on the
fringes still had the naiveté to believe were the issues to which linguists
should address themselves.
Obviously again, there was a need to raise and discuss such issues.
Linguistic theory could not be considered defined until they were
settled—at the very least. But an insidious assumption, never openly
stated but often implicitly believed, crept into the transformational-
ists’ minds over the years, that these questions were really all that was
59
important; solve them and you will understand language, you will have
a full-fledged theory. The idea that a theory might stand or fall on its
formal assumptions alone is what I consider the most dangerous heresy
within the field.1
Ill-advised, maybe, you say; boring, often; but dangerous? Yes, danger-
ous in the sense that it has led transformational grammar down the gar-
den path many times, to bad analyses and untenable generalizations, and
into theoretical positions that are decidedly unhealthy for anyone who
wants to understand how language really works—as opposed to some-
one who wants to study mathematically pretty notation. I assume that
linguistics is to be done by those in the first group, and mathematics by
the second; and I therefore assume that an emphasis on formalism for
its own sake is deleterious to the development of an adequate theory of
language use.
There has, thus, over the years, been a great deal of too-glib generaliza-
tion based on the doctrine of the neatest formalism. The idea is that if you
can formalize an analysis, it’s right, and if you can’t either it’s wrong or it
isn’t within our provenance as linguistic theorists… .
[A]‌pparently beautiful formalization[s] turn out to be inadequate in
many ways. It is true that no perfect solution of the passive problem has
been forthcoming from any quarter, but at least it seems more honest to
admit that your tentative solution is interesting but unworkable (as has
been done by Postal and others within generative semantics who have
made proposals concerning passivization), rather than putting a superfi-
cially attractive solution up as a panacea without regard for the data that
contradict it.
I said too that overdependence on formalization might have the effect
of discouraging research from being done that, though incapable of per-
fect formalization at present, nevertheless could be fruitful, insightful, and
good for linguistics through attracting new recruits.
I refer to two kinds of cases: first, work within the domain, as it is gen-
erally envisioned at present, of “linguistics proper”; then, work that goes
beyond the boundaries of our field. In both cases, of course, what fre-
quently happens is that first the informal work is done, pointing out the
1. I am not saying that interest in formalism per se is destructive; much less that we
should abandon theory and formal rigor in favor of purely informal description and
nontheoretical data collection. Certainly anyone would agree that a theoretical view-
point is necessary, at the very least, to show us what data are interesting and how to
categorize the facts: it is unarguable that formalism is very useful, properly used, in
determining the accuracy of our observations. I am inveighing here only against blind
dependence on formalism for its own sake.
P luralism in linguistics [ 59 ]

60
nature of the problem and why it is of interest; formal work follows upon
this, some time later. But if we insist on formalization at the outset, much
preliminary investigation will be squelched, and we will be cut off from
doing some of the most interesting potential research.
Consider work by linguists such as Georgia Green, Deborah James, Ann
Borkin, and John Lawler.2 These scholars were among the first to point out
that classical transformational grammar was formally incapable of dealing
with many areas of language use: types of speech acts, interjections, polar-
ity assignment, and genericity, to name a few. It is sometimes held by more
orthodox transformationalists that if a solution can’t be formalized, the
problem is of no interest to linguistic theory; but this sort of work shows that
nothing is farther from the truth. The very fact that relationships existed
that were incapable of formalization was sufficient to show that the formal-
isms thought by orthodox transformational theory to be adequate for the
complete description of language were not so by any means. But the facts had
to be presented first; had we looked only at what our formal apparatus could
handle, we would never have discovered the inadequacy of that apparatus.
Now what these and other investigators have found, the more doggedly
they tried to account for the distributions they discovered, was, first, that
syntactic criteria alone could not be used to account for the data; then, that
semantic criteria were not sufficient either; and most recently that, if we are
going to talk about the occurrence of sentence-types, we must extend our
vision beyond even what is normally considered “linguistics” proper, and
erase some of the boundaries that have been imposed on our domain: we
must become engaged in research that we might be tempted to call “psy-
chology” or “psychiatry,” or anthropology,” or “sociology” or “literary crit-
icism”—anything in fact but linguistics. Yet if we are going to be serious
linguists, we must agree that our field is part of theirs (and vice-versa).
However, if formalization within the accepted bounds of linguistics is
unlikely at present, complete formalization within these broader defini-
tions of the field is, presently, impossible. This is not to say that it will never
be possible, but only that its realization demands much more knowledge
than we have available at present. Yet we will never be able to integrate our
knowledge about linguistics with these various other fields and arrive at a
formal system incorporating concepts of the various branches of human
knowledge, unless we first discover what sorts of facts we will have to deal
with: precisely where, and how, linguistic data interact with other kinds
of information. So once again, close examination of the facts comes first,
2. For instance, Borkin 1971, Green (1972) 1975, James 1972, and Lawler 1972.
61
and only after we have dealt informally with the problems for some time
will a means of formalization of what we have found out be likely to arise.
And even here, we ought not to expect a formal system for the integration
of all human knowledge to arise one day full-fledged and complete, from
someone’s mind. The process of devising formalisms will be long and often
frustrating, and we will have to be content with only partial and rather
unsatisfactory representations for some time to come. It will be a process
of trying a formalism, seeing how it deals with the facts, modifying or
discarding—and starting anew. But always we must match formalism to
the facts, which come first—not the other way round.
In these days of declining enrollments and student interest turning away
from pure research to real-world relevance, it behooves us to consider the rel-
evance of linguistics to the real world. And the interaction of pure language
phenomena with these other fields is precisely where linguistics can be rel-
evant, interesting, and fruitful. The majority of students are not apt to be
turned on by Boolean conditions on analyzability; rather, they find it interest-
ing to ask how one’s mental state and status in society are reflected in one’s
use of language. This is where we ought to be exploring; but this is also where
we have no rigorous methodology, and hence, where the traditional transfor-
mational linguist tells us to stay away from, since it isn’t “linguistics.” This has,
in my view, always been poor advice, but now it is becoming utterly dangerous.
Let me talk a little about the kinds of questions I think it is time we
started addressing ourselves to, as linguists: questions about the intent
and total effect of communication—both purely linguistic, and, where nec-
essary, paralinguistic as well: gestures, intonation patterns, posture, and so
on—rather than merely its superficial purely linguistic form:
1. The use of language by minorities, or, more generally, groups who don’t
set the dominant style. Among such groups, we might identify:
a. blacks
b. Chicanos
c. lower and higher (than middle-class) economic groups
d. women
e. academics
f. children
g. hippies
h. militants of various persuasions
This is just a partial list, but I think everyone will agree that each of these
groups possesses a language pattern that stamps it as not within the English

62
spoken by the media and other spokesmen of the “normal” or maybe we

should say “President’s” English. There are several reasons why the language
of each of these groups, individually and severally, is worth studying.
It is obvious that the use of nonstandard English is both evidence that a
speaker is not from the dominant group, and a means of keeping him out
of that group. However compassionate we think we are, it is still the usual
gut reaction to assume that, if someone is not speaking the dominant lan-
guage, he is therefore not expressing his ideas in optimal form. Of course,
this prejudice acts against all Americans for speakers of RP British English,
several of whom have been delighted to tell me how sloppy, inarticulate, and
fuzzy-minded all Americans are, because they don’t speak RP. Naturally, I find
this highly infuriating, and I generally point out that it’s (sloppily speaking)
hogwash, Americans are as articulate as anyone, but they are articulate in
their own way. But what we want to do ultimately is link a person’s speech-
style with his way of life, with the assumptions he makes about how he wants
to appear to someone else. There is no evidence that any speech style is ipso
facto worse that any other, or that if a dialect is nonstandard, it is incapable
of deep intellectual expression. But an emotion that one dialect expresses in
one way, another may express in another. To the speakers of the first dia-
lect, those of the second may seem to lack the ability to express that emotion;
and vice versa. Speakers of RP British English, for instance, seem, both male
and female, to have far greater access to intonation variation than do most
male speakers of American English. Speakers of RP can use their freedom in
this area to express hesitancy and deference. Speakers of American English,
without this recourse, still want sometimes to express the same emotions,
whether sincerely or as a polite gesture: “I don’t want to step on your toes.”
But, denied access to intonational variation, speakers of most dialects of
American English must resort to lexical devices: among these are the much-
maligned y’know, like, and I mean, all of which qualify the communicative act
in one way or another, and express the speaker’s real or feigned hesitancy to
perform that speech-act. It is fashionable to berate Americans, or their lan-
guage, for inarticulateness because they make so much use of these devices;
but when conventional deference may be dispensed with, Americans certainly
can sound as articulate as anyone. It’s just that the device they use to keep
from being pushy is wrongly identified as a mere conversational place holder
to be used when the speaker has run out of what to say. Generally he hasn’t.
It is important, in cases such as these, that linguists lead in telling the
lay public how to react to nonstandard speech styles. And they can only
do this if they have analyzed these speech styles in some detail. Students
have come to me looking for evidence to prove that Black English is more
logical than standard American, and I have told them again and again that
63
no language and no dialect is more logical than any other, as long as both
retain the ability to express what the speakers of the dialect in question wish
to express. It is perfectly possible that speakers of Black English want to
express different ideas than do speakers of the standard language, and that
this is reflected in their language. But obviously it is impossible and stupid
to make value judgments on this basis. The same may be said of women’s
language, and the language of other non-influential groups. The speakers
of these sublanguages should know that they may be misjudged according
to the way they speak; they should know the stereotype that exists in the
mind of the public of speakers who speak this dialect, and decide if they
wish to fit this image or not. (Gloria Steinem obviously does not want to
fit the image of someone who speaks women’s language; Pat Nixon, on the
other hand, might, faced with all the knowledge we could give her, make
quite the opposite choice.) At the same time, we must educate everyone to
realize that there exists the possibility of alternative modes for expressing
similar ideas, and that none is by fiat “better” than any others. This is not
a simple task, but it will be greatly facilitated by an understanding of the
data, hopefully along with both formal and informal explanations.
2. Politeness and its attendant difficulties. We typically expect other

people to be polite, linguistically and otherwise, but often we don’t recog-
nize their politeness when it occurs, because we apply our rules differently
from the way they do (cf. R. Lakoff 1973). In the first place, an understand-
ing of the operation and problems in the operation of these rules, formally
expressible or not, is at the very basis of human interaction; I believe that, if
we can sort out what is going on, linguistically and otherwise, we can make
a start in explaining to people why they have problems interacting, and how
to recover from them and understand other people better. This is true both
within a culture and between cultures. We know that such linguistically rel-
evant theories as Grice’s ([1967] 1975) rules of conversation, David Gordon
and George Lakoff’s (1971) concept of conversational implicature, and a
great deal of the recent work in the theory of speech acts are fundamental
in our understanding of what constitutes a polite action or utterance, and
why an utterance that is polite in one situation is construed as impolite in
another. We can state the rules and their interaction semi-formally at the
moment, but more remains to be done. However, it is necessary to achieve
informal understanding at first. It is also known that politeness interacts
with syntactic phenomena, involved as it is in the predictability of such
rules as tag-question formation and imperative formation, once believed to
be governed by purely syntactic conditions on applicability. For this reason
it is important to formalize these “sociological” rules, so that we may better

64
understand what we mean when we talk about their interaction with the
already formalizable rules of syntax.
3. Pathological and aberrant language. Here I am referring to the special lin-
guistic usages of people in one or another abnormal mental state. I am thinking
most specifically of the language of schizophrenics, but we might also include
aphasia and perhaps also the language characteristics of mysticism and trance-
states. I don’t think there’s much virtue in trying to construct a schizophrenic
grammar; rather, the task before us here is to categorize the ways in which
schizophrenics use language in ways that psychiatrists can use in making
their diagnoses, and then see how they deviate from the “normal” grammar.
What I suggest we will find is that the schizophrenic’s grammar (up till a very
advanced state of decay, perhaps more due to the confines of institutionaliza-
tion than to any inherent mental deficiency) is perfectly all right, but that his
world-view is quite different from the norm. (Laing3 would have us question
the reasonableness of the norm.) And since, as we have seen, normal grammar
is strongly dependent on normal world-view, we should expect to find devi-
ations in the schizophrenic’s use of language that mirror the abnormal way
in which he perceives his world. We might ultimately be able to use language
behavior as a fairly precise diagnostic: if trait x is there, the schizophrenic is
thus-and-so far advanced toward normality, and so on. But we need to view
the aberrancy from norms as a point on a continuum: it isn’t that everybody
who is accounted “sane” speaks one way, and everyone who is “crazy” speaks
an entirely different way. Rather there is an infinity of points along the line,
and some of us fit in at one point, others at others. And certainly we would put
Thomas Pynchon, say, at a different point in our line than we would put Henry
James, but this does not, again, mean that we are making value judgments or
casting aspersions on anyone’s sanity or creativity. It is, however, the business
of the linguistic theorist to try to figure out what the possible sorts of world-
view variations are, and how they are connected to different forms of linguistic
expression. Just as Chomsky showed in Syntactic structures (1957) that it was
impossible to define a grammar without reference to ungrammatical struc-
tures, so it is impossible for the linguist to talk about his real-world-relevant
grammar without referring to alternative views of the real world, normal and
aberrant, possible and impossible. Formalizable or not at present, this infor-
mation must be considered a part of the baggage of linguistic theory.
4. Language used for special effect: literature, advertising and propa-

ganda. We need to ask why stating something in one way is “effective”—it
3. Cf. Laing 1967.
65
gets a reader or listener to believe something, or do something—while

stating it another way is ineffective, although the denotative content of the
utterance is the same. Certainly we are all familiar with the aspirin com-
mercial that says “No other brand of aspirin is better than Bayer” rather
than “Bayer is as good as any other brand,” and the oven-cleaner commer-
cial that says, “Easy-Off cleans as well as another leading brand,” and not
“At least one other brand is as good as Easy-Off.” We need to talk about why
these statements are so easily misinterpreted—as they are intended to be;
probably the techniques of logic will be of use here, but the facts to which
they must apply must be dug up first, and their significance noted.
Similarly, we need to know how literature affects us as it does—why,
to return to my earlier allusion, some people respond joyfully to Henry
James, others to Thomas Pynchon. Surely the reason is linguistic, since it
is through language alone that a writer communicates. But counting pas-
sives, or the number of Latinate words in a writer’s prose, or talking about
the position of adverbs in a writer’s sentences, offers little insight into this
basic question. We must start from a writer’s emotional effect, from the
world he creates, and talk about how that influences his language and pro-
duces his effect on us. Again, this sort of work cannot now be formalized,
certainly not as prettily as one can cite statistics for passivization in James
Joyce, but this seems to be the track to take if it’s insight we want, not
merely superficially attractive formal statements.
These are among the issues I see as attracting the next generation of lin-
guists, even if some of you may scoff and say, “Then they’re not linguists.”
Whatever you call them, these are some of the things people should be
thinking about. The question I address myself to here is: How are we to
train people to handle such questions? To be both precise and flexible, to
work with an eye to the future? For it seems evident that the first course a
student takes in linguistics, or certainly in syntax and semantics, will mold
him permanently into a linguist of one or another specific kind. Our task
is to decide what kind of linguists we want our students to be, and devise
methods of training them accordingly.
I am suggesting, then, that these days there are many more profitable
avenues for linguistic exploration than may have been apparent a few years
back: we have widened our views, and have become capable of dealing with
deeper questions than we had originally. That doesn’t mean that every
linguist has to become a propagandist, or a sociologist, or a psychiatrist.
Obviously we will still have plenty of need for formal work to be done, for
purely linguistic syntax and phonology. It’s just that these will be only a
few of the options, instead of all of them. I think linguistics is becoming
a more pluralistic field. People of widely differing abilities will be able to

66
make contributions—if we train them right and don’t turn them off. I feel
that one of the kinds of damage that has been done in the last dozen years
by the overemphasis on formal description of language for its own sake is
the very business of discouraging people who weren’t interested in formal-
ism, weren’t interested in pure syntax, were interested in “relevance.” For
whatever reason, we tended to sneer at such people, on the grounds that
whatever they were doing, it wasn’t linguistics. And no one will stay in a
field after you’ve told him that he isn’t capable of working on those ques-
tions that the field in its wisdom considers germane, and that the questions
that interest him are irrelevant or trivial.
I am also aware that many of those who have in the past been turned off
by undue obeisance to formalism have been women. It is a well-known and
ill-explained fact that among undergraduate linguistic majors, more than half
are women; among graduate students in their first few years, about half in
most schools; and then, as we get closer to the Ph.D., the number of women
inexplicably diminishes, so that many fewer women than men actually get
that degree, and the number of women teaching in what you might consider
respectable places is a tiny proportion of the total. Something funny has hap-
pened. You can get a lot of different answers, but I feel that it is the emphasis on
formal descriptions of the superficial aspects of language that many of us find
discouraging: hence many women, in an attempt to escape into relevance, go
into psycholinguistics and sociolinguistics and TEFL, but are lost to the main-
stream and often end up dissatisfied anyhow. I don’t know, nor does anyone,
whether there is an inherent indisposition toward formalism among women,
or whether it is a learned trait that may eventually be overcome; I know merely
that it is the case now and is apt to remain so for some time to come. I think it
is criminal to attract people into a field and then waste their abilities and insult
their intelligence by telling them there is no place for the talents they have.
Don’t tell me this isn’t done—I’ve been there. And when jobs are given out,
we all know that no one these days is such a male chauvinist pig as to refuse to
think about hiring women—it’s just that the prestigious kind of work, the sort
of specialization the department wants to acquire, is that typically possessed
by men, not women. The only way to equalize things in the long run, rather
than making token appointments that do no lasting good, is to broaden the
field’s view of itself, to agree that many things can be done equally respectably
within a linguistics department. But we must educate people, first, to be able
to do this kind of work, and second, to appreciate it. …
[Ed. note: For more about incorporating insights from theoretical

linguistics into applied linguistics and vice-versa, see excerpts from
“Linguistic Theory and the Real World,” below.]
67
LINGUISTIC THEORY AND THE REAL WORLD
With few exceptions, linguistic acts are performed by a speaker to apprise

an addressee of real-world facts. But it is equally true that the majority of
sentences uttered give clues, in one way or another, as to how, precisely,
that utterance is to mediate between the speaker’s mentality and the real
world outside. That is, not only does language provide an outlet for infor-
mation about the world outside the speaker’s mind, but it also enables a
speaker to encode internal information. This is not, in the strictest sense,
information, though it is of course communication; but since it carries no
immediately useful denotative content, practitioners of rhetoric are occa-
sionally heard to advise speakers to eschew these internal-state signals, to
restrict their communication to purely external information, and thus to
save time and the listener’s patience. What is realized too seldom is that
this scorned information is at least as valid as the “useful” kind; one should
be aware that one is communicating on both channels, and one should be
aware of the message one is sending via the internal signals, as much as the
external.1
Internal-state signaling devices occur in all languages, though different
languages encode them differently. The second-language teacher therefore
has as one of his functions to make learners aware of the correspondences
between internal-state signals in their languages and those in the language
they are learning. At first glance this may seem a less essential task than
teaching the denotative lexical items—but one thing I hope to show in the
course of this paper2 is that both types of linguistic information are crucial
to adequate communication. It is essential that the applied linguist under-
stand the use and form of those signals, and be able to identify them in
the language he is teaching, as well as in the language of the students he
This paper first appeared in Berkeley Studies in Syntax and Semantics (1974), vol. 1,
XVIII 1–53. Berkeley: Department of Linguistics, University of California.
1. A very interesting discussion of some of these problems is to be found in Larkin &
O’Malley 1973.
2. This paper was prepared for presentation at the 1974 Annual Conference of the
B.C. Association of Teachers of English as an Additional Language and for the 1974
Convention of the Teachers of English to Speaker of Other Languages. Research for
this paper was partially supported by the National Science Foundation, under Grant
NSF GS-38476.
L I N G U I S T I C T H E OR Y A N D T H E R E A L W OR L D [ 67 ]
68
is teaching. Further, the applied linguist, if he does this, is in a position to

make significant and unique contributions to linguistic theory.
Let me talk a little about the sorts of data I mean, where “sociologi-
cal” information affects the linguistic turn of an utterance. I shall confine
my discussion here to English, since that is the only language I know well
enough to produce clear evidence in, but those familiar with other lan-
guages can easily adduce similar forms in those other languages. …
[For example,] all languages have devices to indicate politeness and
formality. But for some languages, politeness must be encoded into every
sentence: there are obligatory markers of status, deference, and humility.
Other languages express politeness less overtly, or differently: perhaps in
choice of vocabulary, perhaps in intonation patterns, perhaps by smiling,
or in the stance, or distance kept between participants in an encounter.
A speaker from one culture translated to another will not, perhaps, know
how to match his feelings to the signals he is supposed to give. If he comes
from the sort of culture where politeness is made linguistically explicit by
markers indicating humility and deference, perhaps he will try to force
deference and humility into his every utterance in the second language—
which may be one where politeness is expressed by smiling and indications
of interest. This may be why, when Japanese come to America, they often
appear overly polite, and why Americans appear to Japanese overly infor-
mal, too quickly intimate. On the other hand, still a third culture expresses
politeness by stiffly bowing, and maintaining a distance between speaker
and addressee. To the American, such paralinguistic behavior will seem
arrogant and stiff. And this may be the difficulty Germans have assimilat-
ing to our culture. So merely teaching “please” and “thank you” is not suf-
ficient if we want non-native speakers to be at home in our language and
culture. It is especially important to be aware of these variations for this
reason: if someone is evidently not a native speaker of your language, and
he mangles the syntax and the phonology and maybe the lexicon, you tend
to be very understanding and forgiving; you figure it’s a language problem,
nothing personal, and you bend over backwards to understand what he’s
saying. But suppose he uses too many deferential expressions, or too few?
Uses direct imperatives where you might have expected a tag, or indirect
request? Asks questions without apology that are, you think, none of his
business, or bristles when you call him by his first name, after you’ve known
him a few days? You then are not so prone to be forgiving—particularly if
you’re just a normal speaker of English, not a language teacher. You won’t
ascribe his odd behavior to a failure to digest certain forms, but rather
you’ll think of it as a personality defect—he’s too shy, too polite, too stuffy,
too arrogant. And as you meet his countrymen and many of then act the
69
same way, you’ll develop a national stereotype—“they’re all like that.”

Actually, they’re not “like that”; it’s their culture that is like that, and they
just haven’t learned how to translate intentions from theirs to yours—it’s
really parallel to learning how to use the pluperfect, or the passive. It isn’t
that cultures are different, it’s that they express similar feelings in different
ways. But socially the results of not having been taught are much more
disastrous, which is why I say that, hard as the task may be, the language
teacher should think about how to teach it.
We need not cross national borders to get into trouble this way. There
are innumerable subcultures within the United States, each with its clearly
defined rules for determining polite, and friendly, behavior. The more iso-
lated a particular subculture remains through discouragement of inter-
marriage and confinement to ghettos, for instance, the more idiosyncratic
their rules are apt to be. It has been noted by Labov (1969) that the rules
of conversational logic are differently applied by lower-class black and
middle-class white children in New York City schools; and since teach-
ers in the school system, whether black or white, are overwhelmingly
of middle-class backgrounds, these teachers assume their black children
don’t know the rules—don’t know any rules, in fact, since their own sys-
tem is all they can conceive of. In reality, of course, both have rules, in fact
the same basic set of rules, but apply different ones under different cir-
cumstances. The same is true of the rules of politeness: they exist and are
followed in all subcultures, but the conditions determining the applicabil-
ity of the different rules differ from one group to another. It is important
for the teacher who deals with cultures different from his or her own to
recognize this, and to learn to interpret unexpected forms of behavior as
conforming to a different set of equally strict rules, rather than indicating
unruliness or chaos.
But this diversity of subcultural politeness systems within our society
creates problems, in turn, for the teacher of English to members of another
culture altogether.
In effect, we have to teach, let us say, a speaker of Japanese, how he
should project his Japanese personality via linguistic and paralinguistic
means, into a particular sort of Berkeley resident, or (quite differently)
native New Yorker, or middle-American. That is, there must be some way
of keeping the personality, and the impression produced by it, constant,
while changing as necessary the surface signs used to represent and recog-
nize that personality type.
I am well aware that this is not a task for the language teacher, or at
least, is a task for the psychotherapist and the language teacher together,
if indeed it is in any sense doable. Interesting attempts have been made
70
to present stylistic differences in conversational behavior, between cul-

tures: Roy Miller’s (1967) Japanese textbook is one very nice example
I am aware of, and I am told by Elena Greavu that similar attempts are
being made in the teaching of English in Romania; but I think that, now
that speech-acts and their superficial representations are becoming better
understood by linguistic theorists, perhaps some of their insights could be
smuggled out of the ivory tower and into the streets.
I have noted that we are entering into the realm of the psychothera-
pist; and I might also note the interest shown by therapists in another area
mentioned above, namely that of paralinguistics. In one sense, everything
I’m talking about here is “paralinguistic” in that it functions as a com-
ment on this message, not generally as the message itself. Certainly we
must start thinking in terms of broader horizons for our field, an erasure
of the boundaries we have tended to guard jealously: this is linguistics;
this is kinesics; this is sociology; this is psychology. If we are going to solve
problems—both applied, like language teaching, and theoretical, such as
how messages are understood—we must be willing to erase these lines,
blur these definitions, and do some serious mind-stretching.
Even within linguistics, it’s evident that subfields cannot be kept pure. …
[Take, for example, the use of well as a particle commenting on the speech-
act to which the sentence in which it occurs is a response, or providing an
explanation for the sentence in which it occurs.] This aspect of language use
would in some theories be relegated to syntax, since it has to do with the
placement of words in sentences. But I have also noted that a specific pitch-
contour is associated with the use of well and, indeed, that sometimes this
pitch-contour alone could substitute for well in sentences. Then we must,
even in a non-tone language like English, assign certain semantic-pragmatic
functions to certain intonation patterns; the only question is, how wide-
spread is this relationship? And how inviolable? Then another intriguing
question rears its ugly head: we know that in many languages (Japanese
and Classical Greek, to name a couple), particles are rife. There seems also
to be a relationship between a language having phonemic tone, and having a
wide assortment of particles. Now consider the case of well in English (and
the same sorts of relationships might very well be found to exist between
other interjections and other pitch-patterns; it is our inability to describe
and precisely and unambiguously represent in writing pitch-systems that is,
in large measure, keeping us from doing much work in this area): either tone
alone, or tone + particle serves to get the meaning of well across. In a lan-
guage in which tone was phonemic, we would be denied the use of tone in
expressing those meanings. So we would expect to find a richer interjection
system in such languages—or at any rate, in languages with phonemic tone,
71
we would expect to find expressed lexically those semantic-pragmatic con-

cepts that in other languages might be expressed by various pitch contours.
This is saying that a purely phonological phenomenon and a purely semantic
phenomenon should sometimes be treated as two sides of the same coin—
something that is anathema to a classical transformational grammarian,
but seems unavoidable if we are to deal with facts like this in linguistic the-
ory. And it is also unavoidable if we want to teach speakers of English the
correct use of Japanese particles, and speakers of Japanese the correct use
and interrelationship between English particles and pitch contours.
As suggested in the last sentence, linguistic theory and applied linguis-
tics share many common interests, and common difficulties. The applied
linguist can, and should, do more in these difficult times than sit by and see
which of the various competing theories looks like it’s winning and pick it;
he has a serious stake in the outcome, and besides, his evidence is crucial
in deciding what sort of linguistic theory must be devised for handling this
shadow-area of extralinguistic phenomena. Let me talk about this a bit,
first noting the course of development these ideas have had in recent lin-
guistic theory, and then mentioning how and why the applied linguist and
the theoretician should be in close cooperation.
At various points in my prior discussion, I have found it necessary to
remark that paralinguistic facts are the business of the linguist to ana-
lyze and incorporate within his theory. I have noted that they partici-
pate in syntactic and phonological rules, as well as having semantic and
pragmatic functions. I have further remarked that in some languages—
hopefully a subset that can be precisely defined by their own morphologi-
cal properties—the concepts that we have been mentioning are regularly
expressed by explicit particles, or interjections; in others, phonological
markers alone, such as pitch or stress contours, can be used for the same
purpose.3 Now facts such as these have got to be incorporated somehow
into one’s linguistic theory, if this theory is to have even minimal adequacy
in the description of languages. If a linguistic theory cannot handle these
facts, it ought not to be considered a viable theory qua theory, let alone
as a means toward a more practical end—a method for second-language
teaching, or for gaining an understanding of child language, or stylistics,
or as a means of penetrating the depths of pathological language use (in
schizophrenia, for instance). For an inadequate theory will virtually force
the applied linguist either to overlook vital facts—since one principal use
of a theory in applied work is precisely in directing workers’ attention to
3. Some problems in the teaching of English interjections are discussed in

Catford 1959.
72
perhaps obscure facts about language—or to misinterpret them, to relate

things that ought not to be related and to miss relationships and gener-
alizations that do exist; and thus, to end up, through no fault of his own,
with a plan that is less useful for the intended purpose than it might ideally
be. Therefore it behooves the applied linguist, of whatever kind, to care-
fully examine a theory he feels is potentially useful to him, to see if it has
the built-in ability to deal with all the facts he, through real-world experi-
ence, has at his fingertips; whether, with extensions, such a theory might be
potentially usable; or whether, finally, the theory is basically unable to deal
with these data, and thus best ignored. Too often the theorist, brimming
with prestige from within his ivory tower, presents the applied linguist
with his theory as a fait accompli: “Here’s the latest stuff, you better catch
up with it and use it if you don’t want to get left behind.” And the hapless
applied linguist, bewildered by the formalism, figures that if it has all this
technical stuff in it, it must have some merit, and willy-nilly, bends reality
to meet the theory, ignores facts he knows about full well when the theory
can’t encompass them, and distorts his intuitive sense of the structure of a
language in order that the structure he uses as a base may look like the one
envisioned by the theorist, with lots of neat boxes, arrows, rules—none
of the messiness he knows is really there. He is humbled by technical bril-
liance; he is afraid to suggest revisions in the theory to meet the needs of
reality. He is afraid to discard a theory that does not match reality at any
point. If he is clever, and a good teacher, and a sensitive linguist, he will
work around the theory, pay it lip-service, use some concepts from it but
not others, and thus be successful in his efforts despite the theory that is
supposed to be his salvation. But it’s silly to be a slave to any theory, espe-
cially one that isn’t appropriate, and it’s silly to twist facts to match some
idealization of the way language ought to be, but isn’t; and it’s silly to burden
oneself with theoretical mechanisms that one must pick his way around,
mechanisms that fight his intuition rather than support it. This has gone
on too long, and really ought to stop. Applied linguists have to start asking
themselves questions about the theories they are thinking of adopting—
this is potentially the most significant contribution anyone could make to
theoretical linguistics. For certainly the theorist gets into trouble by not
having his hands on enough real data, by not having the facts forced to his
attention, by not having people contrast reality with idealism as presented
in his theory. If he ends up with an imperfect model, an unusable theory,
it is at least in part due to his isolation—and only the applied linguist can
dispel this. I know that in the past the theoretical linguist has always sat,
haughty and alone, in his ivory tower. If the applied linguist wanted to sit
at his foot and be enlightened, well, the theorist would do him that favor.
73
But the notion that enlightenment was a two-way street—that each could
learn from the other—seldom occurred to anyone, least of all to the theo-
rists, whose theories as a result tend to get more arcane and less relevant
to the real-world tasks at hand with each passing revision.
We can look at an illustration of what I mean: a theory developed pretty
much in isolation, by theorists trained not in language learning as their
principal interest, but in philosophy, information science, mathematics—
disciplines remote from real-world intervention. And they devised a very
elegant system of language description, most attractive, very compelling—as
long as you didn’t look too hard at the facts you were intuitively aware of.
Intuition was scorned: if you couldn’t formalize a relationship, if you couldn’t
write rules, make it submit to the grammatical theory that supposedly was
a tool for the understanding of language, then you’d do best to forget the
whole thing. And this is just why the system was, superficially, so very ele-
gant: the nasty complicated fuzzy parts, the parts that did not fit neatly into
the scheme, were assumed not to exist, or at least, not to be of interest to the
linguist. Rugs in that part of the world got very bulgy.
I an not inventing a parable or a fairy tale. I am talking about transforma-
tional grammar, as should be no surprise to you. I think the sins committed,
the confusion perpetrated in its name have been legion, over the past dozen
years, and I think it’s precisely because all the input came from the theorists,
most of whom had never taught languages, or were ashamed to admit it if
they had, particularly if comparison among the languages they knew revealed
data that could not be reduced to the neat form prescribed by orthodox trans-
formational theory.
Actually, I am being perhaps nastier to transformational theory than
I ought. There is a time in the life of every theory4 when oversimplifica-
tion is not only inevitable, but necessary for the development of greater
understanding. In order to be able to get an allover picture of the sort that
a theory of language must be, one must first oversimplify, pretend that the
facts are more amenable to organization than, actually, they are. Ideally, it
is better even at this stage not to fool oneself; to admit privately anyway
that the treatment gains its elegance through the overlooking of data that
will have to be dealt with some day. The ostrich-treatment, in which one
declares in effect, “My nice elegant theory can’t handle this phenomenon,
so I hereby declare it nonlinguistic and thus keep my theory correct and
neat,” is less desirable, as should be obvious, because it will prevent any
developments or improvements in the theory, ultimately rendering it too
4. Cf. Thomas Kuhn 1962.
74
rigid to be a useful heuristic tool—which in the end is the very least as well
as the most we ask of a theory. …
It should be clear that … a theory [which incorporates the full com-
plexity of language use] is far more complex, far harder both to formulate
and to understand, far less immediately elegant than was classical trans-
formational grammar. But I hope I have already made it clear that those
losses in beauty are necessary, if the truth is to be approached at all. So
as I detail a few of the assumptions of the beginnings of a theory that
will, if it is ever fully formulated, embody all those desirable character-
istics, I would hope the reader will bear in mind that its awkwardnesses
are necessary—some because of its relative youth, others because of the
scope of its ambitions.
Suppose first of all we question the most sacrosanct assumption of clas-
sical (Aspects-vintage) transformational grammar. Is there a demonstrated
need for a syntactically relevant level of deep structure, and therefore a
level of autonomous syntax, with a semantic component not directly relat-
ing to the syntax, but merely interpreting its output? For obviously if it
cannot be demonstrated that deep structure is necessary, it must be dis-
carded. Its necessity could be proved by showing that there existed at least
one generalization in one language that could be expressed by rules assum-
ing a deep structure, and not without. To date, despite strenuous efforts by
its adherents, no such case has been found. Indeed, cases have been discov-
ered that seem to suggest the very opposite: if we assume a level of deep
structure, there are explanations for phenomena we become incapable of
making.5…
This is the basis of the theory that has been called generative seman-
tics. Very simply, there is no separation of levels: a single, highly abstract,
underlying structure underlies the semantics, the syntax, and the phonol-
ogy, and further, syntactic information may be used in the statement of
phonological or semantic rules, and conversely. (Examples showing the
necessity of this are given in G. Lakoff 1970.)
Such a theory allows, in fact forces, us to take real-world contextual
information into account in determining whether a sentence is well-
formed. Real-world-linked information that a speaker uses to determine
whether a sentence may be grammatically or appropriately uttered in a
given context is, in such a theory, available to the speaker as part of his
grammatical information.
5. For a sample of the argumentation and counter-argumentation on these topics

over the last several years, see Baker & Brame 1972; Bresnan 1971, 1972; Chomsky
1970; and G. Lakoff 1970, 1972a, 1972b.
75
It is often objected by people hearing this whirlwind description for the

first time, “But that means every bit of information a speaker has about
anything must be represented in his grammar!” And this, then, sounds like
a terrible load, an incredible excrescence of redundancy. Actually it’s not.
First of all, all that information is in the speaker’s brain, somewhere.
The only questions are: Where? And how accessible is this information to
his grammar? We might, conceivably, want to say that the speaker’s entire
conceptual system is coded linguistically, that the grammar does indeed
have access to all the cognitive equipment possessed by a person. But the
grammar does not have access to it directly; probably all this real-world
information is encoded into broad categories, and it is on the basis of these
categorizations that linguistically relevant distinctions may be made. The
speaker assigns his knowledge to these broad categories, in order to deter-
mine how to speak of it. The categories are probably assumed by most
speakers to be at least culture-wide, perhaps universal; hence consterna-
tion commonly ensues when it becomes clear to one participant in a dis-
course that another is operating with different categorizations. …
There aren’t really too many categorizations possible—those that have
been identified are quite broad. But the reader will note that it is the
assignment of a sentence to a category that determines syntactic or lexical
choice, not specific sentences one by one. So we don’t have infinite possible
contexts to deal with, but rather, a relatively few context-types, to which
contexts are matched and into which they are fitted. While we can, theo-
retically, subdivide our impressions of the real world innumerable ways,
only a relative few (certainly under 50) of these ways will be linguistically
relevant: that is, affect the superficial form of sentences.6 That is not to say
there are no problems for this theory.
The actual determination of possible categories in universal grammar is
one, and a difficult one, as is the related question: How many such linguis-
tically relevant categories are pertinent in a given language? How much
variation may there be between languages in this regard? Here is where the
isolated theorist can fall into a pit: he needs input from the real world, if
he is to avoid such a pitfall. We know of some idiosyncratic choices certain
languages make; but this should merely show us how much more we need
to know. …
We know that there are languages in which, in every sentence uttered,
the speaker must indicate whether what he says is his own opinion, some-
one else’s, or generally assumed truth: these languages use verbal affixes
6. For discussion of some of the contextual categorizations relevant in English in

determining the choice of lexical items, cf. Fillmore 1969, 1970.
76
called “dubitatives” when the speaker is relying on someone else’s informa-

tion. Some languages, as we know, such as Japanese, must encode mark-
ers of relative status of speaker and addressee into every communication,
sometimes on several levels: choice of lexical items, choice of sentence-
particle, or choice of verbal inflectional ending. Speakers of such languages
feel a void when they speak English: there isn’t anything to put where they
feel the need for a marker; so often they resort to circumlocution, and
this is another thing that gives the English spoken by native speakers of
Japanese an artificially over-polite sound.
The language teacher has had much experience with data of this sort.
He knows what, in one language, is easy for speaker of another language
to learn, and what is impossible. A speaker of English finds one thing par-
ticularly difficult to master, say, in Russian; a speaker of Japanese might
find the major hurdle altogether different. And some things seem to occa-
sion a great deal of difficulty for language learners of all backgrounds: we
would hypothesize that such things were relatively rare phenomena, not
occurring in very many other languages in that overt form; other things,
however morphologically complex they looked superficially, might be fairly
simple to grasp; these, we would surmise, were relatively widespread phe-
nomena in the languages of the world, so that someone learning a language
with such a feature would be less apt to find it unique.
It is of great theoretical importance to us to identify these different kinds
of traits and differentiate among them, since this would give us some clue
as to how to set up universal underlying structures in a non-ad hoc or non-
English-dominated way. We need ways of relating culture-wide psychological
assumptions to linguistic output, as well: How shall we incorporate the fact, in
our grammar, that the Eskimos have six words for snow, that Americans have
ten or so euphemisms for bathroom, none for kitchen; that Germans have dif-
ferent words for eating done by people and by animals? Obviously the catego-
ries in which we arrange our assumptions about reality will differ somewhat
from language to language, although the core groups—like true-false; good-
bad—will probably remain the same. But we need to know the extent of this
variability, and only people who have dealt with several languages in detail
can help us here. The theorist can, certainly, eventually give some help to
the applied linguist: he can suggest ways of making analogies, resemblances
between languages that are perhaps not superficially obvious, shorthand
devices like rules which, if they are nonformally and non-ad hocly stated, may
make language learning somewhat easier and pleasanter, by using the abilities
a speaker has gained in one language in helping him learn another. But the
theorist must first know what to look for, what the possibilities are, where
true similarities exist; and only the applied linguist can tell him.
77
Finally, this dichotomy I have been assuming—here is the applied lin-

guist, out in the real world, he never reads books, he never speculates. Here
is the theorist. He only reads books. He is abstruse. No one can under-
stand a word he says, especially him—is in many respects an exaggeration,
and becoming less realistic by the day. In the days when it was understood
that the theorist studied linguistic “competence”—some idealized version
of the rules that seldom if ever, actually, existed in the real world of real
speech—and the practical linguist, if he did anything, was concerned with
“performance,” and never the twain, etc., there was some justification for
supposing there were two separate realms. But we have been saying that
the theorist’s most recent pressing concern has been with establishing the
rules of language use, intention, success and failure in real communication.
Therefore, as recent work has tended to show, there is no basis any more for
a performance/competence distinction (cf. G. Lakoff 1973); most of what
had been thought of as “performance” is just as much competence as the
passive rule or relativization, let us say. So this hoary argument is no longer
of such interest. Rather, the theorist needs to know—as he doesn’t really
right now—what sorts of deviations from the ideal “competence model”
are found, and how they are related to the “ideal” grammar of a language.
For instance:
1. In second-language learning: What errors are most prevalent?

2. In language pathology—mental disease and aphasia: How is the gram-
mar affected? Shall we call it grammatical degeneration, or problems in
perceiving and encoding real-world context? And what does our answer
tell us about preferred means of therapy?
3. In stylistics: What special idiosyncratic uses of grammar create a “style?”
And how does a style create in its readers a specific, generally agreed-
on emotional and intellectual reaction? We react differently to Vanity
Fair and to Pride and Prejudice—and altogether differently to Gravity’s
Rainbow. Why? Most of the stylistic work done from a linguistic point of
view has dealt only with surface syntactic categories—number of arti-
cles, length and superficial complexity of sentences, passives. But how
is the idiosyncratic world-view of the writer reflected in his work, in its
superficial form, or its style? Is this what we are, most deeply, reacting
to? How is poetry like, and different from, the pathological language
degeneration of the schizophrenic?
4. In advertising and propaganda: How do we “persuade?” How do we use
language to alter people’s perceptions? Is there a distinction we can
make on linguistic grounds between “legitimate” and “illegitimate”
forms of persuasion?
78
So we see that the theoretical linguist must deal with problems of the
intellect and morality, with reality and sanity. In order to do linguistic the-
ory at all adequately or interestingly, he must come to grips with the effect
of language in the world in which it is used, with the intent and effect of
communication. The applied linguist must concern himself with decisions
among possible theories, universals of grammar, relations among gram-
matical systems. The differences between the two types of linguists are fast
becoming less interesting than their similarities.
REFERENCES
Baker, C. L., and Michael K. Brame. 1972. Global rules: A rejoinder. Language

48:51–75.
Borkin, Ann. 1971. Polarity items in questions. Papers from the seventh regional
meeting of the Chicago Linguistic Society (CLS 7). Chicago, 223–31.
Bresnan, Joan. 1971. Sentence stress and syntactic transformations. Language
47:257–81.
Bresnan, Joan. 1972. Stress and syntax: A reply. Language 48:326–42.
Catford, J. C. 1959. The teaching of English as a foreign language. The teaching of
English, ed. by Randolph Quirk and A. H. Smith, 164–89. London: Secker &
Warburg.
Chomsky, Noam. 1957. Syntactic structures. Gravenhage: Mouton.
Chomsky, Noam. 1970. Some empirical issues in the theory of transformational
grammar. The goals of linguistic theory, ed. by Stanley Peters, 63–130.
Englewood Cliffs, NJ: Prentice-Hall.
Fillmore, Charles. 1969. Verbs of judging: An exercise in semantic description. Papers
in Linguistics 11:91–117.
Fillmore, Charles. 1970. The grammar of hitting and breaking. Readings: English
120–33. New York: Ginn.
seventh regional meeting of the Chicago Linguistic Society (CLS 7). Chicago, 63–84.
Green, Georgia M. (1972) 1975. How to get people to do things with words: The
whimperative question. Syntax and semantics 3: Speech acts, ed. by Peter Cole
and Jerry L. Morgan, 107–141. New York: Academic Press.
James, Deborah. 1972. Some aspects of the syntax and semantics of interjections.
Papers from the eighth regional meeting of the Chicago Linguistic Society (CLS 8).
Chicago, 162–72.
Kuhn, Thomas S. 1962. The structure of scientific revolutions. Chicago: University of
Chicago Press.
Labov, W. 1969. The logic of nonstandard English. Linguistics and the teaching of
standard English to speakers of other languages or dialects, ed. by J. A. Alatis,
1–44. Georgetown University Roundtable on Language and Linguistics 1969.
Washington, DC: Georgetown University Press, 1970.
79
Laing, R. D. 1967. The politics of experience. New York: Ballantine.

Lakoff, George. 1970. Global rules. Language 46:627–39.
Lakoff, George. 1972a. The arbitrary basis of transformational grammar. Language
48:76–87.
Lakoff, George. 1972b. The global nature of the nuclear stress rule. Language
48:285–303.
Lakoff, George. 1973. Fuzzy grammar and the performance/competence terminology
game. Papers from the ninth regional meeting of the Chicago Linguistic Society
(CLS 9). Chicago, 271–91.
Lakoff, Robin. 1973. The logic of politeness; or, minding your P’s and Q’s. Papers from
the ninth regional meeting of the Chicago Linguistic Society (CLS 9). Chicago,
292–305.
Larkin, Don, and Michael H. O’Malley. 1973. Declarative sentences and the rule-of-
conversation hypothesis. Papers from the ninth regional meeting of the Chicago
Linguistics Society (CLS 9). Chicago, 306–19.
Lawler, J. 1972. Generic to a fault. Papers from the eighth regional meeting of the Chicago
Linguistics Society (CLS 8). Chicago, 247–58.
Miller, Roy A. 1967. The Japanese language. Chicago: University of Chicago Press.
Pynchon, Thomas. (1973) 1987. Gravity’s Rainbow. New York: Penguin Books.
80
81
Not in our stars, but in ourselves

Introduction to “You say what you are:
Acceptability and gender-related language”
BY MARY BUCHOLTZ
I first encountered the work of Robin Lakoff as a high-school student wan-

dering the stacks of the Tulsa Public Library. I was one of those nerdy
kids who read the dictionary and prescriptive usage manuals for fun, and
I had already worked my way through most of the popular volumes on lan-
guage that the English-language reference section had to offer. But thanks
to the Dewey decimal system, linguistics was literally just around the cor-
ner. I remember the feeling of discovery that came over me as I gazed upon
the shelves and realized how much was still ahead to read and learn. I went
home that day with two linguistics volumes in my backpack: Chomsky’s
(1965) Aspects of the theory of syntax and Lakoff’s (1975) Language and
woman’s place—two books separated by a mere decade in time, but by an
unbridgeable chasm in their understanding of what language is. Aspects
languished for the next two weeks on my bedside table after I repeatedly
failed to get past the first page (in fact, it wasn’t until I was assigned the
book in one of Robin’s graduate seminars that I finally managed to read
the whole thing), but I devoured Language and woman’s place in an hour of
eager reading.
When I eventually rediscovered linguistics as an undergraduate classics
major and a passionate feminist, I knew I had to study with Robin Lakoff
82
at the University of California, Berkeley. As Robin’s graduate student,

I gathered up as many of her wide-ranging publications as I could find in
Berkeley’s campus libraries and local bookstores and pored over them. I was
repeatedly struck by the connections between the feminist insights of her
work and the theories I was encountering in my gender studies courses. Yet
somehow I managed to overlook one of Robin’s most theoretically impor-
tant pieces, from the standpoint both of feminism and of linguistics: “You
say what you are: Acceptability and gender-related language.”
“You say what you are” is Lakoff at her best: acutely intelligent, pas-
sionately political, avowedly iconoclastic, quotably articulate, and delight-
fully wry. The piece opens with Lakoff calling into question the established
Chomskyan linguistic binary between grammatical and ungrammatical
sentences. Although readers not steeped in the history of generative gram-
mar may be baffled by the technical language of the first few pages, Lakoff’s
welcome clarity as a writer—a quality as rare among linguists then as it is
now—is a trustworthy guide through the key points of this arcane theory.
She argues against the all-or-nothing notion of linguistic grammaticality,
whereby decontextualized sentences deemed ungrammatical are marked
with an asterisk (or “starred”). Lakoff instead advocates the more flex-
ible, gradient, and context-centered notion of acceptability, noting that
even the most seemingly straightforward linguistic facts cannot be judged
on linguistic grounds alone but only by appeal to such supposedly extra-
linguistic factors as psychology and sociology.
This discussion lays the background for the heart of the chapter: Lakoff’s
examination of “when a judgment of acceptability ceases to be a linguistic
judgment and becomes a political statement” (p. 89). She addresses this
issue with particular reference to her famous characterization of “women’s
language.” Here the chapter becomes an invaluable sequel of sorts to the
far more widely read (and widely misunderstood) Language and woman’s
place, revealing both continuities and changes in Lakoff’s thinking about
the relationship between language and gender. Lakoff introduces several
additional features of “women’s language” in this chapter, including silence
and hesitation markers, low volume, and pitch variation, and usefully elab-
orates on some previously discussed features. Equally valuably, she offers
a three-way classification of the features of “women’s language”: (1) non-
directness, (2) emotional expressiveness, and (3) conservativism. Each
of these characteristics stands in contrast to what she terms “neutral
language,” which has traditionally been more available to men than to
women. Noting that “linguistic deviation from the norm is but one form
of social deviation from the norm” (p. 94), she raises the important ques-
tion: Whose norms? And if a woman adheres to the norms of “women’s
83
language,” is her linguistic behavior acceptable, based on her gender, or

deviant, based on the male-oriented norm? The fundamental problem,
Lakoff concludes, is not the issue of acceptability itself (or even grammati-
cality, which she brilliantly designates a special case of acceptability unin-
formed by contextual factors). Rather, it is a question of who has the right
to determine the acceptability of another’s speech and what larger political
consequences might result from any such determination.
For feminist linguists, Lakoff’s chapter is an important step in the
ongoing development of her thinking about gender and language. Whereas
Language and woman’s place focused primarily on the ways that “women’s
language” disadvantages women, in the present chapter Lakoff notes the
ways in which women’s ways of speaking may in fact be superior to men’s,
and she remains agnostic regarding whether it is best for women to emu-
late the speech of men, for both women and men to arrive at a gender-
neutral norm, or for the genders to maintain separate speech styles.
In the decades since Lakoff wrote this chapter, feminist linguistics has
shifted its focus from “woman’s place” to “gender positioning,” from “gen-
der styles” to “styles of gender,” from “acceptability” to “ideology.” Yet
changes in academic fashion do not diminish the continuing importance
of Lakoff’s early work on gender, in this chapter and elsewhere. For Lakoff,
as a feminist linguist in the male-dominated domain of theoretical syntax,
the issues she examines in this chapter were not simply academic ques-
tions but deeply personal concerns. A generation later, I have continued
to confront the same agonizing questions myself, and I am reminded of
the ongoing personal and political relevance of Lakoff’s contributions as
I introduce her ideas to my own students year after year. Like all of Lakoff’s
work, this chapter is deeply embedded in the time and place of its writing,
but it is also timeless in the questions it asks, in the problems it confronts,
and in its very human and humane engagement with the complexity of
gender and language.
REFERENCES

YO U S AY W H AT YO U A R E [ 83 ]
84
85
CHAPTER 4
You say what you are

Acceptability and gender-related language (1977)
Her voice was ever soft,

Gentle and low, an excellent thing in woman.
William Shakespeare, King Lear 5.3.272–3
W ithin the model of transformational generative grammar, questions

of grammaticality at first were the ones deemed interesting, impor-
tant and, indeed, answerable; to the extent that the question of accept-
ability was raised at all, it was felt to reflect “performance” rather than
“competence” and therefore to be out of the range of interest of linguistic
theory proper. Today, when our model is much more sophisticated and we
are very much more demanding of our theory, we tend to be a bit bemused
at our insensitivity to the data, a mere decade—nay, seven or eight years—
ago; but there was good reason for it, and it is lucky for us today that we in
the past were so blind to the facts.
Grammaticality implies an either/or (*/non*) distinction; assuming (an
untenable assumption in reality, but the enterprising linguist should be
able to believe six impossible things before breakfast) that one can divide
all the sentences of a language according to such a criterion makes for a
tolerably workable theory. Binary distinctions such as this are relatively
This paper originally appeared in Acceptability in Language, edited by Sidney

Greenbaum, 73–87. Contributions to the Sociology of Language 17. The Hague: Mouton,
1977. Reprinted here with permission of Mouton de Gruyter, a division of Walter de
Gruyter & Co.
86
simple to make, and there will likely be widespread agreement among

linguists as to the assignment of asterisks; indeed, such a theory virtu-
ally entails such agreement, since questions of personal idiosyncrasy and
imaginativeness are thereby ignored. In those far-off days, too, it will be
recalled, binariness of various sorts of features was a feature of transfor-
mational theory: phonological distinctive features, syntactic selectional
restrictions, and semantic markers all were binary. It was tempting to
believe that linguistic markers, like other animals, came in pairs, and it was
therefore natural to assume that grammaticality was an either-or question.
Syntactic rules were formulated on this basis. Either a derivation was
subject to a particular rule or it was not. If a phrase marker underwent a
rule in accordance with its structural description and the conditions on
the rule, the resultant derivation would prove grammatical; otherwise an
asterisk would be assigned, and that was that. The validity of a particular
formulation of a rule was checked this way, too: if sentences were, by this
automatic procedure, assigned stars when the native speaker would declare
them grammatical, or not assigned stars when he would not, the assump-
tion was that something was the matter with the formulation of the rule;
a better one had to be found. The least, in fact, that one could expect of a
theory (one that was observationally adequate) was that it would allow all
and only the sentences of a language to be generated—another way of say-
ing what I said above.
Assuming, again, that this evaluation procedure approximated the
facts of language, it was a simple, streamlined, and elegant procedure. It
is obvious that only under such an assumption could rules of the type that
were being written at that time have been written at all. Of course, we see
with our 20/20 hindsight that this very fact—that binary grammaticality
judgments necessitated as well as permitted the rules of classical transfor-
mational grammar—casts doubt on the entire set of assumptions we call
classical transformational grammar; but at the time, this seemed to us the
way things ought to be in a well-ordered universe, and we were still capable
of believing, with our endearing childlike faith, that the linguistic universe
was well ordered.
It is further true that a new theory can arise only out of an old theory; it
was out of transformational grammar with all its faults that we have con-
structed our present-day theory with its unnumbered virtues. The change in
emphasis from grammaticality to acceptability was forced upon us, around
1967 or 1968, by our recognition of a whole new range of data and our
consequent search for more sophisticated explanations for the occurrence
and nonoccurrence of sentences. Where grammaticality judgments were
87
determinable by purely linguistic criteria, acceptability judgments invaded

the realms of psychology and sociology, greatly increasing the range of
facts one had to look at, as well as the range of possible explanations. But
we found we could no longer honestly restrict the concept of explanation to
purely linguistic determinants; and inexorably at the same time we found
that our judgments were no longer predicated on a binary system of gram-
maticality, but had to make use of a hierarchy of acceptability.
The necessity for this became clear to me when I was looking (R. Lakoff
1969) at the use of some and any in English. When Klima (1964) had exam-
ined the data, he had attempted to make grammaticality judgments about
the use of these forms, to assign asterisks to sentences on the basis of
purely linguistic data. (He managed to sneak psychological assumptions
into his set of criteria by the use of his [+affect] marker, but he never
openly acknowledged that this was the purpose, or the effect, of adopting
his apparently linguistic and apparently binary marker.) When I started
looking at sentences containing some and any, however, I found it impos-
sible to declare many of them purely “in” or purely “out”; one could declare
a sentence “good if one made the following assumptions about the state of
the speaker’s mind” or “generally out, but acceptable in case the speaker is
in a particular and peculiar social situation.” And of course, social, psycho-
logical, and linguistic situations intersect and interact with one another, so
that immediately it was clear we were dealing with a delicately shaded and
rather interminable hierarchy of acceptability.
What also became clear with some more thought was that the hierar-
chical fuzziness was the norm, the clear grammatical judgment rather the
exception, an artificial construct useful for the facilitation of theory devel-
opment rather than an accurate perception of how speakers spoke in the
real world. Even those sentences that had previously been adjudged unex-
ceptionably grammatical could now be seen for what they were—good if
one were uttering the sentence in a particular social-psychological environ-
ment; and in fact, many of the sentences so insouciantly described as fully
grammatical were now, in the clear light of day, seen to be rather bizarre,
like Gleitman’s well-known
(1) I wrote my grandmother a letter yesterday and six men can fit in
the back seat of a Ford.
(Gleitman 1965:262, ex. 11)
That is one obvious case of a sentence that is fully grammatical, that is,
linguistically unexceptionable; and yet, if we are judging acceptability—that
You say what you are [ 87 ]

88
is, the probability of such a sentence being uttered, or the number of con-
ceivable real-world circumstances or the normality of the real-world cir-
cumstances in which this sentence is apt to be used—we find a different
judgment pertains, and we must rank this sentence low on an acceptability
hierarchy.
On the other hand there are sentences which, taken out of context, are
bizarre; some even violate selectional restrictions, such as Morgan’s cel-
ebrated example:
(2) I think with a fork.

(Morgan 1973:732, ex. 106)
As Morgan notes, “fragments” such as (2) are intelligible when interpreted

from the point of view of a larger context, discourse or social; sentence (2),
for instance, is fully intelligible if uttered as a reply to (3):
(3) How does Nixon eat his tapioca?

(Morgan 1973:732, ex. 105)
A sentence like (2) would have been asterisked in any classical transfor-
mational discussion, as it violates certain selectional restrictions. Yet sen-
tences such as this are frequent in ordinary speech and hence would rank
high on any rational acceptability hierarchy. So there are sentences that are
ungrammatical but of relatively high acceptability in context. Thus, given
the correct set of social, situational, and linguistic contexts, both (1) and
(2) might be considered good sentences of English. The difference between
the two in this regard is that, while one can imagine a reasonably high
number of contexts—and “plausible” contexts, too—in which (2) might be
uttered, the same cannot be done for (1). So, defining acceptability in this
way, the “ungrammatical” (2) is a “better” or more acceptable sentence than
the “grammatical” (1).
Another way to view the grammaticality/acceptability distinction is
to say that grammaticality is a special case of acceptability. A sentence
is grammatical if it is acceptable according to purely linguistic criteria.
Grammaticality is acceptability shorn of social and psychological differen-
tiations. Then it seems fairly apparent that grammaticality is a very highly
specialized and not terribly useful concept, outside the realm of strictly
autonomous syntax. As soon as we concur that autonomous syntax is not a
viable level of analysis (as various works written in the last ten years have,
I feel, conclusively proved), we see that a separate notion of grammaticality
is neither necessary nor possible within a coherent linguistic theory. But if
89
we discard grammaticality as a criterion, how are we to talk about cases of

the kind discussed above, where, apparently, grammaticality and accept-
ability do not coincide? Here we must speak of normal extra-linguistic
context. Although a sentence is judged by its appropriateness in its social,
psychological, and linguistic contexts, we may assume that some of these
contexts outrank others in determining whether a sentence is acceptable.
Thus, (1) would have to be judged acceptable in terms of linguistic context,
unacceptable in psychological context (that is, a participant in a discourse
would be hard put to figure out what the two parts of the conjunct had to
do with each other, and hence the conjunct as a whole is psychologically
invalid). But (2) is unacceptable in terms of pure syntactic grammaticality
(selectional restrictions between verb and adverbial phrase are violated)
but psychologically viable (since participants are capable of figuring out
from prior linguistic context what has to be supplied to make the sentence
in this particular form intelligible). Further, we may say that in certain
social situations—e.g., informal discourse—(2) is acceptable. But in oth-
ers, though the participants have the psychological ability to make sense
of the utterance, they are unwilling or unable to use it, and so we must
confine the acceptability of (2) to certain social contexts. These examples
serve to show that psychological acceptability outranks purely linguistic
acceptability.
In any case it is clear that using acceptability rather than grammaticality
judgments forces us into a much more complex theory of syntax and one
with many more variables, but one which, used correctly, makes for more
accurate predictions—which is, after all, what a linguistic theory should
do. It is a far more problematic theory, however; and the assignment of
particular points on the hierarchy to particular sentences is not the hard-
est issue to be solved if one is to make good use of the notion of hierar-
chical acceptability. What I want to do for the remainder of this paper is
examine one particularly vexing problem: the applicability of the concept
of acceptability to one dialect of American English, and the question of
when a judgment of acceptability ceases to be a linguistic judgment and
becomes a political statement.
In earlier work (R. Lakoff 1975), I talked about differences between
men’s language, or rather the standard language, and “women’s” language.
I catalogued a number of features that seem to characterize women’s lan-
guage, in those segments of the American populace that, consciously or
otherwise, make that distinction. (It is useful to bear in mind, here and in
the succeeding discussion, that a speaker’s disavowal of the use of, or even
the knowledge of, women’s language does not mean she does not or cannot
use or understand it. In socially and psychologically charged issues such as

90
whether or not one speaks women’s language, one’s judgments as to one’s

own speech patterns may easily be false; what one says is by no means
identical to what one wishes one says or fears one says. This fact colors all
intuitive observation, as Labov has correctly noted; but it colors the most
strongly those observations where the observer has something to gain or
lose by his (or her) decision. And although some women will find it neces-
sary to believe that they always use women’s language, and others that they
never do, it is probably true that all of us use some of it some of the time,
whether we want to or not, whether we hear ourselves doing it or not.)
Let me recapitulate briefly what I consider characteristic of women’s
language:
1. Special vocabulary—in particular, women seem to discriminate linguis-

tically among colors with more precision than do men.
2. Use of adjectives that principally express the speaker’s feelings toward
the subject under discussion: charming, adorable, divine, and the like.
3. Use of empty intensifiers like so, such.
4. Greater adherence to standard “correct” forms; avoidance of slang and
neologisms, both lexically and grammatically. For example, psychologi-
cal studies of kindergarten-age and nursery-school children indicate
that, even at this age, little girls are guardians of “correct” grammar: they
“drop their g’s” in participial endings (runnin’, talkin’) much less than
boys, use fewer substandard forms (ain’t, snuck), less double negation
(I didn’t do nothin’), use fewer “bad” words, and in general articulate
more precisely. These traits become a part of traditional adult women’s
language.
5. Use of lexical, grammatical, or phonological devices to suggest hesitancy
or deference. Examples:
a. Prefacing declarative utterances with I guess, questions with I won-
der, etc. In this way the speaker mitigates the force of her speech
act, creating an impression of hesitancy to impose her opinions on
other participants in a discourse and thereby giving an impression
of politeness, or deference. Of course things are not necessarily what
they seem, and the politeness, or deference, may be conventional
rather than real. But confusion may easily arise, and the speaker’s
character be judged by her superficial style—marking her as indeci-
sive, inarticulate, and fuzzy-minded. On the other hand, if, in tradi-
tional American culture, a female speaker habitually fails to employ
these devices, she is categorized as aggressive and unfeminine. Until
recently, most speakers have opted for the first of these uncomfort-
able options.
91
b. Silences or interjections like ah, um… .

c. Use of questions where a declarative would be more appropriate (i.e.,
where the speaker is in possession of the necessary information,
if anyone is). In such cases, the question is not a request that the
addressee supply information but rather that he supply reassurance
that the speaker’s speech act is acceptable to him.
d. Lower vocal volume, sometimes a mere whisper.
6. Greater use of euphemism for topics that are considered to be taboo or
unladylike, as well as greater tact in avoiding sensitive topics in the pres-
ence of people they are sensitive to. For this reason (also a part of non-
linguistic behavior, of course) women have typically been considered the
arbiters of etiquette as well as the mainstays of conservatism. This latter
role is also illustrated in point 4. Otto Jespersen was perhaps the first
to discuss the role of women in linguistic conservatism. More recently
Labov has made the opposite claim: that women are linguistically more
innovative than men. As with most cases of conflicting claims, both are
probably partially valid, and each especially valid in the writer’s contem-
porary society. It is also true that a group might be conservative in one
aspect of language use, radical in another. Thus, if the behavioral role
of a particular group were facilitated by an emergent linguistic form,
the group to whose advantage it would be to adopt this form might well
do so more quickly than other social groups and in this regard appear
especially innovative. But in other aspects of language use, the same
group might elect to be more conservative. For example (as discussed
by Edwin Newman (1974) among many others) a relatively recent trend
in American speech favors the use of speech-act hedges like like, y’know,
I mean… . (For some discussion of the role of these hedges in American
English, see my review of Newman (R. Lakoff 1976).) Now as I noted
above, for reasons consonant with their traditional social role, women
have a tendency to use hedges more profusely than do men. It might
therefore be to their advantage to adopt these hedges more quickly and
more profusely than men do. But they might still stoutly resist other
changes, e.g., the use of like for as. Labov, I believe, was talking about
phonological innovation. Often, nonstandard phonological forms
sound “cute,” or nonserious, and mark the speech act in which they
occur as amusing, social rather than informative. For example, we can
think of the shift to Black English among middle-class white academics
when they are feeling linguistically playful or the use of baby talk by one
speaker to show that he isn’t really in sympathy with someone else’s
complaints. The more serious the occasion, the more pompously conser-
vative the style, and bombast seems in our culture to be more available

92
to men than to women. Thus it would be to a woman’s advantage to use

phonological forms that made her sound “cute” and nonserious—for
the same reason that it is to her advantage to have in her lexicon adjec-
tives like charming, divine, and adorable. In other times, women might be
assumed to be more innovative because they were less educated, were
less in touch with a formal norm. And where women are seen as the arbi-
ters of respectability, it is to be expected that they will resist any change
that can be viewed as a lowering of standards, linguistic or moral.
7. Greater variation in pitch and intonation. This difference might be
viewed in either of two ways: as a means of achieving indirectness or
as a way of expressing emotion. By making use of pitch and intonation
variation, one can express thoughts and feelings nonverbally which it
might be difficult or uncomfortable to put into words. But if one has
no intention of explicitly talking about one’s feelings, more emotional
warmth is conveyed by a speech style that allows these variations than
one that does not. So to judge what such a trait connotes for an indi-
vidual’s speech style, one must first examine the rest of her (or his) style.
Then we may say that three basic trends characterize women’s language
as a deviation from the standard:
1. Nondirectness: e.g., 5, 6, 7
2. Emotional expression: e.g., 2, 3, 6, and 7
3. Conservatism: e.g., 4 and 6
Point (1) does not figure in this summary, and indeed it is rather mislead-
ing to categorize “special vocabulary” as idiosyncratic to women’s language.
The deeper point to be made here is that every subculture has its own
vocabulary and that vocabulary will involve terms that are of specific use
to the particular culture in question. Whorf, of course, was the first to raise
this issue when he pointed out that Eskimo has six words for “snow” where
English has just one—presumably because snow is much more important
in the Eskimos’ lives than it is in ours, and minute differences in the quality
and quantity of snow are for them of crucial importance, as they are not for
us. So they need precise words to make these crucial distinctions, as we do
not. As I have said elsewhere, this is undoubtedly the reason for the more
precise color-discrimination vocabulary among women: because it has
traditionally been considered important for a woman to possess this sort
of expertise, for fashion and interior decoration have both been women’s
work. Of course, men have their own special highly developed vocabularies,
93
e.g., in regard to automobiles and sports, to which women traditionally

have not been privy.
What we find, then, in looking at those traits that distinguish women’s
language from neutral language is that we can define them in terms that
cover more purely linguistic behavior. In fact, the only distinctions it seems
reasonable to assert between the two forms of English, rather than aris-
ing directly out of differences in the learning of a linguistically relevant
grammatical system, appear to stem from differences in what is socially
and psychologically expected of women in terms of explicit behavior, both
linguistic and nonlinguistic. Where women’s speech differs syntactically
from the standard, it does not differ in containing more, fewer, or differ-
ently stated rules. I know of no syntactic rule present in one group’s gram-
mar and entirely absent from the other: that is, I know of no case where a
sentence utterable by one group would be totally impossible in all contexts
for the other. In this sense we are not even dealing with differences in the
linguistic conditions for the applicability of rules, or their order, as is true
of dialect differences in quantifier-crossing as discussed by Carden and oth-
ers. It is not out of the question that this might be so, and in other lan-
guages it would not be overly surprising if it were so. How would it look? It
might mirror Carden’s cases. Thus, suppose that all English-speaking men,
when they encountered (4):
(4) All of the boys don’t like some of the girls.
interpreted it as (5):
(5) None of the boys like certain girls: namely, Mary, Alice, Nancy… .
Whereas, faced with the same sentence, all English-speaking women inter-
preted it as (6):
(6) Only some of the boys—Fred, John, Max …—like some of

the girls.
But, as I say, I know of no such cases. Rather, even where syntactic rules
such as question formation are involved, the difference between the dia-
lects lies in the fact that in one a sentence is usable in more social or psy-
chological situations than in the other. So we cannot define the two dialects
in terms of purely linguistic autonomous-syntactic distinctions. Additional
reason for this belief is the fact that even if a woman will not or cannot
use certain forms that are not parts of traditional women’s language, she

94
can certainly understand them and, if she is of a certain personality type,

when she encounters them in other people’s speech, can correct them, i.e.,
indicate what forms in her dialect are equivalent to the ones she has heard.
All this suggests that linguistic deviation from the norm is but one form
of social deviation from the norm. But here we must raise another, and
more troubling, question: What do we mean when we talk about the norm?
Whose norm? And to what extent, when we talk about norms and standards
and deviations, are we invoking value judgments? Psychological writers in
particular often claim piously that they intend no reproof when they speak
of abnormality or aberration; but actually one typically finds the implicit
claim that the standard is better, and if you know what’s good form you’ll
adhere to it. Of course, this then constitutes a self-fulfilling prophecy. The
same has been true too often in the past in linguistic dialectology, which
tended to be prescriptive and, more or less overtly, looked with disapproval
at “substandard” forms. But in the case of women’s language the question
becomes a bit more complex: If a woman in our society speaks traditional
women’s language, and more generally behaves like a traditional woman, is
she conforming to our cultural norm? Or deviating from it? And if we must
be prescriptive, what shall we punish as deviant? More positively, for what
kinds of behavior shall a woman in our society be rewarded? What, return-
ing to the theme of this essay, constitutes “acceptability” for a woman in
this culture—linguistically and behaviorally?
In fact it is less the bare notion of acceptability that causes difficulty
than the fact that it is made into a prescription. In the same way, we can
talk about expectation being a two-edged concept. We expect women to
talk a certain way, which is only partly damaging; but we also expect it
of women that they will behave a certain way, and thereby we impose a
value judgment, either that it’s good for a woman to talk traditional wom-
en’s language because it fits the stereotype, which is by definition good
because it does not force us to readjust our perception of reality; or it’s
bad for women to speak women’s language because it deviates from the
norm, and the norm for society as a whole is viewed as a good thing to
adhere to, and any deviation is to be criticized. And both linguistically and
otherwise, when we say that a certain form of behavior is “acceptable” for
a woman, we tend to be prescribing—both for the woman, that she act
this way to indicate she “knows her place,” and for a man, that he not act
this way, to show he knows his. This kind of prescriptivism is constraining
and destructive. We should be able to think about acceptability in regard
to linguistic gender distinctions without recourse to value judgments, just
as linguists several generations ago pointed out that description did not
imply prescription, that talking about a norm did not imply that deviation
95
from that norm was censurable. I recall that Paul Goodman—no doubt in
good company—made this mistake some years back in an article in the
New York Review of Books (1964), in which he criticized Chomsky’s lin-
guistic work on the grounds that—by distinguishing sentences that were
grammatical from those which were not—he was squashing linguistic
creativity and innovation. But the distinction between description and
prescription is still blurrier, even among linguists, than we might like, par-
ticularly in areas where we have been brought up to make value judgments
before we learned to be disinterested academic observers. The question of
gender-related roles is one of these highly charged areas, and it therefore
behooves us in discussing acceptability as a factor in understanding wom-
en’s language to bear in mind that there is a danger that we will confuse
linguistic norms with social values.
There are other confusions to be avoided. The notion of acceptability,
I have said above, implies a standard against which a speech act may be
judged. It has also been pointed out that, in talking about acceptability as
opposed to grammaticality, that standard is grounded in social and psy-
chological context: an act of speech or behavior is judged acceptable in a
specific context. Now, it would seem at first glance that men and women,
being these days participants in the same activities, similarly educated, at
least superficially raised alike, would share this set of contexts. A male and
a female, participating in a specific kind of behavior, linguistic or other-
wise, would perceive the context in which the behavior was to take place
similarly. But if acceptability implies appropriateness within a particular
social-psychological setting, and as we have suggested men’s and women’s
languages differ somewhat in terms of what is acceptable, then we are faced
with a paradox.
We must, rather, assume, I think, that a given context is interpreted one
way by a male speaker, another by a female. Actually, it will be recalled we
are dealing with a complex hierarchy of acceptability, and it is in principle
not at all unlikely that different speakers will arrange their worlds in quite
different ways. It is not as though every male speaker of English defines
Contexts A-L , let us say, as “situations requiring directness” and M-Z as
“requiring nondirectness” while all women interpret situations A-P in the
former way, Q-Z in the latter. We would prefer to say—continuing for the
sake of clarity to look at the division of social contexts in this extremely
simplistic way for the moment—that women would tend to interpret
situations as requiring nondirectness until further down in the alphabet
than men typically would. But all sorts of variations are conceivable. That
is, what I mean by saying that traditionally women’s language has tended
toward nondirectness is that women will interpret a greater number of

96
social contexts as being appropriate for nondirect expression than will

men; and perhaps as well, that women will, in a situation in which both
men and women would tend toward nondirectness, tend toward greater
nondirectness. But this implies that a woman’s social/psychological con-
text is often, or perhaps always, different from that of a man. Whether
innately, or through early education, a woman learns to perceive social
situations, and interpret psychological events, one way, a man, another.
Hence a setting that would evoke one set of linguistic responses in a man
would be expected to evoke another in a woman.
If this is true, it seems reasonable to say that, if we want to even out
the differences in linguistic behavior between men and women (a goal the
utility, not to say feasibility, of which is in my mind very much open to
question), linguistic behavior cannot be changed directly but only through
somehow educating men and women so that they typically perceive the
same situation in the same way. Of course, many questions are being
begged here: it is also apparent that no two individuals, in all likelihood,
perceive a single social or psychological setting precisely identically. But
we are talking here about somewhat grosser differentiations; the problem
is, how much grosser? Since we have been talking in terms of a complex,
highly individualized hierarchy, at what point do we draw the line and stop
speaking about individual idiosyncrasies and start recognizing broader sex-
linked distinctions? Since it will, predictably, never be so that 100% of the
men react in one way to a situation, 100% of the women a different way, at
what point do we decide we are dealing with “women’s language”?
We tend to consider women’s language as the aberration from the norm,
the standard. To do so raises more problems.
First, there is the implication that to be a woman is to be a deviant.
This has traditionally been true—people have looked at men’s behavior
as the norm, as rational, and men’s language likewise. Hence it has been
typical—to the extent that such a topic has even been considered worthy of
discussion—to talk about women’s language. Men’s language is language
and need not be further specified. But this is the case also, as we would
expect, for other aspects of human behavior. Until recently no one has
lifted an eyebrow at questions like “Was will das Weib?” suggesting that a
woman was a thing apart, something we could analyze. But the question
“Was will der Mann?” would have been unthinkable. But if men’s behavior is
the standard, so normal that it is not even worth investigating in its own
right, then a woman cannot expect equality with men. She simply is not
parallel to a man, and thus cannot expect to be treated similarly.
Moreover, there is another danger in looking at women’s language as
having special and idiosyncratic standards for acceptability. What is a
97
woman to do? If she adopts the frame of reference of a man—supposing

she can—she will be ostracized by traditional society for not conforming
to what is acceptable for a woman. If, on the other hand, she does adopt
women’s behavior, she will be treated nonseriously because her behavior is
not commensurate with the standard, the behavior expected of men.
There is one corner of the real world where, interestingly, the distinc-
tion between men’s and women’s language seems to be blurred: the same
set of social and psychological conditions, or very nearly so, are opera-
tive in determining the acceptability of utterances both of men and of
women. This occurs in academia. The traits I listed earlier as characteristic
of women’s speech are frequent in academic men’s speech, and academic
women’s speech tends at the same time to use these devices less than
does the speech of traditional women. This is not, I think, to imply that
academics are sexless or that academic men and women find themselves
in social settings that cause them to perceive their roles differently less
often than does the general populace. Rather, I think that this difference
indicates that male and female professionals, in academia, regardless
of gender, perceive their roles as similar and hence tend to have similar
perceptions of the social and psychological settings in which they find
themselves. It is interesting that the distinction is not erased in favor
of the masculine form, but rather there is a neutralization toward the
center. What is of interest in this is that in society generally, when there
is pressure to blur sex distinctions in roles, usually women seek to adopt
men’s prerogatives, seldom the reverse. But in academia some of women’s
prerogatives—nondirectness and expression of emotions—are adopted
by men. I have suggested elsewhere that these traits have less to do with
anything inherent in the female character than they do with being at the
periphery of power, or opting out of power. Academics, the British upper-
class men, and women all share this situation, for various reasons and
in different ways, and hence, to a rather surprising extent, share their
language.
My conclusions then are clear:
1. There is a women’s language in American English, if by that we mean

that in a particular context, women and men may not express the same
thing in the same way.
2. This difference is not traceable to the purely linguistic grammar: the
distinctions are not statable in terms of syntactically conditioned rules,
nor are they statable as either/or pairs. The differences involve hierarchi-
cal acceptability—some sentences are better for women to say in some
circumstances.

98
3. Acceptability in language is directly related to social and psychological

perceptions; a sentence is defined as acceptable if it is fitting in the set-
ting in which it is used. Grammaticality, then, is a special case of accept-
ability, in which only the linguistic aspects of the social-psychological
setting are taken into account.
4. Therefore it seems likely that men and women learn to view similar social
and psychological contexts differently and hence will find it appropriate
to utter different sentences in the same setting.
5. Women’s language differs from the standard in being more nondirect,
more capable of expressing emotion, and more conservative.
6. There is a danger in opposing “women’s language” to “the standard,” as
there is in opposing any group’s behavior to a hypothetical “standard”;
and it is by no means clear what is best for women or for society: to per-
petuate the dual standards of acceptability or seek to merge them.
REFERENCES
Gleitman, Lila. 1965. Coordinating conjunctions in English. Language 41(2): 260–93.

Goodman, Paul. 1964. On linguistics. Review of A linguistic introduction to the history
of English, by Morton W. Bloomfield and Leonard Newmark. New York Review of
Books, 14 May 1964, 15.
Klima, Edward. 1964. Negation in English. The structure of language: Readings in the
philosophy of language, ed. by Jerry A. Fodor and Jerrold J. Katz, 246–323.
Englewood Cliffs, NJ: Prentice Hall.
55: 608–15.
Lakoff, Robin. 1976. Why you can’t say what you mean. Review of Strictly speaking, by
Edwin Newman. Centrum 4: 151–70.
Morgan, Jerry L. 1973. Sentence fragments and the notion “sentence.” Issues in
linguistics: Papers in honor of Henry and Renée Kahane, ed. by Braj B. Kachru,
Robert Lees, Yakov Malkiel, Angelina Pietrangeli, and Sol Saporta, 719–51.
Urbana: University of Illinois Press.
Newman, Edward. 1974. Strictly speaking. New York: Bobbs-Merrill.
99
Introduction to “Stylistic strategies

within a grammar of style”
BY DEBOR AH TANNEN
R obin Lakoff is one of the reasons I entered the field of linguistics. In

1973, I was a teacher of remedial writing and freshman composition at
Lehman College of the City University of New York. I attended the Linguistic
Society of America’s Linguistic Institute that summer at the University of
Michigan to get an idea of what linguistics was all about. It sounded like an
interesting way to spend the summer. One of the courses I took was taught
by Robin Lakoff. Her notion of communicative style, which ran through
everything she taught in that course, captured my imagination and has
driven it, and my research, ever since.
I believe Lakoff’s notion of communicative style is one of the pivotal
concepts in discourse analysis or, as Lakoff herself would probably call it,
pragmatics. Therefore, when I teach discourse analysis, I assign this paper
to give my students a sense of this concept. The paper also happens to
be about women and language, but, as the title indicates, gender here is
more a corner-of-the-eye phenomenon than the direct focus, a perspec-
tive that I think is useful not only to the study of communicative style but
also to the topic of gender. When gazed at head-on, gender seems to elude
understanding because everyone’s vision is clouded by the emotional dust
storms this topic stirs. As Lakoff explains here, gender-related patterns of
communicative style are simply one facet of communicative style, which
is also shaped by other social and individual influences. In this paper, gen-
der is simply an example—but a readily recognizable one—of the general
100
notion of style: how it works, that it can be seen as a system of rules, why

it matters in people’s lives. Furthermore, by drawing parallels with David
Shapiro’s work on personal style, Lakoff develops the notion that style is
a broadly coherent phenomenon in individuals’ lives, not narrowly limited
to language.
The main thrust of this paper is Lakoff’s assertion that conversation is
rule-governed, that one can devise a grammar of style. In this sense, one
can see that her work grows directly out of her training and earlier work
in generative grammar. As Grice set out to demonstrate, in the philosophy
of language, that everyday conversation follows rules of logic, so Lakoff, a
pioneer in linguistic pragmatics, set out to demonstrate that communica-
tive style is not random but follows “grammatical” rules: we have a sense
of what goes with what, and violations of that coherence give us a start.
They are marked, call attention to themselves, and can be seen as errors,
misfires, or deliberate signifiers of special meaning.
One reason I like this paper in particular is that it clarifies that the stylis-
tic nodes clarity, distance, deference, and camaraderie are seen as arrayed
along a continuum rather than hierarchically ordered. Even more, the four
stylistic nodes yield stylistic strategies that can be intermingled, a kind of
mix-and-match affair.
For those who are primarily interested in language and gender, it would
be requisite, I think, to read this paper to understand the roots of Lakoff’s
work in formal linguistics and to appreciate the nuances of her claims.
Among these, three stand out. First, Lakoff speaks of “women’s style” as
a description (obviously not a prescription) of a stereotype rather than a
description of every woman’s way of speaking, and yet a stereotype with
some basis in actuality, else it would not be recognizable. Second, she notes
that the very concept “women’s language” hides a destructive assumption
that what men do is the norm, whereas what women do is marked; they are
the other, outside the norm. Third, she observes that it is this very devalu-
ing of styles associated with women that leads some to deny those styles’
existence.
Finally, I should admit to having a personal attachment to this paper
because Lakoff wrote it when I was a graduate student at Berkeley, and I had
the privilege of discussing the ideas with her as she was working them out
as well as reading the article in draft. But I don’t think this bias is the rea-
son I doubt that anyone can study either discourse or gender and language
without taking into account the framework Lakoff lays out in this paper.
101
CHAPTER 5
Stylistic strategies within

a grammar of style (1979)
I nterpersonal behavior is frequently regarded as unpredictable and spon-

taneous. We do not feel that we are following rules or even a preordained
pattern in the way we talk to others, move, respond emotionally, work,
think—all the varied aspects of what, following Shapiro (1965), we can call
personal style. Indeed, some of us might be horrified at the idea that, in
all our actions, we are governed by implicit rules, just as Chomsky (e.g.,
1968) has shown we are in our linguistic behavior (and has thereby himself
aroused horrified responses), and are as little aware of it. In this paper,
assuming the validity of the assumption of generative grammar that lin-
guistic behavior is rule-governed, I want to extend that claim to a variety
of other kinds of behavior that can be subsumed under terms like character,
personality, or personal style. I will argue that the same kind of evidence
that indicates a need for grammatical rules relating two levels of linguis-
tic structure exists for this wider range of human functioning. Needless to
say, the argument that implicit rules guide our behavior denies us neither
This paper originally appeared in Language, Sex and Gender: Does La Différence Make
a Difference?, edited by Judith Orasanu, Mariam K. Slater, and Leonore Loeb Adler,
53–78. Annals of the New York Academy of Sciences 327. New York: New York Academy
of Sciences, 1979. Reprinted here with permission of the publisher.
Acknowledgments: The preparation of this paper has profited from my discussion of
its contents with many people, as well as numerous opportunities to present its ideas
in various stages, to many groups. In particular I would like to thank Linda Coleman
for her very helpful suggestions leading to the final form of the stylistic continuum as
it appears here; and Deborah Tannen for innumerable valuable improvements in both
style and content.
102
autonomy nor creativity: these rules are predictive schemata, descriptive

rather than prescriptive. In this paper, drawing on previous work, I use as a
principal example the distinction between men’s and women’s typical per-
sonal style in American middle-class culture; but the argument holds for
the behavior of any individual or group that is felt by its members and by
outsiders to function as a cohesive unit.
The reader may wonder whether there is any justification—beyond
my personal whim and the convenience of having previous work to draw
upon—for using women’s language as a paradigm or model for the pur-
pose of illustrating the more general thesis of the existence of a grammar
of style. To this I would make the rejoinder that it is as good as any, and
I must choose one. Thus Chomsky (1965) chose the modal auxiliary system
of English as his example attesting to the need for a grammar of the trans-
formational type, although other examples might as well have been used.
But even more to the point, women’s language is accessible to every
member of this culture as a stereotype. Whether the stereotype is equally
valid for all women is certainly debatable; but the fact of its existence, overt
or subliminal, affects every one of us, and its assumptions are generally
agreed on. Hence it provides an especially clear case. And while women’s lan-
guage, and women’s style, have long been recognized and commented upon
as an aggregate of traits, not much has been done toward accounting for
why these particular traits cluster together—that is, toward constructing
an explanatory, as opposed to a merely descriptive, model. I shall attempt a
beginning of such a systematization: that is, I shall try to make predictions,
show that there are constraints on co-occurrence, that there are explana-
tions for the existence of some phenomena and the absence of others—in
short, to construct a predictive system of rules for style, to establish for
style something analogous to what linguists construct for language in the
form of a grammar.
To talk really convincingly about style, one should examine all aspects of
a person’s functioning, much as Shapiro investigated several of the aspects
of four or five recognized neurotic styles to show how each formed a coher-
ent system, each trait being predictable on the basis of certain postulates.
His basic determinant of style was the mode of attentiveness. Obsessive-
compulsive style, for instance, involves a need to pay very close attention
to small details one at a time, whereas hysterical style works just the oppo-
site way: a person with such a style perceives the universe as a kind of large
undefined blur in which everything is seen at once, and nothing is singled
out as outstanding. Shapiro shows how these basic ways of perceiving the
world are involved in one’s mode of cognition, learning, and processing
103
new information; in perception; in ways of using one’s body physically; and

in one’s style of social interaction. This interaction, like the grammar of
a language one speaks, is something the normal participant knows about
unconsciously or implicitly. So, seeing that X is concerned with details in
his work, seldom looking at the larger issues but confining himself to tiny
aspects of the problem, an observer will expect, consciously or otherwise,
to find that X is very precise in his linguistic expression, that he will be
very concerned with using just the right word, and that he is apt to be con-
servative, politically and socially, feeling uncomfortable with surprises of
any sort. We can then informally state, as a rule, “If there is concern for
details, then there will be avoidance of surprises.” If this is borne out in
future encounters with X, and, further, with Z and W, the observer feels
comfortable—his rule works, he is in control. If, on the other hand, X—or
W, or Z—turns out to be flighty in his political affiliations and fuzzy in
his choice of expression, the observer may be puzzled, and may feel and/
or even remark to others that that person is a paradox, you cannot figure
him out. Similarly, if the observer finds that X on one occasion is bound
up with niggling details, he comes to expect this sort of behavior of X on
subsequent occasions, and will likewise be surprised if on their next meet-
ing X affects an extremely general and undefined Weltanschauung. So our
rules are predictive across modalities (that is, from one aspect of behavior
to another) as well as across time (from one encounter to the next). I will
call these predictabilities by the names, respectively, of coherency and con-
sistency, and assume that for something to be recognized as a personal style
it must possess both these attributes. (I assume that personal style may, as
a result of the vicissitudes of life, through conscious motivation or other-
wise, change in several of its salient aspects over time, but typically rather
slowly and not too radically. I leave open here the question of how much
style—or character, or personality—may really change.)
Stylistic coherency, then, is the predictability of one component of per-
sonal style from another, on the basis of more or less apparent similari-
ties between them, genetically or dynamically. This notion, as I suggested
above, is reminiscent of the basic linguistic concept of co-occurrence as a
criterion for grammaticality. Syntactic rules tell us, among other things,
which words, or other elements, may occur side by side with others. If these
constraints are violated, an uninterpretable, or at least unacceptable, sen-
tence results. Such a constraint may involve semantic criteria, as in the
well-known example:
(1) Colorless green ideas sleep furiously.
S tylistic strategies within a grammar of style [ 103 ]

104
in which the problems arise because semantic categories are juxtaposed

contrary to universal meaning constraints. They may also involve syntactic
criteria, as in:
(2) John will leave for Rome yesterday.
where the use of the future tense conflicts with the use of yesterday, refer-
ring to past time, or:
(3) The boys is here.
Of course, we might as legitimately call the last two semantic violations,

illustrating a fact that most linguistic theorists have come to accept, with
sorrow: levels of grammar cannot be dealt with as separable entities. We
are dealing with parallel problems when we want to define normal style.
Thus, we would like to express, in precise and possibly even formal nota-
tion (an attempt that has been made in linguistic theory for some time, and
though it has never really been successful, it has been valuable as a learning
experience for linguists), the intuitively correct generalization that certain
modalities of human behavior can be reasonably expected to co-occur, and
others cannot. As an example, within the linguistic sphere of style alone,
we can state a few co-occurrence constraints:
(a) A person given to pontification on large and abstract issues of great

universal and philosophical relevance, most likely, will employ an elabo-
rately Latinate and convoluted vocabulary as well as a rather Germanic
syntax wherewith to express these thoughts: there is an expected co-
occurrence of lexical, syntactic, and semantic material.
(b) A person given to asking questions where others might use declaratives
can be expected also to employ euphemistic modes of speech on any
topic that might be embarrassing, and might further engage in a lot of
linguistic hedging (I guess; sorta; y’know). Here we find co-occurrence of
syntactic and lexical patterns.
Exceptions occur, of course, but we tend to remember them particularly,

as they are so striking. (“Mary is so polite whenever she wants you to do
anything; how come she uses four-letter words?”) Such exceptions, notable
and memorable as they may be, are the exception rather than the rule, com-
parable to the uses of poets who employ locutions like “A grief ago” (Dylan
Thomas) or “The fog comes in on little cat feet” (Carl Sandburg), thereby
violating the rules of linguistic co-occurrence discussed previously, and
105
thereby creating particularly memorable strings of words, typically called

“poetry.” I do not want to imply that such stylistic incompatibilities are
subsumable under the rubric of behavioral poetry; but it is perhaps not
beside the point to suggest that, when extreme, they are assigned to a spe-
cial category to which there is given a special name, namely schizophrenia.
The relationship between schizophrenic behavior and poetic behavior is
frequently if inadequately commented upon by workers in many fields.
Additionally, co-occurrence constraints on style pertain between differ-
ent modalities:
(A) Someone who dresses conservatively and very carefully can be expected
to use language carefully and equally precisely.
(B) Someone who divides humanity into black and white also, cog-

nitively, divides ideas into right and wrong, with no gray areas in
either case. Again, there are people who violate these expectations,
but they are experienced as aberrant. If the aberration is clearly
purposeful, for effect, it is comic: much of Woody Allen’s humor,
for example, depends on incompatibilities of stylistic modalities.
But if it seems unintentional, we consider it indicative of serious
psychopathology—schizophrenia again.
Co-occurrence constraints are but one aspect of grammatical rules.

Syntactic theory devised over the last couple of decades makes other
assumptions about human linguistic capacities, capacities that are
assumed on the basis of excellent evidence to be found in all human lan-
guages. We might well ask how far to carry the generalization. We know
that language is a form of communication of thoughts and emotions,
perhaps (though not certainly) the form of communication par excel-
lence. But there are other modes employed by normal persons. Language
is surely the principal conscious and intentional mode; but that does not
mean it is the most important or the most frequent or the best under-
stood. Rather, all sorts of arguments and evidence have been adduced
of late that language is but a sort of frosting on the cake; when conflict
exists between what a speaker is expressing linguistically and what he is
communicating, generally unintentionally, by means of other modalities,
other participants in the communication tend, generally unconsciously, to
interpret the nonlinguistic communication as the genuine or truthful one.1
1. The work of Paul Ekman is particularly interesting in this regard, indicating that
subjects, when forced to decide between conflicting interpretations of a contribution
based on facial expression and verbal statement, will opt for the truth of the former.

106
But to say this is to assume that not only do these sophisticated interpret-
ers of discourse (i.e., you and I and everyone else) employ their implicit
rules in the interpretation of linguistic structures (as we have been per-
fectly well aware for quite some time) but that we use the same system
for what is sometimes called extralinguistic or paralinguistic functioning,
and we detect mismatches by our application of these rules. Woe betide
anyone caught in a mismatch (Richard Nixon is a notorious example).
I take it, then, as proved that nonlinguistic style is as much rule-governed
behavior as is linguistic style, and I would further venture—it is not
yet proved by any means—that the same kinds of rules, the same basic
devices, are operative in all human behavioral systems. Style is a unity,
not only in its superficial manifestations in the normal individual but
also in its basic mechanisms.
It is therefore incumbent on the linguist determined to give evidence
for this last contention to show that those traits ascribed to the linguis-
tic grammar can be justified and indeed are essential as parts of a wider
stylistic grammar. If this should be provable, we can be content in that we
have given yet another instance of the working of Occam’s Razor: After all,
doesn’t it make sense that the human mind should function with as little
baggage as it can get away with? If it can be shown to need only one general
rule system, from which all forms of psychic behavior can be derived, isn’t
this preferable to a theory that requires separate theories, and thus sepa-
rate rules, for each subtype of behavior? But even if the former is ideal, it
naturally remains to be proved. I do not intend to make a watertight case
for psychic economy here, but merely to give some preliminary arguments
showing that a unified system is tenable. I use linguistic theory as my tem-
plate, and more specifically a kind of generalized theory derived from vari-
ous realizations of generative grammar. The reason I do so is partly that
I am trained as a linguist, so it comes naturally to me, but also because it is
the only attempt at formalization of psychic structure that I know, the only
attempt, I should say, at a predictive rather than a merely taxonomic or
classificatory framework, such as are found in many psychological theories
of character.
To return to the point: linguistic theory assumes a bipartite syntax
(here I am oversimplifying considerably, but the complexities are not
important for our purposes). In such a syntactic model, two levels of syn-
tactic structure are relevant: a superficial level, which is the utterance as it
appears in spoken or written form, and the underlying or deep structure,
which is related to the meaning of the sentence, and contains in a form
accessible to inspection all the elements needed to account for the mean-
ing of the sentence. Since every sentence in a language that is found in
107
superficial form can be related to an underlying form, we must assume

a set of transformational rules, relating these two levels, explaining for-
mally the relation between the sentence as it appears on the surface and
what it “means.” For often, if we are asked what a sentence means, how
its component parts are related, to answer that question properly we will
have to refer to elements that are not present on the surface, but can be
inferred from the properties of the rest of the sentence. It is frequently
assumed that transformational rules exist to make sentences easier to
understand or quicker to utter; this is true some of the time, although by
no means always.2 So if we want to argue for parallelism between linguistic
and other levels of psychic functioning, we shall have to give evidence for
these two levels in stylistic behavior, and a set of rules linking them, stat-
ing, for instance, in effect, that Surface Structure A always corresponds
to Underlying Structure B, through the application of Transformational
Rule C.
The most persuasive evidence for this duality of structure in language
is in the existence of ambiguity and paraphrase relationships between
sentences, which could not otherwise be explained. A sentence is ambigu-
ous if it has different meanings depending on the contexts in which it is
uttered: such a surface structure can be explained as being related by two
different transformational routes to two different underlying structures.
So for instance with (4):
(4) Visiting relatives can be a nuisance.
In one derivational history, we start from an underlying structure roughly

equivalent in meaning to (5)—not itself an underlying structure:
(5) Relatives who visit can be a nuisance.
and in the other, to (6):
(6) For one to visit relatives can be a nuisance.
In each of these underlying structures, something is deleted to produce the

superficial sentence (4), but in each case something different is deleted.
Hence, starting from structures with different meanings, we end up with
identical structures.
2. Arguments for transformational rules as processing strategies were first advanced

by Bever and Langendoen (1972).

108
On the other hand, sometimes we find two quite different superficial

sentences that are very similar in meaning. Such sentences are paraphrases
of one another. Examples (7) and (8), and (9) and (10) are illustrative.
(7) The ball was hit by the boy.

(8) The boy hit the ball.
(9) John threw the garbage out.
(10) John threw out the garbage.
The members of each pair originate as very similar underlying structures,

but they undergo different sets of transformational rules, leading to two
different superficial representations. Since the underlying structures of
each pair are closely similar, the meanings of the sentences in each pair are
similar as well; the surface forms are different because each has a different
syntactic history.3 Just as the existence of these types of relationships in
grammar is evidence for the duality of grammatical structures, if we could
find evidence of a similar duality in other modalities of human behavior, we
would have persuasive evidence that a similar duality should be postulated
to explain these other modalities.
I have given evidence elsewhere (R. Lakoff 1977) of the existence of
such relationships. To summarize, I have suggested that women’s lan-
guage and (typically male) academese function for the same underlying
or deeper purpose, that of evading responsibility for what is being said;
but they do it by different superficial means, which incidentally are given
very different values by our culture; and these then are essentially para-
phrases of one another, like the examples of sentences (7)–(10) above;
and that women’s style and children’s style, although superficially simi-
lar in many of their aspects, are actually used with rather different aims
and arise from very different psychic situations, so that they can be said
to have different meanings, different underlying structures, correspond-
ing to similar surface structures, and thus are stylistically ambiguous.
These arguments imply that many if not all the modalities of human
behavior are dualistic in their functioning, and therefore are best inter-
preted as governed by rules at the least analogous to, and perhaps identi-
cal to, the kinds of rules that have been identified as existing in linguistic
structures.
3. I am oversimplifying greatly here, and probably distorting the facts more than
I ought. The basic premise is true, but the discussion of the intricacies involved in the
differences of meaning and usage in the members of each pair, however valid, would
take us far afield.
109
We can in fact give arguments that, in the various aspects of human

style, there are analogous processes to those of language. Linguistic rules
are implicit, or unconscious: a speaker follows them, using them to pro-
duce as well as to distinguish grammatical utterances of his language, with-
out being able to state the form of these rules, and typically without even
knowing that such rules exist. Linguists know that there exists a syntactic
rule relating sentences like (8) and (9); but even linguists, with their spe-
cial knowledge, cannot give the correct form of this rule; “naive speakers,”
even worse off in this respect, may be interested to learn in an elementary
syntax course that they have access to such rules. Yet the naive speaker
has no trouble using and processing passive sentences. Similarly, we cor-
relate the various aspects of our style, we determine whether or not a new
piece of behavior on the part of an acquaintance surprises us, we decide
whether the range of a person’s behaviors in the course of an evening or a
month or a year is reasonable fluctuation or evidence of serious psychopa-
thology; we do all this on the basis of some internal mechanism, a predic-
tive device that we may perfectly accurately refer to as a grammar of style,
a set of stylistic rules. (It should not be necessary to point out that these
unconscious rules, linguistic and behavioral, are descriptive and predictive,
rather than prescriptive: they tell us, in effect, what will be encountered on
the basis of what has been found, rather than telling us what we ought to
do to act right.) In talking informally about behavior, we make reference to
distinctions between intention and execution, or intention and perception,
which are analogous to the underlying/superficial structure dichotomy of
linguistics.
Linguistics as a scientific methodology has very real relevance to the
methods of the other social sciences. While it is true that they have achieved
far more success with quantification of their data than has linguistics, lend-
ing credence to the notion that that the latter is unscientifically soft-nosed,
nonetheless linguistics alone among the social sciences has attempted to
represent the data it uncovers in terms of formal rules—predictive and
simplifying generalizations. While generative grammar has not been tre-
mendously successful in writing rules that actually work, nonetheless, we
recognize that mechanisms of the sorts devised by theorists in this field
must exist and will ultimately function as we might desire. But linguistics
illustrates for other social sciences the importance of devising grammars,
predictive rather than merely taxonomic models. So when I speak of a
grammar of style, I mean that we are to transfer the concepts devised for
linguistic theory—rules, co-occurrence constraints, ungrammaticality, and
so forth—to the description of other forms of human behavior, a system
that if adequate will not only categorize what is actually extant, but will

110
also indicate what does not exist, in normal persons on the one hand, and
in general, on the other.
With this definition in mind, let us turn to the examination of stylistic
behavior. I have said that, to be a style, behavior must be able to be seen
as both coherent across modalities and consistent across time, and that
one form of stylistic pathology may lie in the failure to be either coher-
ent or consistent, or both. Further, a style assumes a match between dual
levels of structure—the one that is superficially accessible to participants
in an interaction, and the other, the level of intention, which itself may
be multileveled and accessible not at all or in varying degrees to partici-
pants in their interactions. Also, participants generally assume a one-to-
one relationship between intention and execution: if I perform Behavior
A, my interlocutor is apt to assume that I have intended to communicate
B. There is some possibility for ambiguity, even as there is in language,
but as with language, in actual practice the context, for normal individu-
als, tends to disambiguate most potential ambiguities. So, for instance, if
a female participant does something, others will interpret it one way; if a
male, another—since “female” and “male” are different contexts. As also
with language, some users are much better at realizing that two readings
might exist, and determining which is the one that is correct in that con-
text. And the fact that many people apparently are not capable of doing this
efficiently causes great difficulty in interpersonal relationships—difficulty
that is particularly troublesome because participants are seldom aware of
its cause, or even of its precise form.
While Shapiro described one contributor to style— the nature and
degree of attentiveness—style has many other potential points of differ-
entiation. We could think of it as a grid, with Shapiro’s distinctions being
one column of many. This makes sense, in that we do not typically con-
sider a person completely classified when we have labeled him or her, as,
say, hysterical or obsessive-compulsive. We also define such an individual
as demure or assertive; as reticent or loquacious; and so forth ad infini-
tum. Another point of differentiation is the mode of rapport: the relative
importance for the individual of making the content of the communication
clear versus establishing a personal relationship with others, where this
choice must be made; and, where the latter is determined to be important,
how rapport is to be effected, what mode of presentation of self the indi-
vidual chooses to adopt. There are obvious correlations between Shapiro’s
attentiveness-oriented and my rapport-oriented categorizations, so that
the grammar should allow us to predict one category of behavior from
another: e.g., someone who tends toward diffuse attentiveness is apt to
display rapport-related behavior. But on the other hand, a person whose
111
attention tends to be highly constricted and confined will impress others

as being unwilling to establish rapport.
Shapiro talks about these relationships, in fact, noting that hysterical
characters, people utilizing diffuse attentiveness, tend toward a quick if
shallow emotional rapport, the appearance of intimacy; while a contrastive
type, obsessive-compulsive characters, tend toward impersonal relation-
ships and aloof politeness. But we should like to know why this is so; what
it is in the individual that necessitates these co-occurrence relationships,
which we assume are not coincidental. I am suggesting that Shapiro’s tax-
onomies can be expressed as parts of a grammar, i.e., made into a predic-
tive system. We want to explain why we encounter these juxtapositions and
not others, as character types. And while extremes of these two types are
far from “normal,” and may even be considered psychotic at the farthest
reaches, they are found.
In another way too, character style is reminiscent in its logic of linguistic
grammar. Looking at language, we can place any string of words on a scale,
from “fully grammatical, normal, and acceptable,” like (11), through “intel-
ligible but peculiar,” like (12), to “bizarre and uninterpretable, meaning-
less,” like (13).
(11) The boy hit the ball.

(12) Won’t you please get the hell out of here.
(13) Two boys elapsed.
The determination as to whether the sentences are acceptable and mean-

ingful, or the degree to which they are so judged, rests on the co-occurrence
relations of items in the sentence. In (11), there is no problem. In (12), we
have a pragmatic conflict: please correlates with social situations in which
we are trying to be polite and gentle, the hell with contexts where we are
willing to be brusque and rude, and so it is in some sense inappropriate for
both to occur together. But the meaning of the sentence is not impaired,
only its emotional content. In (13), however, a concrete noun serves as
subject for the verb elapse, which semantically can refer only to abstract
subjects, and more specifically, expressions of time (cf. the perfectly appro-
priate and intelligible Two hours elapsed). The breaking of co-occurrence
constraints in (13) is analogous to that in (12), but of a deeper and more
serious order, affecting meaning.
Character types, too, may reasonably be looked at as linguists interpret
sentences: the more bizarre a personality seems to us, the more serious is
the violation of the stylistic co-occurrence rules that creates it. So (11) is
similar to a more or less normal person in terms of the interplay of relevant

112
characteristics; (12), to, let us say, a hysteric—we are not quite sure what
to make of the behavior, it is not what we expect, but we can learn to make
allowances; (13), to a schizophrenic, whose behavior may be totally uninter-
pretable, and frightening to us as the hysteric’s is not, for just that reason.
Further, just as with linguistic rules, we have an implicit notion of sty-
listic rules and recognize when they are violated, even though we usually
are at a loss to say what it is that we recognize. Perhaps the most strik-
ing instance of this unconscious recognition of character co-occurrence
is found in literature: in the novel, and perhaps most crucially in drama.
What distinguishes a good novelist or playwright, as well as a good actor or
director, perhaps above all else, is the creation of credible characters. And
what we mean by “credible” has largely to do with the co-occurrence of sty-
listic traits, traits drawn from the repertoire of the personality as a whole.
A writer, a director, and an actor probably perceive this intuitive coherence
in different ways, or at least make different use of their intuitions: but what
distinguishes the great from the second-rate is an almost uncanny sense of
which combinations are plausible, which not.
These observations that I have made on style can be described in terms
of a theory of communicative competence such as I have discussed in
somewhat different form elsewhere (1977). What is relevant to the pres-
ent discussion is the part of the system that describes how participants
determine the appropriate mode of presentation of self in discourse with
others, on the basis of their own personal habits, the relationship between
participants, and the subject matter under discussion. On the basis of
this intuitive judgment, the speaker selects— fluent speakers, gener-
ally unconsciously—a matching strategy, dictating the point on the scale
shown in Figure 5.1 that is deemed appropriate by the speaker’s subcul-
ture for the type of interaction in which the speaker assesses himself or
herself to be engaged. It is the Rules of Communicative Competence that
enable speakers to make judgments as to the use and interpretation of sen-
tences. And, like the latter, the implicit Rules I am describing here func-
tion as models both for production and interpretation of the contributions
of others, so that—a significant point, as we shall see later—one under-
stands the contributions of others only in terms of one’s own internalized
strategies. But it is important to bear in mind that the named points on
the continuum represent strategies, or modalities, of interaction, rather
than Rules: the rules themselves, then, can be seen as metastrategies. The
modalities themselves, and the relationships among them, are universal;
but the statement of the Rules about which point is appropriate in a par-
ticular context—as well as the decision as to how the context itself is to be
interpreted—differs from culture to culture as well as from individual to
113
individual. These differences are the basic determinants of personal style.

The strategies themselves and their interrelationships can be expressed in
schematic form as in Figure 5.1.
CLARITY DISTANCE DEFERENCE CAMARADERIE

least relationship most relationship
between participants between participants
Figure 5.1
The strategies are named in terms of the kind of relationship each

assumes as ideal. The relationship-based continuum illustrated as Figure 5.1
correlates with Shapiro’s model, enabling us, for instance, to superimpose
the characteristic linguistic preferences which this theory of communica-
tive competence presupposes upon Shapiro’s attentiveness continuum.
To speak of a continuum from least to most explicit relationship between
speakers is not to imply that necessarily, when one strategy is selected by
the participant as appropriate, they disavow entirely the existence of any
sort of relationship between them indicated by another strategy. What is
at issue is what is explicitly stated, and what is not: whether the speaker
chooses to express, or not, that someone else is involved in the discourse,
i.e., that the “discourse” as a whole is not merely about what is said, but
who is saying it, to whom, and in what linguistic and extralinguistic context.
The continuum represents, from left to right, an increasing awareness of
the addressee’s presence as explicitly manifested by the speaker. A situation
requiring Clarity will entail a rather colorless contribution: lexical items
will be selected that are emotionally drab and associated with the delib-
erative or consultative registers of speech: neither oratorical, nor poetic,
nor slang—essentially safe speech, which does not identify its speaker as
a particular and individual human being, and which does not make any
overt assumptions about the character or needs of the addressee. So the
participants themselves are not made noticeable: the message itself, the
information conveyed, is what is important, and the aim of Clarity-based
discourse is to express this information as clearly and succinctly as pos-
sible. H. P. Grice’s Conversational Maxims (Grice 1975) can be thought of as
exemplary of the expectations of ideal behavior in this framework (though
not, as we shall see, under other assumptions). So when Clarity is involved,
there is no indirectness, circumlocution, or redundancy.
Clarity is often thought of as the ideal mode of discourse, at least in
this culture at the present time. Hence there has been a great deal written
popularly of late bemoaning its apparent demise: it is claimed that peo-
ple are becoming less able to speak and write clearly, and it is generally

114
assumed that it really is a matter of inability, that people would express

themselves in accordance with Clarity if only they could think clearly, as
they used to be able to do. But as I have argued elsewhere (R. Lakoff 1976),
in fact, the reason that people do not communicate directly and forcefully
is that they do not choose to; that such a mode of expression is no lon-
ger viewed as indicative of our culture’s good human being; that, given the
choice between seeming (to oneself as well as others) fuzzy-minded and
seeming impersonal, most contemporary speakers would opt for the for-
mer. But at the same time, older standards, still in force, assert that the
person who does not speak in accordance with Clarity is not to be taken
seriously, at least in a professional milieu. So conflicts arise, which are sel-
dom completely conscious.
The crucial question is what people view as the main aim of an interac-
tion. If, as is frequently assumed, the transmission of factual or real-world
information is the ultimate purpose of any linguistic discourse, we would
have to make some rather surprising judgments. Since Clarity is the only
mode of communicative competence that is specifically concerned with
the transmission of information, and since only what is transmitted in
conformity with the statements of Clarity is maximally informative, we
would, given such an assumption, have to view any failure to use Clarity
as a bizarre, unintelligible, or pointless contribution. And we would find
that a very large percentage of conversations fell into this category. What
we discover in fact is that only a relatively few contexts are appropriate
for Clarity-based contributions, and that anyone who makes use of Clarity
under other conditions, or at any rate does so persistently, is considered
not a particularly admirable member of the culture, still less an extremely
effective communicator—but rather someone distinctly aberrant, or at
worst psychotic. We obviously have to rethink the notion that being clear
is always best, or even always good.
Distance, the next discrete point to the right, appears to invite even
more impersonality than does Clarity, but only superficially. When Clarity
enjoins us to “Be clear,” Distance says “Remain aloof.” But to talk of aloof-
ness and distancing is to assume that the relationship between the par-
ticipants is of interest, even if that relationship consists of staying as far
apart as possible. So the use of this strategy ensures that participants will
tread on each other’s toes as little as possible—and therefore assumes
each has toes to be trodden on. Here we find formal politeness, the rules
of etiquette, diplomatic language, bureaucratese, and professional jargons
of all kinds—all systems designed to maximize the distance between the
participants—and at the same time, to impute authority to the speaker,
whether through status or expertise. Hence what is being communicated is
115
particularly hard for anyone else to question. While Clarity is unemotional,

Distance is anti-emotional: the form of expression is chosen partly on the
grounds that there is something covertly present in the intended commu-
nication that may be unpleasantly emotion-provoking and therefore needs
to be neutralized by the speaker. On the other hand, Clarity assumes that
participants are dealing with cold facts, and there is nothing to conceal.
So as Clarity is the language of news broadcasts and classroom lectures,
Distance is the mode of communication of the more defensive—the poli-
tician, the bureaucrat, the academic delivering a paper. Both Clarity and
Distance lend an air of uninvolvement to a communication.
Distance uses technical terminology, formal polite language, and overly
correct diction and grammar—the opposite of the colloquial. All this
makes it hard to understand, precisely because we do some of our intellec-
tual understanding through our emotions, and the denial of emotionality
makes it very hard for us to figure out what really is being said, and how
we ought to respond to it. Because of the safety in Distance, people are
more apt to lapse into it the more they are afraid of the repercussions of
what they are saying, whether because they fear what they say is unpopu-
lar, or because it may be incorrect. Distance is, furthermore, a means of
avoiding responsibility for what one is saying, while claiming the author-
ity and power necessary to say it. So bureaucratese and its relatives fre-
quently make those who are exposed to them nervous, much though their
users seek to placate—or rather, narcotize. And the use of Distance, since it
implies that the speaker holds power over the person addressed, is abrasive
as the other modalities are not.
Deference too establishes a relationship between participants. But this
modality explicitly recognizes the existence of both participants and their
relationship, where Clarity ignores these issues and Distance denies them.
Using Deference, the speaker is following the injunction: “Don’t impose—
give options.” Thus, its use implies that decisions as to the interpretation and
outcome of the exchange are in the hands of the addressee, whether actually
or merely conventionally. Even if the deference is mere convention, however,
the very fact that the speaker chooses to adopt that convention is a mark of
courtesy, as most people prefer conventional deference to outright and real
brusqueness. Deference apparently leaves the important decisions up to the
addressee—although the speaker really knows perfectly well what he intends
to achieve by his contribution, and gives plenty of instructions overtly or
covertly to the addressee to indicate the desired response. In fact, sometimes
the use of Deference creates friction between participants precisely because
the speaker seems to be giving with one hand and taking with another, offer-
ing autonomy but in fact retaining power as instanced by the very ability to

116
offer autonomy. Then the addressee is often faced with the impossible task of
determining which of the levels of communication to recognize—the overt
offer, or the covert injunction? Either way is risky. Where Distance avoids
uncomfortable topics by the pretense of intellectual uninvolvement, that
is, by denying the emotive force of the contribution, Deference denies the
cognitive content, conventionally of course (or else understanding could not
take place), saying in effect, “It’s up to you to translate my message: if you
decide we’re talking about THAT, we are.” Hence euphemism is characteristic
of this modality.
Camaraderie, necessitating as it does direct confrontation, is the
modality least in accord with what we usually think of as “politeness.”
For Camaraderie explicitly acknowledges that a relationship exists and is
important, whether one of friendliness or of hostility. Camaraderie is the
level of direct expression of orders and desires, colloquialism and slang,
first names and nicknames—much that is considered good and typical con-
temporary American behavior.
The strategies have been described as points on a continuum partly
because they can be used in isolation or in combination: a contribution can
be pure Distance, or mostly Distance, with a little Deference, or half of each.
Some people are very formal, distant, and aloof; others less so; still others,
hardly at all. Figure 5.1, however, makes some combinational possibilities
hard to visualize: a situation such as that exemplified in Figure 5.2 is no
problem—a style falling between Distance and Deference, which are contig-
uous; but this diagram would make it impossible to represent, say, a combi-
nation of Distance and Camaraderie, which is just as theoretically possible.
One way of avoiding these difficulties is shown in Figure 5.3, with
the left–right axis maintained intact, but some means of indicating that
combinations between any of the modalities are possible. We still can-
not graphically represent combinations of three and four modalities, but
Figure 5.3 is at least a step toward more accurate representation. A specific
linguistic entity—whether a lexical item, an intonation pattern, or a syn-
tactic formula—is typically categorizable as assigned to one of the named
points on our diagram, while the overall stylistic behavior on the part of
individuals can fall at nodal points or between them. So Figure 5.1 can still
be thought of as an appropriate model for the subcomponents of which
style is composed, but Figure 5.3 seems a better representation of a style
as a whole.
CLARITY DISTANCE DEFERENCE CAMARADERIE
Figure 5.2
117
DISTANCE
CLARITY CAMARADERIE
DEFERENCE
Figure 5.3
As an example of the operation of this system, let us examine the

differences between women’s and men’s ideal styles in this culture. It is
important to remember at the outset that, when two types of behavior
are put in contrast by a society, they will tend to be polarized—to be
perceived and stereotyped as more different than they probably are. But
these exaggerated stereotypes must, to be intelligible, be based on real
differences.
Let us think for convenience of people who typify our masculine and
feminine idealizations—say, Clark Gable or Marilyn Monroe, who perhaps
seem particularly exemplary to us now since we have had time to abstract
their images from reality. If we were to place these people—in their offi-
cial Hollywood personalities—on our chart, I think we would place Gable
somewhere on the Clarity–Distance axis, and Monroe along Deference–
Camaraderie—about as far away from each other as they could be. Gable
is perceived as the strong, silent type—he says just as much as necessary
to get information across, and is somehow withdrawn into himself, almost
reclusive. Think of Rhett Butler saying, “Frankly, my dear, I don’t give a
damn,” as the epitome of the ideal masculine style that most of us have
come to maturity assuming. When such a man speaks, his contribution is
incisive, precise, and to the point—utterly straightforward—and tells us as
little as possible about the speaker’s state of mind and his attitude toward
the addressee. We expect here, too, an even and low pitch, flat intonation,
declarative rather than interrogative sentence structure, no hedging or
imprecision, and lexical items chosen for their pure cognitive content, not
their emotional coloration. If such a person shows emotion, it will be anger
rather than tenderness or grief. Anger is expressed, ideally, not by rais-
ing the loudness or pitch of the voice, but by becoming ever more precise,
soft-spoken, withdrawn—an exaggeration of the silent characteristics. As
a contemporary example of the type, we can think of Walter Cronkite, at
least in his public person. Not for nothing is he the “most trusted man in
America.” He is trusted because his style is the very epitome of the current
ideal. Cronkite’s style is probably perceived as somewhat closer to Clarity
than Distance on the scale, where Gable’s was the reverse, but they are oth-
erwise similar.
By contrast, Monroe’s public presentation of self was profoundly impre-
cise. There is always a sense that the audience does not really know what

118
she is talking about (nor does she), but that she is very concerned with
whomever she is talking to, concerned with whether he is interested in
her, and whether his needs are being met. Here we have the reverse of
Gable/Cronkite style: Clarity is entirely absent, and there is no evidence
of Distancing behavior. In contrast to Gable’s characteristic poker face, we
have Monroe either smiling or looking sensuous, but certainly wearing an
identifiable facial expression. She uses interjections and hedges freely and
her dialogue is sprinkled with I guess and kinda in distinction to Gable’s
unembellished yups. Her sentences seem not to end, but rather to be ellipti-
cal, as if in invitation to the addressee to finish them for her—classic femi-
nine deference.
These diverse styles are also classifiable on Shapiro’s scale—Gable/
Cronkite’s as representative of obsessive-compulsive style, Monroe’s of
what Shapiro terms hysterical. More accurately, we would want to place
truly neurotic examples of stylistic behavior farther along our lines than
any of these more or less normal examples, but certainly it is arguable that
this culture’s paradigmatic masculine style shares a great many traits with
a “neurotic” obsessive-compulsive style; and our prototypical example of
femininity is exemplary in many of its aspects of neurotic hysterical style.
Shapiro shows too how the communicative behaviors I have just discussed
correlate with cognitive, perceptual, and motoric forms of functioning to
produce the coherency we call personal style.
Now we are faced with some interesting observations. I have remarked
that those linguistic traits that are characteristic of men in our society cor-
relate with a form of functioning that verges on the obsessive-compulsive;
and, feminine traits correlate with hysterical behavior. It is a truism among
psychological theorists that hysterical traits of character are predomi-
nantly found in women, so that it has sometimes been thought they were
exclusively the province of women. And similarly, though less strikingly, a
correlation has been claimed between obsessive-compulsive style and mas-
culine behavior.
Students of style in general, and linguistic behavior more specifically,
have tended to talk in terms of women’s behavior, women’s style—as
opposed to a more general notion, people’s style. Women have always been
classified as the other, the not-quite-human, whether by medieval theo-
logians who claimed women had no souls, or by more modern psycholo-
gists (as demonstrated in the Broverman et al. 1970 study) who would
claim that a “healthy woman” and a “healthy man” are characterologi-
cally very different things: a “healthy man” is identified with a “healthy
human being” or “healthy person,” while a “healthy woman” is not. This
extremely dangerous prejudice infects much of our scholarly behavior
119
implicitly or explicitly. There is nothing wrong with talking about wom-

en’s style or women’s language as long as it is not covertly opposed to
“everyone’s style” or “normal language.” The chart I have sketched indi-
cates quite clearly that both the prototypically masculine and prototypi-
cally feminine variants of stylistic behavior are representable as points at
one or another end of the continuum; each is equally far from the ideal
middle, or norm, which presumably is rather rare in actuality. We have
several prejudices to overcome here: that women’s special manifestations
are further from the norm than men’s, which in this system of thought
is the norm; and that anything nonnormal and hence nonmasculine is
worse, weaker, or degenerate. Both of these covert (sometimes embar-
rassingly overt) assumptions color our perceptions of the stylistic dif-
ferences in the behavior of the sexes. If indeed we are brainwashed into
believing that anything on the Deference–Camaraderie end is no good,
we will object strenuously to the very idea that there is a women’s style,
distinct from men’s. But if we can see that there is no difference in degree
or validity between behavior at the two ends of the continuum, perhaps
we can study both as interesting entities. In the same way, our society
rewards behavior on the “masculine” or obsessive-compulsive side as
being in keeping with the work ethic, logical thought, and so on and dis-
avows the other end as irrational, lazy, or nonserious. It is common to
counter these attacks by claiming the difference does not exist; but there
is another way, to say that it does but that neither side constitutes the
more virtuous human being. Further, both sides could gain from moving
a bit closer to the center at least some of the time.
It seems rather natural, indeed inescapable, that neurotic characteris-
tics should pattern as they do: for obsessive-compulsive style is merely an
exaggeration of our culture’s preferred masculine mode of presentation of
self; and similarly, hysterical style is an exaggeration of the feminine.4 We
might say that neurotic behavior is not just randomly aberrant behavior;
but rather, one tends to become neurotic by exaggerating and calcifying
the mode of response to the environment one would be using normally.
4. It is true that obsessive-compulsive style is more prevalent among women than
hysterical style is, or has ever been, among men. As far as I know, no reason for this
is advanced in the literature dealing with neurosis. One possible line of explanation is
suggested here: if neurosis is another term for the inflexible adherence to a cultural tar-
get, it makes more sense to cling to a target that is given some respectability than one
which is ridiculed. There is a sort of hierarchy, perhaps, in the selection of a neurotic
style: first, choose the one that is the exaggeration of one’s own group’s preferred strat-
egy; failing that, for whatever reason, choose what is culturally valued. So a woman
would most naturally select hysteria, second, obsessional neurosis. A man presumably
has only the latter option.

120
As Shapiro and others define neurosis as inflexibility in adjusting behav-

ior to the environment, this reasoning seems to make sense. Then it fur-
ther follows that more women will be classified as neurotic than will men,
since neurosis is defined as departure from a preferred norm; and the norm
is defined in terms of masculine behavior. So for a man to be diagnosed
as obsessive-compulsive, he must exaggerate those characteristics even
beyond how a man (i.e., a healthy human being) in this culture is supposed
to display them: but to be a hysterical woman, one need only be a “good”
woman—the more hysterical, the better. But the better a hysteric, the
worse a human being by the androcentric definition of that term. By this
reasoning, a woman finds herself stylistically in a double bind. She cannot
be rewarded at once for being feminine and being human as a man can for
being both masculine and human. It is a problem, and really not her prob-
lem, but her culture’s (R. Lakoff 1975).
We can say, then, that a culture has implicitly in its collective mind a
concept of how a good human being should behave: a target for its mem-
bers to aim at and judge themselves and others by. Members of the culture
over the age of adolescence (perhaps earlier; in the past, certainly earlier)—
at least male members—are expected to behave accordingly. Outsiders,
marked typically by their accented speech, a badge of outsiderhood, are
excused from some of the rules, although the behavior of members of other
societies is nonetheless judged by the criteria of the culture doing the judg-
ing: thus we may say, for instance, that the Japanese are “deferential,” i.e.,
employ the rule of Deference more than a male in our culture would, and
the Germans are “arrogant,” meaning more or less that they use Distance
more than we would. We get impressions of foreigners as if they were
members of our own culture. So rather than saying to ourselves (uncon-
sciously, of course), “Although he presents the appearance of arrogance,
I don’t consider him arrogant because that is normal behavior within the
context of being a German, which he is,” we say, “He’s arrogant, but that’s
what I expect from someone who speaks with a German accent; they don’t
know any better, so I’ll overlook it.” While superficially this may seem to be
a hairsplitting distinction, it has considerable import, especially as regards
feelings about feminine behavior.
For just as Germans are aliens in the American culture, because they
have a different target for relating to others, so are women in this male-
centered or androcentric culture which identifies “healthy human being-
hood” with the masculine stereotype. Woman are the other, just as surely
as are Germans or Japanese, and are similarly stereotyped in joke and folk-
lore: women are fuzzy-minded, women are not precise, women are overly
polite and careful in their speech and actions. We have no comparable
121
stereotypes for men, even as we have no jokes based on stereotypical male

behavior (analogous to “woman driver” or “mother-in-law” jokes) nor
expectations based on necessarily male roles. (Compare the prevalence of
assumptions about widows, or for that matter divorcées, in our culture
past and present, with that of widowers and divorced men. It is significant
that the spoken form divorcé(e) will almost certainly, unless clearly disam-
biguated by context, be interpreted as feminine.)
I have already mentioned the paradox that a woman cannot at once
be a healthy woman and a healthy human being: the stylistic ideals are
directly contradictory. There is an additional problem as well, deriv-
ing from another stereotypical view of women prevalent through the
ages: women are manipulative, deceitful, do not say what they mean;
people cannot figure them out, they do not play by the rules. Whether or
not this stereotyping is fair, whether it represents reality, is not relevant
here. It is important, however, to acknowledge the prevalence of this ste-
reotype, in our culture and throughout the world. To change it, we must
understand its prevalence, and give a reason why people (including many
women) have clung to it so tenaciously. It is convenient because it rep-
resents still another way in which women are categorized as “the other,”
“the nonhuman.” For we do not have an analogous deleterious stereo-
type of men, although parallels could no doubt be constructed. But the
crucial question is whether there lurks a reality behind the stereotype.
For stereotypes are important to study only insofar as they are repre-
sentations of reality. And I would argue further that societies only create
and maintain those stereotypes that their members feel hold a mirror up
to reality—though it be a fun-house mirror. Then to tell members of a
group that their stereotypical image of an outsider is unrealistic is to say
something pointless or meaningless; even if they profess beliefs to please
you, they will not really abandon the stereotype that so well suits their
thinking. So our task is to go behind the female stereotype to see why it
has the form it does.
I have already discussed the distinction between real and conventional
use of strategies. It is possible to be truly deferential, to really have as one’s
main concern what the other person wants, and to really have one’s own
choice of what to think or do depend on the other’s—and equally possible
to be conventionally so, as when someone knows perfectly well whether,
for instance, he (or she) wishes to go to the opera (or bowling) and says
meekly, “It’s up to you, dear.” The same duality of interpretation exists in
the enforcement of the other strategies as well: the politician’s apparent
Clarity, the diplomat’s Distance, the salesman’s Camaraderie—all are con-
ventional. And indeed, conventional application of all these rules is found

122
to a less pronounced degree in all our behavior. We need not sneer or pro-
test our total sincerity: were it not for conventional politeness and our will-
ingness to use it and interpret it kindly, civilization would likely dissolve in
no time at all.5
But conventional application of a strategy occurs only in case that
strategy is seen as the target for an individual in a particular cultural con-
text. For the Japanese, male and female, conventional Deference is com-
mon, where for the German it is not. For the traditional American male,
Distance/Clarity has been the target, and so we find in this group’s behav-
ior much conventional Distance and Clarity. But Deference is not targeted,
and is not conventionally expected, of a male in this culture. And even when
overt Deference occurs, we still do not find the extralinguistic indicators
of Deference in the behavior of a normal American man—e.g., downcast
eyes, giggle, an attitude of general helplessness—although we do find such
behavior among Japanese males, as we might expect. We might interpret
this discrepancy as showing that the extralinguistic devices are the most
apt to be used, and interpreted, as conventional. So these are good tests of
the targeted politeness for a group—what indirect, and therefore perhaps
not entirely conscious or intentional, indicators of a strategy are employed.
Now if a strategy is the target for a particular group, a member of that
group will know that behavior in accordance with that strategy may be
interpretable either as real or as conventional. But for nontarget strategies,
only real behavior can be inferred. And although one might think that peo-
ple might learn that other groups make different strategies their targets
and so apply them conventionally, people generally seem to be unable to
transcend their own systems: participants can interpret others’ strategies
conventionally only in case they themselves could use them conventionally
(because their group makes them targets). People who retain perspective
and flexibility in interpretation better than others are felt to have special
understanding of groups and cultures other than their own, but they are
admired as the exception rather than the rule.
American women’s traditional target is Deference, as men’s is Distance/
Clarity. So we find certain kinds of conventional behavior based on these
strategies in women’s behavior, which men do not normally adopt. This is
the reason why men are in general unable to interpret feminine behavior as
conventional, rather than real, and hence regularly misinterpret women’s
intentions. And then, when women respond to their misapprehensions by
5. The same point is made about the social utility of disclaimers of action (e.g., the
same sorts of hedges and indirections discussed here) by Roy Schafer (1976).
123
acting, apparently, unpredictably, this gives rise to the stereotype of women

as fickle, deceitful and untrustworthy, and illogical—all valid perceptions
from the standpoint of male-centered strategies. Since, however, these are
tacitly or explicitly recognized as the strategies, women themselves have
often, shamefacedly or gleefully, gone along with the implied definition of
womanhood as perversity. This view is neatly summarized by the male cry
of consternation: “What do women want?”
The reader might object that what is said here about men’s inability to
accurately interpret women’s style must surely have a parallel in women’s
inability to interpret men’s style. To a limited extent, this parallelism
indeed exists: it is unlikely that women understand men much better than
men understand women. But there the similarity ends. For women do not
make the assumption that their ways are the healthy and good ones, or
the only ones. So women do not, on the basis of their misunderstanding,
construct stereotypes of men as irrational, untrustworthy, or silly; they
merely assume that men’s behavior is beyond their poor comprehension.
Thus, women’s inability to understand men does not cause them to dero-
gate men, as men’s inability causes men and women both to form unfavor-
able stereotypes of women. Although in reality the inability of men and
women to interpret each other’s style is totally parallel, because men’s style
is defined as the norm, only women’s is interpreted as aberrant and only
women’s is subject to stereotyping and being considered as a curiosity wor-
thy of study.
If the American male’s conventionalized style is best distinguished as
Distancing, how are we to describe styles that seem to belong even further
to this strategy? For example, to the American, the typical German—male
or female—seems stereotypically aloof and unemotional, not to say arro-
gant and brusque. Is this not quintessential Distance? Yet to the American,
the German stereotype seems light-years away from his.
This is, I think, but a seeming contradiction. It does illustrate the con-
ceptual difficulty inherent in thinking of style as a two-dimensional con-
tinuum. For there are various aspects of behavior intertwining here, which
we must sort out if we are to resolve the difficulty.
The styles of two different cultures may differ from each other for any
of several reasons within this framework. The cases already discussed are
those in which the targets themselves differ: women prefer Deference, men
Distance. But styles can be differentiated in other ways. Within a group,
participants adhere to the idealized target up till a certain point in a rela-
tionship. Then, more or less implicitly, they may drop the conventionaliza-
tion and behave toward one another as they “really” feel. Some cultures
have explicit institutionalized means of effecting this switch: Germans, for

124
instance, have a ceremony called Bruderschaft, in which the participant of

higher status invites a more lowly acquaintance of long standing to use
du rather than Sie as the term of address. With this use of the familiar
pronoun go other marks of familiarity—first-naming among them. In
American society, first-naming is normal for most relationships within
a very short period of acquaintance, and between persons of very differ-
ent statuses. The implication is that in American society, Distance is the
normal and conventionalized strategy only under some circumstances—
very early indeed in a relationship, and an impersonal relationship at that.
Camaraderie takes over quite early in a friendship for most people. So it
might be that American Distance is just as distant as its Germanic equiva-
lent, but shows up in fewer kinds of relationships because Americans drop
the convention earlier.
Another interpretation of the difference might be that, in fact, the mix-
ture is different, or that two or more strategies are conventionalized in one
culture, corresponding to one in the other. The range of situations covered
in Germanic culture by Distance alone is partially covered, for the American
man, by Distance, partially by conventionalized Camaraderie. The super-
ficial result of either of these two hypotheses would be the same: to the
casual observer, in many circumstances the American would appear to be
open and sociable when his German equivalent would appear aloof. But the
first theory says this is true because the American becomes truly friendlier
faster, the second because he adopts conventional friendly behavior where
the German does not have this option.
Finally, differences can arise because what one culture classifies as appro-
priate Distance behavior, another may consider suitable for Camaraderie,
or, more likely, Deference or Clarity. Refusing a second helping of food may,
for one group, be a form of conventional Deference: it means that the other
person now can take charge, do the urging; the eater will not express his
or her own desires. In this case, the expectation will be that, upon urging,
the second helping will be accepted, perhaps with a show of reluctance. But
refusing seconds may also be seen as Distancing, real or conventional—a
refusal to show pleasure, to react to the feeder’s attempts at interaction
through food. To refuse a second helping is tantamount to saying, “I don’t
want anything from you. Don’t bother me.” Of course, if the feeder is fol-
lowing the first pattern and the eater the second, or vice versa, bewilder-
ment and outright anger are apt to ensue.
The contrast between the German and the American male seems best
described by the second of these three possibilities. Americans appear to
use a split strategy where Germans use a unified one. This difference may
occur because there is motion in the American culture where there is none,
125
or a much less perceptible one, in the German. There would seem to be

some interesting evidence for this position.
In the last generation or so, the target for American men has been
subtly shifting from Distance/Clarity, still prevalent in older people and
people raised in traditional milieux, to Camaraderie, which is the preferred
mode of presentation of self in younger people. It is probably true that
this change originated in the West and is spreading East, with California
as usual at the forefront: normal male behavior in California seems very
informal and at least superficially intimate, compared with Eastern mores;
there is a quicker transition to first names and nicknames, people seem
more ready to discuss intimate facts of their lives, and there is a good deal
more touching and feeling not only countenanced among men, but indeed
expected. This is fast becoming the conventionalized target everywhere.6
Just as Distance-related behavior was conventional when it was the target,
so Camaraderie is conventionalized. The new intimacies are really no more
intimate than the old uninvolvements were truly uninvolved.
If this is indeed the case, we have another argument for assuming that
stylistic behavior is rule-governed and analogous to a linguistic grammar.
Just like the latter, its rules change; or, more accurately, what happens in
both cases is that the context, the environment, in which a rule is appli-
cable changes with time. Camaraderie used to be applicable only in relation-
ships of comfort and long standing, in which case, not being a target, it was
real: when someone called you by your first name, you knew you were at an
intimate stage in your relationship. While this is still marginally normal
behavior in the East (or so I am told), the use of title–last name is obsoles-
cent in the Bay Area in almost any relationship. It is becoming common,
for instance, for doctors to insist on mutual first-name address with their
patients, and the use of mutual first-name is, I am told, almost de rigueur
among psychotherapists where I live, although I am similarly told that it
is much less obligatory in the East. Obviously, this kind of camaraderie is
pure convention since there is no mutuality or intimacy elsewhere in the
relationship.7
Just as the rules of the linguistic grammar change over time, back and
forth, but are perpetually in flux, so there is some interesting evidence that
6. Cf. the essay in Time magazine by Lance Morrow (1977).

7. Actually, the decision as to whether patient and therapist are to reciprocally use
first names or title–last names may have more to do with the type of therapy than its
geographical locale, with psychoanalytically-oriented types tending toward the latter
and human-potential or radical-psychiatry practice toward the former. However, the
point remains valid, as the psychotherapeutic method of choice seems closely related
to geographical region as indicated, undoubtedly by no coincidence.

126
these Rapport rules have changed over time: this is not the first time that
we have switched our targets. This is an important point, since one might
otherwise think that the shift from Clarity/Distance to Camaraderie was
an unprecedented step in human behavior, induced by the media, greater
mobility, and so forth. It is undoubtedly true that it is harder to live com-
fortably in an era where the rules are in flux, as is the case in this society
at present, than in a relatively stable order—say, Victorian Europe. But
our present situation is by no means unique; indeed, just like the linguis-
tic grammar, stylistic grammar must shift and always is partially in flux,
although the changes are seldom evident to a contemporary member of
the culture.
There is an interesting illustration of an earlier period of rule change
in Philippe Ariès’s book (1962). Although Ariès does not address the
issue directly, much of what he talks about can be interpreted in terms of
Rapport rules and change in target. During the thirteenth century, vari-
ous stylistic modalities were in flux in Europe. People were beginning to
develop a notion of privacy: they began building houses with separate
rooms, wearing more constraining and concealing clothing. About this
time too, I believe, last names were being devised and used. All of these
work toward a Distancing form of rapport. So we can infer from this sort of
evidence that European society during this era was shifting from a target
of Camaraderie to one of Distance. We would expect in such a period of flux
to find more discomfort than usual among people as they realized their old,
internalized rule systems were inappropriate but did not know yet what
was expected of them in the new. And so it is not surprising to find in Ariès
a discussion of the vogue of etiquette books, in the form of so-called “cour-
tesy manuals,” in this historical period. Courtesy manuals played a crucial
part in the medieval educational system, and these etiquette books, judg-
ing from Ariès’s excerpts, are designed to teach Distancing behavior: they
contain rules like “Don’t touch yourself in public, nor other people”; “Don’t
be too familiar in address”; “Dress modestly”; “Don’t tell your dreams to
people.” The aim is to appear as unobtrusive and unintrusive as possible—
clearly, the establishment of conventions of Distance.
It is true that similar Distancing etiquette manuals have remained
around up till the present, but seldom have they been accorded the impor-
tance that they were during this period. There will always be some residual
insecurity about the correct formulas for achieving the idealized and con-
ventional mode of rapport, but never so much as when the targets them-
selves are shifting.
Then if it is true that we are now in just such a state of flux, we should
expect to find a similar desperate emphasis on etiquette manuals. If we
127
think only of Emily Post-type etiquette, we shall be disappointed: no one

seems terribly concerned with the right oyster fork any more. But indeed
we do have a veritable outpouring of Camaraderie advice in books, maga-
zine and newspaper columns, TV and radio talk shows, and so on: the con-
temporary analog of the medieval etiquette manual is the psychological
self-help book. All of these “manuals” have to do with becoming a good
human being in a Camaraderie strategy. No longer are we concerned with
being unobtrusive and uninvolved. Rather, our present-day etiquette rules
offer injunctions about how to be assertive but not aggressive, how to show
you care, how to show you like others and how to be liked—all Camaraderie
behavior. So our present situation is parallel to the one in effect several
hundred years ago, only in reverse. At that time the target was going from
Camaraderie to Distance; now it is going in the opposite direction. But our
behavior in the midst of such a change is apparently not significantly dif-
ferent from that of our forebears: we feel insecure and look to authorities
to instruct us in the new propriety.
An enlightening sidelight, relevant to the larger issue of this paper, has
to do with who writes the etiquette manuals, who are the social arbiters. In
the earlier period of flux, the writers of these books as well as their users
were men: etiquette was considered crucial enough for men to bother with
and take responsibility for. But as its importance waned over the centuries,
women became the dominant figures in the world of etiquette: just as with
other concerns that are deemed by a society to be peripheral to survival, like
fashion and grammar, manners gradually came to be a feminine province
so that, most recently, virtually all the arbiters of correct deportment have
been women. Interestingly, the trend is now reversed. As the new etiquette
is felt to be important and necessary to one’s survival, it has become the
province of men at least as much as of women: men are no longer ashamed
to be reading and writing instructions for proper human behavior. Here
again we have evidence that we are in a period of flux, because the learn-
ing of the new system is considered important enough to be the business
of men.
Earlier in this paper I correlated the use of Clarity/ Distance with
obsessive-compulsive symptomatology, noting (as has been frequently
pointed out) that this form of neurotic behavior has been common among
men in our society, just as hysterical character, correlated with Deference
and Camaraderie, has been prevalent among women. More recently, within
the last generation, it has been remarked that these “classical” neuroses—
the symptoms on the basis of which psychoanalysis was devised, and
for which it is considered particularly effective—have become compara-
tively uncommon. First in their symptomatic and more recently in their

128
characterological forms, they have lost ground as presenting symptoms.

Psychotherapists are seeing more of so-called “narcissistic characters.” (In
this connection, there is a most interesting article by Christopher Lasch
(1976).) Just as the obsessional neurotic’s form of rapport is an exagger-
ated Distance and Clarity, and the hysteric’s, Deference (which manifests
itself in seductiveness and the fuzziness of attention discussed by Shapiro),
the narcissistic person’s is conventional Camaraderie. Since narcissism,
as discussed by psychoanalytically oriented writers, involves an inabil-
ity to empathize with or form deep relationships with others, we might
be inclined to consider it the very opposite of Camaraderie. But in fact
superficial friendliness and interest in others—as long as they serve one’s
purposes—are associated with the narcissistic character, who is desperately
concerned with his reception by others and therefore must feign interest in
others. Hence, too, the linguistic correlates: nicknames, four-letter words,
slang, at least superficial directness. And just as obsessional Clarity is a
false clarity, meant to mask perception rather than enhance it, so narcis-
sistic Camaraderie is merely superficial show, and is thus particularly apt
to function as conventional behavior. If indeed neurotic symptoms are
merely exaggerations of an individual’s—or a society’s—ideal mode of pre-
sentation of self, and it is true that contemporary Americans perceive these
traits as desirable, in contradistinction to those of Distance and Deference,
this fact merely provides evidence for our targeting of Camaraderie. For
if all this Camaraderie has become the new target of rapport, then it all
fits together perfectly; narcissism is the obvious choice of neurosis for a
Camaraderie-based culture.
If we accept this idea that society is becoming Camaraderie-oriented,
there is still one question to be answered: we have been evading the issue
of whose system is changing. It is clear that the male stereotype of Clarity/
Distance is yielding to Camaraderie; but what is happening with women?
I said earlier that one thing that makes it particularly hard to be female in
this culture at this time is the fact that one cannot be a good woman (using
Deference/Camaraderie) and a good human being (defined androcentri-
cally, in terms of Clarity/Distance) at once. What then is happening to
women’s style here at this time, and if there is a change, what does it augur?
There are several possibilities. First, women’s style may be coming closer
to pure Camaraderie as men’s becomes so too. In this case our culture
would end up looking in a curious way analogous to that of contemporary
Japan. For the Japanese, Deference is the target for both sexes—but a dif-
ferent sort of Deference, and under more conditions for women. Although
the requirement for exaggerated Deference still makes women subservient
to men, nonetheless it may allow the Japanese woman to experience fewer
129
confusions in defining her ideal role: it may well be possible in Japanese

culture, as it is not in ours, to be an ideal woman and an ideal human being
at once, without paradox. As another alternative, women’s style may stay
the same, with men’s moving around it; and a third, while men’s style shifts
in one direction, women’s might shift in another—to Distance, or Clarity,
for instance. In the third case, we might expect to find in time that Clarity
was no longer viewed as the embodiment of all that is rational and human;
and women, having tried to gain acceptance as human beings by adopt-
ing the “healthy human-being” behavior, would find themselves once again
on the losing end of the stick, since ideal human behavior would still be
defined in terms of men’s behavior. If the second option is the one that is
occurring, we might expect no significant change in women’s image and
perception of themselves. If in fact both sexes are changing to be more
alike, then we may have a chance to see real changes in the way women are
regarded and regard themselves. Ideally, the members of a society should
allow for pluralism and flexibility in style, should be able to realize that var-
ious modes of presentation of self may be appropriate, and none by itself
defines a human being. But, realistically, this is not apt to happen, and we
can only hope for change in women’s image as women become more like
men, in this case both changing to a new mode of behavior.
What evidence exists in media representations suggests that the first
option is winning out, and I think it is the healthiest possible choice. Just
as women are becoming freer in the extent and types of self-expression
open to them, so—perhaps more slowly—are men. Showing care and con-
cern, crying, and expressing a wide range of emotion have not been char-
acteristic of men’s style, as they are the antithesis of Clarity/Distance. By
the same token, wearing severe man-tailored suits and having the latest
stock-market quotations at one’s fingertips have not been permissible for
women as users of Deference/Camaraderie. But both are becoming com-
mon, despite the denunciations and expressions of apocalypse that issue
from the conservative press. We shall have to be patient and await the final
outcome of the stylistic reorganization that we see occurring around us,
just as our forebears had to wait many centuries to see the final result of
the change initiated during the medieval period.
To say that style is changing for all of us implies a grammar of style that
we unconsciously utilize; to say that men’s and women’s styles are becom-
ing more similar is to make a less theoretical point, but one of great sig-
nificance for our future as men, women, and people. In any event, what we
observe about the times in which we are living can tell us a great deal both
about human psychological capacities and about what we are striving to
become.

130
REFERENCES
Ariès, Philippe. 1962. Centuries of childhood: A social history of family life. Trans. by

Robert Baldick. New York: Knopf.
Bever, Thomas G., and D. Terence Langendoen. 1972. The interaction of speech
perception and grammatical structure in the evolution of language. Linguistic
change and generative theory, ed. by Robert P. Stockwell and Ronald K.S.
Macaulay, 32–95. Bloomington: Indiana University Press.
Broverman, Inge K., Donald M. Broverman, Frank E. Clarkson, Paul S. Rosenkrantz,
and Susan R. Vogel. 1970. Sex-role stereotypes and clinical judgments of
mental health. Journal of Consulting and Clinical Psychology 34(1):1–7.
Chomsky, Noam. 1968. Language and mind. New York: Harcourt, Brace and World.
Grice, H. Paul. 1975. Logic and conversation. Syntax and semantics 3: Speech acts, ed.
by Peter Cole and Jerry L. Morgan, 41–58. New York: Academic Press.
Lakoff, Robin. 1976. Why you can’t say what you mean. Review of Strictly speaking, by
Edwin Newman. Centrum 4(2):151–70.
Lakoff, Robin. 1977. Women’s language. Language and Style 10(4):222–48.
Lasch, Christopher. 1976. The narcissist society. New York Review of Books, 30
September 1976, 5.
Morrow, Lance. 1977. A nation without last names. Time, 11 July 1977, 43.
Schafer, Roy. 1976. A new language for psychoanalysis. New Haven: Yale
University Press.
Shapiro, David. 1965. Neurotic styles. New York: Basic Books.
131
Introduction to “When talk

is not cheap: Psychotherapy
as conversation”
BY JOAN SWANN
R obin Lakoff’s paper on psychotherapy as conversation appeared origi-

nally in an edited collection on The State of the Language, published
in 1980. The collection is highly diverse, as are the contributors: there is a
clutch of academics—several professors of English, for instance—alongside
poets and novelists, historians, a lawyer, a politician, and a business execu-
tive amongst others. The collection as a whole brings to bear a range of per-
spectives, traditions, and ideas on language. The individual pieces are often
insightful and/or provocative, allowing readers to dip into different topics
and reflect on whatever arouses their interest. This is the value of Lakoff’s
own piece, I think. Lakoff presents several ideas on psychotherapy, which
can be pursued on their own terms and which also suggest new avenues of
exploration.
A number of researchers have published empirical analyses of psy-
chotherapeutic interviews, most notably Labov and Fanshel’s classic
Therapeutic Discourse (1977). Lakoff’s approach is more intuitive: it draws
on her awareness of and intuitions about what takes place in psychothera-
peutic interviews, and how this can be related to “ordinary conversation.”
Lakoff claims that psychotherapy has borrowed the model of ordinary
conversation, but that it also differs from it in some respects: it ignores
many of the assumptions of ordinary conversation, for instance, and the
132
relations between participants are symbiotic, or nonreciprocal, rather than

reciprocal. Adults who take part in therapy, therefore, are learning how to
engage in a new form of conversation, in which new conversational strate-
gies are required. The learning of new strategies can itself be beneficial,
leading to the recognition that there are different ways of organizing the
world: “The more patients are presented with possibilities of restructuring,
re-evaluating, and redefining the environment, the more they can make
changes in behavior and interpretation in the real world” (p. 143).
A related suggestion, made fairly briefly in Lakoff’s paper but which
I find equally interesting, is that psychotherapeutic discourse has in turn
influenced other forms of conversation. In reflecting on Lakoff’s account
of psychotherapy and conversation, I would like to examine, in particular,
this idea of the interpenetration of different discourses. My own research
has focused, in the main, on the empirical analysis of talk: not surprisingly,
this reflection will entail relating Lakoff’s study to empirical work that I’m
familiar with.
Analyses of talk are often concerned to establish categories of one sort
or another: to distinguish different episodes, different functional catego-
ries, different styles, or discourse types, and so on. But there is also (and
this is perhaps a more recent trend) a recognition that spoken language is
relatively fragmentary, and that any utterance carries within it the echoes
of former utterances. A spoken text may well be an amalgam of styles, and
it is likely to be carrying out, often simultaneously, a multiplicity of func-
tions. Adults who bring to psychotherapy their experiences of “ordinary
conversation” are likely to have a fairly complex model, some aspects of
which will map onto the therapeutic interview more readily than others.
At least some talk, while not designed as therapy, is likely to have an
(informal) therapeutic function. People often confide in particular friends;
certain occupations (hairdresser, bartender, home help) are also associated
with being a “good listener.” This is not to suggest that clients have no new
strategies to learn when they take part in psychotherapeutic interviews,
but that they are likely to have experience of other “symbiotic” conversa-
tions that, in part, have broadly similar functions.
Lakoff’s characterization of ordinary conversation as reciprocal appeals
to an ideal form of talk in which “peers” interact with one another on
equal terms. This is contrasted with nonreciprocal talk, in which partici-
pants have different, usually nonreversible roles (alongside therapists and
patients, there are many other examples, such as doctors and patients, or
teachers and students). In the case of such nonreciprocal talk, one would
need to acknowledge that people slip in and out of roles, and that roles, at
any rate, are not fixed (teachers do not always engage in “teacherish” talk,
133
even when interacting with students in the classroom). But I think it is

also legitimate to ask how often we attain the ideal of reciprocal conversa-
tion: How reciprocal (or equal) is ordinary conversation in practice?
Feminist research has a valuable contribution to make on this question.
A great deal of empirical work on gender and language use indicates that
gender is a (potentially) salient factor in all forms of talk, including every-
day conversation; and that talk between female and male speakers is often
profoundly unequal, with male speakers using conversational strategies
that enable them to dominate mixed-sex interactions in many contexts.
Gender is only one of a number of factors that will affect the conduct of
conversations, of course; it will be emphasized or played down on different
occasions; and it will intersect with other social and personal attributes of
speakers, their changing purposes in engaging in conversation, and their
perceptions of the setting in which they are interacting. A whole set of
social and contextual factors will affect how ordinary conversation is orga-
nized, what roles people take on in conversation, and who has the right
to ask what of whom, so that the notion of “equal” conversation between
“peers” becomes problematic.
Labov and Fanshel’s empirical study of therapeutic discourse suggests
that nonreciprocal talk in this context is also highly complex. They provide
an extremely detailed analysis of just fifteen minutes from a series of inter-
views between a therapist and a patient, dividing this into five episodes,
which they examine in depth. Labov and Fanshel focus on several aspects
of the interaction, but the main point here is that they identify different
“fields of discourse” that have different characteristics: a style of “everyday
life,” in which the patient recounts recent events in a “fairly neutral, objec-
tive, colloquial style”; an “interview style,” in which emotions, behavior,
etc., are commented on and evaluated (used mainly by the therapist in this
case); and a “family style,” in which the patient “switches to an idiom that
seems to represent the style actually used in her family situation.”
Lakoff acknowledges the different interactional styles that characterize
different types of therapy, but is also concerned with general features that
distinguish therapeutic encounters as a discourse type. Labov and Fanshel
are almost at the opposite extreme. Their microscopic approach documents
the repertoire of styles drawn on in just one therapeutic interview. Such
research illustrates the heterogeneity of therapeutic interviews, which will
affect what the patient needs to learn as well as the relevance of what she
brings with her from previous conversational encounters.
Lakoff’s suggestion that psychotherapeutic discourse has influenced
other forms of conversation prefigures more recent work on the interpen-
etration of different discourse types, at least some of which has also taken
W H E N TA L K I S N O T C H E A P [ 133 ]
134
therapy as an example. According to Norman Fairclough, changing social

and economic relations have led to an increasing number of people experi-
encing “problems and crises of social identity” (1989:164), which are dealt
with by a wide range of counseling and therapy services. The discourses
of counseling and therapy have, therefore, become more salient in many
contemporary societies. They have also begun to “colonize” other areas
of public life, such as education and employment. This may happen infor-
mally, but counseling and therapy are also among a number of “discourse
technologies”: they are seen as communicative resources that can be used
strategically in different contexts.
Fairclough sees this as part of a larger process in which conversational
practices that were hitherto associated with the private domain have made
inroads into more public forms of discourse. Within the media, for instance,
one frequently finds the simulation of “private,” face-to-face conversa-
tions for a mass audience. Some parts of industry have also experienced an
apparent democratization, with workers operating as teams and expected
to engage in discussions and other forms of face-to-face interaction. There
is, I think, a parallel in education with an emphasis on students engaging
in various forms of collaborative work, and in informal group discussions.
Relations between different social groups (teachers and students, doctors
and patients, managers and workers) are also frequently marked by more
(apparently) democratic features, such as the reduction of overt markers of
social status and the use of more informal styles.
The increasing salience of counseling and therapeutic discourse, along
with the larger processes of what Fairclough terms “conversationalization”
and “informalization,” has often been regarded with suspicion. Counseling
tends to address the needs of individuals: the solution to problems usu-
ally resides in individuals coming to terms with events or modifying
their behavior in some way. However, individual problems have a social
context: they are related by some to social and economic change, such as
increased geographical mobility, the weakening of community ties, and
resultant strains on families. Fairclough sees therapy and counseling as
potentially in competition with “practices of political mobilization based
on the contrary assumption that social ills can be remedied only through
social change” (1989:225). Feminist researchers such as Deborah Cameron
have seen similar problems in assertiveness training and other practices
designed to help individual women communicate more effectively.
Greater informalization and conversationalization are themselves
somewhat ambivalent processes. On the one hand, they may allow greater
participation and more democratic decision making. But they may also be
used as subtle forms of social control. Lakoff points out that, despite the
135
conversational appearance of interaction in newer forms of therapy, such

interaction is still nonreciprocal. Fairclough asks whether changing prac-
tices in the workplace and other contexts are functioning as alternative
disciplinary strategies. There is also ample evidence from education of how
control is exercised in relatively informally organized classrooms.
Changing communicative practices may benefit some social groups more
than others. The use of informal talk in small groups, for instance, may
seem particularly suitable for female speakers. Positive action designed to
encourage female speakers to participate in debate has included strategies
such as initial discussion in groups, perhaps feeding into a plenary session.
Elsewhere (Swann and Graddol 1995), I have discussed the extent to which
the introduction of small-group or collaborative talk in the classroom may
be considered a process of “feminization.” Here too there are problematic
issues, however: there is abundant evidence that people may collaborate
perfectly to maintain existing power relations.
I’ve travelled some way from Lakoff’s original paper because, as I men-
tioned, the paper’s value lies in the ideas and hypotheses it suggests, which
encourage the reader to question and speculate further. In my own case,
Lakoff’s ideas caused me to reflect on the ideal nature of discrete discourse
types such as “ordinary conversation” (and, for that matter, “psychothera-
peutic discourse”), on the interpenetration of different discourse types,
and on some suggestions from other research about how this might be
related to social change. The paper will, I hope, suggest additional, maybe
alternative, possibilities for exploration to other readers.
REFERENCES
Fairclough, Norman. 1989. Language and power. New York: Longman.

Labov, William and David Fanschel. 1977. Therapeutic discourse: Psychotherapy as
conversation. New York: Academic Press.
Swann, Joan, and David Graddol. 1995. Feminising classroom talk? Language
and gender: Interdisciplinary perspectives, ed. by Sara Mills, 135–48.
London: Longman.
W H E N TA L K I S N O T C H E A P [ 135 ]
136
137
CHAPTER 6
When talk is not cheap

Psychotherapy as conversation (1979)
A s competent speakers of English we know the pronunciation of words,

what they mean, their preferable order in sentences, and what com-
binations of words are permissible. But to be really competent we must
have information of a more abstract and implicit kind. We must know what
constitutes a conversation, and what is appropriate at any particular part
of one. There are general principles that hold for all types of discourse, for
all languages; but there are also particular constraints, some restricted to
speakers of English, and, even more specialized, to members of a single
ethnic group or social class. Research in conversational analysis has shown
that the competent speaker respects these principles and constraints, at
an unconscious level most often, and therefore can distinguish between
appropriate and inappropriate utterances.1
However, to determine what constitutes an appropriate utterance,
we need to explore the nature of conversation, and we cannot do this
This paper originally appeared in The State of the Language, edited by Leonard Michaels
and Christopher Ricks, 440–48, Berkeley: University of California Press, 1980. It is
reprinted with permission of the publisher. © 1980 The Regents of the University of
California.
1. To enumerate and describe, however briefly, a reasonable subset of relevant recent
work in conversational analysis would occupy an inordinate amount of space. To sum-
marize briefly, contributions in this area have been made from the vantage points of
several fields: formal and informal linguistic pragmatics (of which this paper is an
example); ordinary-language philosophy, of which the work of Grice (1975) is of espe-
cial importance here; and ethnomethodology, of which the work of Sacks, Schegloff,
and Jefferson (1974) is exemplary.
138
without specifying the various contexts in which it occurs. What is suit-

able for a two-person informal conversation would be strange in a lecture.
As children we learn a set of rules for proper participation in conversa-
tion, and, more generally, for the production of discourse, by observa-
tion, imitation, correction, and, at the same time, by engagement in what
Gregory Bateson calls “deutero learning” of conversational strategy2—
that is, learning how to learn new and different rules. So when a child
who has known only how to talk to parents goes off to school, he or she
is able to figure out the different system required for conversation with
the teacher. The teacher may lecture, the child not. The kinds of ques-
tions asked and answers given also differ between teacher and child.
But the child has already learned that conversation is divisible into two
broad types: reciprocal and nonreciprocal. In a reciprocal conversation,
between peers, both participants may produce the same kinds of utter-
ances: both may question, and respond; both may make personal state-
ments and may be equally inquisitive about the other’s personal life. Both
may give and receive instructions and injunctions. But in nonreciprocal
discourse, as between child and parent, the options for both are limited.
Parents may question children deeply about their personal activities; chil-
dren may not so question their parents. Parents may give orders; children
may not. Parents may explain; children may indicate a lack of knowledge.
Traditionally, these roles may not be reversed, just as, traditionally, par-
ents call children by first names or nicknames while children address par-
ents by title, rather than name.
How does this discourse-learning take place, and until what age does it
go on? Are adults capable of learning new kinds of conversational strate-
gies? Under what circumstances will they submit to such learning? What
is entailed in the learning of a new conversational system? Are all the old
assumptions cast off, or is it merely the perspective that is changed? If we
could find a type of discourse that is typically learned by adults, we would
have an interesting case to study, within which to explore these questions.
The existence of such a case would imply that there is at least one type
of discourse to which children are not exposed, and only adults are apt
to want to learn; and, further, that learning new forms of conversational
interaction continues into adulthood. The conversational strategy learned
by participants in psychotherapeutic encounters appears to be just such a
case, specifically those types of therapy in which verbal interchange is cru-
cial to the process—from traditional psychoanalytic therapy through the
more recent developments of the so-called human-potential movement.
2. See Bateson 1972.
139
It should be no surprise that these various therapeutic systems have left

their imprint on the English language: the way we talk, along with the way
we think and feel, has been deeply influenced by our discovery that conver-
sation both reveals and influences the workings of the psyche—the basic
justification for verbal psychotherapy.
The most obvious imprint has been on our vocabulary. From classical
psychoanalytic theory alone, we have adopted or given old words new
meanings: complex, object, repression, identity—to give a mere handful.
(If we considered the contributions of newer therapies this list would
be increased exponentially.) More subtly, and very significantly, the
abstract elements of our linguistic repertoire have been profoundly
altered by psychotherapy. Our ideas about what constitutes a person’s
character and how behavior is talked about have been irretrievably
altered; and, most important, yet least observable of all, our notions
about how to hold a conversation, what is necessary or permissible in
ordinary discourse, have been given new form under the influence of the
psychotherapeutic model.
Psychotherapy developed by borrowing (intuitively) the model of the
ordinary conversation; and its rules became the basis of psychothera-
peutic conversation, as Freud first pointed out.3 Now we see ordinary
conversation repaying the loan with interest: psychotherapy has uti-
lized possibilities inherent, but unrealized, in ordinary conversation,
stretching those possibilities to its limits and beyond; and now we take
psychotherapeutic discoveries and make them a part of our everyday
discourse.
In utilizing the mechanisms and assumptions of ordinary conversation,
psychotherapeutic conversation recontextualizes them so that they have
a different meaning. It is not that we learn new rules in the therapeutic
process, but rather we learn new applications for the old rules. So, learning
to be in therapy is not like learning French; that is, not like learning a new
grammar, only a new way of looking at a familiar grammar.
For instance, in any conversation, the participants must feel that they
are getting some benefit from it, or they will not continue to engage in it or
resume it later. This benefit is sometimes purely informative: if I ask street
directions of you, I will be satisfied with that information alone. But more
often, in social settings, we require emotional benefit: the sense that oth-
ers like conversing with us and are doing it gladly. In such conversations,
information is of course often exchanged, but it serves largely as an excuse
3. As outlined in the works collected in Freud 1958.
W hen talk is not cheap [ 139 ]

140
for the real purpose of the discourse—emotional satisfaction. We give sig-

nals, overt or covert, that this need is being met.
We can do this by maintaining an appearance of spontaneity, which
implies that the participants are engaging in the discourse of their own
free will. We must not look as if we are conversing under compulsion—
so making appointments strictly to talk is normally avoided (“Let’s meet
for lunch”); or, if essential, it is mitigated by an effusion of small talk and
shows of camaraderie at the outset. Similarly, leave-taking must never
appear to be pleasurable or desired. One departs from a conversation by
pleading a commitment elsewhere: “I have to get back,” not “I want to go
somewhere else now.” It is significant too that beginnings and endings of
conversations—the stickiest points in attesting to the spontaneity of the
desire to talk to someone else—are ritualized; there are socially sanctioned
forms that must be followed for these parts of the conversation and not
the middle part. This helps soften the awkwardness of establishing contact
and breaking it.
Not only the form, but the permissible topics and the way they are
dealt with are calculated for the purpose of achieving emotional benefit.
The participants must feel that they are equals in the discourse, and that
their participation is reciprocal. Each has equal need of the other’s con-
tributions. For this reason, the conversation itself is sufficient as recom-
pense for holding it; to offer or accept money from others for talking to
them is inconceivable ordinarily. For the same reason, each participant
has the same conversational obligations and opportunities. If one can
ask questions and expect answers, so can the other. And if one can ask
questions about personal matters and expect straight answers rather
than indignation—“None of your business!”—so can the other. It is rude
for one person to ask another about, say, his or her income, be answered
directly, and then, asked the same question in turn, refuse to reply or
hedge the reply. Participants in informal conversation must all follow the
same rules. In this respect the informal conversation is different from, say,
the lecture, where reciprocity is not assumed to be in effect. There the ben-
efit to nonspeakers is assumed to be intellectual rather than emotional.
No need to worry about emotional satisfaction.
One compliment we are expected to pay to other participants is that their
utterances make sense to us; they are rational. To suggest that another is
speaking nonsense or cannot be understood is to risk grave insult. Hence,
interpretation has generally not been considered a permissible option in con-
versation between adults: to say “What you meant when you said X was Y”
is to suggest that the other person did not know what he or she meant, and
therefore was not in control of his or her behavior.
141
Psychotherapeutic conversation ignores all of these assumptions about

permissible discourse, and it is neither intellectually nor emotionally grati-
fying per se to the participants, at least not as we normally perceive con-
versational gratification. Further, the discourse is not reciprocal in either
topic or form. The patient soon learns that each partner may perform only
one particular subset of possible conversational actions: the relationship is
symbiotic rather than reciprocal.
Symbiosis, in the therapeutic setting, means that though the two part-
ners do different things, their roles are distinguished precisely to benefit
both of them. So, while in ordinary conversation it is reciprocity that con-
fers benefit, in therapy it is the clear-cut division of roles, the symbiosis. In
the many therapeutic frameworks devised over the last century, this divi-
sion of roles has been variously perceived and realized. The classical psy-
choanalytical model makes the distinction between the roles and powers
of the participants most clearly—by physical position, terms of address,
and types of permissible contribution, for example—but all therapeutic
systems in one way or another make it clear that the therapist has the
power and the patient or client the need, that the rules are determined by
the therapist, and that the therapist decides what is permissible or neces-
sary. Meetings are held in the therapist’s territory, begun and ended at the
therapist’s convenience.4
In some ways classical analysis resembles the lecture more than the con-
versation. One person holds the floor; the other may intervene with a brief
question or comment, but no more—usually when the first indicates a
readiness for such a contribution. What is strange—bizarre, even—is that
while in the lecture the holder of authority also holds the floor (an intui-
tively reasonable position), in analysis these roles are split.
In newer types of therapy, we find a style more closely resembling
normal conversational turn taking: each participant speaks and listens,
in more or less equal measure. But this similarity to ordinary conversa-
tion is deceptive, since the roles assigned to participants— implicitly
rather than overtly—differ. Patients soon learn that there are questions
they may ask, and others they may not; that there are questions directed
to the therapists that the latter are expected to answer, others they need
not answer. So patients may be asked—with the expectation of a direct
4. Jay Haley (1963) in particular has discussed psychotherapy as a sort of struggle

for power by the participants, with the therapist necessarily holding the “one-up” posi-
tion. It is also arguable that the therapeutic process can be described as a kind of game
in which the therapist starts out in the “one-up” position, and the progress of therapy
can be gauged by the gradual equalization of positions of the participants.

142
response—questions about their intimate lives, while therapists may not;

therapists may be asked practical questions about what patients are to do
in real life; depending on the type of therapy and the therapists’ personal
beliefs, these may be answered directly or not. As long as differences of
these kinds exist, and they do in any kind of therapy, no external appear-
ance of reciprocity—such as the maintenance by both participants of
upright posture, eye contact, and first-name address—changes the fact
that the interchange is nonreciprocal. The virtue of the classical analytical
model is precisely that it makes the nonreciprocity apparent.
The fact that the therapist receives a fee from the patient is clear evi-
dence that reciprocity is not expected within the conversation itself. The
fee is the therapist’s benefit in the encounter. The patient’s is, presumably,
the “getting well” or whatever equivalent phrase may be substituted in the
various models. But this is a long-term goal, and a rather vague one. How
can we understand the willingness of patients to continue—sometimes for
years—in conversational encounters that continually deprive them of the
emotional satisfactions expected of conversation?
In explanation, we have to invoke again the nonspontaneity of thera-
peutic discourse, and the consequent absence of reciprocity. Some writers
on the therapeutic process have found these departures from the norm a
burden and an abomination.5 But in fact, the overt denial of spontaneity is
what makes therapeutic discourse efficacious—precisely because it is dif-
ferent from normal conversational strategy.
The initiation and termination of the therapeutic relationship have
been treated extensively in the literature: there are discussions of forms
and means of effecting both. The bulk of the process, on the other hand,
remains up to the participants. This general framework for the entire
course—weeks, months, or years—of therapy is surprisingly like the struc-
ture of the single ordinary conversation, in which beginnings and endings
have been ritualized, but the middle parts, the major parts of the discourse,
are expected to be, or, at the very least to seem, spontaneous. What this
comparison suggests is that the entire series of therapeutic interviews
is to be taken as the equivalent of the single ordinary conversation. And
just as we determine the satisfaction and the emotional benefit of a single
ordinary conversation from its entirety, so we determine the benefit of the
therapeutic discourse from the efficacy of the entire series—not session
by session, from some of which the patient necessarily emerges feeling
discouragement or pain rather than pleasure or enlightenment. Because
5. For example, Szasz 1965.
143
in therapy we are engaged in one-to-one conversation, we apply the rules

for such discourse that we have known since childhood; but we apply
them in a different context, in which the series is equated with the single
conversation.
Learning the rules of a new language is painful and difficult; and even
learning to apply our usual rules in a new context is arduous. So we may ask
whether it is necessary: Cannot therapy be done as a conversation between
friends, equals and confidantes? Is there any reason for the imposition of
the burden of learning a new format on the patient, who has much to learn
besides?
Conversations between friends, even for the pursuit of therapeutic
goals, are but ordinary conversations based on the normal application
of rules. Friendly conversation is reciprocal, or at least must look that
way. It also must be understood as beneficial to both parties, in itself. It
must appear spontaneous: even if we make appointments to see friends,
we make them with our mutual convenience in mind; meetings may be
arranged on the territory of either one, or a neutral place; we must assume
that the possibility of future meetings will be decided as the need arises,
rather than by prearrangement; the topics we select are at the discretion
of both; and neither may unilaterally make interpretations. Hence we can-
not make the assumptions that are a part of the therapeutic framework,
nor engage in the re-evaluation of the applicability of our ordinary rules
for a new situation.
And it is the learning of new conditions for the application of rules that,
in all probability, has therapeutic effect. If it were something else, one ther-
apeutic system would be more efficacious than another, or one would work
for particular kinds of people, others for others. But, in fact, all therapies
seem to be effective for anyone approximately two-thirds of the time.6
One can even argue that it is the surprise value of the new system that
is effective: the contrast between old and new communicative strategies,
and success in learning the new, suggests to the patient that other tried-
and-true ways of organizing the world and determining responses to it are
not the only ones, and that new ways are safe to attempt. If the linguistic
reorganization—done in an atmosphere free from recrimination—can be
put into successful practice, then so can these and other new strategies
elsewhere. The more patients are presented with possibilities of restruc-
turing, re-evaluating, and redefining the environment, the more they can
make changes in behavior and interpretation in the real world.
6. The survey of Luborsky, Singer & Luborsky (1975) is evidence for this claim.

144
What therapy teaches by linguistic precept and example is that the same
phenomenon can be perceived in more than one way at once. This is what
therapeutic theorists like Gregory Bateson and Jay Haley refer to as learn-
ing to metacommunicate, to communicate about communication, to see
an utterance at once as both itself and a statement about itself. The allu-
sive nature that therapeutic theory attributes to all discourse helps make
understandable another peculiarity of therapeutic discourse: its fondness
for metaphor and parable.
A great deal of psychotherapeutic theoretical writing of all schools is
expressed in metaphorical terms. This is partly because no one has ever
seen a human mind working and we can only visualize this process in terms
of what we have seen. But more pertinent is the fact that stating something
as a metaphor or an allusion is a form of reframing or recontextualizing.
Metaphor often consists in placing a familiar idea in a new context. So
we find metaphor on all levels of therapeutic discourse: Freud talks about
the ego as a rider trying to control a horse that is the id; of psychological
processes as transfers of mechanical or electrical energy. More recently, as
substitutes for these physicalizations or concretizations of highly abstract
processes, we find mental processes described in terms derived from infor-
mation theory, decision theory, or computational theory. And replacing
Freud’s metapsychological metaphors—the horse and rider, or armies
advancing and retreating—we find images of game playing, of parent-child
relationships, of concrete spatial relations, prevalent in the speech of grad-
uates of the human-potential movement (represented, albeit parodically,
in Cyra McFadden’s The Serial): people are “uptight,” “up front,” “laid back,”
or “with it”; they can or can’t get “behind things”; people “feel” or “experi-
ence” rather than think things. A different class of metaphor, certainly: but
what is unchanging is the representation of abstract psychological states
and processes as concrete, physical realizations.
Interestingly, recent work by Roy Schafer (1975), within a quite ortho-
dox psychoanalytic model, has criticized this fondness for metaphor and
urged the use of literal description on the grounds that the use of meta-
phor prevents the patient from acquiring a sense of responsibility for his
or her actions.7 The suggestion is, then, that the allusive conventions,
implicitly a part of traditional therapeutic discourse, are infantilizing; they
keep patients from achieving maturity. And yet—infantilizing or no—
recontextualization is a necessary part of therapy through the agency of
transference, a necessarily regressive relation developed between therapist
7. See Schafer 1975.
145
and patient, in which the patient comes to respond to the therapist as to

earlier figures in his or her past. Transference is not only an inevitable con-
comitant of prolonged therapeutic encounters—it is a crucial element.
It is never clearly discussed in the therapeutic literature why transfer-
ence is present in any prolonged therapeutic contact. One reason might
be linguistic as much as psychological: through metaphor and metacom-
munication, the patient is being taught a new language, or rather a new
form of communication, by the therapist. There is but one earlier model
for this shared activity: the learning of a first language in early childhood.
In therapy, as at the parent’s knee, the process is implicit and unselfcon-
scious as it never will be later. Transference is the reactivation of that first
linguistic relationship, and by re-experiencing that earliest relationship in
a benign setting, the patient can undo the accretions of the past that per-
sist into the present. So it is precisely the peculiarities and complexities of
therapeutic conversation that achieve therapeutic effect, and their removal
would preclude therapeutic change.
What psychotherapeutic conversation or discourse teaches the patient
is new conversational possibilities in ordinary life. In turn, through the
many people who have been exposed to some form of the process, thera-
peutic conversation has become part of the equipment of English speakers
generally.
REFERENCES
Bateson, Gregory. 1972. Steps to an ecology of mind. New York: Ballantine.

Freud, Sigmund. (1911–1915) 1958. Papers on technique. The standard edition of the
complete psychological works of Sigmund Freud, ed. and trans. by James Strachey,
vol. 12, 85–171. London: Hogarth Press.
Haley, Jay. 1963. Strategies of psychotherapy. New York: Grune and Stratton.
Luborsky, Lester, Barton Singer, and Lise Luborsky. 1975. Comparative studies of
psychotherapies: Is it true that “Everyone must win and all must have prizes”?
Archives of General Psychiatry 32:995–1008.
McFadden, Cyra. 1977. The serial: A year in the life of Marin County. New York: Knopf.
Sacks, Harvey, Emanuel Schegloff, and Gail Jefferson. 1974. A simplest systematics
for the organization of turn-taking for conversation. Language 50:696–735.
Schafer, Roy. 1975. A new language for psychoanalysis. New Haven: Yale
University Press.
Szasz, Thomas. 1965. The ethics of psychoanalysis. New York: Basic Books.

146
147
Introduction to “Some of my favorite

writers are literate: The mingling
of oral and literate strategies
in written communication”
BY JENNY COOK-G UMPERZ
I n an era that has seen computers move in with people as household

objects, becoming indispensable not only as aids to daily living but as
our very own scribal helpmates, Robin Lakoff’s article foresees many of the
pragmatic consequences of these quite recent changes in the production of
written text. In noting new stylistic shifts in written texts, her argument
serves to remind us of the inherent pragmatic opportunities available now
that we can use a computer to compose texts such as this one. Computer
composition enables us to achieve more fluid transitions between spoken
and written language because of the ease with which text can be altered,
edited, even designed. Whether in home or office, computers have brought
into our lives a complex of potential stylistic choices that extends our
previous notions not only of how literate texts get made but also of how
they appear on the page. The home computer offers a standard array of
typographical choices, font shapes and designs, sizes, punctuation styles,
and page layouts as well as other design features that potentially makes
everyone into an art printer. Where once we might have spent a decade of
childhood perfecting our handwriting, and considered it an even greater
accomplishment to learn more than one variety (such as being able to write
148
in either italic or gothic script), we can now expand our design capabili-
ties in a moment by choosing a font style as easily as choosing the color of
the paper.
When rereading Robin’s article, I was struck by the relevance of her argu-
ment for some of the changes that computer literacy has brought into our
lives. I had not stopped to consider how much these typographical choices,
while aesthetically communicative, also affect the rhetorical import of the
written communication. The pragmatic consequences of these styles go far
beyond the surface design of the text. The adoption of computer literacy
has in fact brought about a commingling of the styles of the oral, imme-
diate, and telegraphic with the more considered compositions of written
text. Not only is the process of creation of textual products changed, but
the way we undertake written tasks is also different. Writing and creat-
ing text has become more like tinkering at a workbench than like grand
construction; whether writing a book, a journal account, a personal letter,
or an office memo, we undertake the task in the same contextual environ-
ment and with a similar compositional strategy. Text can be written and
rewritten, spliced, edited, and designed until the producer is satisfied. Or
it can be regarded as a finished written text at the very moment it emerges
from thought, through the fast-moving and often inaccurate finger pres-
sure on keys, and in the same way, it can be immediately transmitted to
others. Thus computer email has made written conversationalists of us
all. And in such composing activities as these, the text is seen to mirror
fleeting thoughts with momentary decisions. Composing activities such as
these has led to a shift in the signaling load between written and spoken
communication that has begun to influence many other forms of textual
composition in new ways (Alvarez-Caccamo & Knoblauch 1994).
When I first read Robin’s article sixteen years ago, I was immediately
interested in its implications for understanding the processes of children’s
early literacy acquisition, with which I was involved at that time. I found
most revealing the possibility of seeing a developmental growth of literate
strategies from those that were more oral and depended on spoken lan-
guage to those that were more governed by written textual conventions.
However, later I discovered this was not completely the case. For many chil-
dren, both middle-class and lower-class, the typographical or illustrated
page has a salience from very early in life that shapes the very idea of what
a text is and what it can convey (Cook-Gumperz & Gumperz 1981). The
linguist Georgia Green made a similar discovery when, while watching her
child choosing books for a bedtime story, she noticed that her daughter was
rejecting some stories as having been read before. When she questioned
her, her daughter said that she recognized the story, although her mother
149
knew she could read neither story nor title. Apparently, the child identified
the style of the illustrations as being the same as those in books that had
been read to her before. When Green went on to devise further informal
experiments with other preliterate children, she found that these children
similarly paid close attention to stylistic characteristics of books and even
made predictions about text and content based on styles of presentation
(Green 1982).
Robin’s examples of the textual intermingling of written and spoken
strategies draws attention to the fact that the pragmatic function of text
goes far beyond the salience of its surface effects. These assumptions have
a direct parallel in children’s lives, where the understanding of the rela-
tionship of speaking to writing begins with an early awareness of textual-
ity which then accompanies their move into full literacy. While children’s
primary literacy experience is with the printed text, as books of stories for
the most part written for children, when these are read to them orally they
become not written speech but spoken writing. The notion of a text that
exists independently of any person, that once created can stand alone as its
own textual self, is born at the very moment that a child says, “You missed
a page, read it again.” The Bahktinian notion of textuality is rediscovered
over and over again at many different bedtimes.
Such a notion of textuality, once realized, is a guiding force in early lit-
eracy development. Children very early in their written literate develop-
ment learn to make use of such displays as size of letters, placement of
text relative to pictures, and, later, quotation marks and other punctuation
devices to do the work of building textuality that they will later be able
to accomplish by relying on the accepted lexical and grammatical means.
In children’s early attempts at writing, the typographical conventions of a
written page often take over the communicative power of words that can-
not yet be fully realized because of difficulties in spelling or problems of
word order. Robin’s article makes one realize that the choice of linguistic
conventions and orthographic displays can offer important cues to shifts
in pragmatic function. Successful early literate development can be seen as
dependent on two different areas of language experience that contribute to
children’s pragmatic understanding of textuality: (1) the iconicity of a text
as a visual entity and its relation to the mapping of different verbal forms
into different patterns of meaning, which ultimately should lead to a sense
of genre; and (2) storytelling as it constitutes the narrative creation of a
shared, linguistically mediated reality that orders the temporality of the
universe and can exist as a stand-alone text. Once told and written down,
the story can stand by itself and continue to provide possible solutions to
all manner of new problems (Cook-Gumperz 1995).
S O M E OF M Y FAVOR I T E W R I T E R S A R E L I T E R AT E [ 149 ]
150
However, Robin’s warnings about the myths that guide the interpreta-
tion of what constitutes a literate consciousness and an intelligent under-
standing are still needed. Even though a new iconicity can be said to pervade
the world of the computer-literate, the idea that clear thinking is demon-
strated in an orderly presentation of written words has not yet been sur-
passed. After children’s early encounters with written words and textuality,
their experience of the power of written words as active textual entities
becomes a part of their past. However, the schoolroom literacy experience
continues. In this environment, the injunction is often heard that children
should say “one thing and say it clearly” (Michaels 1986). The blooming
confusion of the multivocality of emergent texts as pictures, words, quota-
tions, and typographical devices, and all manner of meanings being packed
into a page, like a Richard Scary children’s book, becomes, under the ruling
hand of school literacy, the linear presentation of a single idea.
Robin, we still need to heed your warnings if we are to understand the
shifting literacy of the computer age.
REFERENCES
Alvarez-Caccamo, Celso, and Hubert Knoblauch. 1992. “I was calling you”:

Communicative patterns in leaving a message on an answering machine. Text &
Talk 12(4):473–505.
Cook-Gumperz, Jenny. 1995. “Tell me a book” or “Play me a story”: The oral roots of
literacy socialization. Aspects of oral communication, ed. by Uta M. Quasthoff,
275–88. Berlin: Walter de Gruyter.
Cook-Gumperz, Jenny, and John Gumperz. 1981. From oral to written culture: The
transition to literacy. Writing: The nature, development and teaching of written
communication, vol. 1: Variation in writing: Functional and linguistic-cultural
differences, ed. by Marcia Farr Whiteman, 89–109. Hillsdale, NJ: Erlbaum.
Green, Georgia. 1982. Competence for implicit text analysis: Literary style
discrimination in five-year-olds. Analyzing discourse: Text and talk (Georgetown
University Round Table on Languages and Linguistics 1981), ed. by Deborah
Tannen, 142–63. Washington, DC: Georgetown University Press.
Michaels, Sarah. 1986. Narrative presentations: An oral preparation for literacy
with first graders. The social construction of literacy (Studies in Interactional
Sociolinguistics 3), ed. by Jenny Cook-Gumperz, 94–116. Cambridge, UK &
New York: Cambridge University Press.
151
CHAPTER 7
Some of my favorite writers

are literate
The mingling of oral and literate strategies
in written communication (1982)
INTRODUCTION
It is generally acknowledged that written and oral communication involve

very different kinds of strategies: what works orally does not work in
print, and vice versa. We know the reasons for this discrepancy, at least in
part: oral communication works through the assumption of immediacy, or
spontaneity; writing, on the other hand, is planned, organized, and non-
spontaneous. Hence, the devices utilized in the two media for maximum
effect can be expected to be different, and we may further suppose that
the direct transposition of the devices of one medium to the other will not
work, or even result in intelligible communication.
Overlaid on this distinction is a problem of judgment. For the past
three thousand years, more or less, literacy has been in competition with
nonliteracy (or rather, perhaps, orality) for minds and souls: many com-
mentators are not so much interested in the different values, the different
advantages, of each medium, as in perceiving the two as locked in deadly
This paper originally appeared in Spoken and Written Language: Exploring Orality and
Literacy, edited by Deborah Tannen, 239–60. Norwood, NJ: Ablex, 1982. Reprinted
here with permission of the publisher.
152
combat. Lately, this conflict seems to be exacerbated. The verdict in edu-

cational circles, as well as the circles of prescriptive comment on language
use, is pretty much in: literacy is dying, and as a result, civilization as we
know it is doomed. Only with the preservation of literacy can we hope to
preserve culture and civilization. This assumption rests on deeper unsup-
ported beliefs; but much of the current unrelenting attack on nonliterate
media—earlier, comic books; currently, TV and movies—can be traced to
the terror that we are about to regress as a civilization into a new Stone
Age, bereft of logical, linear thought, that literacy alone allows for sensitiv-
ity, intelligence, and complexity of thought. Since there is much evidence
that, in fact, literacy is diminishing, if we make the assumption that it is a
necessary concomitant of civilized culture, indeed we have cause for fear.
In this chapter I want to take a couple of different tacks: to argue that,
first of all, loss of literacy is not the same as loss of culture. I want to argue
this from a specific point of view. In the past several hundred years, the
reverse has been generally believed. In this time, we have been in the thrall
of the assumption that written communication is primary and preferable.
There is much evidence that, in the past couple of millennia in fact, at least
in written media (which of course is all we have, to document bygone ages)
the assumption has been made that the written form of communication
is basic, is more valid than the oral, and that even originally oral discourse
must be represented in terms of the rules of written communication to
be valid and intelligible. But in the last generation or so, there is much to
suggest that this position is being reversed, that the oral medium is con-
sidered more valid and intelligible as a form of communication than the
written, and that even written documents are now tending to be couched
in forms imitative of the oral mode. Moreover, the reasons for this are not
mere decline of education, or mental sloppiness, but are rooted in techno-
logical progress—even as the advent of literacy was three millennia ago.
I will present some discussion about the differences in ways of representing
thought and discourse in the two media, and how the style of representing
these ideas has been shifting, and why.
Lastly, I want to suggest that we must adapt to these changes rather
than stand by and deplore them, nor can we turn the clock back by any
means available to us. Literacy is useful in one sort of technology; but we
may not require it any more. As one whose entire life has been grounded in
the acquisition of literacy, and whose productivity resides in literate com-
munication, I confess to a sense of horror and betrayal as I write these
words, and of course I write them in the hope that there are those who will
read them, so it may be argued that I am—as my readers are—caught in a
153
paradox. But we must recognize the paradox if we are to emerge from our
confusion with honor.
We can see signs of change and confusion in a number of places, if we
correctly interpret what we see. In this paper I want to bring together a
group of seemingly unrelated or random facts that I have been noticing
over the last couple of decades, with the intention of arguing that these
illustrate the shift in our society from a literacy-based model of ideal
human communication to one based on the oral mode of discourse.
This is not, of course, the only such changeover in our culture. We can
be confident that several millennia ago contemporaneous civilization
underwent an equally agonizing shift, only in the opposite direction, as
writing became widespread and overtook oral modes of literature as a
means of recording present and past events. Of course, we have no record
of the wrenching effect of the changeover: the wrench would have princi-
pally affected the Old Guard—in this case, the nonliterate—and they, of
course, being nonliterate, have left us no record of any pain they felt. More
recently, within the past several centuries, a more minor change seems
to have taken place. Within a culture prizing literacy for the recording of
memorable events, there was a shift from a state where most people were
in fact nonliterate or barely functionally literate, and depended on another,
or scribal, class, to achieve their memorability for them, to one in which the
majority of people were, in the worst case, at least expected to be literate,
to make use of the written medium for information and amusement with
relatively little effort. It is not clear whether, in fact, this ideal ever actually
existed, but it certainly has been present as an ideal.
SPONTANEITY VERSUS FORETHOUGHT
Linguists, psychologists, sociologists, and anthropologists who study

communication strategies agree, as a truism, that the way we express and
understand ideas in writing is in many crucial ways different from the way
we express the same ideas in oral discourse. Actually, work on this topic
tends to create a peculiar dichotomy: planned, nonspontaneous written
discourse on the one hand, and spontaneous, direct oral communication,
on the other. It is unarguable that these represent the clearest cases, and
the sharpest distinctions, and are worth studying therefore as ends of a
continuum. But in order to understand how we utilize the various modes
of communication to their fullest advantage, we must understand that
there are other possibilities, and that some of the characteristics we have
S ome of my favorite writers are literate [ 153 ]

154
ascribed to “oral” discourse, for example, are not necessarily characteris-

tic of the oral medium per se, but rather their choice has more to do with
immediate personal contact—eye contact, for instance—or the usefulness
of an appearance of spontaneity, rather than with the use of the vocal chan-
nel itself.
The distinction between spontaneity— real or apparent— and fore-
thought in discourse is often directly equated with the oral/ written
distinction. The relation is not as clear as it seems. True, in oral, conver-
sational discourse we cannot plan in any real way—our utterances are too
dependent on those of our interlocutors and on other shifting real-world
circumstances. In print, on the other hand, we must plan. Our words are
understood as being subject to editing and revision, as representing the
considered preference of the writer and others among perhaps several
alternatives, all of which have been weighed. Intuitive writers may not con-
sciously make these selections, may go by “feel”—but they are certainly
doing some sort of editing, and are free to make substitutions later on.
But this distinction is not necessarily made in terms of the mode of
communication itself; we can easily imagine—indeed, there exist—types
of nonspontaneous oral discourse, and spontaneous written discourse.
Of course, these are special—they do not fit our prototype. Indeed, they
tend to be viewed with some suspicion, as if, by mixing alternatives like
this, someone is trying to get away with something. For example, formal,
old-fashioned, carefully crafted political rhetoric is viewed with some sus-
picion; when stream-of-consciousness prose was introduced as a literary
device before the turn of the century, it evoked shrieks of protest, with the
majority of contemporary readers evidently outraged at authors who dared
to meddle with the conventions of the written medium, and in particu-
lar, the convention that writers appear to carefully and consciously select
their words.
As spontaneity and forethought have their advantages, they have their
equally inalienable disadvantages. Truly spontaneous discourse has an
immediacy, and emotional directness, that is truly exhilarating; at the
same time, it carries the burden of immediacy: lack of clarity, use of the
wrong word or phrase, hesitation, repetition, and so on. These are neces-
sary concomitants of true spontaneity: we cannot be spontaneous and
polished at once. Planned discourse avoids these pitfalls; but at the same
time, it necessarily lacks warmth, closeness, and vividness. These lacks are
sometimes (e.g., by McLuhan 1964) viewed as necessarily characteristic of
print media; but it can be better argued that they are concomitants of non-
spontaneous discourse, of which print is one example. But print—more
than oral nonspontaneous media—exacerbates these difficulties because it
155
lacks many of the devices oral, present discourse utilizes as carriers of emo-
tional tone: intonation, pitch, gesture, eyes, and so on. On the other hand,
nonspontaneous media, by their ability to capture, through planning, the
appropriate mood, the description, may help the reader or hearer form an
ultimately more lasting and more vivid memory in the mind, and allows a
reader as well as a writer to rethink, re-experience, and revise impressions
which, in traditional forms of oral discourse, are lost forever once uttered.
There are additional “meanings” that we attribute to the choice of
medium. Written communication is memorable, the stuff of history and
reliability. Hence—and also because it is acquired and utilized with more
difficulty—it is more formal, the bearer of respectability. Oral discourse
can be colloquial or dialectal; the representation of nonstandard dialect in
writing—as a reader of, say, Mark Twain or George Ade will attest—tends
to give a reader a tired throat after a short period of reading: we cannot
help subvocalizing as we read “dialect”; it exists only in oral form.
Written discourse, then, is respectable; spoken, more heartfelt. A cul-
ture at any point in time has to decide whether the preferred mode of pre-
sentation of self is as a respectable or as a feeling creature. There may, at
some times in some situations, be available the chance to be both at once,
so that no such decision must be made. But in the matter of form of com-
munication, a society must decide whether the ideal is that of writing or
that of talking—reliability or warmth, respectability or ability to convey
emotion. For the last several centuries, we have, where possible, opted for
the first, assuming that the written channel was in some sense primary
or preferable. For various reasons, some social, some technological, we are
at present in the process of shifting, so that we prefer and respond most
appropriately to communications in any mode couched in an oral frame-
work. This switch, like any profound cultural revolution, is creating severe
confusion and dislocation, especially in those who perceive themselves as
holdovers of the old order. But rather than take moral or aesthetic posi-
tions in favor of one or the other, we will do better to examine the evi-
dence of the claim that this change is in progress, and then consider why, if
indeed it is, and what if anything we are to do in response.
If the written medium is primary, we can expect to find that even when
people speak or are assumed to speak “spontaneously,” their contributions
are represented via the conventions of the written mode. Thus quotations
in biography or fiction can be expected generally to sound like “written”
discourse, to utilize its conventions. We might find other, special uses of
an assumption of the primacy of writing: for instance, serious people,
in serious situations, in a work of art, might be represented as adopting
stylized “written” modes of discourse even when they are supposed to be

156
communicating orally; non-“serious” people, in non-“serious” situations,

might utilize more characteristically “oral” modes, again stylized.
We might consider Shakespeare the clearest piece of evidence for the
existence of such a set of assumptions a few centuries back. Shakespeare
has, basically, two modes of discourse: metrical lines and simple prose para-
graphs. It is frequently noted that the former tend to be assigned to “noble”
characters—noble both in a social and a psychological sense. At the most
climactic moments, the ends of crucial scenes, for instance, we frequently
find not only iambic pentameter but rhymed heroic couplets. But when
“commoners” speak, especially those who are comical, they tend to speak
prose. Additionally, of course, the “noble” poetic utterances contain formal,
elaborate language, complex sentence patterns, and other characteristics
of planned and memorable speech, while the prose segments tend to be
informal, colloquial, and even dialectal, and use much simpler vocabulary
and sentences that are not necessarily shorter, but less complex: conjunc-
tion rather than complementation, for instance. (These are tendencies, of
course, rather than absolutes, and have frequently been commented on by
critics.)
In such dialogue, we can look at meter as an idealization of “fore-thought”
discourse. For surely, more than any other type of discourse, metered utter-
ances must be planned in advance: one cannot speak spontaneously and at
length in meter. Rhyme compounds the plannedness, along with complex
sentence structure and formal language. Thus, Shakespeare’s poetic diction
is meant as a sort of ideal model, a signal to his audience, “We are speaking
for history here,” rather than as a literal transcription of real speech. (And
the prose segments serve to accentuate this intention by their dissimilar-
ity.) What I say here is true, of course, not only of Shakespeare but to some
degree of playwrights until relatively recent times.
Or consider a less striking but perhaps clearer case: James Boswell’s rep-
resentation of Samuel Johnson’s speech in his The Life of Samuel Johnson,
LL.D. Reading Boswell, we tend to be struck with awe at Johnson’s ability
to express ideas pithily and perfectly in apparently spontaneous conver-
sation: it is the ultimate fantasy, l’esprit d’escalier, idealizing our feelings,
on leaving a party—about why we couldn’t, at the telling moment, find
just the right words to say just the right thing—rather than the sloppi-
ness, the hesitation we remember with chagrin. Perhaps in the spirit of
sour grapes, I suggest that Boswell’s representation of Johnson is as much
an idealization as it is a faithful transcript. Johnson was no doubt better at
epigrammatic oral conversation than most of us, indeed than most of his
contemporaries, or Boswell would not have immortalized him thus; but he
was probably not as good at it as the written evidence suggests: Boswell
157
must have done a bit of judicious editing. He was enabled to do so—to

delete the searchings, the hesitations and repetitions, the false starts—
because the conventions of his age suggested that a good human being
spoke without them, that the most intelligent or most admirable person
spoke like a printed page. If Boswell were immortalizing a contemporary
Johnson, it is a safe bet that one would find just the reverse. Johnson’s
speech would probably be represented as colloquial, having hesitations and
repetitions and so on, to show him as real, in keeping with our perception
of our culture’s ideal. (“Eloquence,” since it entails planning, is a concept
whose time has passed.)
TRANSFERRING SPOKEN DISCOURSE TO WRITING
Another window into contemporary idealizations of style is in representa-

tions in novels of spontaneous conversation. Here in fact we might con-
sider ourselves faced with a sort of paradox, a paradox an author of fiction
confronts constantly, and must deal with in his or her own idiosyncratic
way. (This is true not only of novels, but of the screenplays of movies, or the
scripts of plays.) Here we have the transfer of the oral, spontaneous mode,
with all its implications, to either a written or a still-oral medium, but
clearly one where there has been forethought. (Unless, of course, we are
considering some of the cinematic works of people like Norman Mailer.)
In one way, we can detect a striking change between novels and plays of
the past and the present, generally speaking (we still have throwbacks, for
instance, Jacqueline Susann or the writers of Gothic fiction). In the past,
novelists seem to have represented spoken discourse by the same rules by
which the rest of the narrative exposition is unfolded. There are occasional
deviations, scraps thrown to verisimilitude: the use of contractions, per-
haps a few “wells” here and there; but otherwise, one would be hard put to
differentiate between the spoken dialogue and the written exposition of
most novels before the mid-twentieth century. (Again, there is sometimes
a divergence between socially or psychologically “serious” and “nonserious”
characters, the latter of whom are more prone to utilize nonstandard forms
of speech.) More recently, we find attempts in fiction to represent “real”
dialogue: more hesitation, sloppiness, errors of various kinds. But, as we
shall see, even when these occur, they do not have the same meaning as
they do in real, spontaneous discourse. The devices of spontaneous (oral)
speech are found, but in different circumstances and, indeed, with differ-
ent meaning, than in ordinary spontaneous conversation or in transcripts
derived from it.

158
This raises the problem that students of conversational strategy keep

being bedeviled with: that conversation, as taken off tapes and represented
in transcripts, is fiendishly hard to understand and very hard to keep pay-
ing attention to—and more, that the participants in such recorded con-
versations, including quite often the researchers, highly educated people
who take pride in their articulateness under pressure, come off sounding
like oafs or morons, or as if they were under the influence of psychedelic
substances. This problem is manifested, for instance, in published tran-
scripts of the Watergate tapes, which irritatingly seem to present a much
more fallible and human—in any case, bumbling and inarticulate—view of
Nixon and his henchmen than most of us are happy to see. Indeed, it is an
article of faith among liberals that Nixon’s very glibness, his profound lack
of spontaneity, did him in, and rightly. Anyone who plans his thoughts in
advance as he is felt to have done cannot be worth saving. Yet the Watergate
transcripts show a very different person: not lovable, but certainly fallible;
not genuine, but not really calculating either. In general, transcripts do not
feel to readers like “real” conversation—they are not immediately intelli-
gible like the dialogue in a novel or a movie, they don’t get to a point, they
don’t really begin or end. Yet they are real, and constructed dialogue is not.
How can we understand this?
One source of our difficulty lies in the reader’s interpretation of what lies
on the page. As long as a written format is used to represent purely written
discourse, there is no danger. But once we attempt to translate oral commu-
nication to the written page, we find ourselves having to translate meaning,
as much as form. The characteristics that work in one medium are not neces-
sarily ideal for the other; direct translation tends not to preserve sense, or
effect. In fact, there are not many valid reasons for attempting to represent
oral modalities directly in writing. One such reason, true since time imme-
morial, is the representation of originally spoken dialogue in print.
QUOTATION MARKS
All writing systems I am conversant with have developed conventions to

mark “spoken” words from the body of the written text—for example, quo-
tation marks. These signal: “Here we are using writing in a special way, to
represent oral discourse. Hence the representation may not be completely
accurate, so be on guard.” Interestingly, we do not restrict quotation marks
to this purpose. In many systems of writing, we find quotation marks set-
ting off forms that are not intended to be understood as “uttered.” The
markings on the last word in the last sentence are a case in point. I don’t
159
mean that someone just came up behind me and uttered “uttered” in my

ear. Rather, such quotation forms are conventionally used to indicate the
writer’s abdication of responsibility for the locution so enclosed, a sort of
ironic lift of the eyebrow in print: “I represent it like this, but do not fully
take responsibility for the sentiments thus expressed.” This usage shares
one thing with the more direct convention: in both, quotation marks signal,
“This is not the writer’s own.” In the one case, it is because the words are,
literally, someone else’s; in the other, the writer is merely not fully sincere
in using them. Hence, too, the use of quotation marks around nonstandard
forms: dialect, slang, nonwritten register, for example. “I really know bet-
ter than this,” the writer is saying. “See, I’m literate—really I am.”1 To the
extent that a writer is insecure, such uses tend to proliferate. Therefore, as
anyone who has spent time grading freshman themes knows, such writings
teem with quotation marks which, to the literate eye, seem inexplicable if
not downright execrable. We have seen, for instance, the use of quotations
around nicknames, or even real names:
On my summer vacation I went with my brother “Bill” to “Boy

Scout” camp.
They are used to enclose, and exonerate, anything that might be consid-
ered nonliterate form, however mildly colloquial:
The “sophs” really “didn’t” do it. They “sure” didn’t.
Or, indeed, they may be used to indicate anything that is, until entrusted
to paper, information that only the writer, and not the reader, possesses.
And that’s how you make “beef bourguignon.”
This use may seem to contradict the claim that these quotation marks
indicate denial of responsibility; but in fact, information not certainly
shared by the reader is apt to make a fledgling writer especially nervous.
One recent use of quotation marks outside of written communication
is particularly noteworthy. Traditionally, of course, quotation marks are
1. A striking example of this use of quotation marks was seen in the television reports
of the freeing of the US hostages by Iran on January 20, 1981. The hostages were wel-
comed to the U.S. military installation in Wiesbaden, Germany, by crowds, some of
whose members were carrying signs. The most prominent of these was one which
said: “Welcome Home!” (with the quotation marks).

160
restricted to writing: they represent, as I said, the insertion of nonwritten

discourse into communication in the written mode. Then what are we to
make of this?
“I’m cynical enough to understand … my name assures more press coverage.

This is a logical place to use me. I’m a woman of a certain age and I’m indepen-
dent. All the ‘safe’ ”—she put down her glass of ice water and used her fingers for
quotation marks—“accouterments.”
(Interview with actress Ali MacGraw, San Francisco Chronicle, 10/20/80)
Here the secondary meaning of quotation marks becomes primary: their

use in achieving nonresponsibility or ironic distance. Not trusting to vocal
inflections or visual cues—traditionally the means of expressing these
meanings—to establish ironic intent, the speaker borrows from written
form a sort of “fail-safe” device: the gesture of making quotation marks
with the fingers is so strikingly noticeable that it cannot be overlooked by
an interlocutor, as inflection or a lift of the eyebrow might. We can assume
that this device is becoming more common—and I have seen a good deal
of it around, particularly on TV talk shows—because we are losing confi-
dence in ourselves and others as interpreters of subtle signs. (I put quota-
tion marks around “fail-safe” above, by the way, partly because I was using
a metaphorical term and was a little afraid about whether the metaphor
was appropriate; and partly, I fear, to indicate that appreciation by the
reader of that witty (or, rather, “witty”) gesture was in order, but of course
I could only suggest that if I at the same time delicately removed it from
my responsibility. Under other circumstances, I would hope to have edited
the whole thing out in a later draft, but I leave it in to show how it works.)
This extension in the use of quotation marks, then, becomes one way
writers can try to personalize their writing by bringing into it the emo-
tional directness of oral speech. It is not the quotation marks per se that
convey the emotional impact. Rather, they are a signal, unlike other analo-
gous devices; they suggest to the reader, “Feel about what is enclosed in
these marks as you would about oral discourse.”
ITALICS
Other devices, on the other hand, are more direct in their communicative
effect, dictating a specific emotional response just as stress or pitch might,
for instance, in oral discourse. This is a principal function of italics. Italics,
like quotation marks, serve two distinct but related purposes. One, their
161
speech-imitative use, and their direct rather than metaphorical interpreta-

tion in writing, is to provide simple emphasis, by means of suggesting the
modes by which oral discourse is made emphatic: rise in pitch and loud-
ness. This is clearly the use of italics in quoted oral discourse:
“Mary is very intelligent,” he expostulated.
Closely related is the use of italics to stress some idea as important,

for instance, in order to make contrasts, which might be expressed orally
either by pitch/loudness changes or by gestures:
No, this is Mrs. Jones, and that is Mr. Smith, with the carnation in
his lapel.
Both of these are direct representations of oral devices, translations into

writing which we understand through reference to speaking. It is interest-
ing to note that, as far as we can tell, there is no tradition of italics or any-
thing analogous in ancient languages. One reason for this might be that
these languages, being free-word-order types, could achieve what English
can only effect by stress and pitch through word order, which can be rep-
resented directly in writing, and in fact is a device that is probably more
usable in writing than in speech, given planning and processing constraints.
But just as quotation marks have a secondary usage in writing that
stems from the fact that they represent emphatic discourse in the oral
medium, italics too have derived meaning. Italics, because their use sug-
gests the tonal and emotional range characteristic of oral discourse, can
be used in writing to suggest something similar: the writing is made to
seem fresher, more spontaneous, more emotionally open and direct. As
with quotation marks, this is playing with communicative fire: it suggests
the writer is not conversant with the devices available to writing to achieve
these effects. Overuse is deadly. The magazine Cosmopolitan is perhaps the
strongest exponent of the genre:
As illnesses go, hypoglycemia has a good deal of flair, but another debility is
almost as status-laden these days: low back pain. No one is sure why an ach-
ing back should have such panache. Perhaps the appeal of this disorder is that
it has traditionally been associated with vigorous, earthy, physical types… .
(Cosmopolitan, July 1980)
The Cosmo italic is really a sort of amalgam of the two conventions I have
identified. The reader is undoubtedly supposed to imagine the writer’s

162
voice rising to a squeal; but at the same time, the words italicized, as in
the quotation above, would not normally receive unusual stress or pitch in
speech. The italics then signal “N.B.”
Italics, however, have a meaning not shared by quotation marks: they are
popularly characterized as a written manifestation of “women’s language.” It is
popularly thought that women, when they write, overuse italics (for example,
in letter writing). This stereotype no doubt helps to account for the prevalence
of italics in Cosmopolitan. Presumably this accords with the wider inflectional
range of women’s speech. As far as I know, no serious research has been done
on the prevalence of italics in women’s written prose. Any correlation with
the wider range of oral intonational possibilities would be rather unexpected
and inexplicable, unless we assume that in writing we literally encode a spo-
ken “voice” in our minds, which seems quite implausible. In any case, whether
because of their association with “feminine” style or because they too heavy-
handedly attempt to manipulate the reader’s emotional response, italics are
much riskier in writing than their counterparts in speech.
CAPITALIZATION
We might compare with quotation marks and italics—conventions whose

effect in writing is related to their invoking of oral equivalents—a third
emphatic device, that of capitalization. This, when used outside of the nor-
mal convention of capitalizing proper nouns, has both a different origin and
a very different feel from the others. Where quotation marks feel adoles-
cent, and italics feminine, capitals feel childlike. But all are used to set the
matter so marked off from the rest of the text, in one way or another. It is
interesting that, while italics and quotation marks seem to have increased
their range of usage in recent history, capitalization flourished in earlier
periods much more than now. Seventeenth-and eighteenth-century writing
seems to have employed capitalization, especially of nouns, either in a man-
ner reminiscent of modern German (all nouns capitalized) or as a marker of
emphasis, not required but stylistically optional. Relics persist, particularly
in children’s writing (which may be why they have a childish flavor, or per-
haps they are used in juvenile literature because they have a childish flavor).
Perhaps the clearest case is in A. A. Milne’s Winnie-the-Pooh books (1926):
“Hallo, Pooh,” he said. “How’s things?”

“Terrible and Sad,” said Pooh, “because Eeyore, who is a friend of mine,
has lost his tail. And he’s Moping about it. So could you very kindly tell
me how to find it for him?”
163
“Well,” said Owl, “the customary procedure in such cases is as follows.”

“What does Crustimoney Proseedcake mean?” said Pooh. “For I am a
Bear of Very Little Brain, and long words Bother me.”
(Winnie-the-Pooh, 50)
Here capitalization does indeed serve to mark the words of greatest sig-
nificance in the narrative, rather like the Cosmopolitan italics. But it does so
without recourse to spoken convention.2
NONFLUENCIES
There are still other devices of spoken language that are carried into writ-
ten language: for example, ellipses, repetitions, and vocalized pauses.
These, though, fall into a different category. Where the earlier types are
intentionally used to create emotional involvement, these tend to be more
or less involuntary. Indeed, encountering oneself on tape using them to
excess frequently causes chagrin: “Do I talk like that?” They do not, typi-
cally, mark emotional directness, but rather simple unpreparedness—the
negative side of spontaneity. For while in writing we can marshal our
thoughts and present them in coherent and rhetorically effective order, in
speech we seldom have the opportunity, and at least currently, if we do,
we try to look as if we are not making use of it. It is conventional—and,
probably, realistic—to assume that the more emotionally involved one
is, the more one’s thoughts are confused, or at least presented linguisti-
cally in incoherent and rambling form. Therefore, the use of these devices,
primarily resulting from the normal inability in spontaneous discourse to
know in advance what to say, secondarily can signify emotional turmoil in
oral communication, at least if carried to excess, or used in unusual places
in the conversation.
Perhaps, in fact, it is this that gives metered dialogue, in older drama-
tists, its sense of seriousness and majesty: if we are working within the
constraints of iambic pentameter (for instance), the form cannot tolerate
any additions or deletions, no vocalized pauses, hesitations, or repetitions.
Distress must, then, be represented lexically and explicitly. In more mod-
ern forms of constructed dialogue, however, we find these devices used for
2. The reverse of A. A. Milne is, perhaps, e. e. cummings, who eschewed capitalization
entirely. If Milne (and others) use capitals as directives to readers as to what is to be
considered important, then we can take cummings’s avoidance of the device as a kind
of anarchy: readers are left to their own devices, with no authorial guidance.

164
very specific effects. Soap opera is one genre that strives for the appearance
of spontaneity, both to manipulate the audience’s emotions and to increase
the verisimilitude of the dialogue. Besides, soap opera can afford it while
other genres can’t: a script aiming for artistic effect must compress and
telescope its exposition, leaving explicit only what is necessary to further
the development of the plot and the characters. Therefore, those aspects
of spontaneous dialogue that are truly random and accidental—the result
of universal human difficulties in organization and memory, rather than
problems specific to the character and his or her situation—are eliminated.
But soap opera runs by different rules. Soap opera tries to provide a literal
slice of true experience for its audience, so that life, and dialogue, in that
genre move just as slowly as they do in real life. Hence there is a time for
ellipsis, correction, hesitation, and so on.
One random episode in a randomly chosen half-hour daily soap opera
(All My Children) illustrates this aspect of soap-opera grammar interest-
ingly. A male character has, in the way common to soaps, unbeknownst
to himself fathered a daughter years ago by a woman who has since (as a
result, we may surmise) become a nun. This girl and the man’s legitimate
son (unaware of course of their true relationship) are having an affair. They
are somehow involved in a murder, and are in jail. The son, it is implied,
has fallen into these bad ways as a result of his father’s neglect, of which
his father is keenly aware. In the randomly chosen segment, the father first
encounters the nun, who indirectly informs him of the identity of the girl,
and then encounters his son and the girl, separately, in jail. Thus, we have
one character who, in the course of the episode, has increasingly more to
feel uncomfortable about, and whose every confrontation with another
character is wracking to him; and several other characters who, while they
have their various guilts, have no guilt with respect to this particular charac-
ter. This character, incidentally, is a lawyer and politician, so articulateness
is expected of him. As the episode progresses, while the speech patterns
of the other characters remain more or less stable, only occasionally laps-
ing into real-speech conventions, the father progressively gets closer and
closer to transcript form.
Another interesting use of “spontaneous” form in nonspontaneous oral
media is seen principally in commercials. The commercial is, in one sense,
the diametric opposite of the soap opera: time goes by in a flash; every
micro-millisecond must count. At the same time, the actors must seem
real, must be people the audience can identify or at least sympathize with.
Because of the time constraint, we see very little use of time-wasting oral
devices. Rather, we get a different set of quasi-oral conventions. In true
spontaneous discourse, sentence structure such as is expected in written
165
prose is not strictly adhered to. We find run-ons and fragments quite typi-
cally. So one way to approximate spontaneous speech is via telegraphic
fragments:
—Mm! Light, crispy! Glad I discovered it!

—It’s as nutritious as Grape-Nuts!
—As Grape-Nuts?
Aside from the question of whether normal adults ever really find them-
selves conversing at length about breakfast cereal, this passage is unnatural
in a couple of respects. If telegraphic fragments do in fact occur in ordi-
nary conversation, they tend to occur in the body of the discourse—not
as introductions, where the topics of discourse tend to be made explicit.
It also seems to me that the third contribution is unlikely: repetitions of
this type normally signal serious disagreement, not merely the need for
a bit of amplification (for which “Oh, really?” normally suffices). But in
the commercial, every syllable must add to the informational content.
Finally, notice the word crispy. This is certainly not a word characteristic
of formal written register—in fact, crispy is not oral colloquial American
English: I don’t think I have ever heard it uttered spontaneously. It is one
of a set of words used only, or almost exclusively, in writing to suggest oral
register, but in fact seldom or never in true spontaneous oral discourse
(tyke is another).
I have contrasted these special narrative conventions with transcripts,
and indeed if we continue this comparison we can begin to understand
why, accustomed to these narrative conventions, we find accurate tran-
scripts of spontaneous conversation especially hard to interpret. We have
been trained to believe that, when we encounter these devices in written
communication, we must translate them as signals of emotional intensity.
But in ordinary transcripts, we find them in every sentence, in every con-
text. Either we are dealing with a feverish emotional pitch—belied by other
clues in the recorded conversation—or we are truly contending with a for-
eign language, or perhaps a pidgin—as indeed a transcript is, half written
and half spoken grammar.
In general, then, the borrowing of a device from one medium into
another is always overdetermined: it carries with it the communicative
effect, or “feel,” of one medium into another (the metacommunicative
effect) and at the same time attempts to utilize the language of one mode
to communicate ideas in another (the communicative effect). It is no won-
der that this sort of translation can create confusion in readers (or hear-
ers), and can also create in them very strong feelings—typically negative.

166
THE COMIC STRIP
The negative impact of such borrowings, particularly from the oral to the
written mode, is exacerbated by the fact that the least “desirable” forms
of communication are the first to show the traces—or perhaps this is the
effect of the derogation of oral style, rather than the cause of its initial
appearance in unrespectable places. One of the earliest forms of commu-
nication to attempt to convey essentially oral concepts in print was the
comic strip. Interestingly, in the newer and more sophisticated strips (e.g.,
Doonesbury by Garry Trudeau), the conventions of the old strips are miss-
ing, and we have, as it were, reverted to the “respectability” of print con-
vention. But the strips I remember reading as a child, though they utilized
print, actually approximated in many ways a blend of oral and literate cul-
ture. It is easy to recall examples: the fact that comic-book sentences never
ended with a period but, if not questions, with an exclamation point, seems
to have been an attempt analogous to the use of italics or initial capitals to
impart importance and emotional immediacy to the text. (Alas, with over-
kill it lost its potential force.) There were frequent attempts to reproduce
oral self-correction devices, for emotional effect, e.g., pauses and repeti-
tion. Additionally, there was the attempt to reproduce nonstandard dialect
and colloquialism. Very commonly, this was done by “dropping g’s” from
the suffix -ing, but rather more interesting were other attempts to create
the same effect.
I remember as a child trying to imagine pronouncing some of the comic
book writers’ attempts at “natural” speech. There were basically two types
of problems. One occurred when the form could not in fact be pronounced
as written by anyone with the normal articulatory apparatus; the other
involved spelling representations that, in fact, represented the only way a
form could be pronounced at all, outside of profound affectation, so that it
was hard to see what was gained by the special “colloquial” spelling. In fact,
it sometimes led to more confusion than necessary. Examples of the first
type include the dropping of vowels to form impossible consonant clus-
ters: T’ th’ store; of the second, spellings like yuh or ya for unstressed you
(always pronounced [jə]); ta for to; and the mysterious (to me for years)
locution Omigosh, which I perceived then as an exotic exclamation [ˈɑmigɑʃ]
(years later, I realized that it was intended to represent the much less inter-
esting Oh, my gosh). But running it together as a single word (which led to
my confusion) served the purpose of making it seem more quickly articu-
lated, hence more exciting as well as more “colloquial.” We sometimes find
phonetic spelling used to represent slang, although the word itself would
not be pronounced differently in standard and nonstandard dialects: wuz
167
for was, for example. In all these cases, then, special spellings are used not
simply as a guide to pronunciation but as a way of indicating, “Since this
representation is different from the ‘formal’ forms of written language, it is
to be taken as ‘oral,’ i.e., immediate, emotional, colloquial.”3
Abuse of the trust thus engendered is always a temptation. I remem-
ber a striking case from the 1964 presidential primary campaign. George
Wallace was running in this campaign, a fact the New York Times, for rea-
sons good and bad, viewed with dismay. Wallace was perceived as a red-
neck, an illiterate—a far cry from the hallowed JFK, and not even up to
the style of the folksy but nonetheless minimally literate LBJ. At one point
the Times attributed to Wallace, in an interview, a locution the Times repre-
sented as “could of.” What is remarkable is that the newspaper was thereby
representing an utterance that could not in fact have been pronounced,
in nonstilted speech, by a native speaker of English in any way other than
[kʊ́dəv] or, even more colloquially, [kʊ́də]. It corresponded, of course, to
the written form “could have,” but would never normally have actually been
pronounced that way. So the Times was engaging essentially in comic-strip
tactics, representing a form via nonstandard spelling not in order to accu-
rately distinguish a special pronunciation from the standard, but for some
other reason. I would argue that the reason was to make Wallace look like
an illiterate redneck; that if someone of unimpeachable intellectual creden-
tials, say Adlai Stevenson, uttered the same phonetic segments, they would
have been represented as “could have”; had they been uttered by the infor-
mal but respectable Lyndon Johnson, perhaps as “could’ve”; but “could of”
marked Wallace as unmistakably of the booboisie, an implicit editorializa-
tion in the paper of record’s news columns.
So anything deviating from the written medium and its ways of expres-
sion has tended to be viewed as suspect, a bad second best, certainly not to
find its way into respectable print. Indeed, a good part of the fashionable
3. On the Phil Donahue show of January 23, 1981, Donahue read a letter from a
viewer whose essence was this: She had always found the show one of the most “liter-
ate” on television. Recently she had received her first transcript of the show, and was
dismayed to discover that Donahue is cited as saying ya and +-in’—e.g.:
“I have to tell ya… .”
“It’s comin’ to the time… .”
There are two relevant points here. First, it is perhaps curious that transcripts of
the Donahue show provide this sort of close phonetic transcription (and even more
curious, their fidelity appears somewhat selective: if ya, why not hafta or haveta?); sec-
ond, it is striking that—as my foregoing discussion would predict—the viewer was
perfectly comfortable with [jə] and [kʌmɪn] in oral discourse, but reinterpreted these
forms when they were encountered in print.

168
outcry against comics of the 1950s (a campaign superseded now by similar

cries of outrage against television) can probably be traced to the fear that
they were corrupting our youth by confounding written and spoken styles
and thereby destroying the culture’s literacy. Then we have a right to be
surprised, perhaps, and certainly to want an explanation when serious styl-
ists, writers of fiction and nonfiction who are reviewed by and taken seri-
ously in reputable intellectual journals emerge in the 1960s and 1970s with
a style reminiscent of nothing so much as the despised comics. In novels,
particularly, this is indicated by the way dialogue is represented. In older
fiction, dialogue appears not very different in style from the rest of the nar-
rative exposition: sentences are finished and not run-on; there are few indi-
cations of hesitation or self-correction. But in more experimental recent
work, often the dialogue is strikingly different in form from the rest of the
text. For example, from Thomas Pynchon’s The Crying of Lot 49 (1966):
“But,” began Oedipa, then saw how they were suddenly out of wine.
“Aha,” said Metzger, from an inside pocket producing a bottle of tequila.
“No lemon?” she asked, with movie-gaiety. “No salt?”
“A tourist thing. Did Inverarity use lemons when you were there?”
(p. 19)
Here we see an attempt at representation of informal (and somewhat

intoxicated) conversation, giving an impression of sentences trailing off
into nowhere. The writer is trying to draw the reader into the complete
context, much as an oral storyteller might by intonation and gestures.
But these are still attempts to represent originally oral discourse. More
striking is the use of oral devices in nonfiction, nondialogue narrative
exposition. Here the line between oral and written communication blurs
irrevocably. We might expect this from inexpert writers, setting down
on paper exactly what they envision themselves as saying—for example,
schoolchildren or students in college remedial-composition courses. But in
fact, this is seldom what we find in such artless or “natural” writers. They,
it seems, are overawed by being faced with the need to communicate for
posterity, on paper, and their style typically is anything but colloquial. If
we would search for the relics of a bygone age, although misunderstood and
misused, we would be most apt to find them in the writings of neophytes.
Informal or colloquial style, in a written text, is not natural, is not the
mark of one unfamiliar with the distinctions between the media. Rather, it
shows up only in the works of writers of great subtlety and skill, who delib-
erately obscure the age-old distinctions—and why, indeed, would they do
that, unless there were some reward? In a writing-based culture, there is of
169
course no discernible reward, and this is why we have encountered writings

of this type only very recently.
The work of Tom Wolfe is representative of the genre, indeed its most
striking exemplar. In The Right Stuff (1979), his style is strongly and inten-
tionally reminiscent of comic books. (We might suppose that the subject
matter, the astronauts, contributes to this choice of action-packed breezi-
ness as a style, except that it is characteristic of his writing in general.)
Scattered plentifully through the text we find italics, quotation marks, cap-
italization, other aberrant punctuation devices, ellipses, fragments, exple-
tives, dialectal and colloquial forms, and much, much, more:
The thing was, he said, the Mercury system was completely automated.
Once they put you in the capsule, that was the last you got to say about
the subject.
Whuh!—
“Well,” said Yeager, “a monkey’s gonna make the first flight.”
A monkey?—
The reporters were shocked … Was this national heresy? What the hell
was it? …
But f’r chrissake …
(p. 100)
or:
And in this new branch of the military, no one outranked you. (p. 108)
Sympathy … because our rockets all blow up. (p. 109)
… he had literally written the book on the handling characteristics of
aircraft … (p. 108–109)
In all the last three examples, italics are used to indicate, almost paradox-
ically, a sort of ironic detachment: “They keep saying this, this is a cliche.”
Sometimes, too, they mirror directly the vocal inflections of incredulity:
… their religious affiliations (religious affiliations?) … (p. 90)
The italics here can represent nothing but vocal intonation, since in
writing italics are never used to underline part of a word, except to indicate,
for instance, misspelling—deviations from written convention.
Examples could be multiplied at will to illustrate the really shocking
change that Wolfe has visited on English narrative prose convention. What
is surprising is that it works. Purists may complain, but Wolfe has had a

170
profound impact as a contemporary stylist. My feeling is that a style such

as his—while it is perhaps the style of tomorrow, rather than today—could
not have been utilized, however experimentally, until very recently. Even
stream of consciousness, as in Joyce, does not bring the conventions of oral
discourse so directly into the narrative segments of written prose. Wolfe is
not only inviting us to share a particular sort of emotional relationship—
with him, as well as with his subject—but is also informing us that the
relation between written and spoken discourse is not as it has always been,
that we have to rethink our ideas about the primacy of writing. For Wolfe,
it is evident, talking is primary, and writing is successful to the extent it
captures the nuances of speech. (For Boswell, two centuries ago, we assume
precisely the reverse was true: a man was a dazzling oral stylist precisely
to the extent that his “spontaneous” oral epigrams resembled the produc-
tions of a careful writer, although they were perhaps valued more than they
might have been had they been original in writing, for much the same rea-
son that Dr. Johnson himself alleged a female preacher, or a dog walking on
its hind legs, would be admired—because the trick was done against heavy
odds, contrary to normal expectation.)
Even in oral media, style changes to reflect our new preference for the
products of “spontaneity”: the tradition of the political orator goes back,
in this country, several centuries. But where once—as in old Fourth of
July orations that have been preserved for our delectation—the orator
strove to deliver a polished speech, every word considered for effect, noth-
ing accidental, now such an oration would be viewed with suspicion, if not
derision. Now a political speaker must seem to be holding a conversation,
even if there is not much turn taking in evidence. He can hesitate, he can
change his mind about words, he can use vocalized pauses. We can see the
change in our own time. Franklin D. Roosevelt, by all the evidence available
to us, was a dazzlingly successful orator of the “polished” school. (Churchill
was even more so, of course.) Eisenhower was viewed by large segments
of the populace as incompetent as a speaker, even incomprehensible. (Yet,
listening to tapes of his speeches today, they don’t sound nearly as con-
fused as they were supposed to have.) He was forgiven, of course, because
he was quintessentially a man of action. Stevenson, on the other hand,
was Johnsonian (Samuel, that is)—a speaker of wit and grace. He lost to
Eisenhower, the bumbler, perhaps setting into motion, or at least giving
the nod to, a profound re-evaluation of discourse strategies. Kennedy, uni-
versally accounted a good and powerful oral stylist, nonetheless ushered
in a new mode. He, and his brothers as well, used a great deal of hesitation
and, even more, vocalized pauses of the er variety. While these were made
fun of, the fun was benevolent, and none of the Kennedys was faulted as
171
an orator on these grounds. Nixon, interestingly, attempts to be a throw-

back to the old “polished” and nonspontaneous style, but is not successful
at it. Jimmy Carter is an exponent of the new simple-conversation style
of oratory, and his resounding defeat may indicate that we are not quite
ready for this cataclysmic changeover at the highest levels of leadership
just yet. However, the style has been tried and shown to be conceivable if
nothing else.
THE PRIMACY OF ORAL MODES
So there is evidence all around us that as a culture we are contemplating—

if we have not taken already—a leap from being written-oriented to being
oral-oriented. Our stylistic preferences naturally are shifting along with our
values, although we have not consciously been aware of a shift in either.
Publicly and outwardly, we are a people who prize literacy: we believe that
all children must be taught to read and write, and we recoil in despair at
the evidence piling up that, in ever larger numbers, they achieve neither.
We cast about for scapegoats: the decline of education, the failure of paren-
tal discipline, the rise of television. But these are better viewed as effects,
not causes. In fact, deep in our hearts, we are no longer a society which
values literacy. Compare the sales of books with the Nielsen ratings, or box-
office attendance at movies; and the only books that do sell are those that
hardly can be said to be read, their style and plot are so simple. Or, more
and more often, best-selling books are those that are reconstructed, under
contract, from the screenplays of movies, novelizations rather than novels.
You read them after you see the movie, rather than the other way around,
as it used to be.
Our first temptation is to view this imminent changeover as profoundly
threatening. We are tempted to try to turn it around—say, by pouring mil-
lions into the teaching of writing, as educational organizations have done
of late. But perhaps this is not a farsighted approach; perhaps all the untold
billions at our disposal cannot make a significant difference. Perhaps, too,
it is just as well, and there is nothing to bemoan. Literacy may be going
the way of the horse and buggy; and while there are those who bewail the
switch from horse and buggy to automobile, it is also argued that, had
the horse not been replaced by the motorcar, cities today would be buried
under horse manure. Just as no language change occurs unless there exists
in the language the potential to express the same range of ideas in another
way, so communicative change—change of communicative style—will not
occur, I would propose, if something were to be irrevocably lost. Language

172
changes nothing before its time. Thus, if we look at what literacy has
achieved for humanity in the past—as well, by the way, as what the advent
of literacy has cost us, for there have indeed been costs—we may well find
that there are now other means at our disposal to achieve these same ben-
efits, perhaps with fewer unfortunate side effects; and that, in fact, the
new mode that is gaining strength at the expense of literacy will enable us
to communicate more beautifully and forcefully with one another than can
be envisioned now. Since the change is barely in its infancy, is in fact still
in gestation, we cannot see the final product yet. But we can at least take a
different tack in viewing the oral/literate dichotomy. Rather than wringing
our hands about the loss of literacy, let us ask what is replacing literacy, and
why this change is occurring, and finally, what the gains as well as losses are
apt, in the long run, to be.
While most of what we encounter contrasting literate with oral culture,
or discussing the apparent decline of literacy among ourselves, explicitly
or implicitly assumes that literacy is the desirable state, and to forswear
it is to invite the return of the Dark Ages, there are other possible points
of view. Marshall McLuhan, for example, has argued in his writings (e.g.,
1964) that a literate culture loses something: immediacy, warmth—the
qualities I have been suggesting we are trying to put back into our liter-
ate productions, albeit not really successfully. McLuhan suggests that the
coming of literacy caused great, and undesirable, changes in human char-
acter and social behavior: we retreated, physically and emotionally, from
one another. But now that nonliterate media are gaining influence at the
expense of literacy, we are beginning to see the return of the old virtues.
McLuhan, then, sees literacy and nonliteracy as heavily influencing charac-
ter and behavior, as well as less clearly social psychological attributes.
On the other hand, it is sometimes argued that scientific progress, in
terms of linear logical thought, can only be found where there is literacy;
that the written word, in fact, and its linear syntactic arrangement into
sentences going from one side of the page to the other (or, for that matter,
up and down, but in any case, in a visually perceptible pattern implying
sequencing) has formed our thought and enabled us to formulate concepts
like logical causality and temporal sequentiality. Without literacy, the argu-
ment continues, we would lose all ability to think logically, and revert to a
state of animalistic savagery.
To my knowledge, no evidence for either position, in any empirical
sense, has surfaced. A fairer representation may be found in the writing of
scholars like Olson (1977), who sets forth advantages and disadvantages
in cognitive style for both states. In particular, they note that members
of nonliterate societies have far better short-term memory capacity—as
173
indeed they must if they cannot rely on written lists or memoranda.

Literature in such societies has a different relation to its creators and audi-
ence than in ours: in some ways, it is continually being created afresh by
each new teller of tales; members of such societies, then, learn to abstract,
to get the gist of ideas, to discover what, in a list of items or a work of art,
is crucial and unchangeable, and what can be altered and embroidered at
will. We do not develop such critical skills; for us, stories exist in books, to
be reproduced aloud only under special circumstances, and then preferably
word for word. A moderately skillful “singer of tales” probably cannot be
said ever to flub his lines—he is supposed to improvise. But an actor in our
culture, reading from a teleprompter or memorizing a written script, is not
expected to deviate from the unchangeable text, and even if he might get
away with it because the audience is not familiar with it, he is nonetheless
apt to show his anxiety at the knowledge that he has altered a line, and
manifests his nervousness so that it is apparent to all.
In any event, the relation between literacy and high culture is not as clear
as many have supposed it to be. The acquisition of literacy clearly entails
loss as well as gain. We can imagine, if we like, rock-ribbed conservatives
of Homeric times grousing as literacy swept like a noxious firestorm over
the young. Only a fad, they said at first, hopefully. Decadent foreign ways.
But as the fad persisted, as the younger generation ceased to pay attention
to the older educational system and found their own sources of enlighten-
ment, the scorn no doubt turned to fear, and rationalizations set in. The
acquisition of this book-learning is a real danger … soon, none of our young
people will be able to produce a decent rendition of the Iliad, by Zeus, with-
out consulting (sneer) a book! What if they were out on a moonless night
and needed to refer to a passage in the epic? They’d be lost—unable to cope,
unable to participate as intelligent human beings and make use of the riches
with which their society has so generously endowed them … . And so on. In
fact, the plaints of the elders would sound astonishingly like the Jeremiads
of some of our own culture critics, mutatis mutandis.
CONCLUSION: A NEW NONLITERACY
I certainly don’t mean, in a written document no less, to denigrate liter-

acy. But I am arguing that all points of view that represent the two posi-
tions in terms of value judgments are dangerously misleading and make it
impossible to understand what is really going on. What is clear is that, for
whatever reason, societies give literacy and nonliteracy moral and intellec-
tual values that there is no real evidence to support; and indeed, one could

174
argue that at different times, in different cultures, the influential powers

could take diametrically opposed views on the issue, and feel just as smugly
and self-righteously guardians of the old, right way of living. It also seems
true that the issue arises in a culture only occasionally: when its members
have reason to be uncertain as to which mode of expression is preferable.
Clearly, in Homeric times, there was no choice: since there was no literacy,
there was no controversy. Then, of course, partly aided by the advent of
literacy, partly by other factors, life and society became more complex, and
by recent times, we came to feel we could exist as full-fledged human beings
only if literacy, in some sense, were more or less universal, at least as an
ideal. There was no other way to assure that the tremendous amount of
information necessary to survive in times like ours could be assimilated
and utilized. But currently, that whole assumption is called into question
by the development of the first new informational technology since the
advent of literacy (perhaps excluding the invention of the printing press).
Now access to all the information one previously achieved through literacy
can be gained by other means, via newer media, and we see that it is the
younger people who are the first to recognize this and become able to take
advantage of it. Literacy shortly will not be essential for simple survival
any more, nor will there be any need to preserve it except as a curiosity
or an atavistic skill, like quilt making, learned and proudly practiced by a
few. Indeed, with sophisticated information-processing and audio-visual
technology, we will have achieved a sort of meeting of the fullest benefits
of literate and nonliterate forms of information sharing. We will have at
our disposal the emotional closeness of the oral channel, its immediacy,
its ready accessibility. And at the same time we will have the preservability,
the historical accuracy, the immortality of print, because tapes, like books,
can be stored. So rather than the epic singer’s approximation to replica-
tion of his tales, we will have absolutely accurate reproduction of works of
art or sources of information, just as we do now with writing. With some
improvement in technology, it is hard to see why literacy per se should be
crucial for survival in the future.
It is this shift toward which our more innovative writers are tend-
ing. They are developing a style for a nonliterate age, although they are
doing so, rather paradoxically, in a literate medium. We might think of
the first writers of prose, as opposed to poetry, as operating under anal-
ogous assumptions—trying to adapt their techniques to a new medium.
Poetry was devised for oral transmission, for memorability. Prose works
best as a literate device, since it cannot be easily remembered nor reduced
to formulas. Similarly, the contemporary introduction of oral devices
into written communication suggests the merging of the oral and literate
175
traditions—although oral-style-on-paper is really only a metaphorical evo-

cation of written-memorability-through-the-voice—and therefore, the
experiments and special uses I have been discussing in this chapter repre-
sent an attempt to come to terms with the future.
REFERENCES
Coleman, Lee. 1980. “In” diseases: A look at the current crop of chic complaints.
Cosmopolitan, July 1980.
Donahue. 1981. Episode first broadcast 23 January 1981 by WGN. Directed by Ron
Weiner.
MacGraw, Ali. 1980. Interview. San Francisco Chronicle, 20 October 1980.
McLuhan, Marshall. 1964. Understanding media: The extensions of man.
New York: McGraw-Hill.
Milne, A. A. 1926. Winnie-the-Pooh. London: Dutton.
Olson, David R. 1977. From utterance to text: The bias of language in speech and
writing. Harvard Educational Review 47:257–81.
Pynchon, Thomas. 1966. The crying of Lot 49. New York: Bantam.
Wolfe, Tom. 1979. The right stuff. New York: Picador USA.

176
177
Introduction to “Persuasive
discourse and ordinary conversation,
with examples from advertising”
BY JANET S. SHIBAMOTO-SMI TH
A vignette. It is midmorning. I am sitting at my desk in the office with my back

to a blank screen upon which I am about to write the introduction to Robin
Lakoff ’s “Persuasive discourse and ordinary conversation, with examples from
advertising” (1982). One of my graduate students is sitting across the desk. We
did not have an appointment and she disclaims any particular purpose for the
visit. We are talking about this and that and—as often happens—we start mak-
ing jokes and laughing (loudly). A colleague passes my open door, stops, pokes
his head in, and asks what’s going on. Simultaneously, my student and I turn
and say, “Nothing, we’re just chatting.” We were having, in other words, an
ordinary conversation.
Well, yes, we were. Weren’t we? I look back at my computer screen, at
the phrases “persuasive discourse” and “ordinary conversation” and begin
to think this over. In her 1982 paper, Lakoff distinguishes two basic forms
of discourse. Ordinary conversation may contain elements of persuasion,
but is not for the primary purpose of persuading, as is “technically persua-
sive discourse” (p. 185). A good ordinary conversation conveys, primarily,
mutual liking and willingness to interact. Having set out persuasive dis-
course as distinct from ordinary conversation, Lakoff proceeds with a con-
trastive analysis along five dimensions: reciprocity versus nonreciprocity,
informality versus formality, ritualized versus novel, spontaneous versus
178
scripted, private versus public. Her focus in the paper is on distinguishing

the technically persuasive from the ordinary conversation; hence the focus
on advertising. But her analysis can be used to look more deeply into many
sorts of verbal interaction. I use her framework here to see what kind of
conversation my student and I were really having.
An ordinary conversation, as characterized by Lakoff in this paper, is
normally reciprocal. Was our conversation reciprocal? Certainly in terms of
turn taking, of being able to ask questions and expect answers, in terms of
how much talking time we could take, it was. Turning to topic management,
interruption rights, and opening and closing logistics, however, there is
clearly less reciprocity. Like it or not, and irrespective of the friendly nature
of our relations, I have more sway over those aspects of the conversation
than my student. This may always be true of conversations across age, sta-
tus, or power asymmetries, even when the purpose of the interaction is
the expression of mutual liking rather than the intent to convince; in this
paper, Lakoff is clearer about the role of power imbalances in persuasive
discourses than she is about their effects in ordinary conversation. On the
other hand, for a few interchanges, my student was clearly trying to per-
suade me that a particular way of dealing with a research problem was the
best way; she had done the background work on developing alternatives
and was the authority, although I remained the power figure. This stretch
of her speech sounded remarkably like a television commercial (albeit for
the best way to do research rather than for the best headache-relief medi-
cine). Had this been the “real purpose” of my student’s visit? And was our
conversation, then, really a persuasive discourse? Could it have been, given
the power imbalance? Or were we having a “semipersuasive discourse” (as
Lakoff terms psychotherapeutic discourse and classroom lecturing)? The
direction of persuasion and the direction of the power asymmetry were
not, for us, in the same balance as in the cases Lakoff cites, but perhaps
this is immaterial. As Lakoff so insightfully points out, in this society we
are uncomfortable with power asymmetries and find it hard not to imbue
them with negative connotations, even in those cases where equality is
impossible, or irrelevant. The appropriately power-imbalanced interaction
is a hard one for us, then, to appreciate.
But what about the other characteristics identified by Lakoff? Well, we
were certainly informal, another distinguisher of ordinary conversation.
In fact, we were loud, highly engaged, and mildly vulgar. This persisted
throughout the “research-o-mercial” segment of the discourse as well, at
least to the best of my recollection. On this dimension, we indeed had a
thoroughly ordinary conversation, even when I was being persuaded. Our
conversation was cloaked in the ritual and custom that Lakoff claims for
179
ordinary conversation; neither of us was striving for the neologistic, syn-

tactic, or pragmatic novelty of persuasive discourse. Were we, as well, spon-
taneous? Hmmm. Yes, and no. The discussion of research directions was
preplanned—scripted would be too strong a term—by one of us, at least.
But neither of us anticipated that we would be laughing over the specific
faculty foibles, news trivia, or local gossip that ended up causing us to create
enough noise to gather an audience, in the person of the faculty-colleague
passerby. Was the conversation primarily for the research discussion (that
is, is that why the visit occurred) or primarily for the opportunity to encode
mutual liking and solidarity in idle chit-chat? And, was our idle chit-chat so
idle? Scroll eight years ahead to Lakoff in 1990: “We are always involved in
persuasion, in trying to get another person to see the world or some piece
of it our way… .” (1990:18). As to the last dimension, public versus private,
we were private enough to gossip and public enough to attract attention.
So my student and I had a conversation. It was not, as we see, simply an
“ordinary conversation”; neither was it a “technically persuasive discourse.”
But the two idealized types presented by Lakoff in the paper to follow give
us a framework within which we can see how a broad range of interactions,
including the everyday, shift from moment to moment, now closer to one,
now closer to the other. And therein, I believe, lies the greatest value of
this paper.
From the perspective of language and gender studies, I see this paper as
an important way station between Language and woman’s place (Lakoff 1973,
1975) and Talking power: The politics of language (Lakoff 1990). Language and
woman’s place set out an agenda for the field that remains current today.
And yet many of Lakoff’s 1973/1975 claims are problematic. Lakoff may
have been, in fact, wrong in her particulars. And with some reason. The
linguistic times were not entirely friendly to the examination of language
in its context of use, to begin with, and we were quite naive about the rela-
tionship of the association of complexes of linguistic features with women
as opposed to men by the speakers of a community, and the variable or fre-
quency differences in sets of linguistic features as they are produced in the
speech of those same men and women (Sherzer 1987:97). Still, something
was missing. Our empirical investigations into the language used by men
and women were annoyingly inconclusive. Take tag questions. Women use
tag question more than men (e.g., Siegler & Siegler 1976); no, they don’t
(Dubois & Crouch 1975); well, they do but only sometimes (Cameron,
McAlinden & O’Leary 1988). Was a “girl” (woman) “damned if she does,
damned if they doesn’t … talk like a lady” (Lakoff 1975:6) or not? And,
for a real woman in a real context, just what would the phrase “talk like a
lady” mean?
P E R S UA S I V E DI S C O U R S E A N D OR DI N A R Y C O N V E R S AT I O N [ 179 ]
180
“Persuasive discourse and ordinary conversation” provides the begin-

nings of a solution to some of these problems. Looking beyond the surface
form of the discourse to its deeper purpose, or function, allows us to see if
there is a match between surface form and function, as well as, if there is
such a match, whether it is the same for all speakers (as Lakoff herself did
in Lakoff & Tannen 1984). Reading Lakoff’s 1982 discussion of persuasive
versus ordinary discourse—with careful attention to her analysis of the
differences in the allocation of conversational power across participants
in each form—highlights the source of some of our confusion. In the lan-
guage and gender field of the 1970s, we debated the “nature” of women’s
conversations as opposed to the “nature” of men’s conversations; in the
early1980s, we had only just begun to realize how seriously we needed to
take questions of function. Could not some of the contradictory findings
as to the constructions of men’s and women’s conversations be resolved
by a better understanding of the kind of “conversation” (that is, discourse
form) being examined? Just as language and gender analysts were begin-
ning to have an understanding of the complexities of verbal interaction,
Lakoff gave us a preliminary and important key to one of the most funda-
mental divisions in a taxonomy of discourse forms. The types of discourse
that Lakoff addresses in this paper do not revolve around gender divisions.
She focuses her attention on economic and political persuaders, particu-
larly advertising, touching also on what she calls the “semipersuasive”
forms of psychotherapeutic discourse and the classroom lecture. But the
application to gendered discourse is clear, and the work done in this paper
to disaggregate the “persuasive” from the “ordinary” lays a foundation for
a clearer understanding not only of the overtly persuasive discourse form
but of the persuasive elements of the ordinary conversation as well. And
we see this most clearly in the central place given persuasion in all verbal
interaction, articulated in Talking power. How this works had already been
“persuasively” outlined in 1982.
Lakoff also considers persuasive discourse in terms of its adherence (or
nonadherence) to the Cooperative Principle and to the Gricean maxims.
With one notable exception, the maxims are violated routinely in persuasive
discourse. Lakoff outlines some convincing reasons why this might be the
case, then turns to Quality, the one maxim that is not violated in persuasive
discourse—at least, not in technically persuasive discourse. The maxim of
Quality, of course, enjoins us to try to make our contributions ones that are
true; advertisers violate this at their peril. Political persuaders are held to be
more likely to violate this maxim, but suffer when caught doing so, nonethe-
less. Lakoff speculates that the maxim of Quality is in some ways prior to
the other maxims, a speculation that deserves more development.
181
Even when it comes to ordinary conversation, I am persuaded that

I would not wish to talk to Robin Lakoff if I were having an off day. No one
hears a conversation like Lakoff, on so many levels simultaneously. From
the most mundane of ordinary conversations to the most loaded politi-
cal or economic persuasive discourse, it would be hard to sneak a hidden
agenda item past this sensitive and linguistically responsive scholar. Don’t
do it; she’ll catch you. Read the paper following this introduction to see
how she does it.
REFERENCES
Cameron, Deborah, Fiona McAlinden, and Kathy O’Leary. 1988. Lakoff in

context: The social and linguistic functions of tag questions. Women in their
speech communities: New perspectives on language and sex, ed. by Jennifer Coates
and Deborah Cameron, 74–93. London & New York: Longman.
Dubois, Betty Lou, and Isabel Crouch. 1975. The question of tag questions in women’s
speech: They don’t really use more of them, do they? Language in Society
4:289–94.
Lakoff, Robin. 1973. Language and woman’s place. Language in Society 2:45–80.
Lakoff, Robin Tolmach. 1990. Talking power: The politics of language.
New York: Basic Books.
Lakoff, Robin, and Deborah Tannen. 1984. Communicative strategies and
metastrategies in a pragmatic theory: The case of Scenes from a marriage.
Semiotica 17:323–46.
Sherzer, Joel. 1987. A diversity of voices: Men’s and women’s speech in ethnographic
perspective. Language, gender, and sex in comparative perspective (Studies in the
Social and Cultural Foundations of Language 4), ed. by Susan U. Philips, Susan
Steele, and Christine Tanz, 95–120. Cambridge, UK & New York: Cambridge
University Press.
Siegler, David, and Robert Siegler. 1976. Stereotypes of males’ and females’ speech.
Psychological Reports 39:167–70.
P E R S UA S I V E DI S C O U R S E A N D OR DI N A R Y C O N V E R S AT I O N [ 181 ]
182
183
CHAPTER 8
Persuasive discourse and ordinary

conversation, with examples
from advertising (1982)
The very work that engaged him [in an advertising agency] … wafted him into a sphere
of dim platonic archetypes, bearing a scarcely recognizable relationship to anything
in the living world. Here those strange entities, the Thrifty Housewife, the Man of
Discrimination, the Keen Buyer and the Good Judge, for ever young, for ever hand-
some, for ever virtuous, economical and inquisitive, moved to and fro upon their com-
plicated orbits, comparing prices and values, making tests of purity, asking indiscreet
questions about each other’s ailments, household expenses, bedsprings, shaving cream,
diet, laundry, work and boots, perpetually spending to save and saving to spend, cutting
out coupons and collecting cartons, surprising husbands with margarine and wives with
patent washers and vacuum cleaners, occupied from morning to night in washing, cook-
ing, dusting, filling, saving their children from germs, their complexions from wind and
weather, their teeth from decay and their stomachs from indigestion, and yet adding so
many hours to the day by labour-saving appliances that they had always leisure for visit-
ing the talkies, sprawling on the beach to picnic upon Potted Meats and Tinned Fruit,
and (when adorned by So-and-so’s Silks, Blank’s Gloves, Dash’s Footwear, Whatnot’s
Weatherproof Complexion Cream and Thingummy’s Beautifying Shampoos), even
This paper originally appeared in Analyzing discourse: Text and talk, edited by Deborah
Tannen, 25–42. Georgetown University Round Table on Languages and Linguistics
1981. Washington, DC: Georgetown University Press, 1982. Reprinted here with per-
mission of the publisher.
Acknowledgments: The ideas in this paper owe much to discussion with many oth-
ers. In particular, Linda Coleman’s (1983) work on advertising, as well as other forms
of persuasive discourse, has been stimulating. Students in Linguistics 153, Pragmatics
(spring 1981), have been invaluable in refining my thoughts, and their forbearance in
serving as sounding boards is gratefully acknowledged. Deborah Tannen’s suggestions
and inspiration are similarly much appreciated.
184
attending Ranelagh, Cowes, and Grand Stand at Ascot, Monte Carlo and the Queen’s
Drawing-Rooms.
Dorothy L. Sayers, Murder Must Advertise ([1933] 1986)
I n the field of discourse analysis, much attention has been focused on cer-
tain forms of discourse, much less on others. In particular, scholars have
been concerned with ordinary conversation, on the one hand, and written
expository text, on the other. While the treatment of these types as sepa-
rate entities has certainly taught us a great deal about the characteristics
of each of them individually, and something about the nature of discourse
in general, it leaves a great many questions unexplored. For one thing, how
do they relate to other types of discourse—which may superficially, or even
more deeply, resemble one or the other? What are the universal charac-
teristics of all types of discourse, and what characteristics are specific to
just one or two? Can we devise a taxonomy of discourse types, a means of
unambiguously differentiating among them?
Classifying the basic forms of discourse in terms of their differing and
similar characteristics seems rather less glamorous than writing a grammar,
a system of rules, for one or all discourse types. But in fact, it is impossible
to write a grammar without knowing the basic units involved: grammar
consists of instructions for the combinations of these basic elements.
It would also be profitable to look at discourse not from its surface form
(as conversation, say, or literary text) but more deeply, with interest in its
deeper purpose. When we look at a range of discourse types, we notice
that they appear quite different from one another. Clearly, this disparity
is due to differences in what each is intended to accomplish, so that a suc-
cessful performance of a type of discourse is one that accomplishes what
the speaker, or the participants as a whole, had set out to do, rather than
merely one that conforms to some particular surface configuration.
This paper is somewhat experimental in nature, as I want to address
some of these questions, and see how far we can get toward at least prelimi-
nary answers. As a start, I want to consider one possible distinction among
discourse types in terms of function or purpose: ordinary conversation, on
the one hand, and something we can call “persuasive discourse” (PD), on
the other. I will come to the problem of the definition of the latter shortly.
I am not suggesting that this is the only distinction one can make among
discourse types, nor that it is necessarily the major one. Certainly, we can
185
divide up the spectrum in many ways, all intersecting: oral/written, for-

mal/informal, spontaneous/nonspontaneous—just to list some possibili-
ties. I am phrasing the question as if we are to inspect and perhaps justify
a dichotomy, but we should keep in mind the possibility that no dichotomy
will emerge—that we cannot divide discourse neatly into persuasive/non-
persuasive realms, that some types may cut across this distinction and it
may prove irrelevant for others.
With all these caveats in mind, however, it still seems useful to begin
as I proposed, since persuasion is a function attributable to at least some
discourse types. I would suggest at the start that ordinary conversation is
not persuasive in the sense of having persuasion as its major goal. That is
not to say that in ordinary conversation (OC) we do not persuade, or try to
persuade, other participants. But persuasion is not what we enter into the
conversational experience for. We do not come away from an informal chat
saying, “Wow, that was a great talk! I persuaded Harry that bats eat cats!”
Rather, an experience of OC is good if we come away feeling that a good
interaction has been had by all, that we all like each other and wouldn’t
mind talking to each other again. Granted that getting these ideas across
is in a sense persuasive, it is not so in the sense that getting someone to
worry about ring around the collar is persuasive.
One important determinant of technically persuasive discourse is non-
reciprocity: discourse is defined as reciprocal only in case both, or all, par-
ticipants in it are able to do the same things, and if similar contributions
are always understood similarly. A classroom lecture obviously is nonre-
ciprocal: one participant selects the topics, does most of the speaking, and
determines the start and finish of the discourse. The power in such dis-
course is held by the person holding the floor, at least to the extent that
that person makes most of the explicit decisions as to the direction the
discourse takes, its start and finish. On the other hand, it can be argued
that the audience holds power in such a situation, for it can go or stay,
be attentive or not—and by these decisions negate the effect and purpose
of the other’s speaking. I return later to the question of what constitutes
power in a discourse, but here it can be noted that it is meaningful to raise
this question only for nonreciprocal discourse.
Discourse that is truly reciprocal is, at the same time, necessarily egali-
tarian, at least ostensibly. Ordinary conversation, for example, is normally
fully reciprocal: any participant has the same conversational options as any
others, and if one can ask a question and expect an answer, so can the others;
if one can ask a particular type of question, or make a certain sort of state-
ment (say, a question as to the other’s financial affairs; a statement about
P ersuasive discourse and ordinary conversation [ 185 ]

186
the other’s personal appearance), the other has the same privilege in turn,
and if one can refuse to answer, so can the other. Violations of this principle
do, of course, occur in OC, but when they do, participants feel a rule has
been violated, that the conversation is making them uncomfortable, while
nonreciprocity in a lecture is expected and reasonably comfortable.
An example of an intermediate, hence problematic, case is psycho-
therapeutic discourse. In many of its forms, there is the appearance of an
egalitarian, reciprocal conversation, but in terms of deeper intention, the
reciprocity turns out to be only superficial. The therapist can ask questions
which the client soon learns not to ask; and if the latter should attempt to
ask such a question, the therapist, rather than give an answer, will usually
treat the question as a tacit invitation to ask another question, or make an
interpretation: “I notice you’re curious about my personal life.” Further,
many of the marks of power that belong to the lecturer in the classroom
also belong to the therapist: the decision when to begin and end, and—
while the client ostensibly picks the topics of discourse—the determina-
tion of what the client’s contributions actually mean. The client, however,
typically holds the floor for the major part of the discourse—an anomaly
in terms of the relation between floor holding and power holding in typical
conversational settings, which makes it especially difficult to acquire profi-
ciency in therapeutic dialogue.
With these assumptions in mind, one can attempt a definition of persua-
sive discourse as a type of discourse that nonreciprocally attempts to effect
persuasion. Discourse, then, is to be considered persuasive only in case it
is nonreciprocal, and the intent to persuade is recognized explicitly as such
by at least one party to the discourse. By persuasion I mean the attempt or
intention of one participant to change the behavior, feelings, intentions,
or viewpoint of another by communicative means. The last is important.
Communicative means may be linguistic or nonlinguistic (say, gestures),
but they are abstract and symbolic. A gun held to the head may indeed
induce a change in someone’s behavior, but it is not communicative in this
sense. Hence I do not consider a direct physical threat a type of persuasive
discourse. Types such as advertising, propaganda, political rhetoric, and
religious sermons clearly do fall into this category. While lectures, psycho-
therapy, and literature might belong here under some interpretations, they
are problematic and are dealt with later: they seem intermediate between
PD and OC. On the other side, while direct physical intervention is clearly
outside of our definition of PD, brainwashing is more difficult to assign to
either category. It is true that very often there is no direct physical force
involved. But in brainwashing, as the term is ordinarily used, there almost
necessarily are physical interventions— whether isolation from other
187
people and familiar surroundings, privations of numerous kinds, physical

discomfort and torture—so that, perhaps, on the grounds that brainwash-
ing is rooted, however indirectly, in nonsymbolic physical means of motiva-
tion, we ought not to consider brainwashing among the types of persuasive
discourse, although it is a demonstrably effective means of persuasion.
As we attempt to make more precise our definition of what is persua-
sive and what is not, we are confronted with a problem that has, itself,
propagandistic overtones. Perhaps because we live in a society in which
egalitarianism is upheld as a paramount virtue, we extol anything that has
the appearance of equality, distrust anything that does not. We are not
apt to ask whether there are certain kinds of activities and situations in
which equality is unnecessary or even impossible, and find it hard to imag-
ine using phrases like power imbalance, inequality, or nonreciprocity without
negative connotations. Hence, to talk about discourse as reciprocal implies
that it is somehow good, or beneficent; to call something nonegalitarian is
to imply that those who customarily utilize it are manipulative and hungry
for control. (This may, of course, be true, and certainly there are situations
where a position of power is misused or abused. It is further true that using
the surface appearance of egalitarian and reciprocal discourse for deeper
nonreciprocal and power-seeking purposes is illegitimate and deserves
censure, unless justification can be given for this deceptiveness. But dis-
course that is overtly and explicitly nonegalitarian seems not to present
any danger, nor to deserve the opprobrium heaped upon it so often.)
Additionally, for numerous reasons, there are certain cultural prefer-
ences in discourse types. Some we are prone to admire and respect; others
are illegitimate, “dirty,” debauched. We would prefer to keep our likes and
dislikes in neat, overlapping piles: what we like for one reason should be
admirable on all grounds, and vice versa. Hence, if we have been trained to
despise one type of discourse—say, commercials or political propaganda—
we would like to believe that it is “persuasive” because we consider persua-
sive discourse, since it is nonreciprocal, malign; and if a type of discourse is,
at least to its practitioners, beneficent and pure—for example, psychother-
apeutic discourse—there is tremendous pressure to deny that it is “persua-
sive,” that there is anything nonreciprocal about it, or that there is any sort
of power imbalance involved, for believing these claims would seem, to its
proponents, to vitiate the claims for benignness for the discourse type in
question.
Uncertainty and conflict arise, of course, when we simultaneously judge
a type of behavior good or bad, depending on which aspects of it we focus
on; but we have to dispense with the idea that the attribution of power
imbalance and nonreciprocity is name-calling. It is not; it is mere definition

188
and should be considered value-free, with the added assumption that some
discourse types must, to be effective, be nonreciprocal and power imbal-
anced, the only issue in this kind of word being whether it is effective, not
the value of the method by which that effect is achieved. We must try not to
heap obloquy on the commercial, and praise on the therapeutic discourse,
because of their purposes, at least not while we are trying to discover their
properties and their positions within a taxonomy, and eventually a gram-
mar, of discourse.
With this problem out of the way, let us turn to the question of defini-
tion. Some of the factors have already been alluded to at greater or lesser
length in this paper, but I summarize here all the relevant points as they
appear to me, in classifying discourse types, determining how they differ
and what aspects are universal, and differentiating between persuasive and
nonpersuasive discourse.
First, and perhaps most important, is reciprocity, about which much
has been said earlier in this paper. Connected with reciprocity is bilateral-
ity. A discourse may be reciprocal and bilateral, like OC; nonreciprocal and
unilateral, in that true participation occurs on only one side, like a class-
room lecture (though I make amendments even to this statement shortly);
or, most complex, nonreciprocal but bilateral, like psychotherapeutic dis-
course, where both parties most typically can make true contributions to
the conversations, but the contributions may be of different surface forms,
and certainly are open to different interpretations. Turn taking is a natu-
ral concomitant of reciprocity, though (as in psychotherapeutic discourse,
which involves turn taking) the two can be separated; but a non-turn tak-
ing, or unilateral, discourse can never be reciprocal.1
Discourse may be spontaneous or not, or rather, can be conventionally
spontaneous or not. Thus, OC is at least conventionally spontaneous: we
distrust apparent OC if we have reason to suspect any of the participants is
working from a script, or has planned significantly in advance. But a work
of art, or a lecture, is not supposed to be spontaneous, and the lecturer
feels no embarrassment about referring to notes. Hence, we find differ-
ences in the use of hesitation devices, pragmatic particles, cohesion, and
so forth. A spate of y’knows distresses us far less in OC than in a lecture—
and an ordinary conversation style without hesitations and other devices
reflective of spontaneity would make most of us very uncomfortable.
1. An odd apparent counterexample occurred at the end of the Phil Donahue show
(Donahue), June 16, 1981. At this point Donahue turned to the camera (not the studio
audience) and said: “We’re glad you were with us. You were terrific!” It is hard to imag-
ine how the TV audience could have demonstrated its terrificness.
189
Spontaneity is, of course, related to reciprocity and bilaterality: it is almost

impossible to plan the flow of your conversation when another participant
has as much right as you to determine its direction.
Another characteristic of persuasive discourse is novelty. Ordinary con-
versation thrives on ritual and custom: while the topics of our conversa-
tion, and the precise way we talk about them, differ from time to time, our
overall style does not shift, nor in general does the way in which a given
society holds an informal conversation change over time. The general mode
of conversation today is not, at least judging from novels and other contem-
porary evidence, significantly different from the way it was done 200 years
ago. Openings and closings—the most ritualized elements—have changed
very little over time: while colloquially we introduce new forms of these
occasionally—Hiya, ciao, and so on—we eventually return to the old stand-
bys, hello and goodbye. The rest of the conversation follows a style that can
best be described as unstylized: it has no set pattern, and hence no new
patterns can be substituted for the old. Persuasive discourse, in all forms, is
different: a defining feature of persuasive discourse is its quest for novelty.
This is manifested on the lexical level, in the form of slogans and neolo-
gisms; syntactically; semantically, in that new concepts are continually
being introduced and talked about; and pragmatically, in the way in which
PD addresses hearers, its register, its directness or indirectness, and many
other factors. What is crucial here is that PD wears out; OC does not.
A feature common to most forms of PD is that there is an audience,
rather than an addressee. Actually, audiencehood goes along with unilat-
erality: an audience is a hearer or group of hearers who play only that role,
and do not take the active role of speaker. The role of an audience is much
more passive than that of an addressee. In some forms of discourse, we
find both, at least ostensibly: in dramatic performances in any medium,
we often find conversations—involving speaker and addressee in ordinary
conversation—taking place in the hearing of an audience, which does not
participate. But much of the dialogue uttered by the participants in the
drama itself differs in striking ways from true ordinary spontaneous con-
versation. Some of the difference, of course, has to do with the fact that
the dialogue is constructed rather than spontaneous, so that many of the
uses of spontaneity are absent.2 But other differences are directly due to
the fact that the dialogue is occurring, in such situations, for the benefit
not of the immediate participants, but of the audience, and therefore many
of the contributions found in OC designed to make the other feel good, or
2. More discussion of these points can be found in Lakoff 1982.

190
inform the other about necessary facts, are absent, and other things are
present which would ordinarily not be found in OC because all participants
are already aware of them and the contribution is therefore redundant. But
the audience is not aware, and needs to be apprised of the information,
and so it is. An audience has a different role than does an eavesdropper in
true OC.
In addition to the major distinction of audience/addressee, there are
distinct types of audience, based on the role the latter is to play in the dis-
course. The role of audience ranges from totally nonparticipatory to a con-
scious and active involvement. Indeed, an audience can exist even in case
the speaker is not aware of its presence—an eavesdropper, that is, whose
role is precisely to remain undiscovered and therefore to give no clues
whatsoever as to its presence. In ceremonial functions—at a wedding, for
instance—the audience is known to be present, and indeed its presence as
witness to the ritual is necessary if the ritual is to be transacted success-
fully, but there its role ends. It does not participate, it is not expected to
understand, or signal its understanding. Typically, the audience at a wed-
ding does not indicate by any means, verbal or otherwise, its agreement or
complicity. In fact, it does not matter to the members of such an audience
if the ceremony is conducted in a language they do not understand (say,
Hebrew at an Orthodox Jewish wedding). Their role is simply to be present,
not to derive anything themselves from what they hear. On the other hand,
an audience at a classroom lecture has as part of its function to understand
and, perhaps, give approbation—in the form of evaluations later, yes, but
more importantly, immediately, in the form of nodding and other non-
verbal backchannels. A good lecturer, however large the audience, devises
ways of discovering and utilizing this nonverbal response, and without it—
say, speaking to a television audience—someone accustomed to lecturing
in the presence of an audience is lost. Hence, we often find television dis-
course being taped in the presence of a live audience, to provide speakers
with this all-important, though nonverbal and unconscious, feedback.
A third level of audience response is seen in certain forms of religious
and political gatherings: the audience is expected to participate with audi-
ble and explicit backchannels: amen/right on! or applause and cheers. The
audience here is very important to the speaker: it signals by its response
whether persuasion is occurring in these maximally persuasive forms of
discourse.
Related to the foregoing, but perceptible at a more abstract level of
analysis, is the use of the Gricean conversational maxims. The maxims
themselves are problematic for conversational analysis in that they were
formulated by Grice (1975) for quite different purposes than providing an
191
understanding of how discourse functions. Actually, in OC Grice’s maxims

are seldom encountered directly. Rather, they are understood via rules of
implicature, and any ordinary conversations that adhered for any length
of time to the maxims themselves would certainly strike participants as
strange, oddly impersonal, often literally unintelligible. By contrast, a lec-
ture is expected to adhere pretty closely to the maxims, and when implica-
ture is utilized it is for special purposes—irony, for instance—which the
audience is expected to appreciate as special. Indeed, Grice’s Conversational
Logic, as he set it out, is optimally applicable to the lecture, not the ordi-
nary conversation. When we attempt to extend the notion to other forms
of discourse, we find still more complicated difficulties, as I show in greater
detail later.
The presence of a power relationship among participants was noted ear-
lier. Important in this regard is the question of how power is determined.
It has been suggested that conversational power belongs to the one who
controls topic and floor, but there is an alternative view that power is in
the hands of whoever has the choice of whether to continue to participate,
whether to be persuaded. The hearers at a lecture may listen attentively
or may whisper and shuffle; or they might even leave, individually or en
masse, and then the lecture truly ceases to exist. Psychotherapeutic clients
may or may not continue to appear for their sessions. The audience for the
commercials may buy the sponsor’s product or not. Still, while this may in
some general sense be power, we can distinguish it from power within the
discourse: the power to motivate the discourse in a certain direction, to
begin or terminate it explicitly. And this power rests with the floor holder
(or with the therapist in the case of psychotherapeutic discourse). As with
the issue of persuasion generally, we tend to consider power an evil, and its
appearance in discourse a sign of the corruptness of the discourse. But in
fact, as long as it is explicitly acknowledged that the imbalance exists, there
is no problem.
Finally, as a distinguishing feature among discourse types, we can think
about the means of persuasion. Again, this is an area heavily laden with
value judgment, which is not helpful for our present purposes. We can make
distinctions among the means of persuasion, in those types of discourse
that fall into the range that we have called truly “persuasive”—propaganda,
advertising, and political rhetoric—and those that have strong persuasive
elements: lectures, psychotherapeutic discourse, and literature. In fact,
it can be argued that what distinguishes the first category from the sec-
ond is precisely the means—ostensible in any case—of persuasion. The
first operates by appeal to the emotions, the second—largely, or at least
theoretically—by appeal to the intellect. We tend to assign a desirable

192
connotation to intellectual persuasion, since it appears to treat us with

respect, take our most crucial human values into account, give us a real
chance to weigh and judge, and so on; while emotional appeals seem to cir-
cumvent our reason and to appeal to our base nature, giving us no chance
to make a real decision. In particular, within the last several decades, “sub-
liminal” advertising designed to present a tachistoscopic image to the mind
so that conscious perception is circumvented, leaving only the appeal to
the unconscious, by definition motivated only by appeals to emotion and
instinct, has aroused particular outrage and inspired a good deal of preven-
tive legislation, despite more recent evidence that it is less effective than its
proponents hope and others fear.
I discuss further on the use of Gricean maxims in persuasion. It is
important here to note that the Rules of Conversation are perhaps decep-
tively applied, or non-applied; but they are certainly invoked, because they
constitute, for the audience, evidence that we are being persuaded by rea-
son, intellectual argumentation. Hence the appearance of conformity to
the Gricean maxims is critical if we as newly sophisticated consumers are
to be subliminally seduced. The appearance of reason conceals the appeal to
emotions and justifies it for the buyer.
Indeed, all forms of persuasive discourse have gone through changes
over time—partly necessitated by the requirement of novelty, but also in
part by the increasing sophistication of the consumer and the need to pres-
ent products as reasonable ones to purchase. The change in the realm of
advertising is especially interesting. In older advertising (say, up till the
1930s, roughly speaking), there was a heavy preponderance of print, of
words, and far fewer and less striking pictures. This was in part due to the
state of reproductive techniques: photography was rather primitive, color
reproduction ruinously expensive, and graphic techniques relatively unso-
phisticated, so that, for the money, words were the most economical means
of persuasion. But the words, if we examine them, were far more directly an
appeal to the emotions than is wording—or, perhaps, even illustrations—
in advertising today. Part of the change is due, of course, to increasingly
vigilant regulation by federal agencies concerned with truth in advertis-
ing. But ironically—and not surprisingly, given our bias toward seeing
ourselves as logical and rational—regulations have almost invariably been
framed in terms of the wording used in advertising, at most encompassing
explicit illustration: marbles in the soup, for instance. You cannot, in an ad,
say in words, “Glotz Detergent will make your marriage happy”: the Federal
Trade Commission (FTC) will come down hard on you. But you can say in
words, “Glotz gets your husband’s shirts their whitest,” alongside a pic-
ture of a young, vital couple glowing at each other, surrounded by cheerful
193
children, a dog, a white picket fence. The irony is that, in fact, the second
approach is far more effective than the first, since it circumvents intellec-
tual judgment (“How can using a detergent make Harry more responsive
to me?”) and goes directly to the realm of the unconscious, capable only of
desires, fears, and needs.
If much of this discussion has the ring of a psychology textbook, many
modern advertising techniques trace their genealogy directly to the far-
famed couch in Vienna. Indeed, much of modern motivational psychology,
the basis of advertising, derives from Freud, directly or indirectly. For it
was Freud ([1900] 1953) who pointed out the basic distinction between
the processes of the conscious and the unconscious mind. The unconscious
works by the laws of the primary process; the conscious, by the secondary
process. Primary-process thought is preverbal—symbolic, nonsequential,
and visual—while secondary-process thought, more directly “rational,” is
auditory and verbal. Hence, if one wants to persuade by circumventing the
processes of rational thought, it makes good sense to emphasize abstract
images—music and pictures, for instance, rather than words in logical
sequence. And if the FTC is concerned with deception in advertising, it
would do better to pay attention to the nonverbal means of persuasion,
which can be much more deceptive. But the problem is, of course, that it is
difficult for the investigator to prove what the picture of the happy family,
or the vital adolescents guzzling Coke, is communicating, since we think of
communication in terms of logical symbols—words.
Edward L. Bernays (perhaps not coincidentally Freud’s nephew), the
inventor of modern public relations and hence of many of the modern
forms of persuasion, spoke of “the engineering of consent” (1952). It is this
that upsets us as consumers and intrigues us as investigators. If we are so
suspicious of persuasion and its techniques, how are we “engineered” to
give consent? Bernays referred not only to the relatively harmless influence
of advertising, but also to the more baneful effects of political rhetoric,
including propaganda. The latter term itself can be given some scrutiny.
Originating in a religious context, with positive connotations (the
“propagating” of the faith), the term propaganda eventually came to mean,
viewed from the perspective of another religion, a form of improper influ-
ence or pressure. Hence propaganda today is exclusively a pejorative term
(though Bernays makes a plea for its rehabilitation).3 One question is
3. Curiously, the Compact edition of the Oxford English dictionary ([1933] 1971) has a
definition of propaganda that is wholly without negative connotations—quite differ-
ent, I think, from its normal use: “Any association, systematic scheme, or concerned
movement for the propagation of a particular doctrine or practice.”

194
whether the term really has true denotative meaning, as a special kind of
persuasion identifiable even when the argument is one with which we are
in sympathy; or whether propaganda simply means emotional persuasion,
when the argument is not one we approve of. We would agree that pro-
paganda is mainly applicable to forms of persuasiveness that utilize emo-
tion, usually of a high intensity, often invoking fear and irrational desires.
But advertising typically does this, yet is not considered propaganda. So
we might want to add the proviso that propaganda concerns changes in
beliefs, rather than concrete buying habits alone. Yet we might argue that
much advertising is propagandistic—not only the “public service” advertis-
ing and institutional advertising by power companies, but even advertise-
ments for soft drinks that suggest indirectly that youth and a svelte body
are minimal conditions for being allowed to exist, or that create and rein-
force all sorts of traditional sexual and other stereotypes. This then sug-
gests another facet of propaganda: it is normally indirect. It is not present
in the explicit message, but somewhere in the presuppositions. When we
are serenaded to the effect that “Coke adds life,” the propaganda is not in
the admonition to buy Coke, but in the inference that youth is desirable
(and young people drink Coke).
This brings us back a bit roundaboutly to the relation between the
Gricean maxims and discourse types. I have argued that the Cooperative
Principle (CP) is more directly applicable to the classroom lecture than to
any other type of discourse—certainly more than to ordinary conversation.
But even OC makes reference to the maxims, if only via implicature. And
in OC, we understand that flouting of the maxims is due to our desire to
adhere to more socially (as opposed to intellectually) relevant rules, rules of
Rapport: when we have a choice between being offensive and being unclear,
we invariably choose the latter, and a majority of cases of OC implicature
can be seen to stem from this assumption.
But in persuasive discourse, the situation is rather different. Whereas
in OC and the lecture, our aim is to inform—at least, it is so, other things
being equal—in PD our aim is, of course, to persuade. The politician does
not especially want a knowledgeable electorate: he wants votes. The adver-
tiser is not interested in educating people about hygiene: he wants to
sell deodorant. It is not that the maxims are violated only in case a more
peremptory need intervenes, as in OC—they are regularly infringed with-
out explanation, cue, or apology.
Indeed, it might be argued that the PD need for novelty alone is respon-
sible for many instances of violations of the maxims, especially in advertis-
ing. For in OC, we have seen, novelty is not especially valued. Part of the
195
reason for this is that familiarity aids intelligibility—that is, aids in the
keeping of the Cooperative Principle. What is new and requires interpreta-
tion is in violation of the maxim of Manner (be clear). But in PD it is this
very violation that is striking, memorable—efficacious.
If we understand this preference for novelty as an intrinsic aim of PD,
we can perhaps understand better why advertisers cling to certain formu-
las despite—or rather, because of—the contempt to which they are sub-
jected by critics. For example, many of us remember a slogan from the
fifties: “Winston tastes good like a cigarette should.” The criticism heaped
on the hapless preposition was staggering, yet the commercial continued
to appear in that form. (Lip service was later paid to linguistic chastity by
the addition, after the infamous slogan, of the rejoinder, “What do you
want, good grammar or good taste?,” which was cold comfort for traditional
grammarians.) We can understand the company’s clinging to the solecism
if we understand it as a Manner violation: focusing the hearer’s (or reader’s)
attention on style tended to obscure content, and thus to flout Manner, if a
bit indirectly. Indeed, anything neologistic will have much the same effect,
and will serve as good persuasion for two reasons: first, because it vio-
lates Manner as just explained, and thus attracts the audience’s attention;
and second, in the case of obvious neologism—as opposed to the simple
ungrammaticality of like for as—it forces the audience to interpret—as any
violation of the Cooperative Principle does.
It is axiomatic among proponents of all forms of persuasive or semi-
persuasive discourse (e.g., psychotherapeutic discourse and the classroom
lecture) that if the audience can be made to participate at some level—
that is, to function as addressee, not wholly as audience—learning, or per-
suasion, will be much more successful. Hence, if the audience is forced to
interpret neologisms, or relate “ungrammatical” forms to their textbook
version, they will probably remember better. And memory is the name of
the game in PD.
We find violation of the Cooperative Principle not only in Manner, but
in other maxims as well. In terms of neologisms and other sorts of linguis-
tic novelty—most easily considered as Manner violations—we find:
1. Lexical novelty (neologism):

stroft, a portmanteau of strong and soft; devilicious.
2. Morphological/Syntactic novelty (in terms of category shifts: like
for as may belong here):
Gentles the smoke and makes it mild
Travels the smoke further

196
The soup that eats like a meal

Peanuttiest
3. Syntactic innovation:
(Some of these can also be viewed as Quantity violations, in that
either not enough or too much information is given for ordinary
conversation understanding. These include such odd usages as the
following.)
3a. Absence of subjects and often the absence of verbal auxiliaries
as well:4
Tastes good! And nutritious too!
3b. Odd uses of the definite article, which is sometimes unexpectedly
inserted and is sometimes unaccountably absent:
Next time, I’ll buy the Tylenol!
Baby stays dry! Diaper keeps moisture away from baby’s skin!
4. Semantic anomaly (other than lexical anomaly):
(These include quantity violations, among other things.)
Cleans better than another leading oven cleaner.
Works better than a leading detergent.
In instance 4 we expect, because of Quantity (or perhaps Relevance), a

definite article in the item being compared with the product being touted.
Otherwise, the information is noninformative, or useless: after all, what
we need to know is which is best. Here is a good example of persuasive
discourse adopting the surface trappings of informative (i.e., Cooperative
Principle–obeying) discourse and thereby leading us to conclude that it
is informative (if it looks like a duck, and walks like a duck …). Hence
we understand it as informative, that is, in keeping with the CP. Notice
that when a maxim is infringed in OC via implicature, we do not interpret
the utterance as being in keeping with the maxim by virtue of its surface
appearance. Rather, in these cases we are given signals by the infringing
speaker that the contribution is in violation of the CP, and we are thus
implicitly directed to put our interpretive skills to work. In PD, on the other
hand, the flouting of the maxim is covert, and we are tricked into assuming
that an act of information is taking place in cases like this, where in fact it
is not.
4. An added benefit of this form of utterance is that it mimics a casual register and
thus suggests folksy informality. Comparison of these segments with true OC casual
register makes it clear, however, that the truncations found in commercials do not
occur in OC. One more advantage, of course: a micro-millisecond saved is a tidy sum of
money earned, given current rates on prime-time television.
197
5. Pragmatic novelty:
(This includes anything aberrant about the discourse form itself.
In particular, since many commercials are framed as minidramas,
we see many unusual bits of dialogue within these thirty-second
segments. One such type much in vogue lately is the following.)
—You still use Good Seasons Italian?
—Not any more!
—No?
—I use new, improved Good Seasons Italian!
Now in terms of ordinary conversation, the second speaker’s contribu-

tion violates Quantity, and would probably be treated not as the joke it
functions as in the commercial but as a rather stupid bit of obfuscation, if it
were to occur in real OC. This is generally true of the “jokes” highly favored
in recent commercial genres, especially coffee commercials:
—Fill it to the rim!
—With Brim! [laughter]
Wife [with camera]: Give me a smile!

Husband [at breakfast table]: Only if you give me another cup of your
coffee.5
Humor, in ordinary conversation, can often be viewed as a permissible

Manner violation. But the humor here is of a rather different order, espe-
cially as it seems both strained and puerile. This may in part even be pur-
poseful: reassuring the audience that the folks in the commercial are no
different from them, they make awkward jokes too, and this just shows they
are good, spontaneous people. That this seems especially characteristic of
coffee commercials, I think, is because the ambiance of these commercials
is intended to suggest a sort of easy-going informality where joking of this
kind is in place. But beyond this, these jokes fulfill another communicative
purpose: if we assume that, unlike OC, the major purpose in commercials
is to ensure that the sponsor’s name is remembered, rather than creating
5. The use of your here is typical of many kinds of commercials (preceding the name
of the sponsor’s product, or its generic category, as here) and is characteristic of the
genre and not OC. Normal here is “another cup of coffee,” perhaps “that coffee.” Your
in this environment (a quantity violation in that it provides unnecessary information)
perhaps is intended to suggest the addressee’s responsibility for the adequacy of the
product.

198
an easy and natural atmosphere of warmth (but, in many types of commer-

cials, the summum bonum is getting the sponsor’s name memorably associ-
ated with an easy and natural atmosphere of warmth—that is, superficial
rapport), these “jokes” see to it that the sponsor’s name gets mentioned in
a prominent position, usually at the very end of the commercial, and in a
loud voice with stress—as we stress the punchline of a joke more than the
end of a normal sentence.
Another pragmatic anomaly is seen in unusual patterns of intonation.
For example (a commercial for Sunrise coffee):
I like it I’ll drink it
with no pause between the clauses. In fact, the stress of this utterance is
what we would expect if we had a syntactic dependent clause at the left
(e.g., “Since I like it so much …”) The intonation here is steady rather than
falling, as would be normal for two syntactically independent sentences.
I think the only way we can make sense of this is as an attention-getting
novelty.
It seems that in the past, most advertisers focused their attention on
lexical and syntactic novelty, while more recently, pragmatic novelty seems
to be favored. What we may be seeing here is metanovelty: the audience
is jaded with the older forms of newness, and advertisers must press ever
onward into the more mysterious reaches of language to get a response
from us at all.
Some of the novelties I have been talking about can be justified on addi-
tional grounds, beyond their usefulness as floutings of some part of the
Cooperative Principle. For example, many of these work for brevity, a real
desideratum when the point must be made in thirty seconds. But I think
that brevity alone seldom justifies any special usage. Rather, all the special
features of PD are overdetermined.
I have illustrated the flouting of the Cooperative Principle with regard to
advertising, in which it is particularly glaring and has, indeed, been exalted
to an aesthetic feature of the medium. But it can be found just as easily in
other forms of PD. Its function in psychotherapeutic discourse is special,
and is not dealt with here. But certainly we are accustomed to it in political
rhetoric—especially in the form of hyperbolic violations of Quantity, and
metaphorical floutings of Manner. It is noteworthy that, of all the forms
of violations of the CP that are found and tolerated in PD, Quality alone
is missing. Quality violations are what bring the FTC down on the spon-
sor, impeachment threats down on the politician. The others seem beyond
reproach. We may want to examine again the assumption that Quality is on
199
the same level as the other maxims, when in fact we feel that abrogation of
this maxim has very different psychological and social implications. Indeed,
violation of the other maxims can always be justified for either aesthetic
or social reasons, without further ado; but we have only a subcategory of
Quality violations that are tolerable—and only dubiously so: the “fib” or
“white lie.” In fact, Quality violations are the only ones ordinary language
has a separate word for—and a word with bad connotations at that.6
I have referred to the importance of Rapport rules and strategies in OC,
and their superseding of the Cooperative Principle when they conflict with
it. Since PD is nonreciprocal and has an audience, rather than participating
addressees, it is not surprising that Rapport as such plays no role. We do
find, in some types of commercials, attempts to establish a one-to-one rela-
tionship with the audience on television: eye contact and twinkling, casual
register, and the like. But in fact, most of the conventions of OC that exist
for Rapport purposes are absent from commercials, and from most forms
of PD as well—openings and closings, for example, are either totally miss-
ing or heavily truncated (as in psychotherapeutic discourse).
All this discussion raises a very troubling issue in discourse analysis of
the kind I am attempting here: What is the role of the CP in persuasive
discourse, and especially in advertising? I have spoken as if the CP were
involved in our understanding of commercials as it is in OC or the lecture—
and yet we saw it was utilized very differently in even these two types of
discourse. Certainly the CP gives us a much-needed and illuminating han-
dle on the workings of commercials and our understanding of them. But at
the same time, invoking the CP as the basis of our understanding of a dis-
course type makes an implicit claim: since the CP enjoins us to be as infor-
mative as possible, to say that our discourse is understood in relation to it
is to imply that the true underlying purpose of that discourse is to inform;
and that, if information is not exchanged in some utterance in an opti-
mal way, there is a special reason for it—politeness, for instance, in OC.
This assumption works splendidly for discourse that is explicitly intended
above all to inform—that is, of course, the lecture. But even with OC, it
gets us into trouble. For the main purpose of OC is not to inform, or even
to exchange information. This can occur, and conversations frequently are
loaded with useful information, but most often the information serves
only as a sort of carrier, enabling the real business of the interaction to get
6. And only with Quality do we find it necessary to differentiate between purposeful

(“lying”) and accidental or neutral (“misinformation,” etc.) violations. For more dis-
cussion of the linguistic and communicative problems about lying, see Coleman and
Kay 1981.

200
done—interacting. Hence Rapport supersedes the Cooperative Principle,

since Rapport is the point in OC. Then we might argue that OC is not really
CP-based, but Rules of Rapport-based, with the CP a mere auxiliary. This
is unsatisfactory, though, in that we invoke the mechanisms of the CP
to account for our understanding of the utterance—implicature is, logi-
cally if not psychologically speaking, secondary to the operations of the
Cooperative Principle.
The problem with persuasive discourse, however, is more complex still.
We are not intended to understand PD through the use of implicature;
there is no apparent reason that might justify infractions, yet infractions
are common and, as we have seen, quite different from the sort we get
with OC. Yet in PD there is an appeal to our knowledge of the workings
of the Cooperative Principle. It is our very awareness of its being violated,
in such unexpected and inexplicable ways, that creates the memorability
and effectiveness of all forms of PD, especially advertising. It is not clear,
then, whether we should say that PD, like OC, is predicated on a base of
the CP, but merely is expected to be in violation of it (it is not really clear
what this would mean), or whether a completely different basis must be
proposed. But if the latter, how do we account for our recognition of the CP
through its flagrant violation in PD? Or our insistence on interpreting PD
as if it were in accord with the CP, although it is not (which, as we have seen,
makes it persuasive)? I am not sure how to resolve this issue, and present
it here as something that will have to be determined by future research, if
we are to devise a taxonomy of discourse that accounts not only for surface
form, but for deeper intentions and the relations between the two.
In any event, we have made a beginning, I think, and some interesting
facts as well as problems have emerged. We have seen that there are valid
bases on which to distinguish between ordinary conversation and other
types of discourse, and between truly persuasive discourse and intermedi-
ate types. I have discussed the meaning of “the engineering of consent,”
and argued that it has to do with the manipulation of our expectations
about the form and function of OC, translated into PD: both surface form
and deeper intention, in terms of the Cooperative Principle, can be turned
to persuasive effect. I have argued that we must take a number of factors
into account in making these determinations: reciprocity and bilaterality;
spontaneity and novelty; power; the function of the audience/addressee;
the means of persuasion; and finally, the use of the Cooperative Principle
and the Maxims of Conversation.
In short, many discourse types which superficially look similar, at a
deeper level of analysis function quite differently; and many types which
look different turn out to have deeper similarities. We cannot understand
201
discourse until our classification includes, and compares, form and func-
tion together, and we can hope to have a satisfactory grammar of discourse
only when we have arrived at a valid taxonomy. This brief discussion of
one parameter—persuasive and nonpersuasive discourse—is presented as
a beginning.
REFERENCES
Bernays, Edward L. 1952. Public relations. Norman: University of Oklahoma Press.

Coleman, Linda. 1983. Semantic and prosodic manipulation in advertising.
Information processing research in advertising, ed. by Richard J. Harris, 217–40.
Hillsdale, NJ: L. Erlbaum Associates.
Coleman, Linda, and Paul Kay. 1981. Prototype semantics. Language 57:26–44.
Donahue. 1981. Episode first broadcast 16 June 1981 by WBBM. Directed by Ron
Weiner.
Freud, Sigmund. (1900) 1953. The interpretation of dreams. The standard edition of the
complete psychological works of Sigmund Freud, ed. and trans. by James Strachey,
vols. 4–5, 339–630. London: Hogarth Press.
Lakoff, Robin. 1982. Some of my favorite writers are literate: The mingling of
oral and literate strategies in written communication. Spoken and written
language: Exploring orality and literacy, ed. by Deborah Tannen, 239–60.
Norwood, NJ: Ablex.
Sayers, Dorothy L. ([1933] 1986). Murder must advertise. New York: Harper & Row.
The compact edition of the Oxford English dictionary. (1933) 1971. Vol. 2. Oxford &
New York: Oxford University Press.

202
203
Introduction to “Techtalk
as group talk”
BY SUSAN M. ERVIN-T RIPP
Members’ knowledge of how Members behave is so organized that items of that knowl-
edge may, discriminatively, be taken as expectably descriptive if no more than a single
referentially adequate category has been asserted to hold for some Member in question.
It is that Members have such categorically localized knowledge that provides for what it
is they employ to make the assessments just referred to.
Harvey Sacks, “An initial investigation of the usability of conversational
data for doing sociology” (1972)
The SPARCclassic system with its super aggressive price points is targeted at the client
portion of the downsizing/rightsizing phenomenon that is occurring in the mission criti-
cal business application markets… . This model advertises the most aggressive street
price possible to gain mindshare in the market… .
From a SUN workstation ad in 1992
I n this brief paper, Robin Lakoff raises the question of how technical
writing becomes incomprehensible and brings to bear two other per-
spectives—the social dynamics of in-group talk and the place of gender in
the social construction of talk.
We all know what technical writing is and teach it around the world
as language for special purposes. Students are to use arcane vocabulary
and the syntax of impersonality—the passive and the generalization—to
sound scientific. We intend this kind of language to obscure the voice and
204
stance of the writer, to prohibit both emotional and evaluating perspec-

tives, to focus on the object rather than the observer. American journalists
separate editorials, opinion columns, and news articles to maintain this
idealized segregation between opinion and impersonal and objective fact.
Lakoff starts with the discrepancy between the goal of rational clar-
ity and the use of both vocabulary and syntax that interfere with intel-
ligibility even when there are familiar choices possible. For example, “It
was hypothesized that in social settings the initiation of dyadic interaction
would increase with the density of interactional opportunities.” Does this
say more than “We think people make friends faster at a crowded party?” In
this paper, Lakoff finds an explanation for these often ludicrous extremes
in social dynamics, in the display of group membership, exclusion, and
power. Then she asks a more speculative question about the relation of tech
talk to gender.
It turns out that impersonality is a dominant cluster of features when
one statistically analyzes a sample of texts (Carroll 1960; Biber 1995:112ff.).
Texts rated more “impersonal” in Carroll’s study had few pronouns and pri-
vate verbs, but many common nouns, prepositions, longer words, Latinate
verbs like problematize, and passives like it was felt that.. Biber’s larger
project also found less that-deletion, a higher type-token ratio, and more
attributive adjectives in texts high in this dimension. So these studies con-
firm that technical language is really distinctive and that its features hang
together.
Biber’s and Carroll’s studies of a wide variety of texts also cast indirect
light on Lakoff’s speculation about gender. Biber found that the emphat-
ics and hedges that Lakoff has suggested were typical of women’s speech
are personal and hence infrequent in technical style. Carroll’s raters were
asked to rate passages on various scales, such as “masculine” and “imper-
sonal.” While impersonality judgments of texts correlated with the linguis-
tic features above, the text features correlating with masculinity judgments
turned out to be “serious topic or key” rather than linguistic style features.
Is there reason to believe that the peculiar vocabulary and syntax of
technical language are affected by the social dynamics that lead to similar-
ity in speech in groups? There are three correlated but empirically separable
social factors in speech similarity (Ervin-Tripp 1973). One is frequency of
communication, accounting for regional features. A second factor is close
social ties, solidarity or cohesion, in what Milroy (1980) calls the “social
network.” Intimate friends who talk a lot can guess what is missing in their
friend’s text, for instance, suggesting convergence in themes as well as in
lexicon and phrasing (Salzinger et al. 1970). A third factor is identity mark-
ing, which explains why people within families may have different speech
205
features according to age and gender. Labov’s famous paper (1963) on why
young people on Martha’s Vineyard who wanted to stay on-island had dif-
ferent vowels than those aiming not to stay, even to the point of hyper-
correction, is a good example of these factors, and more recently Eckert’s
(1989) work on high-school cliques analyzes phonological effects of social
structure. You could talk often to someone in an emotionally neutral, busi-
ness relationship, or be close to a lover or grandparent with whom you
don’t identify, or identify with someone you don’t know personally.
It is a speculation that speech similarity and writing similarity have
the same social dynamics, but presumably modeling at a distance could be
more possible in writing. That is, without meeting a writer, one could read
enough to identify with him or her and emulate stylistic features. Could
the reverse happen; can copying lead to identification? We know students
can copy styles; this is a standard assignment in some writing classes. How
writing is related to identification remains to be studied.
Specialized language develops—and here I mean especially vocabulary—
whenever a group speaks more to one another than to nonmembers, or
whenever there are new activities or contexts. This happens in both closed
networks of close friends and in open networks in cities. We see these sim-
ilarities whether with Valley Girl accents, group slang, or morphological
or syntactic preferences. The spread of such features is primarily through
contact and emulation. Rampton (1991) even notes borrowing of Pakistani
words and phrases for particular functions by English teenagers as a sign
of group solidarity.
As Lakoff points out, the semantic utility of new vocabulary is only a
part of its function. Nobody is surprised by terms like phonology or mor-
pheme, DNA or neutrino. These terms were necessary for progress in talking
about new concepts and means of analysis. Technical terms when needed
are of high value, and hijacking of them by outsiders creates problems. An
example is the theft of neural, which at one time referred to the nervous
system, but now appears in electrical-engineering and machine-tool jour-
nals. Now biologists use the phrase neuronal network instead.1
What Lakoff points out, however, is that the new vocabulary can go well
beyond what we need to identify new concepts, into renaming old concepts
by new names. What does this accomplish? Most obviously, it marks the
boundary between insiders and outsiders.
1. Even word choices have a social component. Quark (from Finnegan’s Wake) entered
physics because of the literary tastes of Murray Gell-Mann. His preference defeated its
competitor ace, which was offered by a physicist of less prestige.
D O U B L E TA L K : S E X I S M I N T E C H TA L K [ 205 ]
206
In-group vocabulary is clearly a boundary marker. In the case of pick-

pockets (Conwell & Sutherland 1937:44–46), there is a very old, rich
descriptive vocabulary that can be summarized. Here’s how the cannon
works. First one of the mob may fan the sucker to locate his wallet, then
one of the stalls prats the sucker and the hook gets the poke while shad-
ing the duke, and then cleans in case he gets tumbled. A person who can-
not understand terminology is an outsider. “When thieves are talking in a
place where others may overhear, they do not use their slang, for it would
immediately attract attention to themselves. They use regular words with
inflections or winks which indicate their meaning” (p. 18).2 Since knowing
argot can be a clue of membership, audience is a crucial deciding point on
whether to use marked vocabulary to display membership openly, or use
dual meanings of ordinary words and maintain secrecy. Beginners’ failure
to attend to who was listening was dangerous. So thieves’ argot is used for
clarity and boundary testing with insiders, but for secrecy is hidden from
outsiders. The major difference between thieves’ argot and technical lan-
guage is the prestige of the experts before outsiders.
Euphemism is the selection of new vocabulary without the emotional
loading of earlier synonyms, as well as larger ways of talking. Jessica Mitford
(1963) found many euphemisms in funeral-home speech, such as the prepa-
ration room rather than the morgue, the slumber room rather than the laying-
out room, the vital statistics form rather than the death certificate. These are
euphemisms aimed at outsiders, the customers. The Nazis had well-developed
euphemisms and circumlocutions, such as saying “the visitors” were to be
“resettled in the East” (Arendt 1963:118, 147). What is not clear is whether
euphemisms designed for the hearing of outsiders also protect insiders from
the implications of their policies and actions. Technical language provides a
similar protection in the case of emotionally loaded content.
These functions of solidarity, secrecy, and euphemism motivate the use
of argots in informal groups. What seems to be different about technical
language is that the users have high prestige and technical language is
explicitly taught. Further, the users do not need to conceal or accommo-
date to audiences, because of the high prestige of technical or scientific
membership. We would normally consider it rude to speak a language in
the presence of non-speakers, but users of technical language treat this as
a right of power, as in Lakoff’s example of the doctor who had supposed a
PhD was an in-group member and then refused to make the effort to trans-
late to the intruder.
2. See also Cory 1951 on gay vocabulary.
207
So something is different about technical language: it has value to dis-

play to outsiders. It is used as a badge of pride; indeed, doctors use it in
writing prescriptions and take pride in illegibility. I have had to explain to
x-ray personnel what was on a doctor’s order that the addressee couldn’t
read! Authority and intelligibility conflict in a number of situations, in reli-
gious contexts and sometimes in public-health contexts. John Gumperz
reports that in India public-health speakers speak the higher-status variety
to be believed, but the local variety to be understood.
Lakoff speculates that technical language is harder for women. Yet
many technical writers are women. My guess is that the clash mostly
has to do with relations to the audience, to this issue of authority versus
intelligibility, so it might show up more in talk. On average, women give
more feedback responses to others and are more likely to accommodate
to audiences. In medical settings, they give more explanations. In con-
versation, they accommodate to another’s topic more often. My guess
is that there are fewer women than men who would be willing to speak
in a style that clearly is unintelligible, to impose control by displaying
authority to make the power play that Lakoff identifies. As writers,
are women more willing to popularize their work, risking the criticism
Margaret Mead got from her colleagues? This is a prediction one could
test. Maybe there is more than one way to “gain mindshare in the mar-
ket.” In this, as in so many domains, Lakoff has identified major issues
for future work.
REFERENCES
Arendt, Hannah. 1963. Eichmann in Jerusalem: The banality of evil. New York: Viking.

Biber, Douglas. 1995. Dimensions of register variation. Cambridge, UK: Cambridge
University Press.
Carroll, John B. 1960 Vectors of prose style. Style in language, ed. by Thomas Sebeok,
283–92. Cambridge, MA: MIT Press.
Conwell, Chic, and Edwin H. Sutherland. 1937. The professional thief, by a professional
thief. Chicago: University of Chicago Press.
Cory, Donald Webster. 1951. The homosexual in America: A subjective approach.
New York: Greenberg.
Eckert, Penelope. 1989. Jocks and burnouts: Social categories and identity in the high
school. New York: Teachers College Press.
Ervin-Tripp, Susan M. 1973. Sociolinguistics. Language acquisition and communicative
choice: Essays by Susan M. Ervin-Tripp, ed. by Anwar S. Dil, 302–73.
Stanford: Stanford University Press, 1973.
Labov, William. 1963. The social motivation of a sound change. Word 19:273–309.
Milroy, Lesley. 1980. Language and social networks. Oxford: Blackwell.
Mitford, Jessica. 1963. The American way of death. New York: Simon & Schuster.
D O U B L E TA L K : S E X I S M I N T E C H TA L K [ 207 ]
208
Rampton, M. Benjamin. 1991. Panjabi in a British adolescent peer group. Language in

Society 20:391–422.
Sacks, Harvey. 1972. An initial investigation of the usability of conversational data
for doing sociology. Studies in social interaction, ed. by David Sudnow, 31–74.
New York: Free Press.
Salzinger, Kurt, Muriel Hammer, Stephanie Portnoy, and Sylvia K. Polgar. 1970.
Verbal behaviour and social distance. Language and Speech 13(1):25–37.
209
CHAPTER 9
Doubletalk
Sexism in tech talk (1983)
L anguage use itself is a technology, if not a science—even though it is

an art at the same time. Language is a tool, as piercing and illuminating
as a laser beam, and with as great a potential for good or harm as nuclear
fission. But language use is changing, and what these changes can tell us
about our own hopes and expectations for change is rather unexpected.
Every field has its own special vocabulary of “technical terminology”;
this involves not only vocabulary but style as well. We see it everywhere—
in the professions, in government bureaucracy, in the physical and social
sciences, in crafts, and in industry. Technical terminology in all fields
shares some salient features: either entirely new words are created or
common words are given new senses, so that the expert is sharply dif-
ferentiated from the nonexpert because only the former can use and
understand the language. And the language itself is specially devised,
consciously (or more often not), to circumvent emotion or subjective
judgment.
The final recourse in this type of language is pure intellectual logic, with
the rationalization that thereby communication is rendered more direct
and clear, and less subject to the confusions and ambiguities that plague
discourse conducted in ordinary language. But this is only apparently so: if
This paper originally appeared in The Technological Woman: Interfacing with Tomorrow,

edited by Jan Zimmerman, 38–43. New York: Praeger, 1983. Reprinted here with per-
mission of the publisher. Praeger is an imprint of Greenwood Publishing Group, Inc.,
Westport, CT.
210
we look at real perception of these special languages both from without

and—more surprisingly—from within, we find that far from facilitating
intelligibility, technical discourse actually impedes it. We see, for instance,
lawyers totally at odds with one another, unable to resolve confusion by
logical strategies in determining the sense of legal documents. We see med-
ical experts in court fencing with one another over fine points of medical
parlance. We see important contemporary books written in academese so
turgid that exegesis is necessary. Thus, we cannot accept at face value the
justification for technical parlance that it is a clear mode of exposition. It
may be, but just as often it is not.
Indeed, governmental bureaucrats make a virtue of obfuscation. If out-
siders, not sure precisely of the meaning of friendly fire or supply-side eco-
nomics or cultural deprivation, are present, insiders can persuade them to
agree to truly horrific policy decisions. And in all these cases it is impor-
tant to realize that it is the very de-emotionalization of the language,
its vocabulary as well as its syntax—for instance, the use of passives or
impersonal expressions like one—that bleeds meaning from the text and
muddies the intellectual content by disavowing the emotive content of the
communication.
LANGUAGE CODES
In fact, technical jargons or “codes,” as sociolinguists refer to these forms

of language, serve three purposes which are so valuable to their users
that codes persist in the face of criticism by outsiders. First, codes have
a semantic use: to fulfill their ostensible function of enabling speakers
to express complicated ideas more clearly, directly, and succinctly. This
is certainly true to some extent. Much special vocabulary does, indeed,
save time and avoid ambiguity in specialized contexts for sophisticated
users. But it is far from always the case, and certainly valid criticisms
are made of lexical preciousness in many codes: bureaucratese, medica-
lese, and educationese are among those most rightfully criticized in this
regard by outsiders, but any code can lapse into murk. Further, it is hard
to see what the nonlexical characteristics of the code—impersonality and
hedging, for instance—contribute to understanding, yet they are virtu-
ally universal.
The other two functions of technical codes are pragmatic rather than
semantic. They concern not the message itself, but the interaction of which
it is one part: the negotiation of a context, and the establishment of a rela-
tionship among participants.
211
Second, the “secret handshake” function of a code language tests and

affirms the unity of the participants. To speak in a code to another is to say,
“We speak the same language, we’re the same kind of people, a cohesive
body.” Codes, then, serve the same functions as professional society meet-
ings; they are socializing forces, uniting members of a discipline.
Third, codes are an exclusionary mechanism, the other side of the coin of
the “secret handshake.” If one person’s recognition of a code phrase affirms
his or her membership in my group, another’s nonrecognition of it tells me
that person is not one of us, cannot join our secret club, and will not be told
the code words.
A colleague years ago, working among many medical professionals (him-
self merely a PhD linguist), was concerned about a serious condition in his
infant son and worried about his own physician’s decision about the treat-
ment. So he telephoned one of the medical doctor’s colleagues, a specialist
in the condition involved. “This is Dr. Z,” he said, “and I’m wondering what
you would do in this case …” and described the situation in the medical
parlance he had, by this time, learned well. He went on that the attending
physician, Dr. X, had suggested such and such, did this man agree? The
physician reflected briefly, then said that well, that was one possibility, but
as for himself, he felt that such a decision was irresponsible, demonstrating
startling ignorance of recent literature, “because, Dr. Y., as you know …”
and launched, at that point, into a spate of pure medicalese. The linguist
at that moment was lost, and asked for definitions. The other doctor was
shocked at such an elementary question. “But aren’t you a physician?”
he asked. “No,” said the linguist, “just a PhD in linguistics.” “Well,” said
the medical doctor, “in that case, forget everything I’ve just said,” and
hung up precipitously. Here is the exclusionary function of the code in its
purest form.
In its exclusionary function, then, a technical code creates in its users
the presumption that nonmembers are not to be treated on the same level
as members. It is language that makes us fully human; if you can’t talk the
language, you will be excluded from serious interchange with those who
can. And those who do the excluding will feel morally justified in doing so.
TECHNOLOGIZING THE HUMANITIES
With this in mind, we can look at academic discourse as a technical code,

particularly the special language of the humanities that has recently
evolved. It has been unquestionably true until very recently, and at its
upper levels it is still arguably true, that the academy is a male preserve
D oubletalk : S exism in tech talk [ 211 ]

212
not unlike the church, the army, medicine, law, or government. We would
expect the chosen code of the profession to reflect this bias, and indeed it
does. In Western culture, the stereotypical conventionalization of behav-
ior toward which men are generally socialized is distancing: a power-based
strategy involving impersonality and nonemotionality, the denial of the
interaction and its effects on the participants.
Women, on the contrary, have traditionally been encouraged toward a
deference strategy, oriented not to power, but to acceptance or love, and
the fear of its loss. Hence, women’s language allows for openness to others’
needs and a reluctance to acknowledge one’s own needs. But as a strategy
concerned with acceptance and rejection, emotional expression is valued
over intellectual clarity in case of conflict.
I am not suggesting here that either of these strategies is in any sense
better, or that the users of either are better people; and less still that it is
male and female biology or physiology that motivates us in this direction.
But I do suggest that these have been, and for most people still are, the
prevalent patterns. And even those of us who have struggled hardest to
rid ourselves of stereotypical behavior often find ourselves confronted by a
society that still holds conventional models up to our children.
What is important in all this is that technical codes, as they were
described earlier, can be understood as exaggerations of masculine distanc-
ing, or power-oriented style. Hence bureaucratese and its relatives make
outsiders angry precisely because they implicitly involve a power state-
ment: “I (the user) am more powerful and valuable than you (the nonuser).”
And since it is implicit in the style, not explicitly present, it cannot be nego-
tiated. So, as outsiders faced with this kind of language, we feel impotent
and resentful.
Academese is no different. As the language of the male power elite, it
continues to be used in a way that excludes women—and probably those
male members of minorities, where the white-male stereotype is differ-
ent from their own. And it does so with the hidden implication that any-
one who cannot use the language is unworthy of membership in the club.
What is illegitimate in this argument is that the selected language is itself
exclusionary on two levels. First, it excludes, in the way any code does,
those who are not yet full members and, second, since it is based upon an
exaggeration of a male style of interaction, it is relatively easier for men
to assimilate. This is, after all, what graduate and professional schools are
largely set up to accomplish, but the burden on women is greater. This isn’t
deliberate racism or sexism, but it works to exclude women and minorities
from positions of influence and authority on what are presented as valid
intellectual grounds. Affirmative action doesn’t touch this problem, which
213
is part of what makes women tend to drop out of graduate programs or

procrastinate in them forever at a relatively high rate.
Basically then, a woman has two choices, neither comfortable, in deter-
mining how she is to use the tool of language in her scientific or academic
work. On the one hand, she can accept the male model, perhaps outdoing
men themselves in the use of the code. But in general, it is more difficult
for a woman to use men’s language comfortably than for a man. Moreover,
it prevents women from using their own “native language” or natural style,
which is invariably devalued in the university as unscientific, imprecise, or
subjective—with the again unquestioned assumption that these are ipso
facto negative characteristics.
Or she can, in defiance of the rules, use her own language. But, as sug-
gested above, she is unlikely to be taken seriously if she does. If she does
not use the code and her work is therefore readily accessible, readers will
not feel flattered and respond with approval. Rather, since they can follow
it with ease, they will assume that she hasn’t had to work very hard. They
could have done it themselves and probably have. Hence formal technical
vocabulary, in a kind of linguistic Gresham’s law, tends to drive out clear—
that is, emotionally punctuated—discourse. The more formal and impen-
etrable the language, the more seriously the content will be taken.
In linguistics we have no end of examples. In the last twenty years for-
mal syntax and semantics have caused the flourishing and expansion of
a field that, before mathematics was introduced as the model, could hold
its professional-society meetings in one small room. Now the Linguistic
Society of America looks more like the Modern Language Association every
year, with as many as twelve simultaneous sessions. What is unintelligi-
ble to laity is gold to the elect. Or consider ethnomethodology, the study
of the surface regularities in conversation. Its body of literature is well
known for its unfailing turgidity of style and vocabulary. Its practitioners
defend its impenetrability, of which they are aware and which they claim
to consciously utilize, by saying that since they are discussing conversa-
tion, which everyone knows about and everyone knows how to do, their
work would not be taken as a scientific inquiry if they did not clothe it in
arcane vocabulary and style. Here we see the exclusionary force of an aca-
demic code used to give a field of study respectability quite apart from any
content—like dressing a beggar in the trappings of a king.
These examples are amusing, but they are not unique nor exaggerated.
The technologization of discourse is rampant in the humanities now, surely
more so than in the past. Writings on language and literature, until very
recently, were almost always fully accessible to the literate nonspecial-
ist. Practitioners argue that, much as they deplore it, technologization is

214
necessary: the subject matter has gotten so complex that language must,

of necessity, follow suit.
But what is curious, if we accept this reasoning, is that modern science,
particularly in the most popularly celebrated fields, like artificial intelli-
gence and atomic physics, has actually done just the opposite. While many
of their concepts are exceedingly complex and not readily accessible to the
untrained, their language, while they invent new technical terms, is quite
the opposite of a new humanistic parlance. Scientists in these fields seem
to strive quite deliberately for the allusion, the metaphor, the term that will
render the abstruse at least partially intelligible to outsiders.
So where humanistic writing revels in Hellenism and Latinity, artificial
intelligence gets back to basic English with terms like input or feedback and
playful uses that delight as well as educate, such as software. Atomic phys-
ics, attempting to explain the inexplicable, offers us quarks and charms,
and black holes. At least, the coiners of these words seem to be saying, “If
we can’t explain to you what’s going on, we admit to you by our playful
choices that we really don’t understand too much either.” They are not play-
ing unquestionable authority, and they are not afraid to admit that they
don’t know everything; their technical vocabulary deliberately invites that
inference, while that of the humanities, at present, explicitly rejects such
modesty.
That is the real problem of technologizing the humanities. Why has it
been done? Why do we adopt this aggressively power-oriented model?
What do we gain? The answer is simple: we are operating out of insecurity.
We are afraid that we really aren’t important or powerful anymore, and
certainly there is enough evidence to that effect; look at the budgets of the
National Science Foundation (NSF) and the National Endowment for the
Humanities if you doubt this claim. But by deliberately cultivating obscu-
rantism, we do not really help. In fact, as we have seen, this creates resent-
ment and suspicion, and leads inexorably to a position of ridiculousness
and total impotence, once people see through the pose.
It is, of course, ironic that at the moment at which affirmative action
is becoming a force to be reckoned with in the universities, male-based
language is taking over with such vigor. But perhaps it is not ironic, but
just the response we would expect, for academese is an exaggeration of
male style. When people are threatened psychologically, their defenses
are strengthened, which means that their style is exaggerated—what
they used to do a little, and under appropriate circumstances, they now
do all the time. So we can understand the proliferation of academese in
the humanities at present as the expression of a male power elite’s insecu-
rity resulting from the decline of cultural influence of the humanities, from
215
their doubts about whether their work is masculine enough, since it isn’t
bringing home the NSF bacon, and about whether they can compete with
women once the barriers are down. Therefore the style of discourse in the
humanities has become scientized without any concomitant scientization
of content. Hence, humanists feel still more uncomfortable: What if they
are found out in their little game? So, men in the humanities are creating
a secret language all their own, to shore up their own sense of being real
and legitimate in the face of doubt. They draw the inner circle ever tighter,
enforcing the shibboleths ruthlessly and continually raising their require-
ments, linguistic and otherwise, for belonging to the club.
Interestingly, the situation we are speaking of as current in the univer-
sity has analogies in older cultures. In the contemporary case, the language
used by both men and women is the same, English. But during the medi-
eval period in Japan, a remarkably similar pattern seems to have existed.
There was an indigenous language, Japanese, which had not long before
been given written form in the katakana syllabary. At the same time,
Chinese was extremely influential in Japan, both in terms of cultural and
linguistic influence, and was considered by the Japanese to be the lan-
guage appropriate to literate expression. But women were not educated
in Chinese during this period; this kind of literacy and familiarity with
another society was considered men’s prerogative. Therefore, there was
no literature in Japanese written by men during this period. What men
wrote were imitative and stilted exercises copying Chinese sources in infe-
rior Chinese. But women, relegated to the vernacular, developed a lively
literature in Japanese, which endures today, recognized as a much stronger
form of writing than the men’s, largely because it was intended as a means
of communication, rather than show.
Returning to the contemporary situation, we find that this divisive and
ominous trend toward the technologization of language is not limited to
academic, bureaucratic, or professional parlance; sadly, it is becoming char-
acteristic of our literature as well.

216
217
Introduction to “My life in court”

BY SUSAN BL ACKWELL
T he processes and outcomes of the judicial system often illuminate

issues of class, race, and gender at their starkest. As one moves up the
hierarchy of the British legal profession, for instance, the representation of
women and ethnic minorities decreases, in roughly inverse proportion to
the percentage of public school and Oxbridge-educated white males.
Such inequalities within the practitioners of the law are mirrored in
the makeup and the experiences of its recipients. In the United Kingdom,
black (African and African-Caribbean) and Irish people are disproportion-
ately represented in the prison population, and the same can be said for
Aboriginal people in Australia and African-Americans in the United States.
Women are less likely than men to find themselves on trial for criminal
offenses, but more likely to be tried for nonpayment of fines resulting from
unpaid bills (e.g., television licenses).1
Some of these inequalities involve police procedures before a person
appears before a court of law: indeed, in the case of deaths during arrest
or in custody, the accused never gets the opportunity to have a trial, fair or
otherwise. Some of the other situations arise while a person is on remand
awaiting trial or after they have been convicted and sentenced. But the
courtroom is nonetheless pivotal to the whole system of “justice.” When
1. Once in prison, women’s oppression is often compounded. Practices which have

been condemned as barbaric in recent years include repeated strip searches, shack-
ling prisoners during hospital visits and even during labor, and removal of prisoners’
babies from them even when they are still being breast-fed.
218
the police make arrests, detain suspects, interrogate them, and—by means
which vary in their legal and ethical legitimacy—obtain statements, it is
with the objective of producing evidence that can be presented in court.
When people are sentenced to periods of imprisonment, it is as a result of
the evidence presented on both sides; the availability and quality of legal
advice, representation, and interpretation available to the accused; and
the expectations and beliefs of the magistrates, judges, or jury, which are
products of their culture and life experiences. If all African Americans had
access to O. J. Simpson’s legal team, would their ethnic group still make up
over 40% of those on Death Row?
Since lawyers and judges are still overwhelmingly male, white, and
middle-or even upper-class, while the defendants are not, there are likely
to be acute differences in language and discourse style between the parties.
This is important, because everything that takes place in the courtroom is
mediated through language.
Of course, not all evidence is verbal in nature, but nonverbal evidence
must be introduced to the court by verbal means. When Sergeant Stacey
Koon and Officers Lawrence Power, Theodore Briseno, and Timothy Wind
were charged in 1992 with beating Rodney Glen King, the prosecutor pre-
sented the amateur video of the incident with the words “What more could
you ask for?” However, police counsel provided the jury not only with plas-
tic overlays outlining King’s body in white on the frames but also with a
set of verbal categories to analyze the sequence of events (Goodwin and
Goodwin 1997). Repeatedly freezing the frame and focusing the court’s
attention on King rather than the officers, the defense invited the jury to
view the tape through the discourse of the police, labeling their weapons as
“tools” and categorizing the beatings into ten distinct “uses of force,” each
with an escalation and a de-escalation triggered by King’s movements on
the ground. The repeated evidence of an expert witness that these move-
ments were “starting to be” aggressive apparently convinced the jury that
the amount of violence used was reasonable: at any rate, on April 29, 1992,
they acquitted all four officers, a verdict that led to the Los Angeles riots
of that year.2 The riots in turn led to further abuses: seventeen thousand
people were imprisoned as a result and hundreds deported to Mexico and
2. A year later there was a retrial. Although the defense again tried to argue that King
acted in a threatening manner and was therefore “in charge of the beating,” one of
their own witnesses contradicted this interpretation of events. On April 17, 1993, after
seven full days of deliberation, the jury found Officer Lawrence Power and Sergeant
Stacey Koon guilty of violating King’s civil rights, while acquitting officers Briseno and
Wind. (United States v. Koon, 1993 U.S. Dist. LEXIS 17926 (1993).)
219
El Salvador. Mike Davis, author of City of Quartz, said in an interview at

the time:
They’re throwing the book at people, demanding maximum sentences. They’re

not plea bargaining … making crimes which would normally be small misde-
meanors into felonies… . It’s clear now—and I don’t think this is well under-
stood across the country—that really the prosecution of all this is becoming
federalized. The Bush Administration has had a direct hand in it. The idea is to
make an example of people here. (Davis, 1992:10)
The admissibility of speech as evidence can be a key factor in a trial. The

jury in the retrial of Koon et al. was not allowed to consider racist com-
ments made by police in computer messages just before the beating of
Rodney King. The jury in the O. J. Simpson trial was—eventually—allowed
to hear some of the racist comments made by retired Los Angeles police
officer Mark Fuhrman.
As early as 1986, long before these more famous Californian trials, and
several years before the current surge of academic interest in the interface
between language and the law, Robin Lakoff recognized the centrality of
discourse processes in the courtroom and wrote “My life in court” based on
her experiences as a juror in a California Superior Court.
The first contribution that this paper makes to the study of language
and the law is the provision of data. Even today, the quantity of courtroom
discourse available to the linguist is meager compared to other genres.
There are four main ways of obtaining such data:
1. To take a tape recorder into court and then make one’s own transcript
from the recordings. This is the most desirable but also the most imprac-
ticable option: tape-recording of proceedings is unlikely to be permitted,
at least in a British courtroom.
2. To make one’s own notes at the time. However, even if the linguist
is competent in shorthand transcription, she is unlikely to be able to
transcribe everything. For psycholinguistic as well as practical reasons,
such items as speech errors, hesitations, discourse markers, repetitions,
and pauses are likely to be omitted, even by a linguist (Blackwell 1996).
Without the aid of a tape recorder, it is impossible to be sure that one’s
transcription is accurate, or to get it checked by a colleague.
3. To rely on the transcript made by the official Court Reporter, which may
or may not be based on a tape recording. While accurate enough for legal
purposes, the same kinds of omissions are likely to occur as in (2) above
(Walker 1990). In any case, solicitors often have to pay substantial
MY LIFE IN COURT [ 219 ]

220
sums and endure long delays to obtain official transcripts (Mansfield

1993:238), while access to them by people who are not legal profession-
als is nearly impossible.
4. To use video recordings of televised court proceedings. Only a very few
trials in Britain have been televised to date. In the United States, far
more viewing hours—indeed, whole channels—are devoted to court-
room discourse. However, unless such coverage is transmitted live and
uninterrupted, the substantial amount of editing that has taken place is
likely to render the data useless for linguistic purposes.
Lakoff does not say how she obtained her courtroom data, but it seems
likely that she relied upon her own notes. With the benefit of what we now
know about the deficiencies of transcription techniques, we might well ask,
for instance, how she could be so sure that she had not missed any simul-
taneous speech:
In six weeks of often-heated, highly emotional discourse, I virtually never

detected any interruption or overlap; there was perhaps one, for which the per-
petrator immediately apologized. (p. 230)
Indeed, the hedges “virtually” and “perhaps” suggest that Lakoff herself
was not so sure. Nonetheless, one gains the impression that she instinc-
tively knew what sort of discourse features to look for. Now that accurate
transcripts of courtroom interaction are gradually becoming available, lin-
guists who have access to them should rise to the challenge and test Lakoff’s
observations on this one case against a larger and more reliable corpus.
The second significant contribution that this paper makes to the study
of language and the law is its attempt to articulate exactly what it is that
makes legal discourse different from ordinary verbal interaction, e.g.:
In direct examination, the attorney asks his witness questions to which he

already knows the answer… . (p. 228)
and
“It is extraordinary that the one participant in the discourse who must be silent
[i.e., the jury] is the one who, in the end, plays the truly significant and predomi-
nantly active role… .” (p. 228)
Any serious attempt to analyze courtroom discourse needs a linguistic

framework within which to operate, and here Lakoff provides the beginnings
221
of such a framework using models that have, by and large, stood the test of
time: those of Grice, Gumperz, Schegloff, and Searle.
The third outstanding feature of this paper is its attempt to illuminate
how inequalities of power in the courtroom, such as those with which
I opened this piece, are manifested linguistically. It is interesting that Lakoff
compares the language of the courtroom to that of the classroom with regard
to the nonreciprocity of speaker/hearer roles in both (p. 227). Other linguists
have noted further, more specific features that the two genres have in com-
mon: for instance, Eades (1996) demonstrates that courtroom interrogation
is typically not dyadic but triadic, with the lawyer repeating part of the wit-
ness’s answer by way of evaluative feedback in a way that is reminiscent of the
Initiation-Response-Feedback structure of a classroom-teaching exchange
(Sinclair & Coulthard 1975). Interestingly, the third lawyer “feedback” move
is one of the items frequently omitted from the official transcript, probably
because it contributes nothing in terms of content from a legal point of view.
Another feature noticed here by Lakoff that is shared with “teacher talk” is
the asking of questions to which one already knows the answer.
Lakoff’s now notorious claim (R. Lakoff 1975) that women use more tag
questions than men, thereby indicating their need for (male?) approval,
has been much criticized by other feminist linguists (e.g., Dubois & Crouch
1976; Cameron, McAlinden, & O’Leary 1989). In “My life in court,” how-
ever, Lakoff makes it clear that while she does indeed view tag questions as
“disempowering devices” in ordinary conversation, their role in courtroom
discourse can be the very opposite of this:
An utterance type normally used to hand power over to the addressee has been
utilized here by the prosecutor to underline and reinforce his own already explicit
power; what is ordinarily an invitation becomes an act of coercion. (p. 230)
Lakoff should be given some credit, then, for acknowledging that tag
questions are a complex phenomenon that can function in a variety of ways
depending on social and linguistic context. That they are even more complex
than she realized is beyond doubt. Michael Cooke, relating his experiences
of interpreting at an inquest in the Northern Territory of Australia, states:
Questions put negatively to Aboriginal witnesses commonly resulted in confu-

sion because they would frequently say yes to affirm the veracity of a negatively
framed proposition in a situation where the English speaker would say no. In
doing so they would be carrying over into English the Aboriginal convention of
answering negative questions by affirming (or denying) the negative proposi-
tion. (Cooke 1995:81)

222
An example:
Counsel for the Police: But the old man didn’t go in the boat,
did he?
Witness: Yes.
Counsel: I beg your pardon?
Witness: Yes.
Interpreter: Yes, he’s affirming [that] he didn’t go in the boat.
Coroner: The old man didn’t go in the boat.
2nd Counsel: He’s answering you exactly on point.
Coroner: You ask these questions that way and that’s what you get.
(Cooke 1995 : 81)
Lakoff claimed perceptively that members of minority groups would

“encounter especial difficulties in the courtroom” (p. 231). While only prob-
lems with elaborated code or public language are mentioned explicitly here,
her remarks on tag questions demonstrate that she believed, first, that, in
court at least, this linguistic device could be used by the powerful as well
as the powerless; and second, that the powerless class in the courtroom did
not only consist of female defendants and witnesses:
Imagine the prosecutor hurling tag after tag at the reluctant respondent, and
think of the latter’s tricky position. If he replies, he damages himself. But if he
doesn’t … he risks labeling himself conversationally inept: he shows he doesn’t
know how to respond to a tag question, which demands a response. (p. 229)3
O’Barr and Atkins had argued in 1980, from their studies of courtroom
testimony, that so-called “women’s language” was actually a marker of sta-
tus rather than gender, being used by low-status men and avoided by high-
status women. It now appears that tag questions are best omitted from any
catalogue of either “women’s language” or “powerless language,” but credit
for leading off the whole debate must go to Lakoff.
This article, of course, has its limitations. The exchanges presented here
for discussion are mere snippets of courtroom discourse, probably not tran-
scribed verbatim (whatever that means: see Eades 1996), and little linguis-
tic or nonlinguistic context is provided with them. Although Lakoff was a
juror in this case, we are not given any glimpse of the jury deliberations,
probably for legal reasons. And of course, as Lakoff herself acknowledges
3. Lakoff does use generic he in her writing; but since the defendant in this case was
male I think the pronouns refer more to him than to a hypothetical generic respondent.
223
at the outset, it is only one case and is not presented as “what necessarily
always happens in court” (p. 226).
One can also, with hindsight, criticize some of Lakoff’s linguistic analysis
and sociolinguistic comment. The position with which I would personally
take most issue is her claim that “we should think long and hard before we
dispense with [the pomp and circumstance of the courtroom and its commu-
nications] in the name of ‘accessibility’ ” (p. 232). An ever-increasing num-
ber of people processed by the courts—the pieces of meat in the “sausage
machine,” to use Goldflam’s (1995) metaphor—do not have English as their
first language, and may not have standard British/US/Australian English as
any of their languages. This does not, unfortunately, mean that they will
necessarily be offered the services of an interpreter, as Goldflam (1995) and
Cooke (1995) explain at some length. For such people, who may well be at a
disadvantage already because of their ethnicity, the strangeness of English
legal discourse may reduce their chances of getting justice still further.
To say that the court system is “a basically conservative structure” is an
understatement. When Lakoff refers to the courts’ “symbolic function as
the upholder of traditional mores, the enforcer of law and order” (p. 231),
I am moved to respond: whose traditional mores, whose law, and whose
order? For the powerless groups I have described above, the archaisms of
courtroom discourse do indeed express “the strength of the justice system
as rooted in tradition,” and for that very reason they intimidate, alienate,
and oppress.
Yet again, however, it is Lakoff who has set the ball rolling by attempting
to characterize exactly what it is that is strange about courtroom discourse.
The insight and analysis that she brought to bear on her “life in court” have
proved extremely suggestive for further research, some of which I have
attempted to describe above. Lakoff is one of a number of scholars who,
over the past decade or two, have sought to bring together the fields of
law and linguistics. The results of that union have been fruitful indeed,
and some of the work now emerging from it should provide the basis for
empowerment of those who, because of their gender, race, social class, or
linguistic background, currently experience the legal system as a “sausage
machine” that mutilates their human rights.
REFERENCES
Blackwell, Susan. 1996. Corrective measures: Some aspects of transcription in the

British legal system. Recent developments in forensic linguistics, ed. by Hannes
Kniffka, Susan Blackwell, and Malcolm Coulthard, 255–76. Frankfurt am
Main: Peter Lang.

224
Cameron, Deborah, Fiona McAlinden, and Kathy O’Leary. 1989. Lakoff in

context: The social and linguistic functions of tag questions. Women in their
speech communities: New perspectives on language and sex, ed. by Jennifer Coates
and Deborah Cameron, 74–93. London: Longman.
Cooke, Michael. 1995. Aboriginal evidence in the cross-cultural courtroom. Language
in evidence: Issues confronting Aboriginal and multicultural Australia, ed. by Diana
Eades, 55–96. Sydney: University of New South Wales Press.
Davis, Mike. 1992. The rebellion that rocked a superpower. Interview by Lance Selfa.
Socialist Review 154:8–9.
Dubois, Betty Lou, and Isobel Crouch. 1976. The question of tag questions in women’s
speech: They don’t really use more of them, do they? Language in Society
4:289–94.
Eades, Diana. 1996. Verbatim courtroom transcripts and discourse analysis. Recent
developments in forensic linguistics, ed. by Hannes Kniffka, Susan Blackwell, and
Malcolm Coulthard, 241–54. Frankfurt am Main: Peter Lang.
Goldflam, Russell. 1995. Silence in court! Problems and prospects in Aboriginal legal
interpreting. Language in evidence: Issues confronting Aboriginal and multicultural
Australia, ed. by Diana Eades, 28–54. Sydney: University of New South
Wales Press.
Goodwin, Charles, and Marjorie Harness Goodwin. 1997. Contested Vision: The
Discursive Constitution of Rodney King. The Construction of Professional
Discourse, ed. by Britt-Louise Gunnarsson, Per Linell, and Bengt Nordberg,
292–316. London: Longman.
Mansfield, Michael. 1993. Presumed guilty: The British legal system exposed.
London: Heinemann.
O’Barr, William, and Bowman Atkins. 1980. “Women’s language” or “powerless
language”? Women and language in literature and society, ed. by Sally McConnell-
Ginet, Ruth Borker, and Nelly Furman, 93–110. New York: Praeger.
Sinclair, John, and Malcolm Coulthard. 1975. Towards an analysis of discourse.
London: Oxford University Press.
Walker, Anne Graffam. 1990. Language at work in the law: The customs, conventions,
and appellate consequences of court reporting. Language in the judicial process,
ed. by Judith Levi and Anne Graffam Walker, 203–44. New York: Plenum.
225
CHAPTER 10
My life in court (1986)
1. INTRODUCTION
In September and early October of 1984, I served on a jury in the Alameda

County, California, Superior Court. From both the nature and outcome of
the trial itself (which I will summarize shortly) and the situation in the
courtroom and jury room, I learned a great deal. I want to talk about some
of this, concentrating here on the area of language in court: the courtroom
as a discourse context. For while in many ways the communication that
formed the trial and the deliberation of the jury resembled ordinary infor-
mal dyadic conversation, in more ways it did not. Some of these differ-
ences are obvious to the most casual viewer of Perry Mason; in other ways,
though, the novelty of courtroom talk repays close observation by opening
a window to ordinary conversational behavior, allowing us to observe and
understand it from a new perspective.
I should make it clear before proceeding further that my comments are
based upon this single six-week observation of one trial, one set of dramatis
This paper originally appeared in Languages and Linguistics: The Interdependence of

Theory, Data, and Application, edited by Deborah Tannen and James E. Alatis, 171–79.
Georgetown University Round Table on Languages and Linguistics 1985. Washington,
DC: Georgetown University Press, 1986. Reprinted here with permission of the
publisher.
Acknowledgments: I am grateful to the many people who helped, in one way or
another, to make my debut into the mysteries of the legal universe meaningful. In par-
ticular, I would like to thank Judge Stanley P. Golde, Deana Logan, Mike Millman, and
Harriet Verbin for aid, comfort, encouragement, and clarification, and in their several
roles, for restoring my faith in our system of criminal justice and its practitioners.
226
personae. So I am not writing to describe what necessarily always happens

in court. Rather, I am commenting as an observer (relatively sophisticated
as a linguist but a neophyte in the courtroom) on what I encountered: what
seemed strange in terms of what I might have expected had this been ordi-
nary conversation. The question I have at the back of my mind, which
I hope my observations will begin to answer, is this: How do we know how
to engage in conversation? Faced with a situation in which at least some
of the rules are new, and contradict or make irrelevant the discourse rules
we had always lived by, how do we construct a new grammar to make sense
of the situation? In a more practical vein, in court perhaps more than any-
where else in society, it is crucial that the neophytes (that is, of course,
the jurors) exposed to the novel situation become comfortable with it, and
able to understand what is being communicated and why, as thoroughly
and quickly as possible. Much has been written about the final and most
formal stage in this process, the judge’s explicit communication of instruc-
tions to the jury as to how it shall reach a verdict, and much as well about
how juries deliberate (for the latter, cf. especially Garfinkel 1967). But if
the days or weeks preceding have been mystifying and incompletely under-
stood, the most careful and intelligent communication of, and following
of, instructions will be of no avail. So there is some ultimate practical value
for linguists and jurists alike in coming to understand how juries come to
understand what is going on around them.
My jury service involved a capital case, though a rather weak one. The
defendant was accused of murdering someone, and attempting to murder
another, in order to steal their car and subsequently sell it. By California
law, the fact that the murder was committed “for financial gain” made the
defendant liable for the death penalty.1 A trial of this kind is conducted in
two parts. In the first, the jury determines guilt or innocence. If the defen-
dant is found guilty of murder with special circumstances, after a short
recess the same jury hears evidence in aggravation or mitigation, and on
that basis determines the penalty: the choices are only life without possi-
bility of parole or death in the gas chamber. In this case, the defendant was
found guilty of all charges after a trial of three weeks, with the jury delib-
erating for about three hours. In the second phase of the trial, lasting two
weeks, the jury voted for a sentence of life without possibility of parole,
after deliberating for about two hours.
Notable in this trial, and part of what made the experience so mean-
ingful for me, was the competence of most of the participants. Aside from
1. California Penal Code §190.2 (1995–96).
227
one member of the defense team (who actually got a good deal better as he
went along), the attorneys on both sides were articulate and organized, and
insulted the intelligence of the jury rather less than I would have antici-
pated. The jurors, drawn from the proverbial cross-section of the commu-
nity, were interested, dedicated, and responsible from the start. During
deliberations, I was impressed with their intelligence in being able to grasp
and use complicated instructions (given only orally); their articulateness in
arguing for their positions; and their compassion. As the presiding official,
the judge in a very real way set the style for everyone else. His lucidity in
explaining things to the jury, coupled with his very real sympathy for what
they were going through—demonstrated in many ways throughout—made
the experience memorable.
As a linguistic event, a trial is unique in many ways. I will organize my
remarks about courtroom discourse under six subheads: (1) the trial as a
kind of discourse situation; (2) special functions of speech acts; (3) con-
versational rules; (4) conversational logic; (5) ritual and spontaneity; and
(6) the creation of sympathy—persuasion.
2. THE TRIAL AS A DISCOURSE SITUATION
In any discourse, there is presumed to be a speaker and a hearer. In dyadic

conversation, these roles are reciprocal: the participants exchange them. It
is reasonable to assume that the more a type of discourse deviates from this
fundamental behavior, the more counterintuitive it will be, and the harder
to learn. Thus, classroom behavior is new for children because discourse is
no longer reciprocal; and for adults, psychotherapeutic discourse is special
for the same reason. In court, too, the roles of the various participants are
strictly defined and not interchangeable.
In any conversation, there are certain kinds of business to be transacted.
In most, outside of the informal dyad, the exchange of information is per-
haps the most crucial. Questions are asked, and answers are given, with
this aim in mind. Normally, in any kind of conversation, the person need-
ing the information will frame the questions, directing them at the person
assumed to possess it. Further, under normal conditions the assumption
is that a question is appropriately asked just in case the questioner has
reason to believe the addressee has the necessary information and is will-
ing to supply it; and the addressee will respond with that information just
in case the foregoing conditions are met and, besides, it is determined that
the speaker needs the information sought.
M y life in court [ 227 ]

228
In court things are very different. In direct examination, the attorney

asks his witness questions to which he already knows the answer; indeed,
I am informed that lawyers are taught to frame questions to friendly wit-
nesses using some where we might expect any, in order to encourage them
to give a positive answer:
Prosecutor: Was there something in that newspaper article that caused you to

do something?
Such a question, obviously, could be asked only by someone quite sure

(through prior discussion with the addressee) of the sort of response that
was likely. Under these conditions, we might normally assume the ques-
tion would be treated as rhetorical and not answered. But, of course, the
witness does answer. Even in cross-examination, the attorney is unlikely
to ask a question the response to which will come as a surprise. So what is
the purpose of this discourse, if speaker and addressee are not exchanging
needed information?
The answer, I think, can be found only if we can see that in this situation
the people actually carrying on the “conversation” in question are not really
the participants—or at least, that the questioner is not playing the nor-
mal role. Attorneys, in questioning witnesses and making objections (their
two roles), are acting in the jury’s behalf: the jury, though silent through-
out, is really the participant who requires the information, who alone can
make use of it. It is extraordinary that the one participant in the discourse
who must be silent is the one who, in the end, plays the truly significant
and predominantly active role. This goes counter to all of our expectations
about how discourse occurs and is understood, and contributes to the mys-
tique of the experience in the jurors’ eyes. As I said, the learning of any new
linguistic behavior takes effort; but learning a role that is anomalous, con-
tradictory to the old, familiar, and by now unconscious rules is especially
difficult and disturbing—even if, as is usual, the jurors cannot themselves
identify the source of their discomfort explicitly.
Likewise, in ordinary conversation, participants all have access to the
complete repertoire of speech-act types (cf. Searle 1969) and can further
expect their contributions to be understood as others’ would be (a declara-
tion as a declarative, and so on, unless signaled otherwise). But in court,
the kinds of utterances each participant may make, and be understood as
making, are severely restricted, in terms of the use and understanding of
speech acts. And this is true (as with psychotherapy) both superficially
(e.g., the attorney can only ask questions of witnesses; the witness can only
give declarative answers) and more abstractly. For example, at a point at
229
which a lawyer may be expected to make an objection, technically a declara-

tive speech act, even what looks like a question may be interpreted by the
judge as an objection.
Defense [with reference to witness’s testimony]: What does that mean?

Judge: Sustained.
3. SPEECH ACTS IN COURT
Similarly, speech acts may play unusual roles in this situation, related to the
special powers and functions of the participants. Tag questions have been
discussed at length, largely from the perspective of their role in ordinary
conversation, where they may function as disempowering devices: their
use invites the addressee to control the conversation, both in terms of
defining content and holding the floor (for more discussion, see R. Lakoff
1985). But if we superimpose the special power dynamic of the courtroom,
we see that even a form such as the tag, with its clearly specified pragmatic
functions, can have quite different implications. Consider that a tag, in its
ordinary-conversation function, is an elicitor par excellence of response
from an addressee. In dyadic discourse, in which participants really or con-
ventionally are equal in power, this is most often felt as an opening up of
the floor. But now imagine a prosecutor cross-examining a defendant who
has already been found guilty of murder with special circumstances. The
attorney clearly has the power; he has no need to invite a response, since
the witness is compelled by law to reply. And the questioner, as we have
seen, doesn’t even need the information himself, so the addressee can’t
do anything for him. Imagine the prosecutor hurling tag after tag at the
reluctant respondent, and think of the latter’s tricky position. If he replies,
he damages himself. But if he doesn’t, not only does he risk a citation for
contempt but (probably worse) he risks labeling himself conversationally
inept: he shows he doesn’t know how to respond to a tag question, which
demands a response.
Prosecutor: You don’t want to die, do you?

Witness [defendant]: You’re treating me like a criminal!
Thus, the prosecutor, by placing him in a double bind, drew from the defen-
dant “crazy” behavior, losing him any possible sympathy from the jury and
earning him a warning from the judge. An utterance type normally used to
hand power over to the addressee has been utilized here by the prosecutor

230
to underline and reinforce his own already-explicit power; what is ordinar-

ily an invitation becomes an act of coercion.
4. CONVERSATION ANALYSIS
We are aware of the basic underlying principle in ordinary conversation: one

party at a time (Schegloff 1972). We are, of course, also aware that in most
informal conversation, this rule is violated rather often. We may surmise
that a principal reason for this rule is that it facilitates the function of the
Cooperative Principle (CP) (Grice 1975): it encourages the optimal trans-
mission of information. (Hence, the circumstances under which it is most
easily violable are informal gatherings among intimates, in which rapport
rather than information is the coin of the realm.) Since the courtroom is
the scene of maximal CP operation, we may expect whatever facilitates it to
be strictly enforced. And indeed this is true. In six weeks of often-heated,
highly emotional discourse, I virtually never detected any interruption or
overlap; there was perhaps one, for which the perpetrator immediately
apologized. (It’s true that the defense attorney rather frequently muttered
to himself, but this was not perceived as a disruption of anyone’s turn, and
he was chastised for it by both judge and prosecutor.)
Also characteristic of the assignment of the floor in ordinary conversa-
tion is its implicitness. Very seldom does anyone have to (or feel able to) say,
“It’s my turn,” or “Let her talk,” unless negotiating with young children. But
in court, the judge often performed this function, largely for the witnesses
(of all the active participants, they alone were assumed to be unaccustomed
to the procedures). “You may answer,” he would say after overruling an
objection; “You may step down,” at the end of their testimony. The attor-
neys had their floor-yielding formula, of course: “Your witness.”
5. CONVERSATIONAL LOGIC: THE COOPERATIVE PRINCIPLE
Of all types of discourse, a trial is one in which “the truth, the whole truth,
and nothing but the truth” is explicitly and exclusively enjoined as permis-
sible. That is, the Cooperative Principle is explicitly and formulaically stated
to be in effect. Here more than anywhere information is what is sought, and
anything clearly noninformational (in some sense of the term) is imper-
missible. Further, more than is ordinarily the case, the participants in the
business normally have no prior relationship with one another, and there-
fore the CP violations and implicatures that intimacy allows do not occur.
231
But although informativeness is a major consideration, it would be incor-

rect to suggest that there are no allowable violations of the Cooperative
Principle. Rather, they are of a different kind than we normally encounter,
and hence are unusually hard for the neophyte to understand. They come
under the heading of “legalese.”
In theory, legal formula has acquired its cumbrous shape in order to pre-
clude the ambiguity to which ordinary conversation is prey. But at the same
time, any violation of the Cooperative Principle is apt to lead to breakdowns
in intelligibility: we have to make an extra effort to process the unexpected.
The language of the courtroom eschews the deletions and compressions we
normally make use of in informal discourse, both to save time and to signal
that we—speaker and addressee—are members of the same group because
we both share information. This brevity, of course, is a significant aspect
of what Bernstein (1973) calls a “restricted code.” By contrast, courtroom
talk can be seen as hyperelaborated. Things are said that would ordinarily
be ignored or assumed, and things are repeated that would otherwise have
been said but once. So the maxims of Quantity and Manner are violated
here as well, but in the opposite of the usual direction.
The automobile she had for sale—did she place an ad in the paper for the pur-
pose of selling it?
With regard to the sale of the truck, was there anything unusual about the sale
of the truck?
Similarly, witnesses tended to answer questions with full sentences rather

than the fragments we would expect in ordinary conversation.
We might note finally that if members of minority groups have espe-
cial difficulty in achieving competence in the use of an elaborated code (as
Bernstein 1973 argues) or in using public language (as Gumperz 1982 sug-
gests), such groups will encounter especial difficulties in the courtroom on
these grounds.
6. RITUAL AND SPONTANEITY
It is an error to suppose that the simple discovery of information is the

court’s only business. Superimposed on this “real” function of the court
system is its symbolic function as the upholder of traditional mores, the
enforcer of law and order. As a basically conservative structure, then,
the court utilizes a language that supports and derives from its symbolic
function. The archaism of legalese and the pomp and ceremony of much

232
courtroom behavior thus have no utility in terms of the court’s fact-finding

function. But as a means of expressing, to all participants, the strength of
the justice system as rooted in tradition, the use of ceremonial forms is
of extreme importance. It functions also as an additional message for the
jury. Jurors are instructed from early on in their service that the modes
of reasoning they will be required to use in reaching a verdict are different
from those in effect in their everyday lives. But it is difficult to learn new
ways of doing what has been second nature, and always a temptation to fall
back into the old. If, however, there are continual reminders that this situ-
ation is special, the rules are different, more is expected of one, it may be
easier to stay on the straight and narrow. And although it is fashionable to
criticize the pomp and circumstance of the courtroom and its communica-
tions, I think that—as long as they do not impair intelligibility—they have
a very real function, and we should think long and hard before we dispense
with them in the name of “accessibility.”
The fact that linguistic behavior is ritualized and nonspontaneous in so
much of court procedure leads in turn to special forms of interaction that
could occur only in such a context. Thus, for instance, it could be assumed
that an attorney would object to certain kinds of behavior on the part of his
opposite number. In ordinary conversation, even if someone’s next utter-
ance is fully predictable, we can’t normally act as if it were: we are supposed
to work on the assumption that we are behaving spontaneously. But in a
situation in which the rules of discourse are explicit and acknowledged by
everyone, the ritualization can permit some (tongue-in-cheek, but none-
theless real and serious) play on expectation.
Defense [in cross-examination]: You mean he just dropped off the face

of the earth, like [the defendant]?
Prosecutor: [Says nothing][Pause.]
Judge: Sustained.
Prosecutor: Thank you.
(We might note here that ordinarily, when the judge sustained an explicit
objection, “Thank you” would have been inappropriate: it was part of
his job.)
7. THE ELICITATION OF SYMPATHY
Finally, we can look at some of the more obvious ways in which lawyers do
their job: winning sympathy from the jury for or against the defendant by
233
establishing identification between them and the defendant, or by facili-

tating an identification of the jurors with “society” and as the agents for
enforcing responsible behavior. Both sides, in their summations, made
much use of the pronoun we. But the defense meant by it, “you (the jury)
and the defendant and I, as human beings”; the prosecution, “you and I as
representatives of society.” To this end, the defense regularly referred to
the defendant by a nickname; the prosecution, by title and last name. To
the same end, both sides tried to establish themselves as authoritative and
trustworthy. A significant argument in mitigation involved the defendant’s
service in Vietnam. Both the male defense attorney and the male prose-
cutor referred in their statements to “’Nam” or “the ’Nam.” In my experi-
ence, these terms are used only by men, and normally only by veterans, or
those who seek to present themselves as such. In using these terms, both
sides were suggesting, to different purposes, that they understood what it
was to serve in Vietnam, that they were one with those who had had that
experience.
Finally, both defense and prosecution made use, where useful, both
of technical legalese even where it was unnecessary, to establish author-
ity, and of slang, to suggest identity with the jurors and impress them as
regular guys. Thus, the defense attorney referred to certain arguments
as “hypotheticals”; the prosecutor, in talking about the defendant’s aber-
rant behavior, suggested several times that he “wasn’t wrapped too tight.”
Ordinarily, the self-presentation of the prosecutor (as befitted the repre-
sentative of law and order) was formal and conservative; by using slang, he
showed he was at heart a regular guy. The defense attorney, on the other
hand, was generally studiously (and often quite inappropriately) infor-
mal. So his use of a bit of legalese was to signify, “See? I really can play
this game.” Besides, the prosecutor’s use of the colloquial term for crazy
helped to undermine the defense contention that the defendant, due to
his Vietnam experience, really was seriously disturbed. No, the prosecutor
suggested, just a little nuts like everyone, no need for special sympathy or
exoneration.
8. CONCLUSIONS
We see, then, from these few examples, that in the courtroom language
is used differently in many significant ways than it is elsewhere, and all
participants must learn new rules and new interpretations of old ones.
But at the same time, we necessarily bring to the courtroom our familiar
grammars; these may be used or abused by the various participants. It is

234
incumbent upon us as linguists and as citizens to understand the interplay

between ordinary language and the language of the courtroom. In this way
we learn more about communication in general, and encourage the best use
of language in a situation in which everything said or heard is of unusual
importance.
REFERENCES
Bernstein, Basil. 1973. Class, codes and control. St. Albans, UK: Paladin.
Garfinkel, Harold. 1967. Some rules of correct decisions that jurors respect.
Studies in ethnomethodology, by Harold Garfinkel, 104–15. Englewood Cliffs,
NJ: Prentice-Hall.
Gumperz, John J. 1982. Discourse strategies. Cambridge, UK: Cambridge
University Press.
Lakoff, Robin Tolmach. 1985. The politics of language. CATESOL Occasional Papers
11:1–15.
Schegloff, Emanuel A. 1972. Sequencing in conversational openings. Directions
in sociolinguistics, ed. by Dell Hymes and John J. Gumperz, 346–80.
New York: Holt, Rinehart, and Winston.
Searle, John R. 1969. Speech acts: An essay in the philosophy of language. Cambridge,
UK: Cambridge University Press.
235
On the Rise and Fall

of Generative Semantics
Introduction to “The way we were; or; The real
actual truth about generative semantics: A memoir”
BY GEORGIA GREEN
“
T he way we were” is a diversely textured social and intellectual history
of fifteen exciting years (1962–1977) in the development of modern
syntactic theory. We get an insider’s view, unabashedly subjective, in a deli-
cious stew of metaphors: theoretical debate seen alternately as combat,
seduction, and religion. It should be required reading in every history of
linguistics course. Of course, I don’t buy every word of it; it’s as hyperbolic
and permeated with procedural metaphors (e.g., replace, change, delete,
choose, operate) as generative semantics ever was, and while this re-creates
the flavor of the times Lakoff1 chronicles, it’s still as likely to lead to misun-
derstanding as it was in the old days.
My working title for this note was “Highly abstract snake oil: An appre-
ciation,” in honor of one of those delicious metaphors in Lakoff’s essay.
But much as the thought of using that title brought back the flavor of the
1. In the absence of instructions to the contrary, I use standard academic style for
personal references here, noting that until I made an issue of it with reference to
the publication of my review of R. Lakoff 1968 (Green 1970), women linguists were
referred to in the pages of Language as “Mrs. Husband’s-last-name.”
236
old days, I was compelled to shelve it; it is too easily misinterpreted as

endorsing the misconception popularized by Newmeyer (1986) of genera-
tive semantics as an irresponsible lark. Generative semantics was a serious
research program, and generative semanticists had a sense of community
and of living to work that hasn’t been seen since.
Lakoff’s essay begins with a detailed history of the development of
generative semantics, and proceeds to the most detailed exposition of
the principles constituting generative semantics since McCawley (1972).
Section 3 is not so much about events in the recent history of linguistics
as a provocative, idiosyncratic psychoanalytic reflection on all of academe.
While section 4 picks up this thread as it relates to what differentiated the
practice of generative semantics from that of competing theories, the next
section contrasts generative-semantic and extended standard theory views
with each other, and with those of structural linguistics before them, on
the role of universals in theory building. The final substantive section ana-
lyzes the demise of generative semantics and reviews its legacy.
VALEDICTION
What did finally become of generative semantics? Did it fold its tents and
steal quietly into the night, as Lakoff implies at the end of section 3? Did it
implode under pressure of self-induced marginalization from topical exam-
ple sentences and cute titles, as Newmeyer (1986:137) has implied? (In
general, it is not true, as Lakoff asserts, that generative semantics evolved
into functionalism, or that the central figures adopted functionalist agen-
das. Some (e.g., Schmerling, Abbott, Dowty, Green, Postal) eschewed the
demonizing of formalism and, finding it empowering instead, went on to
work in formal semantics (incorporating as much pragmatics as they could
into it), or formal syntax constrained by an explicit compositional seman-
tics.) Was it that the effort of trying to represent syntactically significant
pragmatic properties of sentences within the constraints of the available
formalism for syntactic structure caused generative semantics to collapse
of its own weight? The details can be found in articles by Morgan and
Sadock with the sort of typical generative-semantics titles (“How can you
be in two places at once when you’re not anywhere at all?” Morgan (1973);
“The soft interpretive underbelly of generative semantics,” Sadock (1975))
that the curmudgeon-in-chief found so embarrassing. (Wouldn’t I have
liked to be a fly on the wall to see how Fritz Newmeyer reacted to being
called a curmudgeon by a female Presence a couple of years his senior!)
Lakoff characterizes these papers as despairing, but this only goes to show
237
how personal a ground-level view of events must necessarily be.2 Some lin-
guists took inspiration from them3 and embarked on a program to tackle
questions of pragmatics head-on, with an eye to articulating a language-
independent account of pragmatics that would be compatible with a theory
of linguistic structure that was independent of use.4
Science: Values
Perhaps the aspect of this article that I like best is the support it provides
(e.g., at the end of sections 1 and 2.4) for some observations Thomas Kuhn
was criticized for making,5 to the effect that theoretical disputes may be
unresolvable to the extent that values which form the foundations for prem-
ises in arguments are not shared, since what is valued is beyond reasoning
(i.e., not generally subject to alteration as a result of hypothetico-deductive
argument—this is what Kuhn means by saying that “the superiority of
one theory to another is something that cannot be proved in the debate”
(1970:198)).6 Often, of course, it is not acknowledged, or even recognized,
that incompatible points of view, proposals, and conclusions devolve from
unshared values. The combatants in the “linguistics wars” can’t be blamed
for failing to make their goals, hopes, aims, and beliefs explicit, because
as individuals they took it for granted that their particular values were all
universally shared.
Science: Methods
Lakoff is absolutely correct to implicate an increased understanding of the

nature of pragmatics as a major factor in the rapid decline of generative
semantics in the 1970s, though there is still room for a variety of accounts
2. Cf. O’Meara 1996.
3. E.g., Morgan, Green, Horn, and Lakoff herself.
4. See Morgan 1975 for discussion of some of the difficulties they faced.
5. Apparently by scholars who lost track of the premise that science is not a math-
ematical function but an activity of human beings.
6. Cf. also these remarks:
There must also be a basis, though it need be neither rational nor ultimately cor-
rect, for faith in the particular candidate chosen. Something must make at least
a few scientists feel that the new proposal is on the right track, and sometimes
it is only personal and inarticulate aesthetic considerations that can do that.
(1970:158)
T H E WAY W E W E R E [ 237 ]
238
for the details of the cause and effect. In any case, I have to take issue with
the assumption that pragmatics cannot be studied scientifically. As I have
discussed elsewhere (Green 1995), the questions we want to answer dic-
tate the means we will use to eliminate wrong answers, but nothing in the
nature of pragmatics precludes approaching pragmatics questions as scien-
tifically as any others. As human beings, we cannot help but form hypoth-
eses, and once we have a hypothesis, we make a beeline for data that will
test it, and either corroborate it or disconfirm it. (We do this not because
we are following some prescribed scientific method, but just because we
have operational minds and act rationally.) The hard part of research, as
always, is figuring out which questions to ask, and being able to roll with
the punches, and adjust both the question and the means for testing it, as
preliminary results tell us more about the domain of inquiry.
PERSPECTIVE
The sections of Lakoff’s essay all begin with epigrams (in three languages).
My favorite is the quote from Santayana, which has been used elsewhere
(Green & Morgan 1996) for the same purpose. While generative seman-
tics may be today no more than a colorful backwater in the history of gen-
erative grammar, at the time of its ascendancy it saw itself as a natural
development of the standard (“classical”) theory, despite getting a cool
reception from Chomsky. Nonetheless, many of the leading ideas of gen-
erative semantics have been increasingly incorporated in work developing
Chomsky’s more recent elaborations—see Pullum (1989) for a detailed
account of where generative-semantics ideas show up in the more recent
history of current incarnations of transformational grammar. Hold on to
your hat when you read “The way we were”; you’re in for a ride!
REFERENCES
Green, Georgia M. 1970. Review of Abstract syntax and Latin complementation, by

Robin Lakoff. Language 46:149–68.
Green, Georgia M. 1995. The right tool for the job: Techniques for analysis of
natural language use. Pragmatics and Language Learning, Monograph Series,
Volume 6, ed. by Lawrence F. Bouton, 1–18. Urbana: Division of English as an
International Language, University of Illinois.
Green, Georgia M., and Jerry L. Morgan. 1996. Practical guide to syntactic analysis
(CSLI Lecture Notes 67). Stanford, CA: CSLI Publications.
239
Kuhn, Thomas. 1970. The structure of scientific revolutions. 2d edn., enlarged

(International Encyclopedia of Unified Science, vol. 2, no. 2). Chicago:
University of Chicago Press.
Lakoff, Robin T. 1968. Abstract syntax and Latin complementation. Cambridge,
MA: MIT Press.
McCawley, James D. 1972. A program for logic. Semantics of natural language, ed. by
Donald Davidson and Gilbert Harman, 157–212. Dordrecht: Reidel.
Morgan, Jerry L. 1973. How can you be in two places at once when you’re not
anywhere at all? Papers from the ninth regional meeting of the Chicago Linguistic
Society (CLS 9), Chicago, 410–47.
Morgan, Jerry L. 1975. Some interactions of syntax and pragmatics. Syntax and
semantics 3: Speech acts, ed. by Peter Cole and Jerry L. Morgan, 289–304.
New York: Academic Press.
Newmeyer, Frederick. 1986. Linguistic theory in America. 2d edn. San
Diego: Academic Press.
O’Meara, John. 1996. Mayan decipherment. LINGUIST 7:1146. Online: http://
linguistlist.org/issues/7/7-1146.html.
Pullum, Geoffrey. 1989. Prospects for generative grammar in the 1990s. Proceedings of
the Western Conference on Linguistics 89, vol. 2, ed. by Frederick H. Brengelman,
Vida Samiian, and Wendy Wilkins, 257–76. Fresno: Department of Linguistics,
California State University.
Sadock, Jerrold. 1975. The soft interpretive underbelly of generative semantics.
Syntax and semantics 3: Speech acts, ed. by Peter Cole and Jerry L. Morgan, 383–
96. New York: Academic Press.
T H E WAY W E W E R E [ 239 ]
240
241
CHAPTER 11
The way we were; or, The real actual

truth about generative semantics
A memoir (1989)
Infandum, Regina, iubes renovare dolorem

Troianas ut opes et lamentabile regnum
eruerint Danai, quaeque ipse miserrima vidi
et quorum pars magna fui.
—Vergil, Aeneid 2.3–6
I t is now over a dozen years since generative semantics (GS) ceased to

be considered, by its practitioners and by the field of linguistics as a
whole, a viable theory of language. Perhaps now it is possible to look at
its history in a more or less dispassionate way; and evaluate its contribu-
tions so as to understand what the enterprise was really about (which
would have been impossible at the time, had we even been inclined to
reflect upon it).1
I am also moved to take computer in hand as a result of reading
Newmeyer’s (1980) commentary. Further, I have learned in the last few
months of several attempts to write histories of GS, by people who must
have been scarcely babes in arms at the time it was all going on. All of this
moves me to try to get it right, to discourage future distortions of the
Newmeyer type and others, perhaps even further from any reality.
This paper originally appeared in Journal of Pragmatics 13(6):939– 88 (1989).

Reprinted here with kind permission from Elsevier Science–NL, Sara Burgerhartstraat
25, 1055 KV Amsterdam, The Netherlands.
1. I would like to thank Dwight Bolinger, Georgia Green, Randy Harris, George Lakoff,
and Toshio Ohori for their comments and suggestions.
242
Newmeyer’s work has been valuable to me in clarifying other perspectives

and encouraging me to put some years of thoughts on paper. Newmeyer
certainly is to be admired for working through a formidable amount of
complex, contradictory, and often ill-expressed prose in an attempt to
make sense of fifteen years of theoretical developments in linguistics.
At the same time, I feel that his work suffers from flaws which detract
critically from its utility and belie its author’s claim to be presenting a gen-
erally dispassionate history. Newmeyer’s bias is the most dangerous kind—
inexplicit, perhaps not fully recognized even by the author. No one who lived
through the period as a participant in the “linguistics wars” can claim the
status of disinterested observer. By his appearance of doing so, Newmeyer
misleads the reader and distorts the facts. I believe that, when a historian is
in a position to be biased, it is his or her responsibility to discover the form
and extent of that subjectivity, and make it clear to the audience.
In this work, whose subjectivity is rampant indeed, I have chosen to lay
stress on the extratheoretical content of the GS/Extended Standard Theory
(EST) debate, since the theoretical arguments have been, and no doubt will
be, dissected sufficiently by others in the future. Besides, I have come to
realize that theory alone does not make for linguistic schools, much less
linguistics wars: to understand the theory itself as well as the history, we
have to understand the people.
1. THE BASIC HISTORY
Those who cannot remember the past are condemned to repeat it.

George Santayana, Reason in Common Sense
Generative semantics is, or was, an offshoot of what Chomsky was later

to call the Standard Theory and I like to call Classical Transformational
Grammar, or CTG, the model described in Aspects of the theory of syntax
(Chomsky 1965). In that book, Chomsky discussed his model of a two-
tiered syntactic component: surface structures (SS), the sentences we
observe, with information about the grammatical relations among their
parts; and the deep structure (DS). These two levels were linked through a
system of transformational (T) rules.2
2. Readers should note that the intended audience for this paper is diverse. It was
originally written for the first-year graduate students in a proseminar I was teaching,
and some of the background information reflects this origin. On the other hand, I sus-
pect that in some cases I have presupposed knowledge on the reader’s part of particular
243
Aspects theory assumed that DSs and Ts would be strictly constrained as

to form and function, but not all that much was said about what those con-
straints were. Examples were given in Aspects in the form of “fragments” of
a grammar of English. But what was not specified was what kinds of items
could or could not be in the DS; how the DS order related to the surface
order of Ss; and what kinds of processes transformations could and could
not effect. It was impossible to tell, either from written documents or from
oral statements by the author, the limits on the abstractness of DSs, how
universal they were meant to be, and how much Ts could change them to
create surface structures.
On the other hand, in his less technical writings, like Cartesian linguis-
tics (1966) and Language and mind (1967), which were being discussed and
worked on in the immediate post-A spects period, Chomsky implied a lot
in these areas. He talked of DSs as linked to universal human cognitive
structures, Ts as windows into the mind. Now if a DS was to have univer-
sal implications, it was clear to at least some readers of the whole oeuvre,
the model set forth in the various grammatical fragments in Aspects could
not be taken literally, or surely not as an exclusive statement. This version
of DS contained, for instance, articles and modal auxiliaries; DS sentences
were in an NP-V-NP (SVO) order. In this and numerous other ways, these
DSs were English-specific. They did not represent semantic “primes” in
any sense, they were syntactic bases, not semantic (as Chomsky himself
made clear by making the DS the basis of an autonomous syntactic compo-
nent, which was “interpreted” by an autonomous semantic system). The DS
was relevant to, but not a part of or incorporating, semantics. Then if the
Cartesian universals of language that connected all human beings by virtue
of their reason were (as surely seems to be the case) semantic entities, hav-
ing to do with meaning and reference, this notion of DS was irrelevant—
but Chomsky claimed otherwise in his other work.
It’s important for this and much subsequent discussion to understand
the overwhelming influence Chomsky had on his disciples during this
period, an influence both moral and intellectual. He spoke in a soft voice,
persuasive it seemed by reasoned argumentation alone: he carried convic-
tion, in part, because he appeared to be above the academic fray, interested
only in truth for its own sake, not winning in the doctrinal rivalries that
characterized other disciplines.
points of linguistic theory that are not part of the linguist’s current armamentarium.
I would hope that readers would try to piece things together, skimming over what they
already know, or not worrying too much about what is obscure, in order to get the gist
of what I’m trying to say.
T he way we were [ 243 ]

244
A significant part of the attraction of CTG, and its meteoric rise, lay in
its presentation of a formal linguistic theory, the promise of a complete and
rigorous model. It was covertly assumed that all the properties of language
could be described by a system that utilized dichotomous choices, as a for-
mal system must. The “justification” of the idea that a formal system of this
kind was adequate for linguistic description mainly consisted of the sorts
of “fragments” of grammars of particular languages (most often English,
or anglocentric versions of other languages) found in Aspects. While this
showed that, for carefully selected cases, a particular formalism could work,
it didn’t show that it could work for the grammar as a whole, much less that
it really captured what was going on. The assumption was that the frag-
ments we had in 1965 would, in a couple of years, increase and meet one
another to form a complete grammar of English. Students would then take
this grammar and apply it to many “exotic” languages, shortly producing
complete grammars for each and all. The distillation of this effort would be,
in the not-too-far-distant future, an inductively derived universal gram-
mar which would (undoubtedly) jibe smoothly with the deductive version
then being discussed, based on the way language “had to be” due to the
theoretical assumptions of CTG. It was a heady promise and we all believed
it wholeheartedly. But at least some of us wanted to be more precise about
the details of this grammar than Aspects or its author allowed.
Important at this point and later, overtly and implicitly, was the “para-
digm argument” of CTG, the way in which DSs could be justified: both the
idea of implicit underlying structure, and particular realizations thereof.
To avoid phenomenological criticism about postulating forms that were in
some sense interpretively derived, theorists had to offer a rigorous means
of deriving the abstract from the concrete, as well as reasons why this
potentially dangerous step was justified in terms of the increase in gener-
alizing power of the grammar. These arguments were made elegantly in a
1964 article by Paul Postal. He took an example familiar and comfortable
not only to transformational grammarians and structural linguists, but to
grammar teachers everywhere: the idea that, intuitively, English impera-
tive sentences were not, at heart, “subjectless” as they appeared; but, in
order to explain what could and could not occur as imperative sentences,
one had to “understand” a second-person subject not ordinarily accessible
to superficial observation. Postal’s example was inspired, as the presence of
a “you” in our understanding of imperatives was uncontroversial; his leap
was to build that intuition into the syntactic grammar, to justify the for-
mal, systematic existence at a specified level of structure of entities whose
presence could only be discerned indirectly. He did it by arguing that such
an assumption simplified the grammar, allowed its rules to be more general
245
and therefore, overall, fewer in number. So ordinary intuition was linked to

formal theoretical “simplicity.”
His most telling argument was based on the distribution of reflexive
pronouns and tags in imperative sentences compared with other types.
The basic question concerned the most economical statement of a set of
observations. Everyone could agree that the most economical grammar,
the one requiring the fewest rules, was to be preferred. Postal discussed a
class of cases in which the apparently simplest statement required either
a complication in the grammar or, more probably, the inability to state at
least one generalization at all. A theory of grammar requiring more com-
plex assumptions and a more abstract grammatical structure permitted
an elegant generalization and a simpler overall grammar—and therefore,
Postal argued, was to be chosen over its ostensibly simpler rival.
In general (says Postal) English sentences require overt subjects. But
there is one class of sentences which doesn’t, imperatives: Go home, wash
yourself. The simplest analysis of these would be to consider them truly sub-
jectless. This violates an apparent generalization about English sentence
structure, but that in itself is not a deadly problem. But consider this:
There exists in English a set of sentences whose direct objects and sub-
jects are coreferential. The direct-object noun phrases in such cases are
obligatorily replaced by reflexive pronouns: pronouns with -self/selves
added to them. I wash I is replaced by I wash myself; Mary washes Mary, by
Mary washes herself, and so on. As long as the subject is present in the sur-
face structure to trigger reflexivization, there is no problem.
But consider an imperative like Wash yourself, which looks like an ordi-
nary reflexive and occurs in nonsubject position, but not coreferentially
to any superficially observable NP. Besides, only the second person occurs
here. How should we handle this? The apparently simplest solution is to
treat these as special cases, not true reflexives. But Postal shows that this
misses important generalizations and complicates the grammar. Or we
might treat imperatives as originally declarative in form, with a second-
person subject. The reflexive rule could apply at this point—and could
apply only in case the object was second person, of course. Hence only the
second person shows up as a reflexive in imperatives. Later, the subject was
deleted. Imperatives then work according to the same principle as declara-
tives (and other sentence types). Postal makes analogous arguments based
on tags of imperatives and declaratives.
Postal’s argument can be seen as the “specimen case” (or “paradigm” in
Kuhn’s (1962) sense) of Classical Transformational Grammar. By illustra-
tion more than by the explicit statement of rules and regulations, it set out
the ways in which syntactic arguments could be made, hypotheses justified.

246
Syntactic factors alone were permitted to function in these arguments (at

an explicit level, in any event): distribution and co-occurrence of lexical
items in sentences. (CTG always and necessarily was a sentence-level gram-
mar, with S as the starting point in the grammar’s generative capacity and
the unit over which generalizations and connections could be stated. So
larger and more abstract textual units, e.g., the written paragraph or the
conversational turn, were off limits.) But as the specimen case, the article
also provided ammunition for expansion of the theory, extension of the
notion of DS, just in case distributional arguments rooted in superficially
accessible syntactic structure could be offered in justification. At the time
this article was published, it is improbable that its author or his coworkers
at MIT saw in it the seeds of revolution or heresy: it appeared to be more
a propaganda text designed to persuade the outside world of the logic and
reasonableness of the still-threatening transformational paradigm, more
an act of homage to the Master than a gauntlet flung in challenge.
If we are to understand the history of the next few years, we must con-
front a major enigma: as of 1965, and even later, we find in the bowels of
Building 20 a group of dedicated coconspirators, united by missionary zeal
and shared purpose. A year or two later, the garment is unraveling, and
by the end of the decade, the mood is total warfare. The field always was
closed off against the outside: no serpent was introduced from outside of
Eden to seduce or corrupt. Any dissension had to be home-brewed. Yet, at
the time Aspects was published, we detect no trace of disaffection, nothing
published or (as well as I recall) openly discussed involving opposition to
standard theory or proposals of novelty. Where did the split come from?
And why did it take so long to come to consciousness?
It is true that as early as 1963, there were proposals made (George
Lakoff’s memorandum of that year, “Toward generative semantics,” being
one) that would have extended the model to be proposed in Aspects. But
these were not seen as expressions of opposition, much less as theoreti-
cal heresies. Chomsky, then as ever, tended to be vague about what each
version of his theory permitted in terms of abstract structure, what the
limits of his model were. What seems retrospectively to have happened is
this: there were from the start two kinds of people who went into trans-
formational grammar. Each kind went into it with the belief that Chomsky
and his system were valid and worth following—as they stood. But because
of the vagueness of Chomsky’s published and oral formulations (a brilliant
theoretical stroke, first, because it diminished the potential for external
opposition; second, because it made less likely the internal factionalization
of the field), both sides saw in what they read and heard what they wanted
to see. It was a version of the blind men and the elephant, except that the
247
men had perfectly good vision, but the elephant was behind a screen, per-
haps less elephant than chimera.
But the parable applies to the situation: the question of the nature of
language and therefore the linguistic theory needed to capture it. As I said,
two kinds of people entered the field, in those first days.3 One group were
basically mathematicians and logicians, by temperament if not by trade.
Their fascination with language was in seeing it as a quasi-mathematical
system, in isolation, like the systems of topology or algebra. They were con-
cerned with predictable regularities, patterns that recurred, and the for-
malisms necessary to capture those generalizations. To show that language
functioned in this way would be to make a deep claim about the logical
capacities of the human mind, to give deeper and more rigorous meaning
to the Cartesian claim that man was a reasoning animal and that those for-
mal rational capacities were intrinsic to and universal in—that is, provided
a definition of—humankind.
If that was one’s aim in doing linguistics, certain assumptions were
natural. You would tend to search for generalizations, stop with the sim-
pler cases on the grounds that they represented the deeper reality; more
complex examples did not necessarily show the system to be wrong, or
entail more abstract or more complex versions, but merely were static,
interference with the deep patterns based on mathematical logic by sur-
face annoyances—other psychological capacities and incapacities, social
involvements: interesting to other kinds of social scientists, maybe, but
off-limits to linguists, irrelevant and uninteresting. And just as, for the
logician or mathematician, the universe could be fragmented up without
distortion into subsystems, smaller worlds, within which generalizations
were more readily accessible, so too language could be seen as a network
of autonomous systems: phonology, syntax, semantics, interdependent
but not formally interconnected. The rules of syntax did not—could not—
mention semantic criteria, and vice versa. Hence, the existence of abstract
3. This dichotomy, like others, is deceptively sharp. It would be erroneous for example
to suggest that the ESTists had no interest in the cognitive aspects of language. This
of course was one of Chomsky’s motivating concerns, and one of particular interest to
Ray Jackendoff among others. Similarly, it would be absurd to suggest that GSists were
uninterested in formalism: George Lakoff and James McCawley in particular were at
pains to develop formal devices and systems and considered these central to GS. I mean
rather that each group tended toward their own emphasis, the end toward which they
strove. EST was—and its descendants remain—chiefly concerned with the description
of language as an autonomous system (which might, significantly, shed light on other,
autonomous, human psychological processes); GS moved more and more in the direction
of seeing language as the reflex of—and inextricably interconnected with—other human
processes, social and cognitive. It is a matter of preferred emphasis more than anything.

248
elements in deep structure could be justified only on the basis of superficial

syntactic regularities, not semantic criteria like synonymy.
But another group were, at heart, humanists, with the significant parts
of their background in language or literature. For them, transformational
theory promised something rather different—something for which they
had entered the humanities, only to find that the assumptions and meth-
ods of those fields closed the door to their curiosity. For them, the prom-
ise was made in Chomsky’s statements that language was a window to the
mind, a way to enter that black box, to see how people actually worked. To
them, being human entailed more than manipulating formal structures. It
had to do with how we thought, and avoided thinking; how we got together
in groups, and why groups had misunderstandings; language as a logical
mode of expression, but “logical” in the sense of “reasonable,” not “system-
atic”: a means of encoding all those complexities that produced literature,
war, and puns. They read the same articles of Chomsky’s as the other group,
but read something quite different into them than their author (Chomsky
was of course a charter member of the first group) intended—but nothing
that could not rationally be derived from what he said explicitly. For them,
the Aspects model was a sketch of the ideal system, an invitation to go
deeper in order to make the language-specific and rather concrete system
of Aspects into the truly universal, abstract system that would be the “win-
dow into the mind” promised largely in works directed outside the profes-
sion, such as Language and mind. For them, a theory and a grammar that
would link language with the reality, psychic and social, that it reflected
and created, was the point. So more complex and irregular structures were
not only interesting, they were the crucial cases. Their complexity gave
hints to the complexity of the system under study, showed how all parts
were interrelated. If you ignored the hard cases as irrelevant, you would
make wrong predictions—and have a pointless and boring theory to boot.
For both groups, then, Aspects was just a point on the path, not the end.
But its vagueness enabled both groups to envisage different ends and not
perceive their essential disagreement. It’s important first that this disagree-
ment was never made explicit—indeed, was visible only retrospectively; and
that it was not a theory-internal conflict, such as can be resolved in terms
of a Kuhnian paradigm shift based on mutually agreed-upon examples and
counterexamples. Rather, the disagreement was about the subject matter
of the theory: what it should encompass, what language was. Therefore,
no evidence could ever have been devised that would convince one group
that the assumptions of the other group were correct. (Individuals could,
and did, on occasion, cross theoretical lines. But this presumably occurred
because, as individuals, they changed their minds about what they were
249
interested in, not because they determined that their preexisting interests
were better served by the other side’s model.)
2. THE THEORY
La théorie, c’est bon, mais ça n’empêche pas d’exister.

Jean-Martin Charcot
I have touched briefly upon the basic tenets of CTG, and the ways in which
GS differed, and some reasons why the proponents of each felt as they did
and worked as they did. But before we leave the arena of scientific dispu-
tation, we should examine more closely the claims of GS. As I see it now,
although it seemed then that there were myriad aspects comprising a richly
textured conceptual system (as indeed there were), they pattern together
into four major claims, each diametrically opposed to some tenet of CTG,
and all fitting together into a cohesive whole, all interdependent.
2.1. The Base Component
We have seen that, in some of his writings, Chomsky held forth the promise
of a syntactically rooted deep structure that, nevertheless, was based upon
the universal rational capacities of the human mind. We have also seen
that, as he exemplified the model DS in other works, it could not have func-
tioned in this way. GS theorists, troubled by the inconsistency, resolved to
reconcile it (as, indeed, Chomsky did later, with Extended Standard Theory,
though in the opposite direction). They took the less-formal Chomsky at
his word: the basis of syntax was logical and universal. But then, all traces
of English-specific features in the DS, or Base Component as a whole (the
latter term unlike the former including the lexicon) had to be eliminated in
favor of forms that could fit in agreeably with the underlying structure of
any natural language. Problems with the CTG model included:
(a) The lexicon. It was often noted that words in one language did not
generally correspond, in meaning or syntactic constraints, to their “syn-
onyms” in any other. The structures in which a causative verb like kill could
be inserted in English were not identical to their Japanese counterparts.
And, clearly, the more figurative uses of words—kill time, a killer exam—
would not necessarily transfer from one language to another. One solution
might be to assume that, in the basic lexicon, “complete” surface-structure

250
type words did not exist. Rather, what were found were atomic elements,
semantic primes, basic concepts common to all languages which (more or
less language-specifically) could be combined together by transformational-
type processes to form the surface words of each language. Kill in this the-
ory is decomposable into several primes: cause, become, not, alive.
A bigger problem than the composition of individual lexical items was
the nature and number of lexical categories at the basic level. In CTG, to
the extent that the issue was dealt with, the categories in the Base were
essentially those found at the surface. Later, this covert assumption was
explicitly codified by the descendants of CTG: no categories were permitted
in the Base that did not exist in the language at the superficial level; and
no lexical item could change categories by transformational rules (or any
other way).
The first postulate meant that “abstract” entities of various types, with
one or two exceptions, could not exist. Both in Aspects and earlier, Chomsky
permitted two types of abstract markers. One was the “dummy” symbol, Δ,
used to indicate (as for the underlying subject of an agentless passive) an
item that was necessarily deletable because it was semantically not fully
specified, and was not needed to function syntactically at a superficial level.
For instance, the grammar recognized that the underlying subject of the
passive was an NP, and that it was semantically not incompatible with the
verb selected from the lexicon. Since the passive transformation removed it
from subject position, it was no longer syntactically required, and could be
deleted: all relevant semantic information about it could be discerned from
the choice of main verb. The other abstract category was more problematic.
It included a set of items dictating that specific transformations were to
operate on the trees which contained them. Each referred to a category
of sentence types: Imp, Q. Thus, if Q was selected in the DS, in English,
the transformational rule of subject-auxiliary switch was triggered. A late
transformational rule deleted these markers; their only surface trace (mak-
ing them recoverable) was in the transformations they had triggered. It
was never really clarified in Aspects just what sorts of phenomena could be
handled by these mechanisms. By permitting into the DS abstract catego-
ries (both the “dummy” and these categories were chosen from the lexicon,
like nouns or verbs), CTG seemed to promise that other abstract categories
could be justified, as long as they were recoverable from the sentence’s later
transformational history and superficial co-occurrence patterns. These,
then, functioned as an Open Sesame to GS, particularly combined with
Postal’s paradigmatic illustration.
GS argued that abstract lexical items parallel to these types existed, as
long as they could be justified as suggested above. Postal’s discussion of
251
imperatives, his claim that abstractness, and consequent transformational

complexity, could be justified as long as it allowed more general statements
and a simpler overall grammar, was the basis of the argument. First, the
claim was made that the surface categorial assignment of a lexical item did
not necessarily reflect its membership in DS; second, that abstract lexical
items could be generated in the Base, leave traces through transforma-
tional operations, but be deleted before the surface structure.
So, for instance, George Lakoff made the argument that the kinds of
structures that could function as (say) the objects of the verb believe were
the same as those that could follow the noun belief: John believed that bats
ate cats, John’s belief that bats ate cats. In both, what followed the word in
question had to be an abstract nominal form: *John believed that oranges/
*John’s belief that oranges. In the CTG lexicon, two separate lexical entries
were required: one, the verb, one, the noun. Clearly, many significant facts
about where each could be inserted were repeated, a noneconomy and a
loss of generality. But if in the lexicon only one of these was listed (say,
the verb), then those properties need only be listed once; and in specified
syntactic conditions, a later transformational rule would change the verb
into the corresponding noun. Since V-N relationships of this kind, and
analogous types with other categories, are very prevalent in language, a
great deal of duplication would be saved by using this system, a saving that
would more than offset the expense of a few extra transformations (since
the lexical properties would have to be written in individually for each such
lexical item, whereas a single T-rule would apply to a whole class of cases).
Beyond this, just as CTG permitted abstract categories such as the one
that included Imp and Q, it might as well (with no unwarranted increase
in power) include other types which had no possible superficial representa-
tion at all: for instance, the logical operators: ∀ and ∃. In several papers,
McCawley (e.g., 1970a) argued for the existence of such items as parts of
lexical structure, and in so doing made the claim that a level of deep struc-
ture which did not permit such abstractions necessarily lost generaliza-
tions and was thus logically untenable.
Related to these simplifications of the lexicon was one other: the number
of categories to which lexical items were assigned was radically diminished
in GS. In CTG, lexical categories in the Base were identical to those on the
surface in two ways: first, as already noted, a single lexical item could not
change its category transformationally. Second, the categories themselves
were the same: as just stated, no categories were permitted in DS that did
not have possible SS representation; and then, those categories that could
be observed at the surface, in toto, were to be found in the Base. For CTG, to
find the deep (and surface) category assignment of any item, all one had to

252
do was look it up in any dictionary. A noun is a noun is a noun. A preposi-

tion is a preposition. And so on. (Although these constraints are never fully
or explicitly stated, they can be extrapolated from Chomsky 1965, 1970,
and 1971.)
But if, as GS postulated for the reasons already given, lexical items could
switch categorial assignment in the course of a sentence’s derivational his-
tory, this offered the possibility to radically simplify the lexicon. If some
apparent nouns had to be analyzed as verbs in DS to achieve generaliza-
tions, then perhaps it made sense to look at all nouns this way: Was there
really a need for “noun” as an underlying category? That was a problem that
perplexed GSists: some (e.g., Bach 1968) argued that even concrete nouns
like table were really representations of the results of actions or events,
that is, states, and therefore verbal at base; others were happy to accept this
analysis for abstract nouns but not concrete or proper names. For other
lexical categories, falling-together made more sense. George Lakoff (1966)
argued that adjectives were but a subclass of verbs. Properties like stativity
were distributed across both classes, and a generalization was achieved by
considering them as identical. Moreover, it was pointed out in the same
discussion that the adjective/verb distinction, so salient in English, was
much less so in other languages. If one was working toward a universal
Base, surely it made sense to admit as part of that Base only those lexi-
cal categories whose existence could be persuasively demonstrated in every
language. The “minor” categories, too, could be combined with others using
arguments like these. Many types of adverbs could be seen to function as
verbs, as could prepositions and conjunctions. Articles were not present in
the lexicon, but transformationally inserted in English as the reflexes of
presuppositions. So ultimately only two categories could be justified at a
DS level: Nouns and Verbs. The others derived their existence through the
operation of transformational rules.
(b) The nonexistence of VP. CTG followed traditional rules of sentence
parsing in dividing a sentence basically into two major components: NP
(née “subject”) and VP (or “predicate”). This division worked well for an
English-specific DS in which main verb and direct object were more or less
inseparable. But it was obviously less viable for a language in which the
verb was followed by subject, then DO (VSO), or any other such possibil-
ity. As long as the opening rule of the phrase structure (PS) separated sen-
tences into NP and VP, it had to be nonuniversal.
Further, GS began to find arguments for a single underlying word order
(VSO) for all languages quite different from the prevailing SVO order of
surface-structure English. Such an order was essential for a theory involving
253
lexical decomposition and prelexical predicate-raising to produce the com-

plex lexical items of the surface structure. McCawley (1970b) argued for
VSO at the level of underlying structure on the grounds that it permit-
ted a simplified and more understandable statement for several important
transformational rules: various raising operations, there-insertion, and
passive. The GSists argued that the justifications CTG proposed for SVO,
and VP in particular, were weak. What was needed to prove the existence
of VP was (at least) one transformational rule that explicitly mentioned it,
that could not be stated at least as well in some other way, by mentioning
some other category (e.g., V alone). No such rule was ever incontrovertibly
found, and at least to the GS mind, all the rules that ever mentioned VP
were stated at least as well if not more elegantly without it.
(c) Word order. In CTG, the (again covert) assumption was that the word
order postulated for a language in the DS should be as close as possible to
the normal or prevalent word order in surface structure. (In general, CTG
preferred a rather spare transformational component, with as few rules
as possible, working on as few sentence types as possible.) So in the DS,
English was represented in Aspects as SVO, since that is the normal order
of the declarative, assumed to be the “basic” sentence type on statistical
grounds. (Covert again.) But in other languages (as Greenberg (1966) had
demonstrated), there was reason to believe that at a superficial level, many
other either possible or mandatory word orders occurred, and CTG would
treat each of these as having a different DS word order (that is, if the pre-
vailing order in surface declaratives was VSO, that would be taken as the
base order, and so on). One problem CTG couldn’t cope with at all was so-
called “free word-order” languages, where, as we would say now, word-order
was governed pragmatically rather than semantically. (Latin is an excellent
case.) In CTG, such a language could be dealt with only very artificially and
ad hocly: the most “prevalent” surface order was decreed to be the “basic”
order, and generated in the Base—e.g., in Latin, SOV. But the “basicness”
of SOV in Latin was of a very different status from the “basicness” of SVO
in English, since word-order changes in the former did not affect semantic
reference, but rather pragmatic function: vividness, topic, cohesion, etc.
More troubling than the false analysis itself is the fact that CTG theory did
not provide a way to preclude it. A favorite validation for formalism is that
it provides a “garbage detector” for incorrect analyses. This one didn’t.
The postulation of a universal basic VSO order had more than aesthetic
justification. McCawley argued that it was not coincidental that VSO was
analogous to the order of items in the propositions of symbolic logic, where
V = predicate and N (that is, S and O) = arguments. GS suggested that the

254
underlying semantic order of natural-language propositions was to be

equated with that of propositional logic, and that the universality of VSO,
in a truly Cartesian sense, rested on its logical structure.
Additionally, VSO was argued for via the familiar Postalian para-
digm: it allowed for the simplification of the transformational component.
McCawley suggested that a number of important transformational rules
could be radically simplified if they operated on a structure of VSO form.
For instance, CTG assumed two rules of raising: to subject and to object.
Intuitively the processes seemed similar, but if the rules operated on SVO
structures, they could not be handled by a single rule, since the raised NPs
had to move in opposite directions. But if the subject and direct object were
on the same side of the verb, the two rules could be compressed to one,
moving the subject out of the leftmost NP of two, or the only NP in an
intransitive sentence. Thus the structural description (SD) of both forms
of raising was greatly simplified, and the two were related formally as well
as intuitively. Other rules that could be significantly simplified and made
more intuitive were neg-raising, passivization, and there-insertion.
The result of the foregoing claims was an underlying structure quite
different from the CTG DS tree—and, at the same time, from the GS or
CTG surface tree. CTG’s trees—at both levels—were horizontally expan-
sive: lots of categories, lots of divisions into separate phrases: NP, PP, AdvP,
and so on. Because these categories so closely replicated the SS ones, the
transformational component had relatively little work to do: it didn’t radi-
cally alter the shape of trees, just reorganized constituents. But the trans-
formational component of GS did a great deal of constructive work. Even
at the DS level, prelexical T-rules put atomic predicates (e.g., [cause], [not],
[alive]) together to form surface-type lexical items (e.g., kill). And T-rules
had to convert verbs into nouns, prepositions into verbs, and so on. They
had to delete many types of abstract items (performatives, for instance).
But of course, the GS surface tree was the same as the TG model. So, for
GS, transformations had to convert a much skinnier, underlying tree to its
horizontally spreading SS counterpart. One Christmas, Haj Ross’s students
at MIT gave him a mobile made of wire hangers, its nodes represented by
Christmas-tree balls, representing the GS version of the DS of Floyd broke
the glass. In CTG, the tree would have been quite simple, as exemplified in
Figure 11.1.
But the GS tree-mobile stretched from the ceiling to the floor of the
office in Building 20, and this was rather early in the history of GS: by the
time it was over, it would have been necessary to break through to the floor
below, as Figure 11.2 shows.
255
NP VP
Floyd Aux V NP
T D N
past break the glass
Figure 11.1 The CTG DS Tree for Floyd broke the glass.
V NP NP NP
say I to you S
V NP NP
cause Floyd S
V NP
happen S
V NP
not S
V NP
be whole glass
Figure 11.2 One version of the GS Underlying Structure for Floyd broke the glass.
A GS derivation, then, would involve a very much simpler PS component

(there might have been no more than three expansion rules: S → V + NP;
NP → N; NP → S); but in any derivation, there would be more applica-
tions of specific transformational rules, as well as more kinds of transfor-
mational rules: this was, in fact, a considerably more powerful system. The
dispute was not over relative power, but over its necessity, with CTG claim-
ing its system was sufficient to cover all that its proponents required of a
grammar, and GS affirming the same of its. But, as we have seen, the argu-
ment existed because each side had a very different notion about what the

256
grammar ought to do, which in turn went back to a covert dispute about
the nature of what the grammar was supposed to describe—language.
2.2. The power of transformations
From the beginning of the schism, one of the favorite insults the tradition-
alists could fling at the rebels was to accuse their grammar of being “too
powerful.” The proper riposte was that the former’s grammar was insuf-
ficiently powerful; or that the GS version was, in fact, no more powerful
than its competitor, if fully understood. Assuming the validity of formal-
ism itself, the criticism was serious if justified: under the assumption of
Occam’s Razor, the best theory is the one which can accomplish all that is
necessary with the least amount of machinery. A too-powerful theory is
one whose mechanisms allow it to do more than is actually needed, and
is therefore uneconomical. Hence, it was important that transformations
be limited both in number and in type: the fewest possible rules, and the
fewest possible kinds of rules, made for the simplest system.
The problem with this assumption, said GS, is that it is difficult or impos-
sible to gauge simplicity over the grammar as a whole. Maybe you can count
rule applications, or features, in fragments of the grammar, but until you
have a complete grammar of a language (by this time, recognized as not
an imminent possibility), you cannot talk sensibly of economy or power.
What might look like an unnecessary efflorescence in the transformational
component when the latter is considered by itself might arguably effect
significant savings overall, by radically simplifying the Base Component.
Indeed, this is just what was argued.
It is true that in GS, the transformational component did a good deal
more work. It did so most obviously because the DS trees needed more
processing by the same rules, cyclically iterated level by level, than did
their equivalents in CTG. (But it was by no means clear that merely hav-
ing rules apply more often made for nonsimplicity of an interesting kind.)
Then too, GS introduced new rules into the grammar: there were prelexi-
cal transformations, for example, which combined the atomic predicates of
the lexicon into the nouns and verbs of the surface structure. There were
rules deleting abstract elements such as the higher performative. There
were global rules. Additionally GS developed a rich stratum of constraints
on the applicability of rules, of which the profusion of cyclical types is but
one instance. There were Ross’s islands (1967); notions of precedence and
command (Langacker 1969); governed rules and major and minor rules
(G. Lakoff 1970); transderivational constraints; and much, much more.
257
There were category-changing rules, turning verbs into abstract nouns.

There were insertion rules, for instance those adding articles based on
presuppositional information. But, GSists argued, even this did not really
complicate the grammar significantly, since these rules, while new, were of
the same types as already-extant processes: they utilized the same familiar
elementary operations of deletion, insertion, permutation, and substitu-
tion. There were some wholly new processes, to be sure: precyclical and
postcyclical rule application, for instance. But it was argued that any theory
of generative grammar would require them.
But the way GS argued most strongly against the criticism of too-great
power was that, in fact, overall GS was economical. The proliferation of the
transformational component was more than compensated for by the aus-
terity of the Base, the paring of both the lexicon and the phrase-structure
rules. And more: the postulation of VSO, with its consequent simplification
of key rules, provided a diminution of power in the T-rules. In CTG, indi-
vidual Ts could operate quite unconstrainedly: permutations might occur
over a wide swath of structure, as was the case for both passivization and
raising. GS suggested that the operation of Ts be constrained: that opera-
tions only take place between contiguous items in an SD. So GS restricted
the power of Ts by confining their operations on phrase-markers; while
CTG did so by restricting the forms the rules themselves could take.
2.3. Pragmantax versus autonomy
That leads directly to the next point of difference. I have already noted that
CTG and EST were predicated on a notion of autonomous language, as well
as autonomous levels of grammar. In a way there is an irony here. It is well
known that CTG took pains to distinguish itself from structuralism (the
“Bad Guys”) on the phonological level by rejecting the latter’s maintenance
of a level of autonomous phonemics.4 (No such effort was made for syn-
tax, largely because structuralists generally left syntax alone.) In general,
said CTG, levels are nonautonomous, systems are interconnected. But in
fact, CTG (to the extent that it troubled itself with semantics at all) saw
4. It is curious that each side saw the other as relapsing into Bloomfieldian heresy.
While the GSists saw the ESTists as neo-crypto-Bloomfieldians because of their non-
universalist Deep Structure, their belief in the autonomy of levels, and their rejection
of true mentalism, EST saw GS as similarly benighted because of their “empiricism”—a
dirty word in the Cartesian circles of CTG/EST, albeit a curious charge against a group
who were at least as introspective and intuitive in their methods as they were them-
selves. (Cf. Katz & Bever 1976.)

258
semantics as separate from syntax. Like phonology, it interpreted the out-

put of the syntactic component, but did not interact with it.
One of the earliest and strongest thrusts of GS was its persistence in
connecting levels, insisting that there was no logical reason to have an
autonomous syntax—indeed, that such a level was logically and formally
untenable. McCawley (1970a) argued that the Aspects DS itself was logi-
cally untenable, and that, therefore, syntactic mechanisms must be linked
directly with semantic entities. Others gave examples showing that speak-
ers’ assumptions about reality, or the context in which sentences were
uttered, crucially affected their syntactic form; presuppositions, formal
and otherwise, were necessary to predict grammaticality (e.g., R. Lakoff
1969, 1970). Indeed, the purely syntactic */of CTG was already in deep
trouble (as we shall see in section 2.5); what GS showed was that the trou-
ble arose because extrasyntactic context had to be taken into account, and
that the issue was appropriateness rather than pure grammaticality. The
choice of some or any was not determined only by syntactic context (pres-
ence of a Neg or Q), but as much by a speaker’s assumptions about the
proposition: Was it good or bad; likely or unlikely? The outside world was
starting to impinge.
GS first gave to the impingement, or interaction, the semifrivolous
name “semantax,” suggesting the indivisibility of levels. Later, as prag-
matic theory developed and was incorporated, the name transmogrified
into “pragmantax.” Ultimately, formal means were developed to bring to
the “sentence” of TG the idea that it did not exist in a communicative vac-
uum, but was informed by the fact that it was uttered by one speaker to
another, in a particular place, at a particular time. Speech-act theory, as
developed by ordinary-language philosopher J. L. Austin (1962), proved
the magic link between syntax and pragmatics, with Ross’s “On declarative
sentences” (1970) providing a way to incorporate the real-world speech act
into the syntactic form via the incorporation of the performative abstract
verb into the sentence as its highest unit. Ross devised arguments to justify
the postulation of abstract declarative verbs by the usual syntactic meth-
ods: distribution and co-occurrence. In fact, his arguments can be seen as
no more than an extension and refinement of Postal’s paradigm case—
transformational orthodoxy.
Later, arguments were made to incorporate more of pragmatics—the
interactional and contextual component—into syntax, or rather, to make
the two, plus semantics, indistinguishable. Just as natural logic had been
made a part of syntax, or semantax, in the late 1960s, with arguments
demonstrating the necessity for incorporating the propositional structure
and quantifiers of symbolic logic into syntax, so conversational logic was
259
brought in, using Grice’s (1975) theory of implicature. Speech-act theory,

introduced into generative syntax through Ross’s performative argu-
ments, was further refined by David Gordon and George Lakoff’s (1971)
conversational postulates, allowing a formal representation of indirect
illocutionary force. A theory of politeness was connected with these, sug-
gesting a link between syntactic form, pragmatic intention, and discourse
context. From there the theory went in many directions: discourse type
and structure; contextual influence on syntax (involving the influence of
personal and psychological characteristics of participants on language
form); and special contextual functions, like the courtroom and thera-
peutic discourse; and functional grammar, the explicit discussion of the
way in which function governs form. The problem, of course, is that all
of these made syntax much more complex: the possibility of one-to-one
relationships, simple predictions like selectional restrictions, became
unthinkable. While much insight was gained, the development of a com-
plete, rigorous, and explicit theory was rendered highly improbable, now
and perhaps forever.5
2.4. The strength of the Katz-P ostal hypothesis
Let us return, gratefully, to an issue more bound to language-internal

syntax, the Katz-Postal hypothesis (K-P). This theoretical claim goes back
to 1964 with Jerrold Katz and Paul Postal’s monograph, An integrated
5. Newmeyer (1980:168) criticizes GS practice as unnecessarily fuzzy:
This “exuberant cataloguing of … facts” became a hallmark of generative seman-

tics, as every counterexample to a claim (real or apparent) was greeted as an
excuse to broaden still further the domain of formal grammar. The data fetish-
ism reached its apogee in fuzzy grammar. Many staunch generative semanticists
who had followed every step of Lakoff’s and Ross’ up to that point turned away
from fuzzy theoretical constructs. “Of course there’s a squish,” they objected.
“There’s always a squish. It’s the nature of data to be squishy. And it’s the pur-
pose of theory to extract order from squishy data.”
Yes. But “order” is not synonymous with “discreteness,” particularly when imposing
the latter creates distortions. In that case, it is the job of the responsible investigator
to divest his or her mind of outmoded beliefs in what “science” or “theory” must be
(according to whom) and tailor explanation to observations, not vice versa. Granted
that fuzziness is unsettling, even frightening: that is no reason to deny its reality. To
abjure nondiscrete theories because they are unsettling, or because they conflict with
the kinds of formalisms we currently feel comfortable with, is antiscientific in the most
dangerous way: analogous to the Church’s determination that Galileo’s claims were
heretical because they were antithetical to current established wisdom.

260
theory of linguistic descriptions. In it, the authors had a serious problem

to solve with pre-A spects transformational syntax and semantics. In this
early form of the theory, transformations were not necessarily meaning-
preserving. Transformations could, for instance, insert meaningful ele-
ments: a negative, a question. Therefore, semantic interpretation had
to be applicable to derivations at two levels: before and after the trans-
formations had applied. Not only was this uneconomical, it led to some
rather unpleasant results, possibilities for self-contradictions, nondeter-
minacy of meanings, etc. So Katz and Postal proposed that the theory
be reformulated so that the transformations could not change meaning.
The meaning given at the basic level (soon to be christened “deep struc-
ture”) was the meaning to be found at the surface. Therefore the semantic
(“projection”) rules needed to apply only once, at the deep level. But to
accomplish this desideratum, transformations had to be stated so that
they did not change meanings. If, for instance, a sentence was negative at
the surface, the negative element had to be introduced at the underlying
level. If an S was to be interrogative, a Q marker had to be present in the
Base, conditioning the application of subject-auxiliary inversion later in
the transformational cycle.
This was a considerable simplification, but from the start had problems
of its own. Chomsky (1965) noted, for instance, that one of his star trans-
formations, passivization, ran into difficulty as a result of Katz-Postal. It
was a tenet of CTG that passive sentences were transformational equiv-
alents of actives—indeed, this had been a primary selling point for the
theory, that it could relate these types so elegantly. But it was noted that
pairs existed that were not (it was claimed) truly equivalent in meaning.
The application of passivization appeared to change meaning:
(1) Everyone in the room speaks two languages.

(2) Two languages are spoken by everyone in the room.
As a result, passivization, which had been cast as an optional rule operat-

ing on any transitive active sentence, was recast as obligatorily operating,
but only on structures that had some sort of triggering device built into
them. The solution destroyed some of the elegance of the original for-
mulation and introduced some distinctly ad hoc elements, but it allowed
the preservation of Katz-Postal (without which Aspects CTG could not
function).
But more problems emerged. One was that “meaning” was imper-
fectly defined. Under what conditions were two sentences said to be
paraphrases—that is, having the same meaning? When did a derivation
261
change meaning, and when did it not? And if it was doctrine that trans-
formations did not change meaning, was that to be taken to imply that, if
two surface structures were equivalent in meaning, they must (the inverse
of K-P) be taken to have a common underlying source? This was the stron-
gest interpretation of Katz-Postal, and GS essentially adopted it, without
too much explicit consideration. The argument was that it was always in
the interest of economy to derive paraphrases from the same underlying
source: thus, selectional restrictions and other constraints need be stated
only once in the Base. Then, if two sentences were arguably paraphrases
(which was defined as, if neither could be true in a context where the other
was false), Occam’s Razor required and K-P allowed that they share a com-
mon source. But what was a true and complete paraphrase? Was (3) accu-
rately paraphrased by (4)?
(3) John sliced the salami with a knife.

(4) John used a knife to slice the salami.
Was (5) paraphrased by (6)?
(5) John killed Bill.

(6) John caused Bill to die.
These issues turned out to be unresolvable, with GS saying “yes” and CTG/
EST “no,” with no agreed-upon way to decide.6
Moreover, the old active/passive bugbear was understood very differ-
ently by GS and EST. The latter (as discussed by Partee (1971) among oth-
ers) kept to their guns: it demonstrated a deep problem with K-P; and the
introduction in EST of interpretive semantics, essentially the reintroduc-
tion of pre-A spects projection rules operating on surface structures, made
K-P unnecessary or untenable in many cases. GS, on the other hand, saw
6. The Bach-Peters Paradox was an additional complication, arguably making things

worse for both sides. The problem, as it was originally propounded by Bach (1970)
was the existence of surface structures with apparently infinite underlying struc-
tures, which therefore could not be represented either as GS logical structures or EST
DSs, e.g.:
(i) The mani who deserves itj will get the prizej hei wants.
If (as GS argued) SS pronouns represented underlying full (and concrete) NPs, there
was no way this example could be completely accounted for. But EST ran into the same
problem, though it might transfer the problem of the representation of the full NPs to
the level of surface structure, via a Semantic Interpretation Rule (SIR).

262
the problematic examples as special cases that did not cast doubt on K-P,
but only on the CTG statement of passive and its lexical theory, as well as
its difficulties in dealing with multiple meaning and nondichotomy. The
problem with the active-passive pair given above is not with active-passive
per se; the transformation is involved in the problem only in that it moves
subject NPs and direct object NPs over each other. When quantifiers cross
this way (as argued by Postal (1971)), changes of meaning may occur, unre-
lated to the transformational operation itself. Only where quantifiers play
the roles of subject and object do passives display this meaning change, so
it is not characteristic of passivization itself, and therefore, K-P does not
fail with passivization. Moreover, it might also be argued, CTG suggested
that the two sentences in question had distinct and different meanings,
which would cause trouble for K-P. Rather, both sentences have the pos-
sibility for both meanings; but each tends to favor one meaning over the
other. Thus, (1) has (in isolation—always a tricky criterion) the primary
meaning “Each person speaks two languages, but they could be any two”;
(2) most likely means “Two languages (the same two) are spoken by all the
people.” Within autonomous syntax, the meaning preference must remain
mysterious. If, however, we introduce pragmatic, textual, or functional
considerations, things get clearer. There is a tendency in English, other
things being equal, to use the subject position for topics, for focal points.
So (1), Everyone in the room… ., suggests, “I’m talking about these people,
my emphasis is on them,” and there is no reason to think that the lan-
guages themselves are being stressed and pointed out. But in the passive
case, the only reason (in isolation again) why the sentence is passivized (a
marked construction) is that we are being asked to focus our attention on
the languages themselves. It is a great deal easier to imagine doing this
in the case of specific languages that are under discussion, rather than
the vague reference point of languages in general. So functional consid-
erations and discourse expectations motivate one interpretation over the
other, and even so, there is no 100% correlation, as would be expected if
the distinction were based on syntax—that is, on the operation of pas-
sivization. So passivization doesn’t pose a problem for K-P, except within
a theory of autonomous syntax.
This doesn’t resolve the issue about the strength of K-P; but again, this
cannot be resolved theory-externally. If you believe in autonomous syntax
and semantic interpretation of surface structures, then a weak or nonex-
istent version of K-P works best for you; if you believe in a semantics-or
pragmatics-driven syntax, then paraphrase relations are deep and impor-
tant, and K-P must be maintained.
263
2.5. Continuum versus dichotomy
CTG saw language as a candidate for formal description, and therefore,

as a system whose components could be assigned to either-or categories
(Noun/Verb; grammatical/ungrammatical; transitive/intransitive; count/
mass; obligatory/optional; and so on). GS at first fell in with this system,
but as its proponents started looking at more complex sentences and more
intricate relationships, it became gradually clearer and clearer that such a
systematicity, however attractive because easily formalizable and readily
organized, would provide a distorted view of language. In particular, Ross,
in a series of papers (e.g., 1972, 1973), talked about “squishes”: cases where
one category flowed into another, where phenomena were best organized
as continua, rather than dichotomies. There was a continuum going from
“verb” to “noun,” rather than an item being always and unambiguously
assignable to one category or the other. It was probably as well that this
understanding came late in its history, as it would have made any formal
statements profoundly more difficult (and perhaps was partly behind the
ultimate despair): transformations, after all, necessarily mention discrete
and dichotomous categories, and no alternative system was offered.
In a sense, the continuum problem was the first thorn in the flesh of CTG,
but it existed mainly as an unseen irritant until quite late in the history
of the field. The problem surfaced first with the redefinition of grammati-
cality as a subcase of appropriateness or acceptability. The first examples
to be discussed—as is necessary in the development of theories—were
unambiguous in this respect. John admires sincerity was fully grammatical;
*Sincerity admires John, fully ungrammatical, and everyone was in complete
agreement on these judgments.
Over several years, disturbing examples turned up as people began
looking at more complex rules, more intricate constructions. Often there
was real uncertainty in an investigator’s mind over the assignment of an
asterisk to an example. Recall, too, that CTG’s methodology was that of
ordinary-language philosophy: the notion of testing examples on large
groups of informants (or even—perish forbid!—getting examples from
real, spontaneous data) was unheard-of. So one linguist’s intuitive judg-
ment was equal to another’s, and there was no way to discriminate. “That’s
not in my dialect,” you could say to a colleague, but that didn’t obligate
him to change his mind. Hence Ross’s version of the Linguist’s National
Anthem: “Oh, see if you can say… .”
As time went on, things got worse. Not only was an example apt
to be judged differently by different people with different theoretical

264
positions, but it became clear that the simple grammatical/ungram-

matical distinction was an oversimplification. Rather, sentences were
strung along a continuum, from the unquestionably grammatical to the
out-and-out salad; from John admires sincerity through Sincerity admires
John to Admire John sincerity or Run a afterward toast. Sentences ran the
gamut from the totally unmarked through those marked?, meaning that
the linguist him/herself was wavering; ?*, wavering in the direction of
ungrammaticality; ??, not only the linguist but several friends and rela-
tions were uncertain; **, not only bad but absolute, positive garbage, and
so on. Later others were added: %, for “dialectal” in the peculiar CTG
sense—some of us like it, and some don’t (where ? = all of us are unsure);
✡ Yiddish-dialectal, e.g., for Y-movement, as in Trouble you want, trouble
you’ll get. Until quite late, though, the problem was not recognized as
intrinsic to, and highly damaging to, the formal superstructure of CTG.
It fell into the category of observer’s weakness: the linguist’s personal
fault. That context of various types was the crucial factor was of course
ignored.
Other continua began, sooner or later, to emerge from the wood-
work, growling ominously. Early on, Ross’s (1969) “Auxiliaries as main
verbs” paper made the point that an essential (and oft-lauded as bril-
liant) dichotomy of Chomsky’s, Aux versus V, led to wrong predictions.
Rather, an Aux was a verb, but with special properties. And there were
many intermediate points on the Aux-to-V scale: if a modal like can is a
pure Aux (a dangerous assumption, of course), then ought is less of one in
one way, and need and dare in another; and have in still another; transi-
tive verbs are closer to Auxes than are intransitive; verbs taking to as a
complementizer, closer to Auxes than those taking that, and among those
taking to, those which can or must delete it (like let or make) Auxier (as
we began to phrase it) than those which cannot. And so on. This per-
spective made the writing of either-or, all-or-nothing transformational
rules and environment statements seem more and more an exercise in
fantasy or futility. Figure 11.3 lists a few of the recognized continua,
from the theory-internal to the metatheoretical. (For more on this, cf.
R. Lakoff 1982.)
Of all the innovations of GS, continuousness, while the first to be
observed, was the last to receive official and explicit discussion; was per-
haps the only one (maybe because it came late) not to be vigorously resisted
by CTG/EST; and in all likelihood will stand as the most significant contri-
bution of GS to human understanding of language, after all these dispu-
tations are forgotten and the disputers in their graves. For one thing, all
of the other points of argument are in one way or another based on the
265
continuum/dichotomy split; for another, our being forced to see language

as crucially consisting of uncertainties, imprecisions, and indiscretenesses
must ultimately cause us to rethink our hopes for formalizing linguistics
and for assigning it to the realm of “science.” Gazdar and Klein (1978) take
the opposite tack. They argue that, because no current mathematical the-
ory can deal with Ross’s continua, the latter should be discarded as a theo-
retical concept. The thinking is oddly reminiscent of that of the Church
in its confrontation with Galileo: if the current conventional wisdom, or
scientific method, is in disagreement with the facts, deny the facts. Eppur
si muove.
Figure 11.3 Continua.
Figure 11.3 may be helpful as a summary of the positions of the two

sides.7 Some of the listings are a trifle frivolous, but the whole diagram
ought to suggest that any kind of all-or-nothing viewpoint, whether it is
brought to bear on grammaticality judgments or modes of argumentation,
is damaging to a field and its practitioners.
The operative assumptions defining generative semantics were
these: universality, natural logic, abstractness, continuousness. Each was
a gauntlet in the face of CTG.
7. The immediately preceding statement should, of course, be taken cautiously (cf. fn.
3 above). In fact, there were many intermediate positions, between the far left of GS
and the far right of EST. We might think in this connection of Paul and Carol Kiparsky,
Susumo Kuno, David Perlmutter, and perhaps Barbara Partee and Charles Fillmore as
centrists of diverse kinds.

266
3. BEHIND THE RIFT
Facts are stupid things.

Ronald Reagan, speech at 1988 Republican National Convention
By now it should be clear that the theoretical assertions of GS would be prob-

lematic to those preferring a CTG world-view. But this does not explain the
bitter emotional discord nor the rancor with which cross-theory arguments
were carried out, nor does it shed light on another important difference
between the schools: the personality and stylistic differences between their
proponents. Both of these issues (the bitterness and the stylistic clash) are
traditionally ignored by historians of the field, or at best dismissed without
close examination or understanding as irrelevant to the “science” itself, the
theoretical and doctrinal division. I would argue on the other hand that
without an understanding of the differences between the kinds of people
in each group—both in the sense of their bitter divisiveness, and their per-
sonality differences—we cannot really fathom what the two theories were
about. Theoretical models, after all, are devised by human beings to meet
some emotional and aesthetic needs of their own, conscious or not. We
have seen that the two sides went into linguistics with two utterly different
agendas, and to a significant degree these agendas reflected their personal-
ities and both personalities and agendas were in turn reflected in scientific
styles: the kind of people they were governed the way they made science,
which in turn affected the way they wrote up their work. Science, I would
argue—if the linguistics of my generation is any example—does not consist
of theories and data alone: those are the bones of the dinosaurs, but from
their bones we know altogether too little about the creatures, don’t really
know how they lived and worked. In the same way, their theories are what
scientists leave behind of their work, accessible to outside inspection. But
it is their approach, their passion, that informs and enlivens the theories—
and, ultimately, accounts for the success or failure of the theory, as much
as any bloodless assessment of “right” and “wrong.” So I want to reminisce
at some length about these two areas, recollections certainly tinged by pas-
sion and by no means “objective.” But part of what I am arguing is that in
this field and perhaps all fields, pure objectivity is a myth. If we think we
are objective, we are deluded, disengaged, or dead. The best we can do is
acknowledge to ourselves and others our bias and subjectivity, and try to be
as reasonable as we can despite it. Newmeyer makes the “objectivity” error
in two dangerous ways—he assumes that he usually is objective, and that
he should be—so let me try to do better here. I will speak first of the style
of argumentation between CTG and GS, the reasons for it, and the results;
267
and then, I will talk about the personal styles of the two groups, and the
consequences for their theories and for linguistics, with some thoughts on
how to do “science” of this kind, or perhaps better, how not to.
One reason for the bitterness of the fight that eventually erupted lay
in the origin of CTG as a union of empathic souls, like-minded thinkers
fiercely arrayed against a common enemy. The sense of camaraderie was
there from the start, in TG’s “us against the world” format. It should be
noted as well that the earliest adherents tended to be people who had tried
other fields and found them, or been found by them to be, wanting. They
were outcasts looking for a group to belong to and be accepted by, some-
thing larger than themselves to which they could make a meaningful con-
tribution. In other ages they might have chosen the Church or the cloister,
only to be expelled for heresy and found a nonconformist group of their
own. We find, then, an unusual group: unusually close-knit, since until the
late 1960s virtually all had spent significant time at MIT officially or oth-
erwise, and all felt they owed an allegiance deeper than professional con-
nection to Chomsky—it verged on worship; unusually committed—they
were unwilling to stay in fields that didn’t promise to make a difference,
to allow them to do something important. They were, then, ambitious,
as well, but at the same time willing to run risks, as they were willing to
break away from careers in established fields. When they joined the move-
ment, therefore, they tended to be older than beginning graduate students
ordinarily would be, with predeveloped ideas and personalities. And this
suggests too that they were unusually strong-minded, even abrasive: they
could and did tell their superiors in established branches of linguistics and
other fields that the new work they were leaving to do was the “real thing”;
that the old stuff was not worth staying around for or committing your life
to. They didn’t all say it directly in so many words, but by leaving they said
it. By breaking those ties, professional and personal, with their own past
and the history of their field they effectively isolated themselves, making
all the more crucial the relationships that were to be forged within the new
group: it was to be family, world, church.
So the rise of dissension around 1965 was unwelcome and frightening,
not unlike the parents’ divorce to a child. It was different, of course, in that
they themselves played a pivotal role in the disagreement—and so each
side tended to blame the other as the starter of the fight. Then not only
did each side have its sense of its own intellectual rightness and the other’s
wrongness to drive them apart, but the additional rancor based on the feel-
ing that they had broken up an idyllic family.
There was added to the undercurrent one more ingredient that made
the eventual fight nastier than it perhaps needed to have been. Any

268
academic field, over its history, develops or borrows the means to defend
itself: evolves, if you will, teeth and claws. In the humanities, which at that
time would have been taken by many if not most linguists to encompass
their field, the notion of argumentation was scholarly and gentlemanly.
Unlike the hard sciences, the humanities did not have (as they still, for
the most part, do not have) the concept of work being done by competing
paradigms, one of which wins out over the others by dint of demonstrated
superiority. Rather, one achieved repute by demonstrating a fine aesthetic
sense, good judgment, the ability to see many facets of a work, to digest all
that had previously been learned about it and add to that—not overturn
it. So it didn’t make sense to engage in brawls about who was the best: it
was in bad taste, and there was no way to determine the “right” idea in any
case: it didn’t even make sense.
In the “hard” or physical sciences, on the other hand, there was such a cri-
terion: the “best” theory was the one that most economically accounted for
the observed data. Both what was “economical” and what was “data” were
givens: everyone in the field could be counted on to agree, at least in prin-
ciple. One theory could and in time probably would destroy another: pat-
ricide and fratricide were daily events, necessary parts of progress. But at
least according to the mythos of the sciences, these debates were solved
by the data, external to the investigators and equally accessible to all: it
was impersonal and objective, it came out of microscopes and telescopes,
not one’s own mind. (Of course, occasional leaks about the way it really is
in science, such as the autobiographical writings of James Watson, should
encourage us to view this myth with a little skepticism; but there was a
truth to it: it was a feasible position because the data were objectively veri-
fiable and competing positions could be tested in replicable ways.)
The social sciences have always held a problematic place in human
knowledge. As ways of understanding reality, they came last, after the
humanities and the physical sciences. Their very name seems an attempt
to filch some of the glory that the latter fields had achieved. Why “social
sciences” rather than “social humanities”? (I mean, of course, not just the
name but the associated method.) Their proponents will claim, naturally,
that the “scientific” aspects of the fields are extant and valid: quantifica-
tion, hypothesis formation, falsification, replication, lots of impressive-
sounding -ations. The question is what they tell us, what we are enabled
to know with certainty as a result of these methods. What we note in all,
over the last three-quarters of a century or so, is steady factionalization,
into smaller or different, competing, often acrimonious fields. Psychology
begets sociology; anthropology begets linguistics; linguistics splits into
infinite subfields, hyphenated and otherwise. Each generation has its own
269
paradigm, or many, its own revolution, or many; in each case, at least judg-
ing from the perspective of my own field, the new one appears to conquer
because it elegantly handles data that the old one could not, or did not
address itself to. But always overlooked is the fact that the old one elegantly
did things that the new one cannot, or disdains to do. There seems to be no
theory remotely capable of rigorously making order of all the richness of
the data, and one chooses one’s theory partly according to where one goes
to school and what generation one belongs to, and partly, as I suggested
earlier, based on one’s personal preference about what issues are interest-
ing, central, and crucial.
In other words, my feeling at this juncture is that there is no hard evi-
dence that the social sciences really are “sciences” rather than other modes
of organized knowledge acquisition masquerading in the garments of sci-
ence, but no more science than I would be Einstein if I put on an Einstein
mask and talked with a German accent. One can imitate science—as astrol-
ogy does in a somewhat different way—by insisting on its surface fea-
tures: avoidance of “mentalism”; quantification; formalism; discreteness of
categories. But the results do not resemble those of science, nor does the
behavior of social scientists resemble the ways in which the participants
in those fields behave. It may be that we are taking the image of “science”
based, after all, ultimately upon astronomy as the first modern science,
and basing our behavior as linguists upon that. But this might be a fal-
lacy. Astronomy worked according to its rules—as originally postulated by
the likes of Ptolemy, Galileo, Copernicus—and worked well as a predictive
model because it was focused on objects totally remote from ourselves, not
in any direct way connected to us. That disconnection allowed the kinds
of observations that in turn permitted the quantification and discreteness
that underlie the physical sciences and allow them to produce their pro-
found results. But when observation of necessity turns inward, when the
investigator in one way or another must be the instrument that evaluates
the data, or at least some human mind must gauge the meaning of prod-
ucts of some human mind, then those methods become less reliable. When
the data themselves do not come in finite, discrete, and unambiguous for-
mat, but everything is relative to its context, and context itself is highly
subjective, and these are basic realities about language use, crucial parts of
the structure, not annoying encumbrances—then the methods that work
so well for other kinds of data cannot be automatically appropriated for
these new ones. It may be that they could yet be shown to be equally valid,
though probably with profound modifications. But the social sciences have
adopted the methods as unthinkingly as they did the name; and the fact
that they get “results” thereby doesn’t mean they are correct or meaningful

270
results. Of all the social sciences I fear linguistics is the most culpable here,
since it depends the most crucially on the kind of data I am talking about—
the artifacts of the mind.8
Linguistics, like the other social sciences, assumes that it can and should
operate via the “scientific method” developed for the physical sciences since
the fifteenth century; assumes that it is both possible and desirable to treat
the artifacts of language, which are discoverable only through the use of the
investigator’s mind as an instrument (that is, partially or wholly through the
use of intuition), as though they were molecules or stars. But some residual
doubt remains in the social sciences about whether this assumption is rea-
sonable; the problem is that the doubt remains repressed, covert. It seems
to me the question should be asked and investigated, or we should admit
we cannot do so, leave the question open, and call our various disciplines by
some other blanket term than “social science.” Otherwise, as psychoanaly-
sis (another field subject to the same sorts of uncertainties, for many of
the same reasons) would argue, the repressed fear is apt to lead to distress-
ing superficial behavior. For instance, philosophers of science, taking the
unquestioned physical sciences as their model, have proposed definitions of
“science” meant to generalize about what links these fields conceptually. For
practitioners of those fields, if the question is of interest at all, it is of pass-
ing interest, and has not (to my knowledge) sparked much soul-searching.
But the work of these scholars (especially Popper and Kuhn) has been taken
up with passion within a number of the social sciences, where it has been
treated as a litmus test for legitimacy or entry in the club. Therefore, the
claims assume huge symbolic importance: if you can prove that the findings
8. It may be useful here to recognize a distinction between “doing science” and “play-
ing science”—as children play Doctor, or House. Doing science entails utilizing the
scientific method because it has proved, over time, to be useful in facilitating lasting
discoveries and deep understanding of natural phenomena. On the other hand, when
children play at adult occupations, they grasp at superficial behaviors without under-
standing their deeper purpose. We must be very sure that that is not the case when we,
as linguists, call ourselves “social scientists.” There is a danger of a valuing the superfi-
cial manifestations of “scientific” behavior as validating for their own sake: quantify-
ing, formalizing, replicating, and so on make us feel like real scientists, grownup and
responsible—but do they produce lastingly valid results in our field? A corollary of
“playing science” is the overvaluation of theory at the expense of observed data, as
represented perhaps most obviously in the many Chomskyan gibes at “empiricism”
as stupid, culminating in utterances like Newmeyer’s cited in footnote 5, or Gazdar
and Klein’s (1978:666) statement to the same effect. It may also be germane to note
that the paradigm “science” social scientists like linguists are prone to take as a model
is Newtonian physics, with its dichotomies, objectivity, and certainty. But quantum
physics has cast doubt on all these vaunted desiderata, and we might ponder the dubi-
ous advantages of modeling our own theory and method on those of an obsolescent
field. Phlogiston, anyone?
271
of your field are falsifiable, that your field has paradigms, etc., you are
respectable. The implicit assumptions, of course, boggle the mind.
The regrettable surface behavior that arises from these self-doubts is
not long in coming. I recall an article written by Ray Dougherty (1974). In
it he argued that Extended Standard Theory was a science because it had
scientific revolutions; and indeed a better science than any of its competi-
tors because it had more revolutions. I cringed at the time to see this sort
of argument appear in a refereed journal (its very appearance might be said
to disprove its claim); but my chagrin was somewhat assuaged a few years
later (misery loves company) when I encountered, in an equally august psy-
choanalytic journal, an article by Kurt Eissler, one of the giants of the field,
arguing that psychoanalysis was too a science, pace Popper, because it had
lotsnlots of paradigms. Case closed.
Suppose what I have proposed has some truth to it: suppose the social
“sciences” in general, and linguistics in particular, have yet to demonstrate
the appropriateness of the scientific method to their subject matter, the
working of the human mind. Suppose a large part of the work of this field
is, despite our recent disclaimers, still humanistic at heart: dedicated to
figuring out what is individual, how a person creates himself or herself,
and therefore beyond the reach of statistics, of replicable experimentation.
Then what happens if we graft the argumentative techniques that work so
well for science upon a humanistic study? What will we get? Modern lin-
guistics, I suggest—to its misfortune.
Science, with its distinct and contrasting paradigms, works by an adver-
sarial method: only one of us is right, and I intend to show that it has to be
me. Because there is agreement on the basic issues—method and data—if
one interpretation can be shown to be superior in its explanatory capacity,
everyone will sooner or later agree on it, and that will become the prevail-
ing model. While at any moment, in any active science, there are always
several areas of passionate disagreement, there is normally expectation
that sooner or later evidence will transpire that will be persuasive for one
approach. And while scientists do of course take the disagreements some-
what personally and get excited and even angry with one another when
their views are challenged, it does not seem that these passions normally
result in formal rifts. (One can, of course, think of occasional exceptions,
such as the Leakey/Johanson fracas in physical anthropology*. It may not
* In the late 1970s, the discovery of hominid bones (including Lucy) in the Olduvai
Gorge by a team led by Donald C. Johanson and the discovery of fossilized hominid
footprints at Laetoli by a team led by Mary Leakey were each interpreted rather dif-
ferently by the other. Johanson believed both to have been left by the newly named

272
be coincidental that this field is on the border between science and social
science, nor that the necessary data are inadequate, perhaps forever.)
So adversarial argumentation is not a serious problem in fields in which
there is confidence that external and objective grounds are bound to emerge
to prove one side right. And in those fields in which this is not part of the
world-view, this kind of conflictual discussion is not encountered, for it would
clearly accomplish nothing. But the social sciences, and most especially lin-
guistics, are in the middle, with the focus of the humanities but attempting
the methods of the sciences. And Chomsky brought to bear one further tool
of the latter, the better to make linguistics rigorous and respectable: the tra-
dition of contentious and acrimonious adversarial argumentation.
It was first applied, of course, to foes within linguistics or allied fields
(Lees (1960) against Bolinger; Chomsky (1959) versus Skinner). This
served, more than anything, to create the climate in the early years of CTG
of “us against the world,” TG surrounded by vicious enemies. I remember
well the times that nontransformationalists would speak at MIT, in those
early years when the field still saw itself as fighting for survival in a hostile
world. Rather than attempting to charm, conciliate, find points of connec-
tion, the circle at MIT regularly went for blood. Points were made by obvi-
ous public demolition; the question or counterexample that brought the
offender to his knees was repeated for weeks or months afterward with
relish. TG did not win, then, by gradually persuading its opposition, but
mostly by waiting until they retired or died. Since the field had been quite
small, this didn’t take very long. Those who were not won over or gotten rid
of were rendered ineffectual. There was no place for pluralism.
This habit of victorious battle felt very good to these young people,
ardent and ambitious. But by 1964, certainly, the battle was won. No more
opponents came riding into Cambridge eager to joust with the champion.
Could they let their lances rust, slide into a gentle middle age? Impossible!
So when the time came that dissension arose within their own ranks, they
were primed for blood. Everything led inevitably to conflict and implacable
hostility: (i) the fact that people had entered TG with two very different
agendas, and didn’t know it; (ii) the fact that linguistics was caught between
the methods and data of science and humanities, and wasn’t cognizant of
Australopithecus afarensis, an ancestor of both Homo and later Australopithecines,

while Leakey believed the two genera to have diverged earlier, with the remains at
Olduvai being Australopithecus and those at Laetoli being more closely related to mod-
ern humans. The often-rancorous debate between various members of the Leakey fam-
ily and Johanson and colleagues was part of a larger (still ongoing) debate in physical
anthropology about the relationship between Australopithecus and Homo.—Ed.
273
this; (iii) the fact that early transformationalists had emerged, rather late
in life, from other fields that had disappointed them, and they were thus
feisty, ambitious, and in a hurry to prove themselves; (iv) the fact that they
acquired a taste for blood early on, then ran out of prey, and had no one to
turn their weapons against but one another. Thus the bitterness, the divi-
siveness, the insolubility of the struggle. The seeds were sown long before
dissension appeared; once any source of serious disagreement manifested
itself, the outcome would be inevitable.
The extramural hostility had one other unfortunate result, at least for
one side in the eventual struggle. It was seen as very important not to let
the Bad Guys see your weaknesses, if any: rather than be honest, acknowl-
edge that TG had its flaws, things it couldn’t do, the requirement of adver-
sarial discourse was that one present a pose of perfect poise and complete
certainty. Within, at least sometimes, one could groan about the failings
of one’s field, the immensity of the labor (although there was normally an
atmosphere of buoyancy, the sense that all apparent problems were capable
of solution, and very soon); but to the world, one presented one’s best face.
Not only did this serve to further alienate the outsiders, the brighter of
whom could clearly see the weaknesses inherent in the model, but were
hooted at by the faithful if they attempted any critique; but it meant that
generations of students were educated to believe that success involved PR
as much as insight or hard work. If you could overlook your own inadequa-
cies, maybe they would go away, and at least no one else would tease you
about them. The latter might be true, but they didn’t go away, and the atti-
tude created closed minds and an unwillingness to question the established
doctrine, to explore, to combine their theory with others. Along with the
adversarial argumentation, it guaranteed that CTG would remain closed to
outside influences—ensuring a short life, or at least a tedious one.
Because they had closed themselves off to acceptance by the outside
world, it was critically important for transformationalists that they feel
loved and accepted by each other. It was of equal importance that they be
able to see their colleagues as worthy of love and respect. Both of these atti-
tudes were especially true with regard to Chomsky, and became more true
with regard to him as time passed and his reputation soared in the outside
world. In the early days, now barely remembered folklore, before his fame,
it is related that he was readily available—physically and even psychologi-
cally. You could wander into his office, and he’d take time to talk to you.
He might even be persuaded by something you’d said (Postal managed to
persuade him of the need for deep structure, quite a change in theory!),
and maybe if you were favored you’d get a footnote or bibliographical ref-
erence in his next article. As time went on, both kinds of availability were

274
lessened. One would think that as someone’s repute grew, they would feel
more secure, would be able to be more open to reasoned discussion, new
ideas… . One would think that, but in some cases one would be wrong.
It might work the other way: such a person might feel insecure despite
the laurels, and feel that only by maintaining a firm hold on the doctrine
could he maintain his influence. Or someone might feel that his glory only
proved that he alone had possession of the truth, and therefore to listen to
anyone else was injurious to the field. In either case, the result was closure
of the mind.
Not really paradoxically, as Chomsky himself became less available in
mind and body, his status among the students as cult figure rose. He had
always been seen as a figure of towering intellect and integrity; these per-
ceptions increased. Hence, to have Chomsky’s approval meant even more
than it had before, even more than a prominent professor’s blessing means
to graduate students generally. Yes, to be in Chomsky’s good graces meant
mentions in his writings, getting your work published, getting a job; but it
also meant that you were worthy of him, you partook in some small way in
the godhead. For that to be withdrawn was equivalent to banishment from
the Kingdom of Heaven. I realize I am again straying into hyperbole and
religious imagery, and I can only assure the reader that it is employed to
capture a mood. To lose the goodwill of one’s fellow-transformationalists
was less serious than losing Chomsky’s but was still painful in this close-
knit and indrawn community. For all these reasons, then, when the split
came, it had to be highly unpleasant.
The worst of it was that, when Chomsky finally did address himself offi-
cially to the arguments of GS (as early as 1967 in lectures, later (1970) in
print), the level of hostility and close-mindedness was truly disillusioning.
It wasn’t even the sense of having fallen from favor that stung the most,
though that certainly smarted; worse was the sense that an idol had fallen,
leaving nothing else to believe in. The late 1960s was an era of idealism
and hope for a better world, and to see for the first time that the person
one was following to create the New Jerusalem had serious human foibles
probably hit worse, at this moment in history, than it might have earlier or
later, in more cynical times. So people unconsciously blamed Chomsky not
only for being unreasonable as a scholar (which was legitimate), but also
for destroying their illusions (which scholars are not supposed to carry into
their professional lives).
The result can be seen as instantiating a process first described by
Elisabeth Kubler-Ross in those facing death, and then extended to people
in situations in which their sense of self and their relation to their real-
ity is shaken: divorce, serious illness, job loss. It happened here too. First
275
there is a period of denial: Chomsky didn’t say that, didn’t mean it that

way; it isn’t the case that he is refusing to see me or talk with me, it’s
just he’s terribly busy; he didn’t deliberately distort my position in his
paper, he just didn’t read what I said fully—probably because I wasn’t
clear enough. But as incidents multiplied, the denial became untenable
and the next stage was reached: anger. This occupied a great deal of the
next several years, with hostile argumentation back and forth in public
and private. Once Chomsky was seen not to be an idol, he was recast as
satanic, the Enemy. A great deal of the history of GS occurred within this
framework. Then there is a period of bargaining: if I am good, if I am rea-
sonable, maybe this horror will go away. GS entered a time of persuasion,
attempts at rapprochement with the official successor to CTG, named by
Chomsky Extended Standard Theory. The assumption was that if we could
but find the perfect, clear, telling example, analogous to the passive trans-
formation ten years earlier in persuading the heathen to religion… . If
we could but show them that we were looking at things in a way that was
not entirely incompatible with theirs, but ours was better… . And so on.
But since the GS/EST split mirrored the real world/autonomous system
split that had existed underground from the beginning, persuasion was
out of the question. And then, finally, the end: acceptance, the realiza-
tion that there were finally and irreparably two schools, no more unity,
no more us-against-the-world. Camelot had fallen. Not surprisingly, given
this perspective, not long after GS realized the futility of the battle, it dis-
integrated on its own. Much of the fun was in the fighting, and without
fun, there was no GS.
4. THE ROLE OF THE PERSONAL IN

THE THEORETICO-P OLITICAL
Bliss was it in that dawn to be alive,

But to be young was very heaven!
William Wordsworth, The Prelude, bk. 11
Which brings me to my next point: the personalities of the two sides, how

these are reflected in their work and their presentation of their work, and
what that implies about how scholarly work (“science” if you insist on that
term for linguistics) is done.
If you watch movies or television shows in which scholarly persons play
a role, you can get some notion of how a scholarly person is supposed by
the outside world to be, and by extension, how scholarship is supposed

276
to be done. Scholars are often depicted in the media as abstracted from

reality, and either humorless or possessed of a fey humor lost on more
practical people. They babble on unintelligibly about things no sane per-
son could possibly care about. Whatever it is they do, they don’t have fun.
Indeed, scholarship as a whole is distinguished by being that area of human
endeavor that is neither fun nor useful in any way. And, just as (as I sug-
gested earlier) linguists took having Kuhnian revolutions as the litmus test
of science, and thereby could consider their field a science, so (one could
argue) at least some of them, still insecure about their place in the uni-
verse, took the stereotypes of not having fun and of being uninvolved in real-
ity as yet another litmus test: if you Had Fun or Dealt with Reality, you
weren’t playing the game right, you were not a member of the club. It is
useful to bear in mind here too that a generalization, even if it were accu-
rate, is not the same as a test of membership; and only an insecure group
would claim it was.
I have already said that the early proponents of transformational gram-
mar came into the field with two very different implicit agendas. This should
suggest that, in other aspects of personality, they were very different kinds
of people. It is often argued by the Pure Science Club that theory forma-
tion, or preference for one theory over another, is based on purely intellec-
tual criteria: just the facts, ma’am. But in a field like linguistics in which (as
we have seen) both the very identity of the facts and what they proved were
open to dispute, obviously the criterion of factual accuracy was not suf-
ficient (and maybe not necessary either, judging from some of the claims
made over the years about what was a grammatical sentence of English and
what was not). Although most of us would not have admitted this under
torture, I think it probable that, as often as not, we select our theoretical
positions based on other, more personal biases: the way we want the world
to work. We justify these positions ex post facto by finding the right kinds
of facts for our preferred positions. Conveniently, in linguistics, it gener-
ally happens that there are enough facts to go around: some facts that fit
neatly with Theory A, others that tend to support Theory B, still others
that dovetail with Theories C … N. No theory yet known, of course, covers
more than a fragment of the observed and collected data without a lot of
Procrustean pulling and chopping, which tends to do the facts something
of a disservice. In any case, the incipient generative semanticists gravitated
by force of personality into the kind of work they did and concern for the
kinds of data they dealt with, and likewise for the proponents of EST. Thus,
there were three basic facts about the personal styles of generative seman-
ticists which had great influence on their work and the way they talked
about it: (i) disorganization; (ii) nonhierarchy; (iii) nonformalism, or at
277
least, an interest in formalism only as a descriptive convenience—not as

an end in itself. These stylistic issues, when they are discussed at all, which
is seldom, are treated dismissively: trivia which at best just underscore the
theoretical perversity of the lot; or, occasionally, as much more seriously
thought-out decisions than they were, based on doctrine rather than—as
is actually the case—personal caprice. But rather than doctrine molding
style in this situation, it makes much more sense to say that underlying
personality style molded doctrine.
These three significant traits were shared more or less by most of the
people who formed the hard core of GS, and thus gave shape to the organi-
zation (even as the same was true of their counterparts in CTG and EST).
Actually, only one can properly be said to be characteristic of the individ-
uals themselves; the others have to do with the idiosyncrasies of the GS
organizational structure itself. They were all interconnected, though.
First, and perhaps most significant: GS was, organizationally, disorga-
nized, or at best, collaborative rather than hierarchical. There were leaders,
but they were seen not so much as masters but primi inter pares. The spirit
was democratic. (The leaders are, as they were sometimes called sardonically,
the Four Horsemen of the Apocalypse: George Lakoff, James McCawley,
Paul Postal, and Haj (“John Robert”) Ross.) Doctrine, therefore, was not
determined by one Personage at the top of the hierarchy, who alone deter-
mined what was in, what was out; but was arrived at in debate, discussion,
anguished late-night phone calls. This made for a lively group and a lively
field, but one that must have seemed chaotic to the outer world, especially
those used to the rigid ipse dixits of MIT. Chomsky had disciples in a strict
hierarchy: there were the inner circle, the various outer circles, Limbo, and
Bad Guys. GS just had a bunch of people who got together at conferences
to make puns and play Fictionary and smoke funny cigarettes. Theory, and
fact, changed rapidly, depending on whom you asked and when you asked
it. There was no central bureaucracy to tell people the doctrinal Flavor of
the Month. It took a certain sort of mind to tolerate this chaos and flourish
in the climate. Disorganized, said the outside world. Wild and crazy, they
themselves would have said had they given it much thought. They thought
of themselves as rebels, young Turks—but so had Chomsky and his first
disciples, a mere dozen years before. The latter, however, turned into the
Old Guard at the first convenient moment.
There was no GS Establishment, no Congregatio de Propaganda Fide, to
keep GS pure. I think in retrospect that at least some of us, sometimes, saw
GS as presenting a united front, and its adherents as people with similar
beliefs and ultimate aims. But there was never a shibboleth equivalent to
CTG’s assumption of a syntactically based DS, and the various adherents

278
of GS had very different notions about practically all the tenets that
Newmeyer identifies as Generative Semantic Orthodoxy. Indeed, judging
from where we each stand now, years later, it seems fair to say that GS
represented at best a loose coalition of interests: we joined together only at
the points where our interests happened to coincide.
It makes sense, I suppose, that people with a high tolerance— or
preference—for chaos would tolerate and delight in a theory which pre-
sumed and necessitated chaos, one which encompassed as its domain all
human endeavor. And those who preferred a tighter organizational struc-
ture would also prefer a more rigid theory that drew strict limits on what
was a part of its realm, what could be part of the theory and what could not.
One might also argue that these basic personality structures colored other
aspects of the theories of each side. If your preference is an organization in
which one prestigious person controls what is believed, then there will be a
certain tendency to authoritarian argumentation; a desire not to make one’s
case too simply, in such a way that it might be intelligible to hoi polloi, the
nonelect. In such a system profundity or brilliance might be directly equated
with turgidity; preferred style would bristle with arcane references, innu-
merable vaguely relevant footnotes, untranslated quotations in fifteenth-
century French or nineteenth-century German: anything to intimidate and
stun the reader (something no representative of the GS mentality would
ever do, as the epigraphs scattered herein make clear). Examples were
shunned, to be used sparingly only when utterly unavoidable: otherwise
the opposition might see what facts you meant to allude to, and test your
theory out on them. In short, orderly minds too often display a fondness for
authoritarian systems. Mussolini made the trains run on time.
CTG and EST, then, as well as their current descendants, are hierarchical
in organization. In this respect as well as in their dependency on formal-
ism, they are masculine.9 I had mentioned in earlier work that formalism
was a male perspective, and Newmeyer (1980:169) has seen fit to make fun
9. Once again the reader is adjured to beware of easy dichotomies, including M/F. But
although individuals display a range of behaviors and orientations along a continuum
between those points, societies tend to see the sexes as polarized: every behavior,
every physical and psychological trait, is identified as either Male or Female, Masculine
or Feminine. It is in this sense that we might say that GS reflects certain stereotypi-
cally feminine properties, EST more (stereotypically) masculine ones. The fact that
the majority of both groups were males (as was to be expected in academia of that
period—and this) is not especially relevant in this perspective. We would say rather
that GS allowed both its male and its female members to have access to aspects of their
psychologies that this society has identified as typically feminine; and the opposite for
EST. I do find it surprising though that more than one of my colleagues of the GS per-
suasion, and of both genders, have responded to the suggestion that GS was in some
sense “feminine” as though it were an insult. Rest assured, it isn’t.
279
of the statement, apparently without asking what lay behind it. I meant
more than the obvious fact that men tend to be overrepresented in the for-
mal end of the field—the more formal, the more masculine—and women
at the informal, data-oriented end. Of course there are counterexamples—
it would be stupid, especially for a generative semanticist, to claim that
humans are irrevocably dichotomized by gender. I am speaking, as we must
in talking of human possibilities, of general tendencies. But I meant more
than this. There is now an impressive array of evidence that, from earliest
infancy, males respond to stimuli differently from females. Male neonates
are more responsive to their external physical surroundings: light, warmth.
As they get older, they are more active—interactive with their physical
environs, more eager to play with things, objects. This involvement with
inanimate externals is evident again in the games they play, having to do
with external goals, manipulation of objects. Little girls, on the other hand,
are immediately more responsive to people, being held, voices, eyes. They
smile sooner, recognize others sooner. Later, their games are interactive,
concerned with reaching out to others: House, Doctor, and so on.
We can see formalism as maximally noninteractive; and autonomous-
language theories as treating language as an external, impersonal object.
On the other hand, a theory that is concerned with language as an interac-
tive strategy, linking people with one another more or less successfully, is
closer to the way women tend to approach the world. And a hierarchical
theory is masculine as well in that it tends to recapitulate the structure
of male institutions: government, the military, the university, and so on
are and have always been organized in a hierarchical fashion, with a single
authority at the top—like CTG. On the other hand, female institutions
or groups, in those cultures in which they exist, tend to be more collec-
tive, cooperative, or collaborative: there is no formal leadership structure.
This tends to make such groups more fluid and impermanent, and some-
times more disorderly, but they usually get things done. I am—it should be
clear—not saying that EST was the field for manly men and only those, and
GS only for womanly women. But I am equating the theoretical preferences
I wonder, though, whether the “feminine” nature of GS and the “masculine” one
of EST is what has led to the perception of the former within academic linguistics as
less successful than the latter. After all, a “feminine” perspective is antithetical to the
hierarchical, dichotomizing tendencies that are characteristic of academic discourse
and masculine world-view. (And the academic perspective is inextricably masculine,
as it has been for the last couple of millennia in the West.) Here, as often, to under-
stand what the reality is, we may have to wrest ourselves from the prevailing ideology.
Several books have recently appeared within a feminist framework offering reassess-
ments of some of the culture’s unexamined verities: “objectivity,” “science,” “reason,”
and so on (Belenky et al. (1986); Gilligan (1982); Keller (1985)).

280
of each with its organizational structure, suggesting the connection was

not coincidental; and suggesting that there are models of each in typical
gender-oriented behavior.
Generative semantics dealt with a much more convoluted world, but
attempted to make sense of it without reducing it to orderliness. We were
not all born writers, some of us were clearer than others. But we tried to be
intelligible, did the best we could to overcome the stylistic handicap of being
academics. We tended to keep down the number of footnotes, and avoided
long foreign quotations. Above all, we liked examples. Actually, we loved
examples—couldn’t have too many. They tended to proliferate on the pages
of GS articles, an embarrassment of riches, more than was needed to estab-
lish the point being defended. The reason, really, was that there was always
another that proved the point in a slightly different way. The real reason,
though, invokes another point in my discussion: the examples were fun.
Fun was what we were in the world to have. So there would be examples.10
In fact, it is arguably and regrettably the case that, all too often, we
would sacrifice force and clarity for fun in compiling examples. In choos-
ing between a boring sentence that was unquestionably grammatical and
clearly made its point, and a droller one that was dubious or murkily rel-
evant, we tended to choose the latter. Political references, arch invocations
of sex and drugs, weird names, peculiar behavior—we had ‘em all, and
wouldn’t have traded them for the world. It is in part for this reason that
vintage GS papers are often so hard to comprehend today: we have lost the
context (ironically enough): both the topical references and the spirit that
pervaded the writing are impossible to recover.
While we might have made things a little more unclear by our choice
of examples, I think in the long run their frolicsome nature worked in the
direction of intelligibility. It is axiomatic among educators that the live-
lier the text, the more likely it is to stick in the student’s mind. Readers
10. There is more to the GS romance with examples, though I don’t think we saw it
at the time. Examples are egalitarian: they allow each reader to form his or her own
conclusions, based on direct access to the same evidence used by the writer. The her-
meneutic Master, who controls the disciples’ exegesis, in disciplines as diverse as CTG
and deconstructionism, makes decisions on an ipse dixit basis: he (and it is by no acci-
dent virtually always a he) decides what his pronouncements meant, mean, and will
mean; he controls the development of the theory because he is the source of under-
standing. By contrast, a tradition relying on prodigality of examples is saying to the
reader, insider or outsider: Here are the facts. Make what you will of them. The reader is
thereby empowered, relative to the author(ity). One can see why this style was natural
to a group that came to maturity in the late 1960s, and why it might be threatening
to those, younger or older, who preferred a more authoritarian relationship between
Master and disciples.
281
will be more apt to make sense of an argument if its examples are fun,
they’ll be less apt to get drowsy, and less apt to forget: if they at least
remember the examples, they may be able later to reconstruct from them
the arguments they supported. I don’t want to suggest that we invented
those examples from such laudable motives, but they can be defended in
this way. Certainly I would take issue with those curmudgeons (I think of
Stockwell and Newmeyer)11 who object to GS writing on the grounds that
examples were funny, names for rules or principles were frivolous (“Pied
Piping,” “Flip,” “WH-iz deletion”), or they wrote using slang or colloquial-
isms, or heaven forfend, puns. Their argument seems to be that it is inde-
corous, unscholarly—scholars don’t want to Have Fun, and so anyone who
is observably Having Fun is not a scholar.
If GS had had fun at the expense of accurate description, that would be
grounds for criticism, of course. But this is not anyone’s claim. Just as the
establishment made assumptions about the scientific status of linguistics,
it made the same unexamined assumptions about what constituted respon-
sible scholarship. One can certainly ask which stylistic strategy is the more
preferable: murky unintelligibility or quirky frivolity. I don’t know how to
decide, but I do know which I’d prefer to read.
Their collective high spirits point up another salient trait of GS as a
field: the tendency for its writings to point inward, for arguments and
claims often to be implicit. I know I just said that GS writing is especially
accessible. But I ought perhaps to qualify that statement, and shed more
11. Stockwell 1977:131, fn. 2:
In the history of transformational grammar, several scholars have given humor-

ous names like Pied Piping, Though-Movement, and Sluicing to syntactic rules.
Unfortunately, such names, a few years later, are neither mnemonic nor trans-
parent in their meaning.
We might note in rebuttal that many CTG/EST names are no more mnemonic: Root
Transformations, Strict Subcategorization, the A-over-A principle being just the first
that spring to mind. But the real point is that there is no reason for names to be mne-
monic: one associates the names with processes, just as one associates human names
with faces. Stockwell doesn’t especially look like a Robert, but that doesn’t make it
harder to recognize him.
In the same vein, Newmeyer (1980:171f.) grouses:
Such stylistic traits [of GS: he is referring to its practitioners’ “whimsical style of
presentation”] only served to give extra credibility to the charge of lack of seri-
ousness… . Indeed, it is tempting to speculate that generative semantic style is
but a classic example of content both shaping form and dominating it.
A comment which reflects a lack of understanding of both the style and the
content of GS.

282
light on another claim I made above, about the evanescent quality of GS

writing. On a sort of local or superficial level, GS papers tended to be quite
clear and aboveboard. But at a deeper level, it was different. This was not
apparent to the original writers and their primary audience, but surfaces
embarrassingly when we reread our writings now. It is often very hard to
discover just what claims are being made, with what ultimate theoretical
purpose: How do these arguments fit into a more general scheme? What
prompted them? Why are they important? The feeling is that the writer
knew, but didn’t care to say.
If GS was in fact turned inward, there is a historical reason for it, the
same one as informs the style. TG, under Chomsky’s tutelage, always saw
itself as oriented to the outside world, persuading the heathen. They wrote
for others, and were generally careful to define their positions relative to
opposing ones, and state their premises explicitly. They assumed, in other
words, that the reader was an interloper, one who had not been present
throughout the development of the thought processes represented in the
paper: everything of importance had to be spelled out therein.
The generative semanticists, on the other hand, had originated as a
tight-knit group within MIT. Their earliest communication was with one
another. They did, of course, wish to persuade others to their view, but
they always saw their primary audience as one another. They talked to
one another continuously, and their papers often seem to be written as
offshoots of these conversations, intelligible only if you were a party to
the talk. Or they represent the distillation of many discussions and argu-
ments, and skip over crucial steps in the argument because they had all
been through them before—orally—and it would be tedious to spell them
out again. The field remained solipsistic in this way, especially since its pro-
ponents felt rejected by establishment TG. Nobody else wanted to talk to
them, they thought; so they wouldn’t try to talk to anybody. Hence, too,
their colloquial, whimsical style: it was their personal and interpersonal
mode of self-presentation; since they envisioned their writings having the
same audience as their oral communication (whether on the phone or in
papers at CLS), why shift the style? We knew our readers would under-
stand. Indeed, the style itself became a kind of secret handshake. You could
always tell a GS paper: by its title, its breezy style, its funny examples.
You knew who belonged, who your people were. It was cozy comfort in a
heartless world.
While there are philosophical and psychological explanations possible
(such as the ones I have proposed here) for the GS style, it is useful to bear
in mind, when we try to understand how theories are born, live, and die,
that the generative semanticists themselves during their heyday didn’t
283
attempt to account for their assumptions, style, or behavior in these terms.

First of all, they were too busy arguing with each other and outsiders to
worry about how they were arguing, much less why they chose to do it that
way. It didn’t occur consciously to them that they were devising a different
style, separating themselves by style as much as content from their coun-
terparts in CTG or EST. And this, finally, is because the style (even more
than the content) was not a conscious accretion: rather, it arose out of their
fundamental selves, was inseparable from them. They could not have done
otherwise, being who they were, at that time. For we must also recall that
GS arose as an independent theory in the late 1960s, when experimenta-
tion with lifestyle and personality style was encouraged—as it had not
been in the 1950s, when CTG arose, or at present. We cannot separate the
theory from its time.
5. UNIVERSALITY AND RELATIVITY
He thought he saw an Argument

That proved he was the Pope:
He looked again, and found it was
A Bar of Mottled Soap.
“A fact so dread,” he faintly said,
“Extinguishes all hope.”
Lewis Carroll, Sylvie and Bruno Concluded
Before we take final stock of the GS contribution, it may be useful to reflect

upon another point of controversy then and now: the relative importance,
in a theory of language, of universals and typological diversity. Are both
of these aspects of language? Significant aspects? How should a theory of
language incorporate each of them?
To a degree, this was a bone of contention between CTG and EST, though
even more so between generative grammar of all forms, and its immedi-
ate predecessor, structural linguistics; and between the former as the
prime example of “autonomous” linguistics and the latter with its various
hyphenated relations, e.g., anthropological linguistics and sociolinguistics.
CTG and its lineal descendants, as we have seen, tend to attract people
whose background is introspective: humanists, philosophers and math-
ematicians, “armchair” types, who see the systematicity of all human lan-
guage as interesting in itself, or as a way of demonstrating Descartes’ view
that all human beings are basically in possession of the same system of
logical thought, with language as its observable representative. The other

284
fields are peopled by scholars who have spent time in the field, have seen
with their own eyes, heard with their own ears, the diversity among human
languages, and see this as the interesting fact. One of the structuralists,
Martin Joos, once made a statement to the effect that, as far as he knew,
languages could differ from one another in innumerable and unpredictable
ways12—an extreme form of the relativist position, to be sure, and one the
transformationalists seized upon with glee as attesting to the imbecility of
the Bad Guys. On the other side, the anthropological types found amus-
ing, or horrifying, the TG tendency to abstract “universals” from thin air: if
a phenomenon was true of English and German, it was said, TG would
declare it a universal—an overstatement, but not by much. The dispute is
far from dead. Recently Dell Hymes (a representative of the anthropologi-
cal linguists) has written at length excoriating the universalist tendencies
of current linguistics (1986), pointing out the impossibility of proving most
such contentions in terms of any current theories, and the misstatements
and wrong turns they have led to. His examples are in many cases cogent
and sobering, but I think his arguments on the whole miss the point.
Language is neither pure surface diversity, nor pure deep identity. Both
exist as vital parts of linguistic activity, and it is the connection between
the two that composes grammar and indeed necessitates it. There must
(empirically) be universals, or second-language learning would be impos-
sible or tremendously difficult; there must be typological differences or
else it would be trivial, or nonexistent. But even if we grant these proposi-
tions, questions still remain that divide theorists: which of the two is more
important (if either); and if there are agreed to be universals, what are they
and how are they discovered?
Anthropologically minded linguists would say that the only way to prove
the universality of a phenomenon is to demonstrate it empirically, for
instance as Greenberg (1966) did in his work on universals of word order.
To make his case that there were statistical correlations between basic sen-
tence word order and other intrasentence orderings, he and his investiga-
tors examined superficial sentences in a large number of diverse languages.
So their universality claims had a firm empirical base.
Transformationalists, on the other hand, tended to reason deduc-
tively: a principle would be identified as necessary for the description of
12. Lest we be too quick to laugh condescendingly at this overenthusiastic expression

of an unfashionable mythology, let us reflect upon an equally misguided more recent
version—almost the reverse of Joos, yet demonstrating the same overly simple view
of the nature of language and the task of the linguist. Stockwell (1977:xv) says, “It
is my belief that language is really simpler than any linguistic theory comes close to
suggesting.”
285
English syntax (the A-over-A principle, the cycle); arguments would be

given very largely from English, though cases from other languages, geneti-
cally related or otherwise, would be supplied if informants happened to
find any. (Informants were not expected to be completely fluent in the lan-
guage.) If no one within telephone range of Building 20 could elicit any
counterexamples within a reasonable period of time, the putative universal
was established.
It should be noted that the universals in which TG took an interest were
of an intrinsically different kind from the Greenbergian-typological. The lat-
ter could be determined upon surface inspection of actual data. The sorts of
phenomena examined did not require of the investigator a deep knowledge
of the grammar of the language—merely a knowledge of the lexicon (what
was a noun, a verb, etc.); and rudimentary grammatical sophistication (the
ability to determine subjects, verbs, objects, etc.). But the phenomena that
TG weighed for universality could be studied only through acquaintance
with underlying structure, and therefore required deep and reliable intu-
itions about a language, as well as considerable theoretical sophistication,
since the phenomena were not superficially accessible. Hence, a typological
approach was seldom feasible. The problem was that claims for formal uni-
versals (as Chomsky called these, as opposed to the substantive universals
favored by the anthropologists) tended to be made on the basis of English
and then applied a priori to other languages: the implicit argument was
that, if it worked for English, it must work for other languages; if it must
work for other languages, then the data of Language X could and should
be fitted into the model; and if the investigator of Language X were clever
enough, (s)he could find a way to accommodate recalcitrant data to theory.
Exciting work, if dangerous.
Perhaps it will be useful to represent this debate in the form of a dia-
gram, representing the two threads of modern linguistics (Figure 11.4).
Some of the points of distinction were explicit and acknowledged by both
sides, others less so. One thing seems clear: this dichotomy, like the others
I have discussed, is a misconstruction of the nature of language and must
ultimately be resolved by a theory that incorporates procedures for respon-
sibly eliciting and testing both universals and typological differences, and
a model that relates them.
Where does this fit into my present topic? Certainly, here was one place
where GS and CTG/EST could present a united front against the forces of
empiricism. But even here, there were sharp differences in the way each
defined its terms. Both sides paid homage to the importance of identifying
universals of grammar. But it became clear fairly early on that what each
meant by “universal” was different.

286
#1 #2
Sources anthropology philosophy
Patron saint Bloomfield Descartes
Level of observation surface/concrete logical/abstract
Philosophical stance empirical rationalist
World-view relativistic absolutist
Professional identification social science “hard” science
Scientific method quantification formalism
Discovery procedure empirical intuitive
Perspective diversity universality
Interaction theory sociolinguistics pragmatics
Figure 11.4 Two threads of modern linguistics.
Chomsky in Aspects spoke of universals of form and substance, clearly

indicating that the former were the more interesting. But we have seen
that he was unwilling to consider alterations in his concept of deep struc-
ture so that its constructs—PS rules, lexical items—could be universal;
the Aspects DS was in many ways English-specific, and even more so
Indo-European-specific. GS took the Chomsky of Language and mind and
Cartesian linguistics at his word: if language was a window to the mind, if
the mind’s capacities represented logical structures shared by all human
beings by virtue of their common humanity, then the deepest level of
grammatical structure should be directly linked to or (as they said later)
fully identified with those cognitive and logical capacities: the base of
syntax (the deep, or rather, underlying or logical structure) should be
semantics itself, not merely accessible to and interpreted by the semantic
component. Then, the basic forms and mechanisms of syntax, including
the lexicon, were the same across languages. Chomsky, of course, would
not follow where these suggestions led, and the question of the universal-
ity of the base became perhaps the most divisive issue between the two
camps. It is a bit ironic, seen through GS eyes, that Chomsky should be
attacked for his universalist proclivities on the one hand, while on the
other, he appears rather closer to the surface-typological school (rather
the way so many aspects of his theory, seen through GS eyes, come closer
to structuralism than to GS).
6. WHY GS FAILED, IF GS FAILED
Errare malo cum Platone quam cum istis vera sentire.

Marcus Tullius Cicero, Tusculan Disputations 1.17.39
287
In the end, we must ask: Why did GS fail—for all its hope, its optimism,
and its energy? If we accepted the question as valid, we would have to
answer: because it tried to do too much, dared too much. It tried to encom-
pass what could not be formally or rigorously controlled, and this was intol-
erable to those who see linguistic investigation, or scholarly work generally,
as taming the universe, getting things under our control. It’s a ruggedly
masculine image: knowledge, like nature, must be broken, tamed, bent to
our will. We cannot tolerate slippage or disorder. GS reveled in disorder,
and in its ability to tolerate it lay its contribution—and, if we accept the
myth, its downfall.
I have offered disclaimers in lamenting the downfall of GS. For I think
there is an excellent argument that GS never died. The conventional wis-
dom is that the Chomskyan branch of the theory continues in a straight
line, with only minor modifications separating the theories of Syntactic
structures, Aspects, early lexicalism, EST, and Government and Binding—
and on into the future. For Chomsky’s writings imply that position: had
we, the readers, but read Aspects intelligently, we would have understood
how fully it presaged EST… . Nothing really changes, it’s just made more
explicit by the Master, pushed by the stupid misunderstandings of the
opposition. So we have a sense of growing, flourishing, success. But the GS
people, as we have seen, were of a different kind. Their distaste for author-
ity brought them together, and it also made them quick to acknowledge
when some premise or formula fell apart. As Chomsky can be said to be
too slow to acknowledge paradigm shifts, GS may have been too quick to
despair—and to make its despair known to all, as loudly as possible (cf.
Morgan (1973) and Sadock (1975)). So the legend has it that GS finally
exploded, of its own impossible convolution, about 1975. But I think a bet-
ter understanding is that, at that time, we began to move away from the
idea that social and psychological context could be represented as the basis
of syntax in a syntax-central grammar such as GS was. We understood that,
to deal with the phenomena we had uncovered, the relationships between
form and function that our work had made manifest and unavoidable, we
needed to shift the emphasis of the grammar from syntax to semantics
and pragmatics. No longer did we use pragmatics to “explain” the central
point, syntax. More and more we started from function, and saw syntactic
devices as the servants of that function—so function still explained form,
but form depended on function, not vice versa. Yet, there is no sharp break.
I would say the difference between early-1970s GS and what its descen-
dants practice today is less strong than the difference between Syntactic
structures and Aspects, the latter and EST, or the latter and Government
and Binding. It’s just the salesmanship that makes it seem otherwise. One

288
should never underestimate the power of salesmanship—that, if anything,

is the message that the history of GS has for us.
In the nondogmatic spirit of GS, I will present here two contradictory
arguments. One: GS died, and serves it right. Two: GS never died, and
thank heaven for it. I believe absolutely in both.
(a) GS died, and serves it right. Seen from the CTG/EST side, GS failed
because it was chaotic, and that chaos was the direct outcome of a theory
that strayed too far from autonomous syntax, tried to incorporate the
whole world, and choked on it. I would argue rather that GS did die: but
“GS” in its literal acceptance, in the form its practitioners specified up
until the mid-1970s. GS in this sense died because it was too conserva-
tive, its practitioners too much enslaved by their early training as classical
transformationalists.
The problem was not the attempt to connect syntax, semantics, and
pragmatics in pragmantax; the problem was that whatever we called it,
syntax was still central and syntactic modes of argumentation and proof—
distribution and co- occurrence— were still the only options. But this
doesn’t really make sense: Why should syntactic form be what drives the
grammar, if meaning and function were basic to form? We gave lip ser-
vice to the concept of semantic and pragmatic conditioning for syntactic
rules, but never really got away from the idea that everything had to be
syntactically justified; that is, that the other two levels existed to serve
syntactic form.
Then, too, although late in the history of GS we began to realize the
importance of continua, we never followed through; the grammatical
devices and categories GS developed remained as dichotomous and discrete
as any within CTG. And as the meanings and functions we sought to cap-
ture were of course continuous, we went crazy finding more and more and
finer and finer subcategorizations in an attempt to account for the com-
plexity and richness of the data we wanted to incorporate into “linguistics”
and “language.” Because we clung too desperately to the mindset of our
youth, we made GS untenable.
There is another way we can account for the perception that GS failed
and its competitor survived, and that is by considering the role the devices
of public relations and salesmanship play in the ivory tower. GS is pre-
sumed to have failed because its practitioners said it did: how often, pub-
licly and privately, we bemoaned the fact that our theories didn’t fit the
data! That everything was getting too complex for our understanding! That
the phenomena of language went beyond the realm of science and into the
realm of aesthetics and even the supernatural! All this was interpreted as
289
a statement that we gave up, it was no good. We didn’t really intend it that
way (Newmeyer cites such statements as evidence of our despair, but as so
often he misunderstands), but that was what was understood, and eventu-
ally we began to wonder if they maybe weren’t right. In any case, we never
made a vigorous defense.
Newmeyer suggests that salesmanship was the style of GS, with CTG/
EST peopled by earnest Young Doctor Pasteurs, concerned only for the
truth, while GSists ran around the world making conversions and selling
highly abstract snake oil. But the premise is contradictory; at least, if we
were salesmen, we were bad ones. We had no journal; we controlled few
departments, if any; we couldn’t even keep our own gang in order.
CTG, on the other hand, had in its progenitor a superb persuader,
whose buoyant optimism about his theory swept others irresistibly
along. Chomsky’s style (and that of his disciples) always was to down-
play sharp revisions in the theory. One reads Aspects without encoun-
tering any suggestion that deep structure represents a revolutionary
departure from the TG of kernel sentences; one reads “Remarks on
nominalization” without a clue that the description there of the lexicon
and of deep structure existed nowhere before and was developed only
because GS pushed Chomsky to redefine his position—quite radically.
The fissures separating early TG, Classical TG, EST, and GB are at least
as deep as those dividing GS and what its developers are doing today.
But the Chomskyans present themselves as coherent, united, and essen-
tially unchanging—a reassuring position for the neophyte looking for
a wing to shelter beneath; GS shows itself as disorganized, chaotic, and
unfocused—not a position to attract new adherents, or even hold ones
who are afraid of uncertainty.
(b) GS never died, and thank heaven for it. What GS strove to do, its ideal
version of itself, is still healthy, and if anything can save linguistics and
make it once again a rational field, I think it is this perspective. It is a highly
ambitious program, as well as (currently) a rather nebulous one, not fully
articulated by any of its practitioners and practiced by people with a wide
diversity of agendas who do not necessarily see eye to eye (any more than
the old-time GSists did). But the contributions toward which GS was reach-
ing when it fell into crisis are still part of all of our active agendas, though
we approach it in diverse ways. Basically, I think the lasting contributions
GS made, and its successor(s) will continue to make, to our understanding
of language are these:
First, the realization that the phenomena of language are continuous,
and that this continuity extends across all the data as well as the theoretical

290
artifacts that describe it; as well as in the metastructure, our understand-

ing of the relation between language and what it describes. Any formal the-
ory that relies upon dichotomous representations is doomed to fail in the
long run; and until formal devices are developed that are truly nondiscrete,
we are wise to avoid any and all attempts at so-called “rigor.”
Second, and actually a part of the first: the full awareness of the inter-
connectedness of three parts of the human experience: language, the mind,
and the world. While there are aspects of reality that may not be linguisti-
cally encodable, most of what we perceive as reality is colored by the forms
of the language we use to understand and describe it; and the forms lan-
guage takes are determined in large measure by the functions to which it
is put—the understanding and communication of reality, or perceptions of
reality. Any theory of language must begin with these functions, and see
linguistic form as an artifact thereof.
7. CONCLUSION
And the new sun rose, bringing the new year.

Alfred, Lord Tennyson, Idylls of the King
What are we finally to make of the upheavals of those years and the com-
plexities of feeling and belief that still remain? For surely there is no reason
to dwell at length on a moment’s misguided squabbles and the bad feelings
they engendered, unless there is something to learn from the experience.
It seems to me now that the “wars” were an unfortunate outgrowth of a
failure on the part of all of us to make explicit—to ourselves and others—
our covert hopes, aims, and beliefs concerning language and its analysis.
Not surprisingly, the unaskable and unanswerable questions festering a
decade ago remain largely unmentionable. Only by bringing them out into
the light, as I have tried to begin to do here, can we hope to achieve some
resolution and perhaps ultimately reunify the field, or at least make it pos-
sible for us to hold rational discourse with one another.
I see this as the major, unaskable, festering question: Is linguistics a sci-
ence?; and if not, What happens? Before you clench those muscles in your
jaws, dear reader, consider awhile: Do we want our field to be a “science”
because that name makes our enterprise feel prestigious and worthwhile;
or are there really justifications?
Let us drop back a bit and ask: What does it mean to be a science? Does
linguistics work that way? Does language fit the model of the proper object
of scientific inquiry?
291
Some areas of understanding are, necessarily, not scientific nor scientiz-

able, though attempts have from time to time been made to deny this. The
understanding of artistic products and their effect on us, I think, is a clear
case. One reason is that the function and effect of art is individual. Each
person can and must respond uniquely—to Paradise Lost or Les Demoiselles
d’Avignon. Critics can sharpen and define our responses, creating deeper
resonances for each of us as we experience or re-experience the work. But
the tools and methods of natural science—experiment, replication, quan-
tification, formalization—are impossible or ridiculous in this arena. Who
cares how many college freshmen think Satan is more attractive than God
in Paradise Lost? All that matters is whether I do: the humanities are about
me; science, about us or (more probably) them, or it.
The clearly scientizable modes of investigation are different: they look at
objects outside of us, or at least they address objects that are not part of the
very apparatus we use as the principal investigative tool—the mind. While
the ideal of scientific “objectivity” has been called into question (both as a
possibility and as an ideal) in even these cases, at least it makes some sort of
logical sense there. One can count stars, or determine statistically the pos-
sible interactions of elementary particles. Much of the data is discrete and
objectively verifiable; perhaps even replicable. But once one is dealing with
the products of the mind, the products of several minds acting together, the
products of the mind acted upon by external reality, and so forth, possibili-
ties change radically. Replicability is highly improbable when exact context
can never be reproduced in each instance. Discreteness is unlikely when any
form in question is the result of the interplay of an indeterminable number
of factors, each with its own subtle effect. Sheer statistical probability is
uninteresting, if the reasons individuals choose particular forms in specific
circumstances are what is relevant. Formalism is futile if there are so many
subcases, subvarieties, vicissitudes that no sets or categories can be reliably
differentiated, when the irregularities and counterexamples outweigh the
presumably “regular” cases predicted by the rules. And all of this, I would
argue, is the case with linguistic data, more often than not.
The first temptation is to give up in despair, to assume the data are so
various and complex that no systematicity could capture them, and there-
fore, that there is nothing of interest to be said. But this conclusion makes
sense only if we agree that the methods and results of hard science are the
only ones to emulate. If we see that humanistic models must be a part of
our system, then we see that only part of our work can be concerned with
recurring regularities for which formal rules can be written—the rest is
more like exegeses of Hamlet. The two must operate in tandem on the same
body of data.

292
We have tried for most of this century to force language into the
Procrustean bed of “science,” and the chaos and dissension that we have
experienced in the field are the result. If we are a science, we must assume
that only one paradigm has access to the truth, and it had better be our
own. But the impossibility of getting everyone in the field to accept a sin-
gle paradigm, to settle down to Kuhnian “normal science,” demonstrates
that we have been seeing things incorrectly. Just as every serious literary
critic who has had something to say about Hamlet has added to our under-
standing of that work, although each sees it in a very different way, so each
linguist, or each theoretical perspective, captures a different vision of the
linguistic reality, and all, though incompatible as scientific theories, have
something to add to our knowledge. But we can no longer require that per-
spectives be combinable into one single theory: we must settle for differ-
ent, but equally valid, viewpoints.
We must see that our work is not, on the whole, “objective,” since we
can only view language through the filter of our own individual minds,
themselves working through a lifetime of diverse and unique experiences.13
Language use is subjective, and much of our treatment of it, if it is to repre-
sent it accurately, must mirror that subjectivity: talk about how the inves-
tigator reacts to the data as a human being rather than speaking as though
that unique experience were somehow that of all of us.
The introduction of pragmatics into linguistics in the early 1970s brought
(without our realizing it) this subjective and mentalistic aspect of language
use into focus, where syntax-based approaches allowed the problem to be
glossed over. Therefore, it was at the time of the introduction of pragmatics
into GS that the dispute between the two sides got most heated and bitter.
Not because one side wanted to see linguistics as “scientific” and objective,
and the other didn’t; but because both did, and neither could reconcile that
desire satisfactorily with the data that were now turning up. Both felt frus-
trated with the inability of their own theories to deal with these facts fully;
one turned away completely, and the other attempted to incorporate the
problems within the domain of linguistic science. The internal frustrations
created irritation—for which each side blamed the other.
If we now take seriously the implications of the existence of a pragmatic
aspect of language, we must recognize that the nature of language is not as
13. It should be noted throughout this discussion that when I talk about the nec-
essary objectivity of linguistic investigation, I really mean to confine my comments
to the domain of syntax, semantics, and pragmatics. Phonetics and phonology, being
grounded in instrumentally examinable, instrumentally verifiable actualities, might
certainly be scientizable.
293
we had been assuming, and the way in which it has to be studied is much
more complex and sophisticated than any method available to scholarship
at present. We have two choices, as I see it: to continue to pretend to objec-
tivity, discreteness, the artifacts of Newtonian physics; and as an inevitable
concomitant, brace ourselves for another generation of noncooperation,
frustration, and bad feelings. Or, we must put aside all of our previous
ways of looking at the linguistic universe, and carve out a new perspec-
tive: toward language, toward knowledge, toward one another.
REFERENCES
Austin, J. L. 1962. How to do things with words. Oxford: Clarendon Press.

Bach, Emmon. 1968. Nouns and noun phrases. Universals in linguistic theory, ed. by
Emmon Bach and Robert T. Harms, 90–122. New York: Holt, Rinehart and
Winston.
Bach, Emmon. 1970. Problominalization. Linguistic Inquiry 1:121–22.
Belenky, Mary Field, Blythe McVicker Clinchy, Nancy Rule Goldberger, and Jill
Mattuck Tarule. 1986. Women’s ways of knowing. New York: Basic Books.
Chomsky, Noam. 1959. Review of Verbal behavior, by B. F. Skinner. Language 35:26–58.
Chomsky, Noam. 1966. Cartesian linguistics. Cambridge, MA: MIT Press.
Chomsky, Noam. 1967. Language and mind. New York: Harcourt Brace Jovanovich.
Chomsky, Noam. 1970. Remarks on nominalizations. Readings in English
184–221.Waltham, MA: Blaisdell.
Chomsky, Noam. 1971. Deep structure, surface structure, and semantic
interpretation. Semantics: An interdisciplinary reader, ed. by Danny Steinberg
and Leon A. Jakobovits, 183–216. Cambridge, UK: Cambridge University Press.
Dougherty, Ray. 1974. Generative semantic methods: A Bloomfieldian counter-
revolution. International Journal of Dravidian Linguistics 3:255–86.
Gazdar, Gerald, and Ewald Klein. 1978. Review of Formal semantics of natural language,
ed. by E. L. Keenan. Language 54:661–67.
Gilligan, Carol. 1982. In a different voice. Cambridge, MA: Harvard University Press.
seventh regional meeting of the Chicago Linguistic Society (CLS 7), Chicago, 63–84.
Greenberg, Joseph. 1966. Universals of language. Cambridge, MA: MIT Press.
Hymes, Dell. 1986. Discourse: Scope without depth. International Journal of the
Sociology of Language 57:49–89.
Katz, Jerrold J., and Thomas G. Bever. 1976. The fall and rise of empiricism. An
integrated theory of linguistic ability, ed. by Thomas G. Bever, Jerrold J. Katz,
and D. Terence Langendoen, 11–64. New York: Crowell.
Katz, Jerrold J., and Paul M. Postal. 1964. An integrated theory of linguistic descriptions.
Cambridge, MA: MIT Press.
Keller, Evelyn Fox. 1985. Reflections on gender and science. New Haven, CT: Yale
University Press.

294
Kuhn, Thomas. 1962. The structure of scientific revolutions. Chicago: University of

Chicago Press.
Lakoff, George. (1963) 1976. Toward generative semantics. Notes from the linguistic
underground (Syntax and Semantics 7). ed. by James McCawley, 232–96.
New York: Academic Press.
Lakoff, George. 1966. Stative adjectives and verbs in English (Report NSF-17).
Cambridge, MA: Harvard University Computation Laboratory.
Lakoff, George. 1970. Irregularity in syntax. New York: Holt, Rinehart and Winston.
45:608–15.
Lakoff, Robin. 1970. Tense and its relation to participants. Language 46:838–49.
Lakoff, Robin Tolmach. 1982. Rule and role in second-language teaching. CATESOL
Occasional Papers 8:37–50.
Langacker, Ronald. 1969. On pronominalization and the chain of command. Modern
studies in English, ed. by David A. Reibel and Sanford A. Schane, 160–86.
Englewood Cliffs, NJ: Prentice-Hall.
Lees, Robert B. 1960. Review of Interrogative structures of American English, by Dwight
Bolinger. Word 16:119–25.
McCawley, James. 1970a. Where do noun phrases come from? Readings in English
166–83.Waltham, MA: Blaisdell.
McCawley, James. 1970b. English as a VSO language. Language 46.286–99.
Morgan, Jerry 1973. How can you be in two places at once when you’re not anywhere
at all? Papers from the ninth regional meeting of the Chicago Linguistic Society
(CLS 9), Chicago, 410–47.
Newmeyer, Frederick 1980. Linguistic theory in America. New York: Academic Press.
Partee, Barbara. 1971. On the requirement that transformations preserve meaning.
Studies in linguistic semantics, ed. by Charles Fillmore and D. Terence
Langendoen, 1–22. New York: Holt, Rinehart and Winston.
Postal, Paul. 1964. Underlying and superficial linguistic structure. Harvard Educational
Review 34:246–66.
Postal, Paul. 1971. Cross-over phenomena. New York: Holt, Rinehart and Winston.
Ross, John Robert. 1967. Constraints on variables in syntax. Cambridge, MA: MIT
dissertation.
Ross, John Robert. 1969. Auxiliaries as main verbs. Studies in philosophical linguistics,
ed. by W. Todd, 77–102. Evanston, IL: Great Expectations.
Ross, John Robert. 1970. On declarative sentences. Readings in English
222–72. Waltham, MA: Blaisdell.
Ross, John Robert. 1972. The category squish: Endstation Hauptwort. Papers from the
eighth regional meeting of the Chicago Linguistic Society (CLS 8), Chicago, 316–28.
Ross, John Robert. 1973. Nouniness. Three dimensions of linguistic theory, ed. by
Osamu Fujimura, 137–257. Tokyo: TEC.
Sadock, Jerrold. 1975. The soft interpretive underbelly of generative semantics.
Syntax and semantics 3: Speech acts, ed. by Peter Cole and Jerry L. Morgan,
383–96. New York: Academic Press.
Stockwell, Robert. 1977. Foundations of syntactic theory. Englewood Cliffs,
NJ: Prentice-Hall.
295
Introduction to “Review essay:

Women and disability”
BY SUZET TE HADEN ELGIN
A s I read Robin Lakoff’s review essay, I remembered a day in the late

1970s, in my doctor’s office. He was annoyed with me. That being the
perpetual situation I faced, I wasn’t surprised, but it seemed worse than
usual, and so I asked him what I’d done. He was angry because a few days
previously I had checked out of the hospital where I’d been a patient and
had given a keynote address at a conference. “You couldn’t really have been
sick!” he said, with both bitterness and contempt.
I waited a few seconds for the sting to fade, and then gathered my cour-
age and told him that he needed to give some careful thought to his reac-
tion. “If I had cancer,” I said, “and I had gotten up out of my hospital bed
and carried out an obligation that I felt I could not set aside, you would be
sitting there now saying how brave and noble that was, and how much you
admired me for it.”
There was a very long silence. And then, to his credit, he nodded and
said, “You are absolutely right—that’s exactly what I would be doing.” He
didn’t go so far as to apologize or to take back his claim that I couldn’t
“really” have been sick, but he did acknowledge his bias.
“What you call someone is what that person becomes,” Lakoff writes in
this paper (p. 300), and she carefully sets out some of the linguistic prob-
lems that disabled persons face. As she says, the name we are given cre-
ates the reality in which we have to function, because our name assigns
us to a group. With that assignment comes a variety of privileges and
296
responsibilities and limitations. The assignment ranks us, too, on the

continuum from “truly and seriously disabled, entitled to help and com-
passion” to “just a wimp, needs to pull herself together and get on with
her life.” Others’ behavior toward us is the behavior our culture considers
appropriate for that group, not for us as individual persons.
Every disabled woman, especially if she has a family to look after, faces
an impossible dilemma. To carry out her duties she has to be able to request
certain concessions, and those concessions will only be granted if she
plays the disabled role—she has to act crippled. However, if she copes well
enough to fulfill her obligations, she risks losing the concessions that make
it possible. She may have to put in twice as much time and effort to accom-
plish her tasks as a woman who is “normal” would; she may have to pay
for each obligation achieved with hours or days of pain and perhaps with
days confined to bed. That ought to win her respect and admiration; after
all, if she can cope that well in the face of such difficulties, think how much
more she could accomplish if they were eased. But the respect and admira-
tion occurs only in those cases, like paraplegia or quadriplegia, where her
disability is overwhelmingly obvious. The armless woman who does her
housework and shopping and child care with her feet and legs never has to
hear, “There isn’t really anything wrong with you!”; for many of the rest of
us, that sentence is a perennial threat. “There’s nothing wrong with you”
ought to be a compliment; unfortunately, it means “You’re not entitled to
the concessions your label requires; you don’t qualify for that group.” And
you wonder how you could manage to meet the requirements. What should
you do to get around the semantic barrier? Faint? Limp? Drool? Stammer?
Scream? You wonder what item or items you could add to your behavioral
repertoire that the “able” culture would accept as credentials for the help
you need, without at the same time making it impossible for you to carry
on with your life.
It’s painful for me to write about this, painful for me to speak or think
about it; I do stammer, in both speech and writing, when I must take
up the subject. Lakoff notes that when she began to read some of the
writings in the books reviewed she reacted with a kind of disdainful dis-
tancing, because “we expect polish in published writing” (p. 306). But
then she is won over by the passion and the immediacy of “the rougher
essays” (p. 306). I am here to tell you that it’s difficult to write other
than roughly about disability, especially if you are a woman, especially if
you have one of the “hidden” disabilities. There’s no vocabulary for our
experiences. We grow up hearing our mothers say, “Nobody wants to
hear about your problems, dear—if you can’t say something pleasant,
just keep still.”
297
And we know that’s true. Our doctors are no exception; they make it very
clear that they don’t want to listen to us either. It’s very very hard—when
you have worn a gag all your life, when you have been endlessly shamed
for trying to go gagless—to remove it and use language that is elegant and
polished. I know how rough this introduction I am writing is; several times
I’ve almost given up and told the editor I can’t do it. The fact that I’ve pub-
lished dozens of books and written a dissertation helps very little. The gag
I wear, like my twisted spine and my distorted neural wiring, is invisible to
most eyes, but I can’t get away from either one, not even for a few minutes.
Like Lakoff, I have no solution to offer for the linguistic problems. Some
of the other problems are closer to solution today, I think, than they have
been in the past. I could no longer be forced into retirement for lack of a
handicapped parking space or an office in the building where I taught my
classes, for example—I would have to need much more exotic concessions
than those today, and that’s an improvement. But the linguistic problems
remain the same, and no one is interested in solving them. The interest
ends, predictably, at finding a “name” with which those who must fit us
into the society of the Whole-and-Able could feel comfortable. There is still
no interest in learning how we cope, or what might be done to make it eas-
ier for us to do so. There is still no interest in finding a way to increase soci-
ety’s tolerance for the way disabled women often look. (I am quite certain
that if Stephen Hawking were a woman, that woman would never appear
on television.)
My personal vote is for the old-fashioned word cripple. Not “I was
crippled by spinal polio,” the passive participle that Lakoff rightly identi-
fies as demeaning, but the straightforward “I am a cripple.” She says that
she would prefer “Are you Jewish?” to “Are you a Jew?” because the lat-
ter question suggests that Jewishness is “the totality of her … identity”
(p. 301). But there is no “Are you cripplish?”; in many important ways, the
crippledness (no word for that either, you perceive) is your whole identity.
None of the euphemisms—disabled, handicapped, differently abled, physi-
cally challenged—expresses the simple truth of my situation in the way that
cripple does. If I’d insisted on that word all these years instead of being
politically correct and “nice,” I think some of the most painful parts of my
life could have been avoided. The reaction of men to seeing my naked back,
for example. If I’d said to them beforehand, “Be warned—I’m a cripple,”
perhaps they’d have been less startled and better able to disguise the fact
that they found me repulsive; perhaps those who knew they wouldn’t be
able to disguise it would have been wise enough to decline with thanks.
I am grateful to Lakoff for raising these issues, as she has raised so many
others, and for making it explicitly clear that there are linguistic problems
W O M E N A N D DI S A B I L I T Y [ 297 ]
298
associated with disability. I am grateful to her, at a time when best-selling

books by linguists call the Sapir-W horf hypothesis nonsense, for her state-
ment that “words alter reality” (p. 300). There is no area of our life in which
that is more true than in the territory ruled by MDeities, where the natural
course of every woman’s life after roughly the age of fifty can be transformed
into a disease by naming it “hypoestrogenemia,” instantly shazamming
into reality a necessity go to a doctor and seek a treatment. Overweight,
defined for most women as “more than you weighed at fourteen,” has gone
the same route, as have the time periods immediately before menstruation
and menopause. Aging is next, and will shortly be defined as a deficiency of
something or other; perhaps we old women will be told to get ourselves to
a specialist in “hypopediatricism” or “hypojeunessics” forthwith.
It is language—a disgracefully large part of it medical language—that
has made Barbie the standard by which every woman is measured. It is the
lack of language, the absence of a vocabulary (and an accepted mode of dis-
course) with which we could discuss “differently-abledness” efficiently and
effectively and “with polish” that makes fighting that process so hard. That
lack of language silences us in a way that no legislation ever could. If a solu-
tion can be found, it will, I believe, have to be found by linguists. Linguists
who are willing to come down out of their ivory towers and slog around
in the mud with people like me. Linguists who understand that we can no
longer justify leaving the application of linguistic science to the fields of
medicine and advertising. I look forward to working with these linguists,
and would greet them joyfully.
299
CHAPTER 12
Review essay
Women and disability (1989)
W omen with disabilities: Essays in psychology, culture, and politics. Ed. by

Michelle Fine and Adrienne Asch. Philadelphia: Temple University
Press, 1988. Pp. xv, 347. ISBN 0877224749.
With the power of each breath: A disabled women’s anthology. Ed. by Susan
E. Browne, Debra Connors, and Nanci Stern. Pittsburgh: Cleis Press, 1985.
Pp. 354. ISBN 0939416069.
Plaintext: Essays by Nancy Mairs. Tucson: University of Arizona Press, 1986.
Pp. xii, 154. ISBN 0816508925.
With wings: An anthology of literature by and about women with disabilities.
Ed. by Marsha Saxton and Florence Howe. New York: Feminist Press at the
City University of New York, 1987. Pp. xv, 167. ISBN 0935312617.
The disabled are a group that the rest of society often prefers to ignore.
Yet, in recent years, like other groups in analogous circumstances, disabled
people have begun to demand equity and visibility. And, as with other
groups, although the efforts to achieve these rights have been valuable,
there are complications. The books under review encapsulate, by their very
existence, the force behind the struggle; their form and content speak of
courage and determination in the face of a society that still seems to wish
the authors and people like them didn’t exist or, at least, would keep their
This paper was originally published in Feminist Studies 15(2), 1989, 365–75. It is
reprinted here by permission of the publisher: Feminist Studies, 0103 Taliaferro Hall,
4280 Chapel Lane, University of Maryland, College Park, MD 20742.
300
existence a guilty secret. The strength, poignancy, and beauty of these writ-
ings communicate the error of such attitudes, which have caused a great
loss to all of us over the millennia. And how fortunate it is that the barriers
are beginning to be torn down, although barely beginning, as many of the
testimonials make painfully clear.
“The disabled,” I said at the outset; and most of the books incorporate
that term into their titles. As clear as it is, the term is in some respects
problematic. The basic dilemma is familiar to most nondominant groups.
By what name are its members to be called, and do they have the right
to name themselves? What is the consequence of any particular choice?
(Are they to be Negroes, African-Americans, or blacks? Ladies or women?
Old people, the elderly, senior citizens, or golden agers? No choice is without
difficulties.)
A first reaction may be dismissive: “What’s in a name? Let’s get to the
realities!” Our culture is ambivalent about the power of language, swinging
proverbially between “Sticks and stones can break my bones, but words
can never harm me” and “The pen is mightier than the sword.” But words
are actions; like actions, words alter reality. What you call someone is what
that person becomes—most often for both of you. Naming has too often
been the privilege of the powerful, further reinforcing their power. To
name oneself is to take back one’s power, to assert the right to determine
one’s own identity.
The problem for the group under discussion is complicated by the
fact that membership is seen, almost universally, as unfortunate, if not
accursed. Therefore, the term disabled tends to elicit reactions of pity and
fear and patronizing attitudes and requires, all the more, a new terminol-
ogy to suggest a new point of view. But because of the strength of the nega-
tive image, it is especially hard to find a name that carries the appropriate
denotation and connotations and none of the undesirable ones. The oldest
term in modern English is cripple, unsatisfactory in many ways, because of
the Tiny Tim overtones, the implications of ineffectuality, and the associa-
tion with pity.
Around the turn of the present century, an attempt was made to find
a more neutral (and “objective”) term; and handicapped was substituted.
More recently, though, this term has been criticized as implying helpless-
ness. But it has been harder to find a term that both achieves descriptive
accuracy and avoids negative implications. Among those I have seen in
print, probably the one closest to universal acceptance (as all of these writ-
ings suggest), is disabled. (I have also encountered differently abled, which
I think we could do without, and challenged, which I like, suggesting as it
does the eventual transcendence of the difficulty.)
301
The problem remains that even disabled carries a negative connotation

not far from that of cripple. It is arguable that disabled is more euphemistic
than cripple; the former, like the other substitutes, is an adjective, the lat-
ter a noun. I think here of the preference for Jewish, the adjective, over the
noun Jew as a descriptive term. I would rather be asked, “Are you Jewish?”
than, “Are you a Jew?” Probably the reason adjectives are preferred for sen-
sitive identifications like these is that, unlike nouns, they impute only a
partial identity; one who is Jewish might be other things as well. But to
call someone a Jew, in many contexts, is to suggest that that is the totality
of her or his identity, all one needs to know about the person. The case of
cripple, versus disabled, is parallel.
Equally problematic, these adjectives are passive participles, implying
the existence of an unspecified agent who did the “disabling” or “challeng-
ing.” Does this suggest a helpless victimization by forces beyond one’s
control? If that sense is indeed present, disabled—like lady or Negro—is a
euphemism, a term coined to circumvent the embarrassment of reference
to the unmentionable.1
Beyond the choice of name are other dilemmas. For many nondominant
groups, membership is clear; society at large is in agreement about what
observable trait(s) mark someone as “one of them.” But there is much more
ambiguity as to who is “disabled.” What defines the category? Is there—as
the name implies—a simple dichotomy, the able versus the disabled? Or
do the words represent points along a spectrum or a continuum, mean-
ing that everyone has a “disability”—whether it causes them, or whether
they choose, to define themselves as “disabled.” A might be physically para-
lyzed; B agoraphobic; C subject to allergies; D poor; E elderly; F a woman.
All might so describe themselves—or none of them. Does it make sense to
try to draw a line?
The decision has implications, both for the individual and for society
at large. If we agree that, in some way, everyone is “disabled,” we begin
to include the more obviously and severely disabled among us, no longer
consigning them to separateness and invisibility. We can, perhaps, begin
to empathize, to share more than we have shared. On the other hand, isn’t
it dishonorable to suggest that your tribulations are greater than, in fact,
they are, to say that, because (for instance) you’re a woman, you belong in
the same category as a quadriplegic, entitled to the same privileges, the
same special considerations? The extension of the category could have that
effect. I am not about to propose a solution, Solomonic or other; I raise the
1. For discussion of the concept and uses of euphemisms, see R. Lakoff (1975).
W omen and disability [ 301 ]

302
point merely to show that these books, both by their very existence and by
the points they raise, force us to reflect, to rethink, to recognize that ambi-
guity is very much a part of our world.
As women and feminists, readers of these books may feel particularly
at a loss for solutions. All the books specifically address the situation of
the disabled woman. As several authors point out, she is in significant
ways worse off than a correspondingly disabled man, although not in every
way. In terms of policy, the disabled woman, like other women, tends to
become invisible or to get the raw end of the deal; she receives less sup-
port, economic and other; more paternalism; more interference with basic
human needs, especially reproductive rights. Just as women have fought
for visibility, for public existence, so have the disabled, and disabled women
have had to fight twice as hard. But there is a sense, as Michelle Fine and
Adrienne Asch note in Women with disabilities: Essays in psychology, culture,
and politics (p. 23), in which a disabled man is in a worse position than his
female counterpart. Disabled man is a self-contradiction, because men are
stereotypically supposed to be “able,” strong, and powerful. Hence, society
is less concerned with helping disabled women to achieve competence and
autonomy—as women, they aren’t supposed to want them, much less get
them. Disabled woman, on the other hand, is redundant. She is a super-
woman in an ironic sense: superchildlike, dependent, incompetent, even
beyond the stereotypical ideal female.
The women’s movement and the disability-rights movement have many
common aims. Both are dedicated to extending opportunities and increas-
ing options for everyone. What women have struggled to achieve over
the last century—equality of education, access to employment, economic
equity, opening of the professions, reproductive autonomy—the disabled
have also historically been denied, often with the same excuses. There are
also strong parallels to the civil rights movement. Thus, a century ago,
women were denied access to higher education because they weren’t capable
of it, because they wouldn’t use it productively, and because their presence
in a coeducational classroom would be distracting to the dominant group
(whose education matters). Similar things were said about black school
integration half a century later. And the same arguments flourish today in
opposition to the mainstreaming of the disabled in public education. There
is little apparent commonality among blacks, nondisabled women, and the
disabled. But society’s insistence on treating them similarly and subjecting
them to analogous stereotypical judgments suggests that all nonpowerful
groups serve similar functions for the dominant. Their existence gives the
dominant group a comfortable sense of its own superiority, as long as it is
not forced to share its privileges with the inferior group.
303
Fine and Asch speak of disabled women as on a double “pedestal,”

comparable to the one on which women have been placed. Like nondis-
abled women, but to a much greater degree, disabled women are seen as
asexual, perhaps even saintly. And like nondisabled women, they have a
darker side; they are evil, depraved. They function as symbolic scapegoats
for the dominant group, and hence the latter may feel horror and disgust
and avert their eyes—or stare. (Just so, some societies require that women
make themselves invisible behind veils; others consider any female alone in
public a fit object for stares and other marks of unwanted attention.) For
both women and the disabled, looks are disproportionately important. The
sight of both is disturbing, each in a different way. So legally, historically,
socially, and symbolically, the two groups and their respective movements
share a great deal.
But not everything. The two groups have different perspectives on
reproductive rights, an important item in the agenda of both but meaning
something different for each. Feminists are concerned with bodily auton-
omy: the right to choose or refuse to have intercourse or engage in other
sexual behavior; the right to bear a child, or not to; the right to bring up a
child, or not to.
Bodily autonomy includes, for the nondisabled woman, the right to be
free from sexual harassment on the job or on the street. But the perspec-
tive of the disabled shows that the issue is less clear-cut than one might
think. In their introduction, Fine and Asch tell a story about a women’s
conference where participants complained about sexual harassment.
Disabled women retorted: “Try two weeks or thirty years without that sex-
ual attention. Then tell us if you would hate getting some, even in the form
of harassment!”
More serious are issues like abortion and others summarized under the
heading “right-to-life.” None of the writers argues with the premise that a
woman has the right to choose abortion. But the question becomes inflam-
matory when abortion is sought because the baby is likely to be “defec-
tive”—a word to which several authors take objection. What is the message
sent to the disabled (and everyone else) by a philosophy that implies that
a less-than-perfect child is garbage, unworthy of life, disposable, a burden
to family and society? Should amniocentesis be available at all? If we coun-
tenance its use to allow women to decide whether to abort fetuses with
Down’s syndrome or spina bifida, should we be equally sanguine about
using it to support the abortion of fetuses bearing the XX chromosomes?
In the same way, “quality of life” positions, supported with good reason
by many feminists, take on a more complex coloration when the disabled
are seen as autonomous human beings. When a severely disabled newborn

304
child needs extensive medical procedures in order to survive (procedures

that would be automatically performed if the child were otherwise “nor-
mal”), what should be government, and hospital, policy? Where should the
parents’ rights end? If the parents see refusal of surgery as a way out of the
burden, should the state step in and force them to consent? The authors
argue that such children should have access to the same life-saving proce-
dures as any other child, but they also recognize that this policy is humane
and enforceable only if the government is ready to provide equally heroic
lifelong support to the family and the child, rather than pusillanimously
abandoning their advocacy of the “right to life” at the hospital door. The
right to life is meaningless without a guarantee of quality of life. Fine and
Asch make the point that no woman should be forced to bear a child against
her will. But education to bring about changes in attitude, and sane public
policy, can make the bearing and raising of a less-than-perfect child no lon-
ger the horror that it seems to many today. If a government, by a strategy
of benign neglect, conveys to the American people that the disabled are less
than human and undeserving of special services, then a significant part of
the blame for prospective parents’ horror and dismay must be laid at that
government’s door.
But, other social theorists argue, such liberal compassion is misplaced. By
saving imperfect specimens, we dilute the quality of the gene pool; by sup-
porting them, we strain already scarce resources. We can’t afford it.
I would suggest that we can’t afford to do anything else. If the species is to
survive, we must develop empathy for “the others” whoever they are. With
our scientific expertise we will surely be able to devise ways to compensate for
disabilities; but there is no prosthesis to offset the inability to empathize. If
the human race continues to produce members with this condition, we are on
the brink of World War III—from which few will emerge nondisabled.
There are other entrenched ideals we share as a culture that a close
look from the perspective of the disabled forces us to re-evaluate. What,
as Freud might have said, do the disabled want? Like Freud, the culture
has tended to understand that question and frame the answer on the
basis of its own needs, conscious and unconscious. Professionals (who
are by definition outsiders) have claimed that the disabled (like women)
want to be “normal,” they want to “adjust,” they want to blend in as
much as possible with the normal, even at the cost of greater inconve-
nience. They don’t, above all, want to be conspicuous.2 Professionals
2. Cf. Freud (1916) 1955. Women, Freud says, feel shame because of their lack of the “nor-
mal” male sexual organ; this defect gives them the status of “exceptions,” including the dis-
abled. He assigned to this category Richard III and women in general, on similar grounds.
305
argue for the segregation of the disabled in special schools so that

they won’t be made to feel “abnormal” by comparison with the others
and advocate the wearing of debilitating and cumbersome prostheses,
because they make the wearer look “more normal” even as they reduce
mobility and competence. Here, as elsewhere, society has to get used
to the idea of variety, of a spectrum of human possibility; to get over
its shame, loathing, and discomfort at anyone appreciably different or
“other.” For too long we have dealt with that embarrassment by denying
it and attempting to segregate it from us; it is time to acknowledge that
this won’t work, that it probably does as much harm to the “normal” as
to the “nonnormal.”
In his Stigma (1963), Erving Goffman speaks of the socially stigmatized,
those with “spoiled identity.” He includes a diverse assortment of human-
ity, all those who, for whatever reason, don’t belong, are not allowed to
participate fully in the world—criminals and their families; those whose
occupations render them noxious (e.g., executioners and prostitutes);
those whose race, religion, or ethnic background is unacceptable; and those
who are disabled. All inhabit a world of their own, a place apart. That is
mostly by society’s choice, but, Goffman notes rather surprisingly, it may
also be the wish of the stigmatized. They don’t necessarily see “normality”
or adjustment as the summum bonum.
The books under discussion represent a broadly unified point of
view. All are eloquent advocates for the disabled, and each contribu-
tor makes the case in her own way. But within this larger unity, many
perspectives are represented. Three of the four books are collections
of writings by different authors; the fourth consists of essays written
by Nancy Mairs alone. Each volume approaches its task in a slightly
different way.
The three anthologies could be organized along any of several conti-
nua: between social-science and humanistic approaches; objective to sub-
jective; scholarly to literary; personal to impersonal. The collection by Fine
and Asch represents the social-scientific, objective, and impersonal end
of the spectrum, although many of the contributors refreshingly sharpen
and clarify their points with anecdotes, personal and otherwise. The col-
lection is balanced among a variety of academic fields and perspectives.
It is divided into three parts: the disabled woman’s self-image, disabled
women in personal relationships, and public policy for the disabled. Some
of the essays are sociological, bristling with facts and figures—how many,
what percentage? Others are anthropological, seeing disability as a minor-
ity culture with rules and rituals of its own; some essays are psychological,
assessing the comparative merits of Freudian, object-relations, and other

306
psychological viewpoints on disability; one contribution might be classified

as humanistic, considering the dearth of disabled women as role models in
literature. I found the combination of approaches stimulating; there was
no sense of confusion or disunity, although some of the articles are, as one
might expect, more readable than others. A collection like this makes its
deepest point explicitly, and each of the authors makes it in her own way—
the disabled are capable of full lives, given the opportunity.
The other anthologies, as well as Plaintext: Essays by Nancy Mairs, stress
the personal anecdote, the subjective experience of disability. But there is
a wide range within the field. Most of the writers are themselves disabled
women, or occasionally, friends or family of a disabled woman. All talk
about the personal experience of disability—but in how many voices! We,
the larger world, tend to lump “them” together, see “the disabled” as a ste-
reotypical set to which we can attribute whatever sort of character and atti-
tudes we find convenient. But it should not be surprising that there are as
many ways to experience and talk about disability as any other form of the
human condition—wittily, ironically, angrily, despairingly, courageously,
anxiously, matter-of-factly, to name a few. Many of the writers are pro-
fessionals whose polished and literate testimonials are a pleasure to read.
Nancy Mairs, in particular, both in her own book and in her contributions to
Marsha Saxton and Florence Howe’s With wings: An anthology of literature by
and about women with disabilities, is a writer of depth, sensitivity, and power.
Her description of her experience as a woman and a disabled person stays in
the mind. But others, particularly in With the power of each breath: A disabled
women’s anthology, edited by Susan E. Browne, Debra Connors, and Nanci
Stern, are amateurs who seem uncomfortable with the pen.
My first response as I started to read was one of a certain aesthetic
distancing, or even disdain; we expect polish in published writing. But as
I continued to read, I realized that the rougher essays were often the most
telling and vivid. An accomplished style gives the reader pleasure, making it
a little easier to glide past the writer’s pain; and in books like these, experi-
encing the pain—and the other emotions—is as essential to understanding
as comprehending the statistics would be in a social-science article. Finally,
the choice to include these pieces, relatively few in number, is a way of say-
ing that disability is a leveler: it brings together people of disparate talents.
Although the books cover a broad array of experience, from a wide
selection of perspectives, as Fine and Asch point out, they represent just a
beginning. There is much we need to know. The authors note a few topics
for further research, such as women with “hidden” disabilities (those that
are less obvious, such as agoraphobia); the difference between being born
disabled, becoming disabled early in life, and in later life; disabled women
307
and their partners; race, class, gender, and disability; disabled women who
have achieved satisfying work (how and what?); disabled women as moth-
ers; and the disabled in institutions (most of the articles in these volumes
are by or about women living at home).
Even more than women and blacks, the disabled have lived in shadows,
if not invisible, certainly unmentionable; very little is written about them,
still less from their own perspective. Much that has been written, most
often from a vocational-rehabilitative perspective, is (to judge from the
excerpts quoted in the articles) insensitive when not downright paternal-
istic. And, of course, society considers it neither possible nor quite proper
for the disabled to speak for themselves. I have heard many descriptions by
disabled women and men of their treatment in service interviews—even
though they are not visually or aurally impaired, waiters and store clerks
address them through their nondisabled companions: “And what does she
want?” These books begin to give the disabled a voice, with which, ulti-
mately, will come visibility and full participation.
Collections like these serve two functions seldom combined elsewhere
(as academia tends, improperly, to find such combinations improper),
analysis and advocacy. Some of the writings stress one over the other or
develop one explicitly, the other as subtext. But it is impossible to read even
the coolest of the descriptive articles without asking, “What is to be done?
Why did this happen?” and the reader comes away from even the most stir-
ringly persuasive contribution with deeper intellectual understanding.
One great benefit of reading all these books in all their variety might
be to move the reader past one undiscussed, but highly toxic, disability—
dichotomization. Humanity as a whole takes comfort from polarizations—
he male, she female; this right, that wrong; I able, you disabled—forgetting
that all these properties function as continua, not dichotomies. Academics
are, if possible, more susceptible to this disease than most people. We
like our neat disciplinary categories. Most universities make it difficult if
not impossible to teach interdisciplinary courses or do interdisciplinary
research. Social scientists like writing by social scientists, in social-scientific
vocabulary, with nicely contorted social-scientific syntax; humanists pre-
fer a more graceful, if less general or definitive, approach. It is considered
unseemly—a form of miscegenation—for the two to meet. In fact, aca-
demics agree on only one thing: that nonacademics should be kept out of
the discourse, as irresponsible and unscholarly.
Therefore, these collections serve a variety of purposes. For the disabled
reader, they provide ideas, role models, possibilities. The nondisabled are
afforded a view of another way of life, more demanding, certainly different.
Both groups achieve an understanding of the connectedness between the

308
disabled and the rest of society, of the need for demolishing the barriers
that create and enforce invisibility.
And finally, the reader may learn something else as well. The content of
the books, which stresses the commonality of the human experience, and
their format, combining disparate methods, approaches, and perspectives,
make, finally, one profound statement. We come away with the knowledge
that full understanding of anything is possible only if we achieve these
combinations. Just as the juxtaposition of all these collections, and the
varieties of writing within each, add up to a richer whole, so by including
everyone within the sphere of “us-ness,” including, finally, the disabled, we
will achieve together the complexity, dignity, and wholeness that define the
property of being human.
REFERENCES
Freud, Sigmund. (1916) 1955. Some character-types met with in analytic

work: (I) The “exceptions.” The standard edition of the complete psychological
works of Sigmund Freud, ed. and trans. by James Strachey, vol. 14, 311–35.
London: Hogarth Press.
Goffman, Erving. 1963. Stigma: Notes on the management of spoiled identity.
New York: Simon and Schuster.
309
Introduction to “Pragmatics
and the law: Speech act theory
confronts the First Amendment”
BY SUSAN C. HERRING
W hat happens when a linguist with keen powers of observation

regarding language use encounters at first hand the workings of her
nation’s legal system? In the case of Robin Lakoff, the result has been some
insightful research that expands the discipline known as “applied socio-
linguistics” and sheds light on the complexities involved in attempting to
carry out justice via the courts. That language plays a key role in legal delib-
eration has always been recognized. Just how it works, however, and in
particular, the extent to which it relies on pragmatic (as opposed to struc-
tural or formal) considerations, has only recently begun to be understood,
thanks in part to Lakoff’s work.
I remember when Robin first began thinking seriously about language
and the law. In the mid-1980s, when I was still a graduate student at the
University of California, Berkeley, Robin somehow managed to get selected
as a juror for a rape/murder trial. (This was no mean feat, given the anti-
intellectual bias of court lawyers, then as now—usually just admitting
that you had anything to do with a university was sufficient to get you dis-
missed from a jury.) She presented her observations about the linguistic
and criminal aspects of the trial in chilling detail at an evening colloquium
that I attended, and I remember thinking, here was the real thing: linguis-
tic analysis that helped one to understand the compelling mysteries of life,
310
of crime and punishment. Her next court appearance was in the role of an
expert witness, in a 1989 case involving allegations of “false advertising,”
and I also went to hear her speak about that experience. It seems she had
given the court an introductory lesson in presupposition and implicature,
complete with illustrations on a chalkboard brought in specially for the
purpose. I inwardly delighted at the thought of a linguist teaching judges
and lawyers about pragmatics—maybe linguistics could contribute some-
thing practical to the real world after all! In this way, through public pre-
sentations (and, later, in published articles as well), Robin transformed her
courtroom experiences into linguistic insights.
The paper reprinted here was first delivered in April 1991 as a plenary
speech at the annual meeting of the Chicago Linguistic Society. In it,
Lakoff draws on various legal interpretations of the First Amendment—
the constitutionally guaranteed right to “freedom of speech”—in order to
critique speech-act theory. Particularly targeted are two legacies of clas-
sical speech-act theorists: a tendency to focus mainly on the speaker’s
intent, or illocutionary force, to the exclusion of an utterance’s perlocution-
ary effect, and a tendency to treat speech acts as either performative or
not, and if performative, to treat them all as equally performative, that
is, capable of acting on the world with equal force. As Lakoff points out
in the article, these tendencies are problematic enough within linguistics
itself. But they become even more problematic when confronted with the
real-world complexities involved in legislating cases involving the First
Amendment.
At the root of the problem is the fact that not all forms of speech are
protected. The U.S. government has always reserved the right to limit indi-
vidual expression when it goes against the common good, as, for example,
anti-draft rhetoric during wartime, or when it inflicts harm on others, as
in the case of defamation, or “fighting words” whose sole purpose is to pro-
voke conflict. That is, the legal system recognizes that some language use is
merely expression, while other language use constitutes action—and lan-
guage as action is not protected by the First Amendment. The problem is
and has always been: How can we tell the difference between the two?
As if this were not difficult enough, there is a second problem: what
has harmful effects on others may not have been intended as action by
the speaker, but rather as simple expression. (Such, for example, is often
claimed to be the case in defenses of “racist” and “sexist” language use.)
Whose perspective should prevail in determining whether the speaker
is guilty of a “language crime,” or whether he was merely exercising his
First Amendment right to free speech—that of the speaker or that of the
offended party? Lakoff argues insightfully that these are false dichotomies,
311
both in legal theory and speech-act theory. Rather than classifying every
speech act in terms of either action or expression, intent or effect, she bases
her analysis on the notion that language use is gradient and complex. That
is, words constitute action (they are performative) to varying degrees, and
the effects (perlocutionary force) words have on the hearer are part of the
socially situated meaning of an utterance, along with the intentions (illocu-
tionary force) of the speaker.
Lakoff is not the first to critique speech-act theory on the grounds that
it falls short of capturing the complexity of actual language use (see, e.g.,
Stubbs 1983). She is, however, the first to do so with evidence from court
cases, and the first, to my knowledge, to address linguistic issues sur-
rounding the First Amendment. In this, she has targeted a key issue of our
times—the First Amendment is increasingly evoked in contemporary pub-
lic discourse in defense of controversial behaviors ranging from flag burn-
ing to posting offensive speech on the Internet.
Here I find myself once again inspired by Lakoff’s work, this time not
as a graduate student seeking reassurance that linguistic research can be
socially meaningful, but as a professional linguist interested in how social
inequality is perpetuated through language use on the Internet. In the
course of my investigations into gender-based asymmetry on the Internet,
I have observed a disturbing correlation between the use of misogynistic,
harassing language by some individuals (all of whom appear to be male) and
explicit appeals to the First Amendment. Lakoff’s article provides a useful
perspective from which to understand and critique this phenomenon.
I should state first off that defense of misogyny is but one manifesta-
tion of a more general civil libertarian-influenced ideology on the Internet
in which the concept of “free speech” plays a crucial role. In this view, “free
speech” is conceptualized in the strongest possible terms, as absolute
and unrestricted. Such an extreme stance is justified, in part, in terms of
the noncorporeal nature of computer-mediated interaction, which alleg-
edly ensures that all that takes place is “just text,” even when the words
imitate actions. It cannot physically harm you (the argument goes), and
if you don’t like it, you aren’t forced to read it—you can delete the mes-
sages on your computer screen, or avoid participating in electronic forums
containing the kinds of messages you find offensive. Reasoning of this
sort was evoked—successfully—in 1995 to defeat the first introduction
of the Communications Decency Act in Congress which aimed to restrict
the transmission of pornography via computer networks (Herring 1996).
(The Act was later amended and passed in 1996.) According to this view, all
communication on the Internet is, by technical definition, expression and
not action, and therefore is protected by the First Amendment.
P R AG M AT I C S A N D T H E L AW [ 311 ]
312
This view is at odds with three important points made in Lakoff’s arti-
cle: (1) that words can constitute actions; (2) that the effects of the words
on others are relevant in assessing their harmfulness; and (3) that “free
speech” is not protected when it harms others. Of course, Lakoff did
not write her article with the Internet in mind—perhaps the electroni-
cally mediated nature of computer communication takes us beyond such
considerations, rendering them obsolete. Some postmodern theorists of
cybercommunication would no doubt wish to claim this. But does com-
puter mediation really mean that people don’t “do things with words” on
the Internet, and that they aren’t accountable for their words if others are
adversely affected by them?
Revealing counterevidence appears as soon as we set theorizing aside
and look at actual computer-mediated interaction. Consider, for example,
a much-publicized event that took place on the Internet several years ago
involving a “virtual rape” (Dibbell 1993). A male character in a recreational
MOO (an abbreviation for “multi-user dungeon, object-oriented,” a type of
real-time electronic forum) publicly posted graphic descriptions of a vio-
lent rape having as its primary object a female character in the forum. The
descriptions were clearly performative in structure, making use of immedi-
ate present tense verbs in sentences of the type “X does Y to Z.” However,
the man responsible for the actions of the male character claimed after-
ward that his words were not intended to be taken seriously; rather, he was
experimenting with the freedom to construct different textual personas
in the new medium. That is, he appealed to the prevailing libertarian view
that “anything goes” on the Internet, because it is “just text.” In contrast,
other members of the group noted the psychological distress of the woman
whose character was raped, and the disruptive effect the event had on the
group as a whole, and maintained that the behavior constituted a harm-
ful action. After much discussion, the group voted to deny the man future
access to the MOO.
In this case, a community of users decided that computer-mediated
words were more than “just text,” that their meaning was determined in
part from their effects on others, and that because the effects were harm-
ful, the speech was sanctionable. In other words, the virtual community,
like courts of law in the physical world, concluded that in order to preserve
the common good, some restrictions on absolute free speech were neces-
sary. This outcome is fully consistent with Lakoff’s claims, but is inconsis-
tent with the arguments of male cyberspace libertarians, which in light of
this and similar incidents, seem not so much “postmodern” as self-serving
and pragmatically naive.
313
Although the group legislative process that transpired in the MOO case
was an isolated event, it opens the floodgates, it seems to me, to the same
kinds of messy deliberations faced by courts of law in the “real world”
regarding the First Amendment. That is, the Internet, too, will ultimately
have to confront in a systematic way the fundamental problems identified
in Lakoff’s article—expression or action? intent or effect?—and when it
does, it would do well to avoid false dichotomies.
REFERENCES
Dibbell, Julian. 1993. A rape in cyberspace. Village Voice, 3 August 1993, 33–37.

Herring, Susan. 1996. Freedom of speech or freedom of harassment? The College
(magazine of the University of Texas at Arlington, College of Liberal Arts)
1(1):7–8, http://ella.slis.indiana.edu/~herring/cola1.html
Stubbs, Michael. 1983. Discourse analysis. Cambridge, UK: Cambridge
University Press.
P R AG M AT I C S A N D T H E L AW [ 313 ]
314
315
CHAPTER 13
Pragmatics and the law

Speech act theory confronts the First Amendment (1992)
I n this paper,1 I want to discuss a serious problem for two fields seriously
concerned with language use: pragmatics, and therefore linguistics in
general, and the law. It seems to me that speech-act theory, as framed by
J. L. Austin (1962), an essential underpinning (in some form) of pragmat-
ics (as well as other areas of linguistics), is in direct conflict with current
interpretations of the First Amendment of the Bill of Rights of the United
States Constitution (cf. Tribe 1988). While it is possible to imagine ways
out of the morass, they will probably be anathema to legal and linguistic
theory.
Linguistics and (in some respects) the law are social sciences; both
deal with the meaning and intention of language—the understanding
of human motives as represented in linguistic form, language as a link
between the mind and reality. Additionally, linguistics and legal schol-
arship are both interpretive disciplines. Both have access only to super-
ficial representations, and both achieve their results by making sense of
those surfaces: by finding recurrent patterns, deeper rationales, and gen-
eralizations. Linguistics, it is true, places its emphasis on description and
This paper originally appeared in Papers from the Twenty-Seventh Regional Meeting of
the Chicago Linguistic Society (CLS 27). Vol. 1, The Main Session, edited by Lise M. Dobrin,
Lynn Nichols, and Rosa M. Rodriguez, 306–23. Chicago: Chicago Linguistic Society,
1992. Reprinted here with permission of the publisher.
1. I would like to thank Elizabeth Bader, Barbara Bryant, and Jesse Judnick for their
insightful discussion and other help with this paper.
316
prediction, the law on prescription, but both seek to relate form and func-
tion. Both linguistics and the law have theoretical and applied branches.
But in linguistics, theory has always been preeminent, with concern for
real data secondary and practical application still further downgraded; the
law, on the other hand, begins with the reality of statute and precedent,
and derives theory therefrom, with the sense that the theory should clarify
and simplify the task of those who are engaged in real-world application
(trial and appellate attorneys and judges).
So if a crucial tenet of linguistic theory is directly at odds with a basic
aspect of American constitutional law, and if we agree that theories of lan-
guage use, meaning, and function should be consistent across disciplinary
boundaries, we are faced with a paradox. In this paper, I will offer one pos-
sibility for a rapprochement, more as a first attempt, to see if it’s feasible,
than as a complete and cohesive model. I assume that, in the long run, a
unified understanding of the functioning of language among the social sci-
ences (including, as here, linguistics and the law) is not only desirable but
essential.2
Central to the problem is speech-act theory, as originally developed by
Austin and developed by many others in linguistics and the philosophy
of language, perhaps most notably John Searle (1969, 1979). While there
has been a great deal of rethinking and revision of speech-act theory since
Austin’s proposal, its basic claims retain their validity, and it is those essen-
tial claims that are troublesome in this perspective.
Austin’s aim was a critique of theories of language that equated truth,
or truth- functionality, with meaning: a sentence whose truth- value
could not be computed (anything other than a declarative), was, for
the purposes of logical positivism, technically meaningless. To Austin
as an ordinary-language philosopher, this conclusion was dissatisfying,
2. I am aware that the suggestions I am making here and below could not readily
be amalgamated into legal theory and much less into legal practice. That is why I am
making these proposals in a linguistic, rather than a legal forum. It is not that linguists
are necessarily more openminded than lawyers, or more willing to abandon treasured
theoretical beliefs. But changes in linguistic theory do not require changes in reality
(for instance, in the way we speak). But if legal scholars were to re-evaluate their inter-
pretations of the First Amendment (and other linguistically based legal formulations)
and practitioners were obliged to implement their proposals, the application of the law
would be radically altered, with resultant chaos.
As linguists, we have the luxury of being able to concern ourselves with matching the-
ory to reality, rather than vice versa. It is our task to ask how well a theory accommo-
dates itself to real language observations, simplifies real-world complexities (without
oversimplifying), and relates phenomena that have something in common. We may
wish the realities of linguistic form molded themselves more readily to the theoretical
devices we have at our disposal—but we cannot require that they do so.
317
removing as it did large parts of language from the realm of the analyz-
able, the interesting, or even the meaningful. Yet it had its charms. By
reducing complexity to dichotomy, it allowed “meaning” to be formalized.
Thus it offered an opening to the scientization of language—to many,
then as now, a consummation devoutly to be wished and eventually to be
achieved.
Austin offered an alternative, less formalizable, but permitting clari-
fication and taxonomization— themselves prerequisites to scientific
treatment. In his framework, truth was reduced to a condition on the
appropriate usage of one class of verbs, rather than the major criterion of
meaning. Austin also shifted the central focus of investigation from mean-
ing, that is, semantics, to function, or pragmatics. (Where meaning is com-
puted in terms of truth, function is judged by appeal to appropriateness in
context.) Austin saw language as functioning, that is, doing things, altering
reality rather than merely describing it.
Austin begins his argument rather blandly. Some utterances, which he
called constatives, are evaluated in terms of their truth-value; others, a
smaller category, performatives, cannot be assessed for truthfulness, but
only for appropriateness, or as he called it, felicity. They are judged by their
success or failure in bringing about real-world changes. Such utterances
are speech acts: they perform actions (alter reality) through the medium of
words.3
It ought to be the case that utterances with similar meanings, func-
tions, etc. are analyzable as members of the same category (performative
or constative). But there exist pairs very closely related in meaning, but one
member constative and the other performative: I promise to return this book
on Friday, a performative; and I will return this book on Friday, a constative.
A performative/constative split creates a paradox. Austin’s remedy was to
make all utterances performative: some, like the first example, explicitly
so; others, like the second, implicit. Then truth becomes one of the felicity
conditions on the usage of a subset (verbs of asserting) of expositives, one
class of performatives. So, for Austin, all language is action: all utterance is
performative, and all equally performative.
But many of us had nagging doubts about Austin’s resolution of the
problem, even as we found it a source of significant insight, though in our
3. A constative, in Austin’s original discussion, is a descriptive statement: The sky is

blue; It will snow tomorrow; The Allies won World War II. Performatives create, rather
than describe, reality; they bring new situations into being by language alone. As such,
they must be first person and present tense, and their main verb must represent an
action that is performed linguistically: I christen this ship the Queen Victoria; I’m telling
you to get out of here right now; I promise to return this book tomorrow.
P ragmatics and the law [ 317 ]

318
youth we tended to relegate those doubts to the linguistic position of honor,

under the carpet. One particular fact that was assiduously ignored is that,
as performatives or world-changers, not all verbs of linguistic communica-
tion are equal. For one thing (as I said some years ago; R. Lakoff 1980),
an utterance containing an explicit performative is not really equivalent
to its “paraphrase” without it. But even more disturbing for the theory,
saying, that is, assertion, does not have the same “world-changing” clout
as, for instance, “apologizing,” “ordering,” or, especially, “appointing” or
“excommunicating.” It isn’t that verbs of saying are nonperformative—just
less performative than others. But Austin’s (and our) theory forced us to
make a dichotomous decision: fully performative or fully not. There was no
way to represent the ambivalent reality as a scalar relationship,4 and so we
opted for the better of the two choices, declaring all communicative verbs
equally capable of performative status.
Discussion of felicity (under various names) further specified the work-
ings of performatives. In general, the speaker of the utterance was held to
be responsible for its meaning and function (its illocutionary force), with
the addressee or hearer as a more or less passive recipient. Austin’s theory
did include a complementary perlocutionary force, what the hearer made of
the utterance both in terms of understanding and responsive action (the
“uptake”). But theoretical interest has focused on the illocutionary force,
on the argument that it can be more rigorously defined and determined.
(That assumption may be formally true, but factually false: we tend to have
a much better conscious grip on what has been said to us than—beneath a
relatively superficial level—on what we ourselves have meant by what we
have said. We are seldom fully aware of our deeper intentions. At the same
time, we tend to believe that we know what we mean, we mean what we say,
and we say what we mean. In the gap between the reality of nonresponsibil-
ity and the myth of responsibility lie many of the reasons for the existence
of fields as disparate as psychotherapy, literary criticism, the law, military
science, and politics, not to mention pragmatics).5
Then pragmatics6 includes a theory of speech acts in which all language
is action, and the same level of action; and in which the speaker bears most
4. Ross’s discussion of squishes (e.g., 1972, 1973), somewhat after the period of
strongest interest in performatives, offers us a way to begin to deal with these prob-
lems more lucidly.
5. Another way to frame this apparent contradiction is to say that, at a general and
theoretical level, the illocutionary act is superficially apparent, much easier to deter-
mine and define than its complement. But on a case-by-case basis, looking at actual
utterances produced by real people and heard by others, the perlocutionary act seems
much more reliably determined.
6. I told you not to mention pragmatics!
319
of the responsibility for the import of the message. These assumptions

are problematic enough within the field itself, but become more so when
placed alongside current interpretations of the First Amendment, in con-
stitutional theory and appellate practice.
On its face, the meaning of the First Amendment is simple enough. It
reads, in its entirety:
Congress shall make no law respecting an establishment of religion, or prohibit-

ing the free exercise thereof; or abridging the freedom of speech, or of the press;
or the right of people peaceably to assemble, and to petition the Government for
a redress of grievances.
For about the first 130 years of its existence, the First Amendment led a
quiet existence (cf. Kairys 1982). The occasional cases that invoked it before
the Supreme Court tended to be settled in a way that restricted its domain,
rather than extending it.7 Its scope was first directly tested in the years
immediately following the first World War, in a series of cases pitting the
First Amendment against the government’s right to restrict speech in situ-
ations of “clear and present danger” such as occur most often in wartime.
Most of these cases involve persons who, during wartime, urged the United
States to withdraw from the conflict or counseled young men to resist the
draft. The first such case, Schenck v. United States,8 was decided against the
appellant, Schenck, by a unanimous court, in which Justice Oliver Wendell
Holmes wrote the opinion, including as part of its justification the famous
analogy: “The most stringent protection of free speech would not protect
a man in falsely shouting fire in a theatre and causing a panic.”9 But in
a rather similar case decided later that same year, Abrams et al. v. United
States,10 Holmes, along with Justice Louis Brandeis, dissented from the
majority affirmation of the verdict because the wording of the pamphlets
involved does not allow the inference that the defendants acted “with
intent by such curtailment to cripple or hinder the United States in the
prosecution of the war,” as the statute under which they were prosecuted
requires. Holmes and Brandeis required proof that the speakers had an
intent to do what their language appeared to suggest. In other words, the
majority of the Supreme Court, in accordance with normal judicial behav-
ior of the times, was requiring only a demonstration of perlocutionary force
7. Cf. especially Davis v. Massachusetts, 167 U.S. 43 (1897).

8. Schenck v. United States, 249 U.S. 47 (1919).
9. Id. at 52.
10. Abrams et al. v. United States, 250 U.S. 616 (1919).

320
to affirm a guilty verdict. Holmes’s and Brandeis’s dissent strengthened the

First Amendment by requiring a proof of relevant illocutionary force. “I am
aware of course,” writes Holmes,
that the word intent as vaguely used in ordinary legal discourse means no more
than knowledge at the time of the act that the consequences said to be intended
will ensue… . But, when words are used exactly, a deed is not done with intent
to produce a consequence unless that consequence is the aim of the deed.11
This two-man dissent gradually developed, over the next dozen years or so,
with later changes in Supreme Court membership, into a majority which
helped to give the First Amendment more or less its current understanding.
The issue or responsibility for the interpretation of an utterance was
not, however, fully resolved by Holmes’s discussion: later courts continued
to argue over whether responsibility for meaning resided with speaker or
recipient. Another problem, for both pragmaticists and legal scholars, lies
in the interpretation of even simpler language: What does the “free speech”
clause of the First Amendment actually refer to?
The clause says that Congress shall make no law abridging the free-
dom of speech. The words italicized are simple, but their range of meaning
broad: they are vague, or ambiguous. And, as is frequently the job of appel-
late courts, this vague language must be reduced to dichotomous precision
in order to conform to the needs of legal reality, to set precedents and allow
appropriate remedies to be taken.
The first of these terms, freedom, has been seen by the Supreme Court
throughout its history as relative rather than absolute, guaranteed as long
as it does not conflict with another constitutionally recognized right or
obligation. In the case of “clear and present danger,” as most typically in
wartime (or its moral equivalent, for example the Red Scare of the 1950s or
today’s “war on drugs”), the duty of the government to protect its citizens
from enemies foreign and domestic is frequently deemed by the Supreme
Court to supersede the First Amendment right of citizens to free speech.
Likewise, in its obligation to ensure domestic peace and tranquillity, the
government has been adjudicated to have the right to restrict speech in
order to prevent breaches of the peace, so that “what men of common
intelligence would understand [as] … words likely to cause an average
addressee to fight …,’’12 that is, “fighting words,” do not receive constitu-
tional protection. But the line between potentially troublesome utterances
11. Id. at 627–28.
12. Chaplinsky v. New Hampshire, 315 U.S. 568 (1942) at 573.
321
that are unprotected “fighting words,” and those that are not, and there-
fore protected by the First Amendment, has been a tricky one for appellate
courts and constitutional scholars to draw. And one traditional distinction
on which both “fighting words” and “clear and present danger” interpreta-
tion has been based conflicts directly with Austinian speech-act theory.
Jurists and scholars have, since at least the 1920s, wrestled with this
problem: how to draw lines between the kind of speech they want to pro-
tect, and other kinds of social behavior that should be subject to criminal
or civil prosecution. They have frequently drawn those lines on the basis
of hypotheses about the Founders’ original intention in putting the First
Amendment into the Constitution. One frequently cited justification is
Alexander Meiklejohn’s (1948) doctrine of the marketplace of ideas, based
on Holmes’s dissent in Abrams: the First Amendment exists because, in a
democracy, citizens need to have access to a range of information in order
to vote intelligently. Thus, any restriction on the flow of relevant informa-
tion is destructive to the functioning of a democratic government.
This rationale allows a distinction to be made between conduct and
expression. Language does express ideas, the articles of trade in Holmes’s
marketplace. But it also can be used simply to let off steam, or to abuse
others, not unlike the ringing of bells or a punch in the nose. Such language
does not materially aid the business of informing the citizenry; indeed, it
may hinder it. That sort of behavior, the reasoning goes, is tantamount
to action, indeed it is action. And the law has never recognized protection
for actions per se. There are uncontested laws on our books dealing with
disturbance of the peace, physical assault, and other crimes of aggressive
interpersonal conduct. When language is used as action, it, too, can expect
no constitutional protection.
In the abstract, as is the case with most jurisprudential reasoning, the
argument makes a great deal of intuitive sense. But while the law (like too
much linguistic theory) is predicated on dichotomies such as this, the real
world does not return the favor: it organizes its phenomena as continua. So
trial courts instruct juries hoping to reduce the blooming, buzzing confu-
sion of the conduct they are called upon to judge into the either/or catego-
ries mandated by law: guilty or innocent, intentional or unintended, sane
or insane … expression or conduct. What trial juries frequently do, in this
impasse, is craft a compromise: find the defendant guilty of a lesser offense
than the prosecution demands and the evidence permits, as a way of say-
ing, “Guilty, but… .”
Appellate courts adopt analogous stratagems, albeit of a more explicit
and sophisticated form. Often uncertainties about the plausibility of the
legal dichotomy take the surface form of sharply divided courts, with

322
plethoras of dissents and concurrences. In this reading, a 9–0 Supreme

Court decision is a way of saying something like, “We have no trouble con-
struing the situation in the dichotomous terms the law provides,” while
5–4 with separate concurring opinions on both sides often corresponds to,
“We feel the situation is too complex to be covered by the law, or any law
written as laws must be; so the division of the Court means that we are as
a whole uncomfortable with the way we have to operate as jurists.” Not
always, by any means; but often.
The neat conduct/expression dichotomy was problematic for the Supreme
Court which, in 1970, handed down another famous opinion, Cohen v.
California.13 The appellant, Cohen, was arrested for entering the Los
Angeles Superior Courthouse wearing a denim jacket on the back of which
were inscribed the words FUCK THE DRAFT. There is no allegation that
he intended to create a disturbance by so doing, nor did any disturbance
result. Yet he was charged with and found guilty of violating a section of the
California Penal Code that prohibits “maliciously and willfully disturb[ing]
the peace or quiet of any neighborhood or person … by … offensive con-
duct… .”14 So a statute originally intended to punish breaches of the peace
was used in this case to prosecute Cohen for using “fighting words,” and
he was convicted on the grounds that his behavior constituted conduct, as
opposed to expression. He appealed on the basis of his First Amendment
right to free speech, claiming that the wearing of the embellished jacket
was expression, rather than conduct. A bare 5–4 majority agreed with
the appellant’s reasoning and reversed. Writing for the majority, Justice
Harlan noted,
The conviction quite clearly rests upon the asserted offensiveness of the words
Cohen used to convey his message to the public. The only “conduct” which the
State sought to punish is the fact of communication. Thus, we deal here with a
conviction resting solely upon “speech” … not upon any separately identifiable
conduct… . At least so long as there is no showing of an intent to incite disobe-
dience to or disruption of the draft, Cohen could not, consistently with the First
and Fourteenth Amendments, be punished for asserting the evident position on
the inutility or immorality of the draft his jacket reflected.15
So the majority in Cohen based its reversal both on the lack of intention,
and on the fact that Cohen’s jacket was an example of expression, rather
13. Cohen v. California, 403 U.S. 15 (1970).

14. Id. at 16.
15. Id. at 18.
323
than conduct. The dissenting opinion (written, incidentally, by Justice

Blackmun) argued vehemently against this latter view. That would suggest
that, as of twenty years ago, no clearcut distinction between expression
and conduct could be counted upon, as indeed it cannot to the present day
(as the recent Supreme Court decision on flag burning,16 holding the lat-
ter again by a bare majority to be expression rather than conduct, would
indicate).
We have thus far identified two points at which First Amendment inter-
pretation intersects with, and comes into conflict with, speech-act the-
ory: on intent (or illocutionary versus perlocutionary force); and on the
distinction between expression and conduct. Let us deal with each of these
at greater length.
Any theory of language meaning and function, as well as any theory
of jurisprudence, must deal, each in its own way, with the problem of the
responsibility for the meaning of communicative behavior. The issue is cur-
rently in contention in many of the social sciences and the humanities.17
It also underlies two of the bigger political antagonisms currently rife on
campus, so-called “political correctness” and the promulgation of edicts
forbidding the use of “fighting words.” In the first case, the assignment
of responsibility raises questions about authority and the existence of a
“canon.” If there is no preordained canon, if instructors do not have abso-
lute authority to decide what to teach and how to teach it, many feel the
whole notion of the university, and indeed knowledge itself, is in terrifying
disarray. The second problem lies in the right or ability to determine the
existence of a verbal “insult”: Is legal authority to look to the utterer’s pro-
fessed (or demonstrated) intention, or the sensibilities of the hearer? Who
determines what an utterance “means”?
In an interesting recent article (1987), Peter Tiersma considers the
problem of the assignment of responsibility of meaning from the legal
perspective, with reference to defamation law. Tiersma shows, first, that
historically defamation (a blanket term that includes libel for written lan-
guage and slander for spoken) was determined by assessing the effect on its
object: damage to business or reputation, as the recipient himself or her-
self perceived it: that is, the perlocutionary effect of the utterance. Tiersma
discusses the problems inherent in such a theory: most obviously, effect
is difficult to assess with accuracy; and what is injurious to one person is
16. Texas v. Johnson, 109 S. Ct. 2533 (1989).

17. It is, and has been over many years, disputed under rubrics as diverse as antimen-
talism, phenomenology, hermeneutics, ethnomethodology, deconstructionism, and
postmodernism.

324
harmless to another. The law is too generous in its assumption that a nor-
mal speaker can make an accurate assessment of a hearer’s sensibilities; to
do so accurately would require knowledge of, and the ability to compute,
the hearer’s frame of reference, the status of hearer and target relative to
speaker, the physical and psychological setting in which the speech act took
place, and much more—far more than it is reasonable to suppose is avail-
able to the average speaker, especially in oral discourse.
On the other hand, an entirely speaker-based theory of defamation (or
language crime generally)—that is, one based purely on the computation
of illocutionary force—is equally untenable. Especially in an adversarial
situation where the utterer has much to lose, the latter’s assessment of
his/her intention is not wholly trustworthy. The same applies in cases of
harassment, “fighting words,” and linguistic crimes or torts in general. It
is all too easy to plead “only joking,” or “never realized,” or myriad other
bluntings of the responsibility for the force of the speech act. And it is fre-
quently impossible to prove otherwise.
A case of sexual harassment that recently made its way through the fed-
eral appellate courts illustrates the problem. In this case, Ellison v. Brady,18
a woman is suing a coworker for harassment, charging that he pursued her
with bizarre notes, although she asked him several times to desist, and
denied interest in him.
The trial court granted the employer summary judgment (dismissed the
plaintiff’s claim) on the basis that the conduct of the coworker was not
“sufficiently severe or pervasive” to constitute sexual harassment. Based
on the standard of what would be threatening and offensive to the “reason-
able man,” the trial court found the harassment isolated and trivial.
The Ninth Circuit reversed, on the grounds that the “reasonable man”
standard was inappropriate in this case; that women were in a more vulner-
able position, physically and psychologically, than men, reasonable or oth-
erwise, and that to invoke a hypothetical “reasonable man” was to adopt
a standard inappropriate to the plaintiff in this case. In other words, the
Circuit Court held that to some degree and for some cases at any rate, the
particular world-view of the addressee had to be taken into account, by
requiring that the jury consider a standard based on a “reasonable woman.”
The US Supreme Court came to a different conclusion in a criminal
case, Rhode Island v. Innis.19 A suspect was arrested on suspicion of murder.
18. Ellison v. Brady, 91 Cal. Daily Op. Service 605 (9th Cir. 1991), as discussed in
Bryant 1991.
19. Rhode Island v. Innis, 446 U.S. 291 (1980). My summary is taken from Tiersma
(1987:315 n. 52).
325
Read his Miranda rights, he asked for a lawyer (that is, refused to waive his
rights). The law is firm here: at this point, all questioning should stop; any
statement obtained without counsel present is legally inadmissible if it is
elicited via a question. The arresting officers were, though, understandably
eager to discover the location of the murder weapon. As they were driving
to the police station, one officer said to the other that there was a school
for the handicapped in the vicinity of the crime scene, and “God forbid”
that one of the children should get hold of the weapon. The suspect then
revealed where he had hidden the gun. The confession was deemed admis-
sible by the trial court, and the defendant was convicted. The Supreme
Court was asked to determine whether the officer’s communication had
the force of a question, which would have required the verdict to be over-
turned. The Court decided that it did not, defining interrogation as “words
or actions on the part of police officers that they should have known were
reasonably likely to elicit an incriminating response.”20 The dissent, writ-
ten by Justice Stevens, argued on the other hand that “a statement that
would be understood by the average listener as calling for a response is the
functional equivalent of a direct question, whether or not it is punctuated
by a question mark.”21 The opinion of the court in this case, then, is to the
effect that the determination of meaning or intent rests with the speaker
of the utterance.
Cases like Ellison suggest that the likely interpretation of language by a
hearer in a particular situation, not only what a hypothetical “reasonable
man” might deduce, and certainly not only what the speaker might claim to
have meant, must be part of any decision in a linguistic crime or tort action.
On the other hand, Innis comes to the opposite conclusion (and is the judg-
ment of a higher court, too). So, as they say, the jury is still out on this one.
Another case currently before the Supreme Court will also raise the
issue of responsibility for interpretation. The Court is being asked by
Jeffrey Moussaieff Masson to decide whether there are grounds for a
libel suit against reporter Janet Malcolm and the New Yorker magazine.
Masson alleges that, in two articles published in 1983, Malcolm libeled him
by grossly distorting the content and implications of his remarks during
interviews with her. At issue is the definition of paraphrase.
It is part of a journalist’s normal job to “fix” the spontaneous oral
remarks of subjects, putting them into the form readers will consider
articulate and intelligible in writing.22 People who have had the misfortune
20. Id. at 322.
21. Id. at 309.
22. Cf. in this regard Malcolm (1990) herself.

326
to encounter unedited transcriptions of their speech will appreciate the

necessity for this charitable act: without it, the most articulate among us
come out sounding like unintelligible oafs. It is further (more tacitly) rec-
ognized that such emendations may appear in print between quotation
marks, even though the words were not exactly what was said. But how
much liberty is appropriate? When does a paraphrase cease to be that, and
become the reporter’s fictive creation?
In the Masson case, both parties agree (in some instances at any rate)
on exactly what Masson said. But the disagreement (and the basis of the
lawsuit) hangs on Malcolm’s notion of paraphrase: how much right did she,
as journalist and hearer, have to determine what Masson really “meant”?
Some of Masson’s allegations, if true, stretch any definition of paraphrase
as linguists use it to the breaking point. Yet a trial court and several lev-
els of appellate review have determined that he does not have grounds to
sue for libel. The U.S. Supreme Court is expected to hand down a decision
within the next couple of months, as I write.*
For instance: Masson said on Malcolm’s tape, speaking of the psycho-
analytic authorities: “They felt I was a private asset but a public liability.”
Those words are all that show up in the interview transcript. The published
story reads: “[They regard me as] an intellectual gigolo. You get your plea-
sure from him, but you don’t take him out in public.”23 Malcolm, according
to Masson, argues that the meaning she represented as gigolo is equivalent
to Masson’s distinction between public and private. A semanticist might
argue that there are other significant components to gigolo. But it will be
up to the Supreme Court to decide whether the numerous such instances
in the text are sufficiently problematic as to justify a lawsuit. If they are
fair paraphrases, there are no grounds for legal action. This may be the first
postmodern Supreme Court decision.
Thus we find, with mingled delight and unease, that some of the most
important and unresolved issues of pragmatics have the same status for
constitutional scholars. A recent Supreme Court decision seems to blur
another dichotomy important to defamation law, that between opinion
(which is not defamatory) and factual statement.24 The opinion of the court
* The Supreme Court decided in June 1991 that Masson’s lawsuit against Malcolm
should be allowed to go forward, stating that the issue of whether Masson’s statements
had been materially altered in meaning was sufficiently in question to be a matter for
jury trial (Masson v. New Yorker Magazine, 1991 U.S. LEXIS 3630 (1991). After a series
of jury trials and appeals, the suit was eventually decided in favor of Malcolm (Masson
v. New Yorker Magazine, 1996 U.S. App. LEXIS 13326 [9th Cir. 1996]).—Ed.
23. Oral communication with J. Masson, March 18, 1991, and column by James
J. Kilpatrick in the San Francisco Chronicle, October 12, 1990.
24. Milkovich v. Lorain Journal Company et al., 110 S. Ct. 2695 (1990).
327
comments that the distinction seems impossible to maintain: Can prefac-

ing (e.g.) “I think” to an utterance move it automatically from fact to opin-
ion? This is the same problem linguists face in dealing with indirect speech
acts and conveyed meaning.
If we hope to achieve a rapprochement between linguistic and legal
treatments of language, there are sharp differences in points of view and
emphasis between the fields; but some of the problems constitutional
interpreters and appellate courts are struggling with are reminiscent of
our own worst nightmares. Is there any way the two fields might—ideally,
if not practically, speaking—work toward solutions of these problems sat-
isfactory to both?
The dichotomization so beloved of both the natural and the social sci-
ences has, on the one hand, enormous practical value: it clarifies, it reduces
complexity, it seems to provide generalizing power. But language and lin-
guistic behavior (and human behavior in general) by their nature may not
be reducible to yes/no decisions, nor amenable to being assigned to only
one pile among many. For language and probably other human behaviors,
dichotomies are too often attractive oversimplifications; to get anywhere
near a deep understanding of the forms, functions, and properties of lan-
guage, a theory which includes the possibility of continuous classifica-
tion is necessary. This applies as well to the law in its language-analytic
functions—including of course First Amendment issues.
This suggestion is relevant for both of the problems I have discussed. In
the first place, responsibility for the meaning or effect of an utterance needs,
as Tiersma has suggested, to be spread between speaker and hearer—or,
more accurately in some cases such as defamation, among speaker, imme-
diate addressee(s), and target/victim. Under different circumstances, the
assignment of responsibility for meaning would be assigned to one par-
ticipant or another, or parceled out appropriately among them. It is not
sufficient to look to perlocution alone, as defamation law has tended to
do: some people are more vulnerable, some more sensitive, and some sim-
ply crazier than others. At the same time, it is inappropriate to place all the
responsibility for meaning on the author of the text, as was done in cases
like Schenck and Abrams. In an adversarial situation, speakers would be too
free to disavow or mitigate their original intentions. Some notion of “com-
munity standards” might helpfully be developed here: what an ordinary,
rational speaker, in roughly the identified context, can be counted upon to
make of an utterance; as well as what an ordinary, rational speaker, operat-
ing by Gricean and politeness principles, might be expected to say, and with
what intent, in a given circumstance. Courts cannot depend on any of the
involved participants to be objective, and must recognize that language is

328
protean: its meaning changes depending on context, and the courtroom-

adversarial context changes it yet once more.
Secondly, the problems raised by the legal distinction between expres-
sion and conduct forces linguists to focus on the equally problematic, and
equally attractive, conflation of all language into performatives, that is,
conduct. As the uncertainty and impermanence of Supreme Court deci-
sions and linguistic formulations in this area suggest, neither dichotomy
is tenable or deeply useful. Words can have the effect of actions, be tan-
tamount to action sometimes (as with fighting words, which by definition
have no ideational content, although even this is often arguable); and like-
wise, nonverbal behavior under certain conditions can be largely or purely
expressive, that is, symbolic and non-iconic (so, for instance, flag burning
is not merely, or even at all, the act of incinerating a piece of red, white, and
blue fabric). Words have consequences and actions have meaning.
Austin’s identification of all speech acts as performative is attractive, not
least because it implies that language is real, it does things, it makes a dif-
ference. But not all utterances are equally “active,” that is, reality-changing.
Recasting the constative-performative dichotomy as a continuum may help
account for the perplexing behavior of speech-act verbs.
The basic generalization is this: the more natural and typical it is for a
verb of linguistic communication to be represented explicitly, and the less
additional meaning explicit expression of a performative verb adds to an
utterance, the more “actionlike” the verb, and the behavior it describes, will
be. This connection suggests a scale, from the least “active” performatives,
verbs like say, in which explicit mention tends to create strained and arti-
ficial structures, to the most, like appoint, whose actions frequently can-
not be done without explicit mention. We might attempt to represent the
range of possibilities as in Scale A of Figure 13.1.
The legal distinction between expression and conduct can be represented
along an analogous continuum, as in Scale B of Figure 13.1. While the legal
range between expression and conduct does not coincide exactly with the lin-
guistic continuum between constativity and performativity, the dotted lines
drawn between the two lines of Figure 13.1 represent points of similarity.
Scale A. The linguistic constative/performative continuum

1 = hedged assertions
2 = descriptions; generic statements; obvious truths
3 = controversial assertions
4 = ask, order
5 = apologize, promise
6 = appoint, excommunicate
329
constative performative
1 2 3 4 5 6
expression conduct
1 2 3 4 5 6
Figure 13.1 Language–action continua.
Scale B. The legal expression/conduct continuum

1 = informative speech
2 = persuasive speech
3 = inflammatory speech: “clear and present danger”
4 = acts as speech: flag burning
5 = language as action (with informative content): defamation
6 = fighting words (language as action, without content)
I have devoted considerably more time here to expounding problems

than to solving them, and the problems certainly remain considerable. But
I hope to have shown that the difficulties linguists and lawyers face in talk-
ing about language form and function are not unrelated, and are equally
serious. They arise on the one hand out of the real complexities of language
use and the relation between language and reality, and on the other, from
both fields’ persistent desire to dichotomize. The realities of language are
sometimes ignored by both fields in order to effect “scientific” rigor. But
both fields must find ways to see language as continuous, yet rigorously
definable and ultimately predictable; and since both fields are working with
the same material, and with some or the same goals, I am hopeful that they
can develop consistent theory together.
REFERENCES
Austin, J. L. 1962. How to do things with words. Oxford: Clarendon Press.

Bryant, Barbara S. 1991. Sexual harassment now defined by the reasonable woman.
California Courts Commentary, 31(3):6, 12.
Kairys, David. 1982. Freedom of speech. The politics of law: A progressive critique, ed. by
David Kairys, 140–71. New York: Pantheon.
Kilpatrick, James J. 1990. When a quote is not a quote. San Francisco Chronicle,
12 October 1990, A22.
Lakoff, Robin Tolmach. 1980. How to look as if you aren’t doing anything with
words: Speech act qualification. “Speech acts theory: Ten years after,” ed. by
Julian Boyd and Alessandro Ferrara. Special issue, Versus 26/27:29–47.

330
Malcolm, Janet. 1990. The journalist and the murderer. New York: Vintage.

Meiklejohn, Alexander. 1948. Free speech and its relation to self-government.
New York: Harper.
Ross, John Robert. 1972. The category squish: Endstation Hauptwort. Papers from the
eighth regional meeting of the Chicago Linguistic Society (CLS 8), Chicago, 316–28.
Ross, John Robert. 1973. Nouniness. Three dimensions of linguistic theory, ed. by
Osamu Fujimura, 137–257. Tokyo: TEC.
Searle, John R. 1979. Expression and meaning: Studies in the theory of speech acts.
Cambridge, UK: Cambridge University Press.
Tiersma, Peter. 1987. The language of defamation. Texas Law Review 66:303–50.
Tribe, Laurence H. 1988. American constitutional law. 2d edn. Mineola,
NY: Foundation Press.
331
Introduction to “The rhetoric

of reproduction”
BY L AUREL A . SUT TON
I n the time since Robin Lakoff wrote “The rhetoric of reproduction”—

almost 25 years— the rhetoric around abortion, reproduction, and
women’s health care has become ever more heated and divisive. Although
Roe v. Wade, the 1973 U.S. Supreme Court ruling that legalized abortion,
remains in effect, it has sustained attacks from many angles and many
quarters, unified in pursuit of one goal: to make abortion illegal again.
The search for this paper was the genesis of this book. At the time, it
wasn’t available through the University of California library system, nor
any other electronic database. (A quick Google search today shows the
same results: nothing.) Robin herself kindly provided me with a paper copy
of Conscience, the news journal of Catholics for Choice, in which it origi-
nally appeared. As with so many of her writings on social justice topics,
Lakoff expertly analyzes the language underpinning the logic used by both
sides, and offers strategies for countering the “deceptive and hypocritical”
anti-choice rhetorical practices she sees being used.
At the time I originally read this piece, Lakoff’s call to action (“If the war
is not to be lost, it is time to act decisively and nondefensively”) seemed nec-
essary but non-urgent. The worldwide abortion rate declined substantially
between 1995 and 2003,1 due to the increased availability of contraceptive
methods, and especially those which are long-acting, such as Depo-Provera
1. Guttmacher Institute (2015); report published in partnership with the World
Health Organization.
332
injections, intrauterine devices (IUDs), and subdermal contraceptive

implants. The discourse of the pro-choice side shifted to addressing the
prevention of unwanted pregnancy, and combating religiously motivated
“abstinence only” sex education, since proven to be spectacularly ineffec-
tive at preventing pregnancy in teenagers. A generation of women could
hope that abortion might one day be “safe, legal, and rare,” to quote the
policy of the Clinton Administration.2
What we didn’t count on was the ways in which the anti-abortion forces
would chip away at aspects of Roe, placing more and more restrictions on
abortion so as to make it practically impossible for many women to obtain
one. The tactics remain the same as when Lakoff was writing: “sloganeer-
ing, semantic obfuscation, and slick advertising (verbal and pictorial).” The
anti-abortion side has worked hard to control the discourse, the culmina-
tion of which was passage of the Partial-Birth Abortion Ban Act of 2003,
a federal law prohibiting a rarely used medical procedure called “intact
dilation and extraction” (a type of abortion performed in the second tri-
mester). The oxymoronic phrase “partial-birth abortion” was first coined
by Douglas Johnson of the National Right to Life Committee and has been
used since in numerous state and federal bills and laws. The name is clearly
meant to evoke the horrifying image of a full-term baby that is first born
and then killed; the fact that the term became attached to federal legisla-
tion shows just how persuasive this type of semantic obfuscation can be.
Anti-abortion sloganeering, too, became more direct. While the Roman
Catholic Church continues to use the phrase cited by Lakoff, “The natural
choice is life,” protestors at Planned Parenthood Health Centers and abor-
tion clinics now hold signs reading “Babies Are Murdered Here,” “Abortion
Hurts Women,” “Stop the Slaughter,” and “Mommy, Why Do They Want
To Kill Me?” (Many are accompanied by graphic images of bloody “fetuses”
which often turn out to mislabeled photos of stillborns.) Again, words
are chosen for maximum impact, regardless of their accuracy: referring
to fetuses or embryos as “babies” (or, more fancifully, “preborn babies”),
equating a medical procedure with “slaughter” and “murder,” and giving a
fetus the voice of a living, thinking child, pleading with its mother to live.
Lakoff noted all of these false equivalencies in 1992 and attributed them
to “intellectual scam artists … prey[ing] on human frailty.” Not much has
changed.
The term “abortion” itself sometimes seems out of place in the abortion
debate. As Lakoff pointed out, rather than risk the negative associations
2. “Abortion should not only be safe and legal, it should be rare.” Source: Speech by
President Bill Clinton at the Democratic National Convention, August 29, 1996.
333
with the word “abortion,” the side supporting the choice to abort went
with, well, “pro-choice.” The Center for Reproductive Rights, a global orga-
nization focused on securing women’s rights to abortion, studiously avoids
the word “abortion” in their mission statement, instead using “reproduc-
tive freedom,” and listing abortion as only one of many aspects of women’s
health care to which they are dedicated. Planned Parenthood, a nonprofit
organization that provides reproductive health services, similarly avoids the
word “abortion” in its profile, positioning itself as a “trusted health care pro-
vider.” And the National Abortion Rights Action League, usually shortened
to NARAL, has officially changed its name to NARAL Pro-Choice America;
their domain name is simply “prochoiceamerica.org.” The White House, in
its annual statement marking the anniversary of Roe v. Wade, hasn’t used
the word “abortion” since 2011, preferring the “choice” language instead.
News outlets have become uncomfortable with the soi-disant “pro-life/
pro-choice” dichotomy. In a 2010 survey of the terminology used by major
news organizations (NBC, CBS, CNN, the Associated Press, the New York
Times, the Washington Post, and the Philadelphia Inquirer), the National
Public Radio ombudsman found that not one of them used the terms
“pro-life” or “pro-choice,” most preferring “anti-abortion” and “abortion
rights,” respectively, as they are seen as more neutral.3 Yet most impor-
tantly, regardless of what either side chooses to be called, the majority of
Americans (55%) continue to support the legality of abortion, according to
a Pew Research Poll in late 2014.4
Of all of Lakoff’s suggestions for abortion-rights slogans, “forced moth-
erhood” has actually gained some traction and spawned another compelling
phrase, “forced pregnancy.” Both are used to refute the notion of women
as vessels, and to evoke a system of near-slavery of women akin to that
in Margaret Atwood’s dystopian novel The handmaid’s tale. This rhetoric
appeals strongly to American anti-authority sentiments while at the same
time acknowledging the consequences for women of being denied abor-
tion—something which previous abortion-rights language had skirted, as
it focused on abortion as a medical procedure quite divorced from the life
and experiences of the woman who chose it.
In the 21st century, the continued attacks on abortion rights, such
as attempted federal defunding of Planned Parenthood clinics and the
3. NPR itself uses “pro-choice” and “pro-life,” because it’s “more conversational and
not as cumbersome” (Shepard 2011).
4. This information was gathered as part of a Pew Research Project on the influence
of religion in the United States (Pew Research 2014).
T H E R H E T OR I C OF R E P R O D U C T I O N [ 333 ]
334
introduction of state legislation that would assign full human rights to fer-
tilized eggs5 (Associated Press 2012), have led to a new type of discourse
that could not have been foreseen from Lakoff’s vantage point in 1992—
that of social media, and the online social networking service Twitter, in
particular. Here, women felt free to discuss their own abortions openly,
rejecting the stigma of abortion itself, as they shared their stories with
each other publicly. In contrast to the type of organized, marketable rheto-
ric analyzed and suggested by Lakoff, the Twitter hashtag (searchable key-
word) #ShoutYourAbortion arose organically as a response to the efforts
to cut Planned Parenthood funding by the U.S. House of Representatives.
The American feminist and activist Lindy West explained her motivation
for catalyzing the moment: “I set up #ShoutYourAbortion because I am not
sorry, and I will not whisper” (West 2015). Tens of thousands of women
responded with the experiences of their own abortions, reclaiming the
word—not with a quiet voice, but with a shout.
The war on women’s rights is far from over, but it is clear that the new
channels of social media have provided an important alternative to the
type of rhetorical discourse Lakoff examines here. The same questions
remain: What is life? What is a person? How shall we speak of such things?
Perhaps the new access to public discourse will allow women to confront
these questions, as Lakoff emphasizes, “directly, intelligently, and fear-
lessly”—and in their own words.
REFERENCES
Associated Press. 2012. “Va. House GOP muscles through abortion curbs.” USA Today,
February 14, 2012, http://usatoday30.usatoday.com/news/nation/story/2012-
02-14/virginia-abortion-legislation/53097654/1
Guttmacher Institute. 2015. Facts on induced abortion worldwide.
New York: Guttmacher Institute, http://www.guttmacher.org/pubs/f b_IAW.
html
Pew Research Center. 2014. “Public sees religion’s influence waning.” September 22,
2104. Washington, DC: Pew Research Center. 37.
Shepard, Alicia C. 2011. “In the abortion debate, words matter.” NPR Ombudsman,
March 18, 2010, http://www.npr.org/sections/ombudsman/2010/03/in_the_
abortion_debate_words_m_1.html
West, Lindy. 2015. “I set up #ShoutYourAbortion because I am not
sorry, and I will not whisper.” Guardian [UK], September 22, 2015,
http://www.theguardian.com/commentisfree/2015/sep/22/
i-set-up-shoutyourabortion-because-i-am-not-sorry-and-i-will-not-whisper
5. Although the Virginia State Senate shelved this bill in 2012, it has been reintro-
duced every year since.
335
CHAPTER 14
The rhetoric of reproduction (1992)
SEMANTIC FLASH VERSUS HONEST TALK
In the propaganda war over reproductive rights, the forces arrayed against
choice generally represent themselves as the moral side, their way as the
high road. They portray the pro-choice position as a slippery slope of
expediency. Yet, while flashy, deceptive rhetoric seldom is considered the
ethical or moral approach, opponents of choice make regular use of such
illegitimate rhetorical strategies. That they claim to act out of religious
motives only exacerbates the dilemma such strategies must pose for them.
Confronted with this paradox, they have argued that because rhetorical
razzle-dazzle works, and the cause is worthy, the end justifies the means—
an argument rarely associated with the high moral ground.
The advocates of reproductive rights, often accused of expediency and
slippery ethical stands, are in the opposite position. Their rhetoric is above-
board and eminently reasoned; but it has been, considering the significance
of its message, singularly unsuccessful in persuasion. When this problem
is discussed, suggestions for more enticing rhetoric often are dismissed, on
the grounds that persuasive techniques that circumvent the intellect are
dishonest and dishonorable—an argument not typically associated with
expediency.
It is time we become concerned with the ways in which anti-choice rhe-
torical strategies are deceptive and hypocritical, and time we find means to
This paper first appeared in Conscience 13(2):4–12, 1992. It is reprinted here with
permission of the publisher. Conscience is the newsjournal of Catholics for Choice,
Washington, DC.
336
overcome them. This task will require an understanding of the real reasons
for many people’s enthusiasm for that side. Further, an effective opposing
strategy must discover why the persuasive attempts of reproductive-rights
advocates have been less than successful, and it must devise ways to change
the balance of the debate.
Rhetorical skill never has been more important in this struggle. As the
U.S. Supreme Court dismantles Roe v. Wade,1 abortion is thrown more and
more into the political arena, to become the business of legislatures in the
fifty states as well as the Congress. It is not so much that effective rhetori-
cal strategies are needed to influence legislators; rhetoric seldom influences
legislators—votes do. But voters can be affected by rhetoric, as indeed they
have been by pro-life rhetoric. Now it must be our rhetoric.
Following the Supreme Court’s Roe opinion in 1973, the anti-choice
side wasted no effort in developing and putting into practice hard-hitting,
gut-wrenching forms of persuasion. As many of its arguments have been
disproved or have become ineffectual through overuse, even more compel-
ling tactics have taken their place. Among these are sloganeering, semantic
obfuscation, and slick advertising (verbal and pictorial). While these are
not independent of one another, it is useful to discuss each separately to
see how they work independently as well as together.
SLOGANEERING AND SLICK ADS
Slogans have always been in the propagandist’s armamentarium. Like

poetry, they circumvent reason or overwhelm it by emotion. They sound
good, through alliteration and euphony, and they feel compelling, playing
into our deepest hopes and fears. They are memorable because of catch
phrases, even if they don’t make much sense. (Of course, not much adver-
tising does.) If money is the mother’s milk of politics, slogans are its pabu-
lum. So it is no surprise to find them providing much of the stuff of this
emotionally charged debate.
It is a little more startling to have found the Catholic Church in league
with Madison Avenue in devising some of the more dazzling recent
slogans—the ultimate union of God and Mammon. The primary slogan is
more than a little reminiscent of commercial advertising campaigns: “The
natural choice is life” links two generations of Coca-Cola campaigns. Older
readers will remember “Coke is a natural—yessirree!” and we all know that
1. Roe v. Wade, 1973 U.S. LEXIS 159 (1973).
337
“Coke adds life.” So what is more natural (to borrow a technical term) than
to unite the two?
Other slogans are close relatives, for example, the refrain of the slick
new Arthur S. DeMoss Foundation television spots, “Life. What a beautiful
choice.” Notable in right-to-life campaigns is the deliberate cooptation of
the language of the opposition—“choice”—which would, in the world of
Mammon, be actionable as trademark infringement.
But that is the least of the ethical quagmires into which such slogans,
and the Madison Avenue expertise behind them, lead their sponsors. In
the first place, it has been noted often that advertising dollars are spent
most freely to promote those products that respond to no real need.
Advertising slogans suggest reasons to choose among many products that
are either identical to each other or irrelevant to real needs. Advertising
offers apparent, but unreal, choices to consumers. The illusion of choice
hides the fact that we have few real choices to make as consumers. In
this sense, advertising itself constitutes a lie or at least a false statement
(quite aside from specific truth-in-advertising problems like marbles in
the soup)—behavior that is immoral on its face. For religion to be mean-
ingful, it must ally itself with the forces of morality. For religion to form
alliances with immoral or at best ambiguous forces is to make a mockery
of its reason for being.
The irony is compounded and deepened in the debate over reproductive
rights. Here not only does the advertising obfuscate the reasons for which
people choose abortion; so far, this is merely the business of advertising.
But advertising in the reproductive arena also goes further. A fake choice—
“choosing life” for lack of any other option—is offered in exchange for the
ability to make a real and meaningful choice. Beguiled by the slogans, the
pictures of dewy infants, one renounces the human right to control one’s
personal destiny, to make informed decisions about issues affecting one’s
very life and body.
Juxtaposing “life,” “natural,” and “choice” is a masterstroke. Every adver-
tiser knows that those are among the ten words that provoke the strongest
positive responses. But what is “life” in this context, and who has it? When
two “lives” are at odds, whose is the more “natural”?
Worse still, the sloganeering obscures a delicate psychological problem
and theological problem. That is, do you in fact make a “choice” for life,
natural or not, if you have been driven to that decision by fear or emotional
blackmail? If you “choose life” under these circumstances, can your deci-
sion be considered an act of virtue or piety? And if abortion is made a crime
and is already a sin, can “life” still meaningfully be called a “choice”?
T he rhetoric of reproduction [ 337 ]

338
If one function of a religion is the inculcation in its practitioners of a

meaningful morality, denying them the ability to make decisions that
might be wrong precludes their ever making real moral choices and indeed,
then, becoming moral persons. Consider as an analogue the case of a priest
who is celibate only because he happens to be impotent. Most theological
opinion would argue that in such a case the sacrifice that is celibacy has
no special merit. As with reproduction, the “choice” is a choice only if it is
freely chosen. A responsible anti-choice proponent must raise and debate
these questions. But the slogans prevent them from being raised, much less
discussed rationally.2
SEMANTIC OBFUSCATION
Debaters who believe their positions are right will choose language that
does not intentionally create ambiguities or otherwise mislead listeners.
Such debaters will be confident that (a) their position is sufficiently strong
as to prevail on its own merits, and (b) listeners are intelligent enough to
be trusted to select the winner on rational grounds. The anti-abortionists
show, by their preference for slippery semantics, that they disbelieve one
or both propositions.
Partial equation masquerading as total equivalency is one favorite tech-
nique, especially when buttressed by another propagandist’s trick, presup-
posing the equivalence rather than stating it overtly. An example of both
tactics is the use of words like “child” or “baby” to refer to a first-trimester
fetus or even an embryo (the accurate but rarely used term during the first
eight weeks of development). In anti-choice statements, the equivalence of
“fetus” and “child” is taken for granted, so that something else is asserted,
depending on it: “Baby killing is murder.”
A smooth new use of this unstated equivalence is the latest DeMoss
slogan, “Everyone deserves a chance to be born.” Everyone presumes the
personhood of its referent. The verb it agrees with, deserves, strengthens
that presumption; personhood is normally a precondition for “deserving”
anything. So the slogan both presupposes and asserts the purported sen-
tience of the fetus—a form of argumentation that is both logically invalid
and ethically dishonest.
2. George H. W. Bush added passion but little light to the discussion in his 1989
inaugural address, in which he spoke of “young women … who are about to become
mothers of children they can’t care for and might not love … though we bless them
for choosing life.”
339
Suppose the creators of that slogan were to preface it by saying, “We

believe that fetuses and embryos are equivalent to babies and therefore
possess full human capacities and deserve …” Then at least a reader or
hearer could direct her attention to that statement, considering whether or
to what extent it was true. Similarly, when a proposition uses “baby killing”
as a simple and unanalyzed synonym for “abortion,” violence is done to the
semantic structure of English: words are severed from their usual mean-
ings, even as, in Nineteen eighty-four (by George Orwell, 1949), Newspeak
was engineered to prevent its speakers from criticizing the party line.
The development equation is as complex as it is devious: embryo =
fetus = human = baby = person. Once you accept one equivalence, the oth-
ers slide noiselessly into place. But each deserves examination, as each has
underlying it highly controversial, indeed currently unresolvable medical,
legal, linguistic, and moral ambiguities. To slide over them is not merely
to make use of a slippery slope; it is to push the reader or listener down a
mountain of sheer ice.
A fetus is not a “baby,” although it shares characteristics with a baby. But
the “pro-life” propagandist juxtaposes concepts that are partly similar in
the hope that they will be interpreted as totally identical. The tactic feeds
on the human horror of indeterminacy. We feel comfortable when things
appear to be clearly assignable to categories—A or B, not both. So if you are
not on your toes and I say that in some respects (helplessness, potentiality
for full human behavior, etc.) a fetus is like a baby, you may well believe you
have heard me say that a fetus is a baby.
Only an intellectual scam artist would prey on this human frailty to win
an argument. As much as a first-trimester fetus is like a baby, it is even more
unlike one. In a first-trimester fetus, the central nervous system has yet to
achieve the level of development and integration necessary for sentience or
pain.3 It has the potential for these things, but potential is not actuality, or
I could write this article by merely thinking about it. Yes, if the fetus goes
to term, and the baby survives, eventually the child or the adult will have
the capacity to feel joy and sorrow, to reason, and to experience pleasure
and pain. At that point, we are unequivocally talking about human life, and
many of us would say, equally unequivocally, it is wrong to take that life.
But to say that, because the fetus has this potential, we cannot destroy it,
3. Anti-abortionists make dramatic use of pictures and other representations of late-

term fetuses, which of course are much better candidates for “personhood.” But in the
United States only 9% of abortions are done after the first trimester, and if informa-
tion about abortion and access to it were easy and free of guilt, I suspect that that small
percentage would decrease virtually to zero.

340
is equivalent to arguing that because an infant will in time become able to

speak and make intelligent choices, we should allow babies to vote. Human
life has its stages, and it is illogical and perverse to confuse them.
Indeed, what we mean by person, in the sense of the U.S. Constitution,
or by human being, in our laws that define murder as “the unlawful killing
of a human being with malice aforethought,”4 is something very different
from fetus. Persons are those with legal rights and, in most cases, obliga-
tions. Babies have some legal rights and no obligations; as they age, they
acquire more of the former and several of the latter. Fetuses qua fetuses
have neither; they can only be claimed to have one or the other (usually, in
these arguments, only rights) via semantic slippage. Humans (besides being
biologically members of the species Homo sapiens) are those who think and
feel in ways with which other humans can empathize; they experience and
evoke compassion, a word that etymologically means “feeling-with.” But
one can only “feel with” someone capable of feelings. Something with the
major gaps that a fetus has in its nervous system does not fit that category.
It may be potentially human, but is not yet human, and to destroy it is not
murder. Arguably, to destroy that which cannot sustain life even briefly on
its own is not even “killing.”
Another semantic slippage is employed more covertly, because if it were
made explicit it undoubtedly would stir up dissension. That is the ancient
perception of women as “vessels,” inanimate receptacles for fetuses whose
lives count as their carriers’ do not. We can glimpse that equation in an
argument against reproductive rights attributed by Peggy Noonan to
President Ronald Reagan:
“The argument is over when life begins,” [Reagan] said. “Well, look, if that’s the
argument: If there’s a bag in the gutter and you don’t know if what’s in it is alive,
you don’t kick it, do you?” (Noonan 1990:159–60)
In other words, you forbear to hurt a woman not because it might injure
her, as a fellow human being or fellow creature, but because she might be
carrying a life you wouldn’t want to hurt.
A third problematic equivalence, also typically implicit, is the equation
of abortion with slavery. Everyone agrees that slavery is evil. Most also
would agree that extreme measures (including civil war) are justified in
order to extirpate it. The rhetoric of the anti-slavery movement was full of
passion, passion justified by the cause it supported. So the anti-abortion/
4. 18 U.S.C. § 1111 (1940).
341
anti-slavery equation works in two ways. It identifies morally ambiguous

opposition to legal abortion with morally righteous abolitionism, and it
thereby justifies the use of heated rhetoric. Hence Operation Rescue uses
as its theme song “We shall overcome” and frequently refers to its cause as
“abolition.”
THE HYPOCRISIES OF “LIFE”
A serious moral cause should use its terms literally. Suppose we take “pro-
life” literally, that is, assume that its proponents are unequivocal support-
ers of “life” in all its manifestations. (Let us grant them a little latitude,
interpreting “life” as “human life” in their extended sense, so as to exclude
animal rights.) Then it would stand to reason that any person or organiza-
tion claiming a “pro-life” position would have to take a public, explicit, and
unambiguous stand that was:
1. Anti-war because war kills. One hundred thousand Iraqi children cannot
be less valuable than one unborn American embryo.*
2. Opposed to capital punishment. It is argued that execution is the taking of
“non-innocent” human life. But who is to be the judge of that? Moreover,
the simultaneous belief in capital punishment and intolerance of abor-
tion leads to a serious moral paradox. Murder is the unlawful killing of
a human being. Capital murder, in those states that employ that statu-
tory term, is premeditated murder committed under circumstances so
heinous as to justify execution. One such circumstance under these laws
is “killing for financial gain,” which subjects both parties in the hiring
agreement to capital punishment. If a fetus is defined as a human being,
then abortion becomes premeditated murder and, further, capital mur-
der, as the doctor usually is paid by the woman. Then, once it is legally
established that an abortion has taken place, two human deaths beyond
that of the fetus are mandated. Is this “pro-life”? Prosecutors might
agree tacitly that the death penalty never would be sought in such cases,
but that would be piling hypocrisy on hypocrisy.
3. Absolutist. All abortions would be forbidden. While perhaps abortion
to save the mother’s life would be debatable (but by no means a guar-
anteed right), the termination of pregnancies incurred through rape or
incest would be precluded; those fetuses are as human and “innocent” as
*
This article was written in the wake of the 1991 Persian Gulf War, in which a U.S.-led
coalition invaded the nation of Iraq.—Ed.

342
any others. But the majority of pro-lifers would permit abortions under
these circumstances, leading to the suspicion that they see pregnancy
as punishing a woman for having “fun”; as long as the sex demonstrably
wasn’t fun, abortion is countenanced. (The anti-abortionists also realize,
to be sure, that the absolutist position would turn many in the middle
against their position. But if this motivates their position, they are trad-
ing moral consistency for a few votes—again, not the higher choice.)
4. In favor of contraception and sex education. If the aim of the movement is
to prevent the destruction of human life, then high on its agenda should
be avoiding the conception of unwanted human life. Many anti-abortion
groups claim to be “neutral” on these issues. But others—notably the
Roman Catholic bishops and Randall Terry (founder of the anti-abortion
group Operation Rescue)—actively oppose contraception and compre-
hensive sex education. Moreover, even neutrality is inappropriate and
hypocritical here.
. Opposed to violence, including clinic-blockade tactics—such as obstruc-
5
tion, jeering, and shoving—that incite violence, as violence tends to
lead to injury and death.
Further, a position that was strongly pro-child and based on the pas-
sionate love for children that anti-abortionists claim as their motivation
would have to take certain explicit positions, to avoid a charge of hypocrisy
or at least confusion:
1. Support for prenatal, neonatal, and early-childhood care, so that medical

and other services are made available to all in need.
2. Commitment to welfare for mothers in need; no child should be denied
what is essential to “quality” human life.
3. Support for education, because education makes human life more fully
human and allows all persons to avail themselves of opportunity.
As many anti-choice individuals and groups at best ignore and at worst

oppose these positions, their advocacy of “life” and love for “children” must
be considered hypocritical.
THE FEAR BEHIND THE SLOGANS
How can we account for the undeniably genuine passion of right-to-life

rhetoric and action? In psychoanalytic terms, I would suggest that there is
a “displacement” of emotion from an unacceptable object, which remains
343
unconscious, to one that is false but can be acknowledged publicly and pri-
vately.5 I assume that the majority of “pro-lifers” are sincere and intelligent
people who are, nevertheless, driven by forces they cannot control or admit
to consciousness. Then it becomes clear why their dubious rhetoric has the
power to sway many people and why, when pro-choice rhetoric rationally
addresses the ostensible arguments of the anti-abortionists, it has little
persuasive force.
The underlying emotion of “pro-life” people is not love but fear, which is
at least as strong a motivator but is less comfortably admitted. It is a fear
of change in a rapidly changing world, of loss of control, of no longer under-
standing what is expected, of no longer being able to count on actions hav-
ing predictable consequences. The fear arouses a desire for “business as
usual” or, even better, a return to a golden age, when men were men and
women were women, with sharply polarized roles: men held power and
women didn’t have it or desire it. There was certainty: the authorities of
church and state set forth unambiguous standards for behavior, with seri-
ous punishments for falls from grace. Children respected their parents; vir-
tue was honored; and individuals had relatively few choices, making life (if
less pleasant, in some respects, than we are used to) quite simple.
Many innovations over the last century or so (especially in the United
States) have rendered this myth inoperative. Perhaps the most significant
is the gradual acquisition by women of control over their bodies, beginning
with laws against wife beating. Women’s control over their bodies came
much closer to reality with the mid-20th-century development of reliable
contraceptives, culminating in the Pill. Much of the social ferment of the
1960s and the ensuing social changes stem from the availability of effective
contraception, which freed women not only sexually but also profession-
ally and politically. The Supreme Court’s decision in Griswold v. Connecticut6
ensured that contraception was a legal right: the genie was out of the bot-
tle. But it took many years for these changes to have their full effects on
American life.
In fact, the ferment peaked just in time to coincide with Roe v. Wade,
which thus became the scapegoat for all the distress the earlier changes
evoked in those unready for change. The Pill was a reality, and Griswold was
firmly enshrined in law and enjoying too high a degree of public approval to
be criticized. But Roe was different; unlike Griswold, it could be construed
5. Displacement is defined as “the process by which the individual shifts interest
from one object or activity to another in such a way that the latter becomes an equiva-
lent or substitute for the other” (Rycroft 1973:35).
6. Griswold v. Connecticut, 1965 U.S. LEXIS 2282 (1965).

344
as morally ambiguous. Two parties arguably were involved in any deci-

sion, the fetus as well as its carrier. So opposition to all that is new and
strange and uncontrollable and unpredictable has, since 1973, used Roe as
a scapegoat.
From this perspective, male opposition to abortion makes perfect sense.
Men lose (at least in the short run) from women’s full equality. But how are
we to account for the presence of women, in greater numbers than men, on
the right-to-life side? If not love for fetuses, what impels them? They, too,
are displacing their true passion. But for them, the word that evokes fear is,
unhappily, the standard of the other side, choice itself.
Americans like to think that we love choice. We like having choices (at
least those that don’t matter), and we like having the option of not choos-
ing in matters that make a difference. That is particularly true for people
who have been brought up lacking self-esteem and self-confidence. Such
people fear that, if they had real choices to make, they would make the
wrong ones, incurring punishment, obloquy, and stigmatization. Choice is
anathema to them.
Women have been brought up in a society which teaches them from
infancy that they are less worthy and less competent than males. Therefore
it is not surprising if a great many women would rather have as few serious
choices to make as possible. It is for this reason that most women did not
advocate female suffrage before they got it; they didn’t want to make real
choices. But guaranteed access to full reproductive rights is far more liber-
ating than the ballot, and far more essential to true autonomy. If women
can fully determine their reproductive fate, then every woman is answer-
able for how she leads her life and for the choices she makes. No wonder
many women are eager to recriminalize abortion. It is their only hope for
comfort, though if they should achieve it, that hope soon would prove vain.
If we do not accept this interpretation, the connection between motive
and passion cannot be made, and the “pro-life” arguments become irratio-
nal, particularly when they support only fetal life. If we do accept it, they
make perfect sense—as displacement or as hypocrisy unworthy of a posi-
tion claiming for itself the moral high road.
On the other hand, the pro-choice arguments seem consistent and
direct. The lives of sentient women outweigh the “life” of a nonsentient,
incomplete embryo or fetus. The needs of families, including children, for
basic nurturance—physical, intellectual, and emotional—outweigh any
demand that each conceptus be gestated because of the potential in its
DNA. The equation of fetus, baby, and person, tendentious at best, must
take second place to the equation between woman and person (as opposed
to vessel) or child and person. And while everyone agrees that abortion is
345
an unsatisfactory method of birth control, it must remain available until

all women have better means, entailing the end of the double standard
and sexual shame, as well as the availability of thorough and correct sex
education.
THE CHOICES FOR “CHOICE”
What is the “pro-choice” side to do, rhetorically, at this pivotal moment?

The first question is the name of the movement itself. The term “pro-
choice” is problematic if many to whom it is addressed find it terrifying to
be given real choices. Besides, choice is weaker than life. If the robber says,
“Your money or your life,” you hand it over; but choice does not make the
adrenalin flow.
Unfortunately, few other words have the semantic and phonetic power
of life. Either they’re less imperative, or they’re just longer and clumsier.
Reproductive rights is a noble concept—but not one for persuasive chanting.
Similarly, any term that favorably alludes to abortion (such as pro-abortion
or abortion rights) is too dangerous, given the negative connotations of
the word.
Attorney Elizabeth Bader has made an intriguing suggestion to reconsti-
tute the debate by putting the negatives where they belong. The erstwhile
“pro-life” forces would be referred to as those of “forced motherhood”; of
“choice,” “voluntary motherhood.” I think forced motherhood provides a nice
hobnailed-boot–cum–Nineteen eighty-four image. We have lived with baby
killer. Let them deal with forced motherhood.
A MODEST PROPOSAL
Beyond the movement’s name, reproductive-rights advocates must do

something about our rhetorical strategy as a whole. We have been charged
often enough with expediency—let’s consider being expedient. I hear pro-
tests: it’s not honorable, it’s disrespectful of our audience’s intellect. Yes,
but if we continue to be respectful and legitimate, we soon may be scouring
the closet for metal hangers.
There is plenty of passionate rhetoric for the grabbing. If their side airs
commercials with cuddly babies, how about a few with terrified teenagers—
children like your own—encountering filthy back-alley abortionists? Or
ads using narratives by women who underwent illegal abortions, with hor-
rendous results? Not sufficiently dramatic? How about hooded Inquisitors

346
stretching women on the rack: “Who did it? Where do we find him?” Or

Gestapo types breaking into a sterile operating room, destroying science?
We have ways, if we choose to use them. If anyone can argue that the
ends justify such rhetorical means, it is the forces of “expediency.” And
because the other side did it first, they can’t object. We at least might talk
about how women are being lied to, and suggest why; bringing the true
fears into the open may demystify them.
We certainly could use a few good slogans: “Life is love” or “Choose to
live.” Planned Parenthood couples the refrain “Don’t wait until women are
dying again” with such lines as “Will the Supreme Court’s next decision be
carved in stone?”—a reference to the accompanying picture of a woman’s
tombstone. We could go further:
• Women are not slaves (with the Battle Hymn of the Republic in the
background)
• Remember the Middle Ages? (with a picture of a woman being burned at
the stake)
• Forced motherhood—the “choice” of the past
A REAL CHOICE
Those possibilities are entertaining, and I would not dismiss them out-
right. But before we rush to hire our own propaganda thugs, let us try once
more to do the right thing. We must appeal to Americans, male and female,
as rational and caring people, people who will take the right public-policy
position if the false attractions of the wrong position are made manifest.
Our rhetoric should compel recognition of the fact that life is full of gray
areas, which make personal decision making (rather than government
dictates) all the more important. Choices are hard precisely because there
are pros and cons, ifs, ands, and buts, and modern life brings with it ever-
harder choices. That is why flashy rhetoric is dangerous. It makes it impos-
sible to think rationally through the options. We must expose the cynicism
of the other side, always emphasizing that women are smart enough and
brave enough and moral enough to choose.
We must acknowledge that there is a vast gray territory: What is life,
what is human life, and when does it start? What is a person, a human being,
and what are a person’s rights, duties, and entitlements? The mindless
intoning of slogans temporarily makes those difficult questions seem easy,
but in the long run, the questions must be confronted directly, intelligently,
347
and fearlessly—not through rhetorical razzle-dazzle. If the war is not to be

lost, it is time to act decisively and nondefensively.
REFERENCES
Noonan, Peggy. 1990. What I saw at the revolution. New York: Ivy Books.

Orwell, George. 1949. Nineteen eighty-four: A novel. New York: Harcourt, Brace
and World.
Rycroft, Charles. 1973. A critical dictionary of psychoanalysis. Totowa,
NJ: Littlefield, Adams.

348
349
Introduction to “True confessions?

Pragmatic competence
and criminal confession”
BY LINDA COLEMAN
W hen I first read this piece, I was transported back to the classroom,
listening to one of Robin Tolmach Lakoff’s extraordinary lectures.
It’s all here: the clear prose style, the memorable turn of phrase, the
straightforward presentation of stance and theory, the careful laying out of
analytical tools, the precisely mapped-out line of reasoning, supported by
just the right examples, and, throughout, the perfectly timed wit. Also here
is the deceptive simplicity of presentation. Like many other gifted lectur-
ers, Lakoff moves with such ease from point to point that listening to her
go over some complex analysis can leave you feeling that, really, the whole
thing is perfectly obvious, and perhaps even wondering, with a touch of
hubris, why you hadn’t thought of it yourself.
Such clarity has its pitfalls for the unwary. When I started taking courses
from Professor Lakoff, I was a sophomore, and it was in that first year or
so of the 1970s that still counted as part of “the sixties.” We returned to
the earth and boasted of our peasant roots, whether we actually had them
or not. Handicrafts, along with granny glasses and prairie skirts, became
popular. In fact, knitting or crocheting in class became something of a fad,
and in many classes you could find a couple of students, usually but not
always women, working on a sweater or a cap or pieces of an afghan. I don’t
recall that anyone noted at the time that it was, at the very least, extremely
350
rude to sit in class and look otherwise occupied. The knitter wouldn’t admit
to being distracted and probably didn’t think that others might be.
Rudeness aside, an accomplished knitter could probably knit easily
enough while absorbing the lecture content. I, however, was not an accom-
plished knitter. Nonetheless, in the second or third class I took with Lakoff,
I started work on something or other and, recalling how very clear her lec-
tures always were, foolishly imagined that I would be able to knit and keep
up with the analysis. It was an act of rudeness for which, Robin, I belatedly
but no less shamefacedly apologize.
It was also a serious mistake. I learned quickly—not quickly enough—
that I had to pay attention, and that it wasn’t going to work if I was going
to be counting stitches at crucial points in a carefully constructed argu-
ment. After a few weeks of frantically increased study time to make up for
my lack of concentration in class, I concluded that the knitting would be
best left at home. Besides, I realized, it was silly to distract myself from the
class I enjoyed most.
There is my “true confession.” As I look back, I wonder: What could I have

been thinking?
And therein lie some of the issues Lakoff is dealing with in “True
confessions,” for the article is about a man whose language behavior
suggests that he quite literally does not know what he could have been
thinking. His situation brings into focus two closely related issues that
have often been the subject of Lakoff’s work. The first has to do with
identity—specifically, the ways in which language reflects our beliefs
about ourselves. It is a special part of Lakoff’s genius to notice the details
of language, the oddities—and the banalities—of phrasing that can so
easily slip by, and to peer beneath them to see what might actually be
happening in the speaker’s mind.
The second issue is the relationship of individuals to the speech com-
munities in which they find themselves, and specifically their need to
adjust to the assumptions and expectations of their audiences. In the case
cited in this article, the pseudonymous “Virgil Reilly” is apparently at least
reasonably competent in some interactions: there is indication that he
understands many of the principles of ordinary conversation and that he
expects the aim of each speaker to be an essentially Gricean cooperation
and helpfulness (although the match between the two notions of relevance
held by Reilly and by his audience is far from perfect). In the interactions
presented here, however, he finds himself operating under the restrictive,
and, to him, apparently unknown, rules of the speech community formed
around the activities of law enforcement.
351
There is, of course, much that could be said, if space permitted, about
the peculiarities of a “community” in which the competent members are
law-enforcement personnel and experienced criminals, while at least a
significant number of those who become involved with this community
are not competent in its interactional patterns. This is one of the issues
explored in Tom Wolfe’s Bonfire of the vanities (1987), which follows the
story of Sherman McCoy, a successful investment banker who gets in trou-
ble with the law, unwillingly develops expertise in the roles and behaviors
of the world of the accused suspect, and, somewhat to his own surprise,
finds himself taking pride in his new competency and, indeed, in the new
identity it gives him.
This development of identities, how we speak of ourselves, how we cre-
ate and sustain our public personae, is not a simple issue. It is delicate
enough when we are interacting with our peers on equal grounds. It is
more complex when our peers are judging us. And it becomes overwhelm-
ingly problematic when we are presenting ourselves to people who perceive
themselves entirely as judges rather than peers—that is, when our disposal
is in the hands of those who believe themselves to share very little common
ground with us. Such is Virgil Reilly’s case.
For purposes of comparison, we might look at the case of Karla Faye
Tucker, a confessed ax murderer who became a born-again Christian in
prison (Pressley 1998). Tucker was ultimately executed on February 4,
1998, but her case attracted widespread attention, in part because many
proponents of the death penalty, including many evangelical Christians,
supported reduction of her sentence to life imprisonment. Tucker’s appeal
hinged on issues of identity: when she committed the crime, she was, in
the words of one news account, “drug-addled”; now, it was argued, no lon-
ger using drugs and having experienced a religious conversion, she was
in essence a different person. More than the simple facts of the case, the
appeal was based strongly on the construction of the audience as peo-
ple who believed in the power of religious conversion. This may explain
why claims of personal change that moved members of the evangelical
Christian community may have been less moving to the Texas State Board
of Pardons and Paroles, which has doubtless heard more than a few claims
that a religious conversion has given an inmate what amounts to a new life
and indeed a new identity.
Nonetheless, Tucker did gain a hearing from some normally reluctant
members of the broader society. And, while some of that may be attributed
to her status as an attractive, articulate, and young white woman in jeop-
ardy, still, that achievement was to a significant degree the result of her abil-
ity to divorce herself rhetorically from her earlier “murderer” identity by
TRUE CONFESSIONS? [ 351 ]

352
demonstrating that she identified with the values and beliefs of the larger
society and stood with them in condemnation of the “old” Karla Faye Tucker.
The problem is, of course, that it is difficult to find any sort of clear divid-
ing line between Reilly’s confused and possibly pathological disconnection
from his own actions and Tucker’s carefully constructed (and religiously
sanctioned) rhetorical self-distancing. Nor is there a clear boundary between
those and the kind of casual and commonplace self-distancing of normal
interaction. Imagine a situation in which you and I meet at a conference (let
us say) and enter into a conversation, part of which involves trading stories
about our favorite professors, about stupid things we did when young, or
about how behavior changes with time—the “true confession” with which
I began this introduction could fit into any of those categories. If I told you
that story in that kind of casual context, you would be unsurprised, in all
likelihood, by my self-distancing “What could I have been thinking?”
But that is another element of Lakoff’s work, for her longstanding inter-
est in special types of language use (language associated with the law, with
advertising, with politics) works in both directions. That is, while showing
how linguistics can inform our understanding of legal (or advertising or
political) issues, she demonstrates that the special demands and circum-
stances of legal (or advertising or political) contexts can serve as a labora-
tory in which to test our assumptions about day-to-day communication.
As a result, her work has located itself on the boundaries linguistics
shares with psychology, in its concern with the representation of identity;
with anthropology, in its work on the construction of community; and with
rhetoric, in its efforts to account for decisions made by individual speakers
selecting from the available choices in order to create a specific effect in the
audience. In short, Lakoff’s work was interdisciplinary before interdiscipli-
narity became quite as popular as it is now, and the students she trained
became, as a result, well equipped to handle interdisciplinary issues. It is in
fact in these interdisciplinary connections that we are most likely to find
answers to the kinds of difficult real-world questions that Lakoff’s work
consistently forces us to examine.
REFERENCES
Pressley, Sue Anne. 1998. For first time since Civil War, Texas executes a woman.
Washington Post, 4 February 1998, A1, A10.
Wolfe, Tom. 1987. Bonfire of the vanities. New York: Farrar, Straus and Giroux.
353
CHAPTER 15
True confessions? Pragmatic

competence and criminal
confession (1996)
1. INTRODUCTION
There are two especially difficult problems for pragmatics and sociolin-
guistics. One is the connection between purely linguistic form (phonology,
syntax, morphology) and discourse function. Eventually we must, however
reluctantly, consider to what degree and in what ways real-world situations
and communicative needs of speakers govern syntactic form; and the sense
in which syntax, semantics, and pragmatics are truly, bidirectionally, inter-
dependent. The second is the definition of “pragmatic” or “communicative”
competence: What does the normally competent speaker know?
The case I will discuss here raises another issue of interest to linguistics
and other social sciences. Over the last fifteen years or so, there has been
much discussion in several fields concerning the nature and/or reality of
the “self.” Some of the data we will be examining provide more evidence of
the tenuousness and fuzziness of that concept.
The examples I will use are drawn from a pair of criminal confessions in
a death-penalty case currently (June 1993) under appellate review. I was
This paper originally appeared in Social Interaction, Social Context, and Language: Essays
in Honor of Susan Ervin-Tripp, edited by Dan Isaac Slobin, Julie Gerhardt, Amy Kyratzis,
and Jiansheng Guo, 481– 93. Mahwah, NJ: Lawrence Erlbaum Associates, 1996.
Reprinted here with permission of the publisher.
354
asked as an expert by the defense to review the confessions to assess the

linguistic (or communicative) capacities of their client. Since I was asked by
counsel to use a pseudonym in referring to the defendant, I shall refer to
him as Virgil Reilly (or VR).
2. FACTS OF THE CASE
The facts of the case are briefly these. When the crimes were committed,
in November 1984, Virgil Reilly was a black male of twenty-five, who had
never been in any previous legal trouble.
VR’s IQ falls between the mid-sixties and mid-seventies—borderline
or “dull normal.” Psychological testing reveals an array of impairments—
particularly relevant to our concerns are verbal deficits, including “severe
impairment in attention, memory for verbally presented material, verbal
fluency, cognitive flexibility, and adaptability to novel cognitive tasks.”1
He is a native speaker of English, and has attended high school. He
reports, prior to receiving a Miranda warning during one of his confes-
sions, that while he “know[s]‌how to read,” he doesn’t read “that good you
know” (SB 2:5,7).2
VR’s life prior to the crimes to which he is confessing was typically
dreary for a defendant convicted of a capital crime. His mother used
alcohol rather heavily during her pregnancy. She was twenty when VR
was born, three months prematurely, after a difficult pregnancy, the
first of eight children born over a twelve-year period. VR was regularly
abused by his father and other relatives (physically and sexually) and
neglected by his mother. He also displayed, from early childhood, a vari-
ety of physical and psychological problems, created or exacerbated by
the abuse.
In October of 1984 he had begun to use large quantities of over-the-
counter and prescription analgesic drugs (especially Tylenol with codeine)
to self-medicate a variety of complaints, including but not restricted to per-
sistent headaches and pain from dental procedures. He was also drinking
large quantities of wine.
1. Report of UCLA Neuropsychiatric Assessment Laboratory, September 26, 1985.

2. References to the two confessions will be indicated throughout the text as fol-
lows: SB = San Bernardino confession, November 24, 1984; R = Riverside confession,
November 22, 1984. Numbers preceding a colon refer to pages of each confession;
numbers following the colon refer to conversational turns on each page. Thus, the
foregoing quotations are taken from page 2 of the San Bernardino confession and rep-
resent the fifth and seventh turns on that page.
355
On the night of November 16 in Riverside, he abducted two children,

raping the girl and killing the boy. The next night, in San Bernardino, he
abducted and raped two young women. He was arrested by the Riverside
police on November 22 and was interrogated later that night; he was inter-
rogated by the San Bernardino police on November 27. In both interroga-
tions he ultimately made full confession.
There is little doubt that VR is guilty of the crimes for which he was given
the death penalty: physical evidence incontrovertibly connects him to both
crimes. So the problem with these confessions is not that an innocent man
may die for crimes he did not commit. Rather, the ethical question that
my discussion poses concerns a possible conflict in our present culture
between our desire to see ourselves as a humane people who require that
criminal confessions be extracted humanely—that is, fully in compliance
with the intentions and instructions of the Supreme Court in Miranda,3
that is, with the subject’s informed consent—and our equally pressing and
valid need to feel secure from violence in our daily lives, and therefore the
government’s obligation to pursue and convict malefactors with dispatch
(or risk a general feeling of the erosion of the social contract, with conse-
quent chaos). What constitutes informed consent to a criminal confession?
Does VR’s situation meet that standard?
3. COMMUNICATIVE COMPETENCE IN THIS CASE
As I noted, I was asked by the defense to evaluate the defendant’s prag-

matic competence to make a valid (under Miranda) confession. Did he
understand what a confession is? What confessing meant for him?
We can define full pragmatic competence as including (but not limited to)
a knowledge of the felicity, or preparatory and essential, conditions neces-
sary for the appropriate performance of any speech act in which the person
being evaluated is involved (in any capacity), an appreciation of conversa-
tional logic, and an understanding of the rules and forms of dyadic conver-
sation.4 It is essential that a suspect understand not only the literal wording
3. Miranda v. Arizona, 384 U.S. 436 (1966). At 479, Miranda states that the suspect
“must be warned prior to any questioning that he has the right to remain silent, that
anything he says can be used against him in a court of law, that he has the right to the
presence of an attorney, and that if he cannot afford an attorney one will be appointed
for him prior to any questioning if he so desires.” Receiving these warnings will enable
him to “knowingly and intelligently waive these rights.”
4. For the definitions and functions of these terms, cf. Austin 1962 and Searle 1969
and 1979; for conversational logic, Grice 1975, and for conversational structure, Sacks,
Schegloff & Jefferson 1974.
T rue confessions ? [ 355 ]

356
of a Miranda warning but its discourse function as marking a boundary

between essentially harmless conversation and discourse in which serious
self-damage can be done by helping the interrogator. The Supreme Court
has been quite clear on this: several opinions since the original 1966 ruling
have required that a suspect demonstrate explicitly a fluency in English
equal to eighth-grade capacity (necessary to understand words like attor-
ney). They have acknowledged that this capacity may be fatally compro-
mised either by non-native speaker status or by demonstrated subnormal
mental capacity. In either of these cases, a confession would be inadmis-
sible at trial, and if such a confession was used by the prosecution as a cru-
cial part of its case, a guilty verdict (and/or a death sentence) would likely
be overturned.
To demonstrate full pragmatic, or communicative, competence in this
discourse context, VR would have to possess several forms of commu-
nicative knowledge, which could be assessed by observing, from tapes
or transcripts of his verbal productions, his conversational behavior. The
evaluator would look for (1) an ability to respond to interrogations, direc-
tives, and other speech acts in a manner consistent with a normally fluent
speaker’s understanding of speech act theory and conversational logic;
(2) an understanding of his role in the discourse more generally—what
it means to be under criminal interrogation; and (3) evidence of com-
petence, active and passive, in the rules of conversational interchange,
i.e., indications that his turns represented preferred, or at least dispre-
ferred but permissible, seconds. Additionally, as essential underpinnings
of pragmatic competence, VR would have to demonstrate other forms
of linguistic competence: knowledge of relevant English vocabulary, the
ability to construct and parse sentences of the appropriate complexity,
knowledge of the relationship between linguistic forms and their real-
world referents. It is my belief that VR demonstrated anomalies or fail-
ures of all of these capacities to some degree, though not so utterly as to
show unequivocal absence of the communicative competence to make a
meaningful and consensual confession. Rather, these confessions repre-
sent an intriguing and ultimately unresolvable gray area, showing that
language and communicative capacity are infinitely complex and ambigu-
ous. We must confront these uncertainties as such: we cannot dismiss
examples like these as uninteresting or meaningless merely because we
cannot be sure what they mean. Here as often in linguistics, it is the shad-
owy cases, the borderline examples, that have the most to teach us about
our subject.
357
4. CONFESSION: WHAT THE CONFESSOR NEEDS TO KNOW
Several specific sorts of knowledge are needed to make a fully meaningful,

informed, and consensual criminal confession, including an awareness of
the following properties of the speech act of confession:
1. Because confession is a truth-seeking discourse genre5 (different, there-

fore, from an informal conversation), contributions must be directed
toward a mutually agreed-upon point: Did you or didn’t you, and if you
did, how did you? Expatiation and irrelevancy are marked as (at best)
requiring special interpretation (beyond the normal bounds of conver-
sational implicature) in a truth-seeking discourse genre. That is, devia-
tions of these kinds tend to suggest to hearers that the speaker either
has something to hide or is seriously communicatively incompetent.
2. The participants in a confessional speech act are unequal in power, before,
during, and after the discourse. But the making of a confession increases
a suspect’s vulnerability still further. The interrogator has significantly
more power through institutional status, as well as the ability to use the
confession in ways that will hurt its producer in the future. Additionally,
interrogators often get extra power in more subtle ways: the interroga-
tion is done on their turf, at their chosen time (often late at night, when
the suspect has been isolated in a holding cell for many hours); they ask
the questions, and are free to withhold benefits of various sorts (water,
coffee, food, sleep, access to family members) until satisfied. They have
training and formal experience in this discourse genre; they make judg-
ments on suspects’ productions. They are often of a higher social class,
better educated, more likely to be speakers of the standard dialect.
3. The confessors (a term I will use here to mean “speakers of confessions”)
must be acknowledging behavior committed previously by themselves
(or at least someone closely connected to them). A confession must
therefore acknowledge an identity between speaker and performer—
explicitly or otherwise.
4. In order for the confession to be informedly consensual, the agreement
must be fully responsive. That is, the suspect must explicitly indicate
understanding of the circumstances and willingness to participate. Mere
indication of compliance is insufficient.
5. For discussion of this term, as well as the notion of truth-seeking versus other
discourse genres, cf. Lakoff 1990.

358
5. To demonstrate communicative competence in this context, a speaker

must show an understanding of conversational structure: specifically,
turn-taking and preference organization.
6. Finally, to confess appropriately, the confessor must agree with the hearer
(and the culture at large) that the behavior in question was bad. This does
not necessarily imply a sense of remorse: the confession need not be, in
this sense, “sincere,” any more than an apology. But if we were to find that
a subject was totally, and genuinely, unaware that the act that he acknowl-
edged committing was one that was considered bad by his society, we would
have to set the confession aside; indeed, under current definitions of crimi-
nal insanity, if the suspect could be shown not to recognize the badness of
the act, he would have to be declared “not guilty by reason of insanity.”
The texts suggest failures by VR in the first five categories, to be illustrated

in the remaining sections of this paper.
5. THE TRUTH-S EEKING NATURE OF CRIMINAL CONFESSIONS
VR repeatedly flouts or violates Gricean maxims, in a discourse context in

which strict adherence should be the norm. Actually, the latter statement
should be qualified. It is not infrequent (indeed it is expected) that a suspect
violates certain of the maxims (especially Quality), especially early on in the
process. But conversational implicature makes sense of these floutings, with
the understanding that they are to the speaker’s benefit (that is, politeness,
or at least defensiveness, here as elsewhere, may supersede strict informa-
tiveness). But what is unusual about VR’s performance is that the particular
ways in which he flouts the Maxims cannot be seen as working, or even
intended to work, in his interests: they are violations largely of Relevance,
and neither supply requested information nor attempt to withhold it.
In example (1), the interrogator (RM) is trying to establish VR’s where-
abouts on the morning after one of the crimes:
(1) VR: What time I get up Saturday mornin’? Uh Saturday

mornin’ ‘round about uh nine or ten o’clock.
RM: Nine or ten o’clock. What time did you go to LA?
VR: Time I went to LA? Close to twelve. I’d say about quarter
to twelve, usually.
RM: Did you get up in between there and take off and get
something or do something?
VR: No.
359
RM: Before you went to LA?

VR: Before I went to LA?
RM: Uh huh.
VR: Well the only time I went and took off is when I woke
up uh with me and my wife both woke up and uh we sat
down in the livin’ room. I had the TV on.
RM: What time was that?
VR: It was ‘round about uh ten o’clock. We was watchin’ the,
uh, what’s the show called? Uh, she was watching uh,
what’s that show? Uh not uh I watched the Gong Show,
that come on around eleven, and movie usually come
on after that. The uh thing where they mash down, that
game goes around uh you get so many points, you get the
whammy, uh you get the whammy. You gets money. If you
don’t get the money, you get the whammy. That, the show
come or it and it usually stays on around a half hour and
the Gong Show come on after that. I have a swelling here,
in the top on down behind my ear.
(SB 28:11-29:1)
And another in the same vein, from the Riverside confession:
(2) Cornejo [an interrogator]: Okay. Tell us what you have on your

mind [Virgil].
VR: Okay. Uh, going to work on the evening, I guess that last
night… . [crying] … I woke up this morning [unintelligible
for several seconds] guarded my hand… . I didn’t know where
it come, you know, I had it out in the truck or in the house.
Half the time I don’t know where, how I got the gun. And
I get these, I don’t know if it’s tumor, or pains in my head.
I tell my wife everytime it happens and she always say,
“Check it out, dear.” “Check it out.” So I… . they haven’t
tell me anything about … pain still there. Before I came to
work, when I was down at the place, the store, okay, I fell
I had an attack or something, and when I left the coffee… .
you know I had coffee, I was shaking and was spilling half
the coffee and all I remember was getting in the truck.
I was going back to the place and I had this gun in my hand.
I don’t know if I walked in the store with it or what. I don’t
know. And when … next thing I know I was back at the site,
and I was hearing shots … some banging noise. And I don’t

360
know if I was scared, or what. I was just as calm, you know,

like I was earlier. And different people come to me, you
know, and I was there. And they were [unintelligible] wasn’t
there. You know, I don’t know if I killed anybody or not.
I never killed anybody. You understand?
(R 26-7:14-15)
And shortly later:
(3) Bowen [another interrogator]: = Where did you shoot him [Virgil]?

VR: I don’t know. All, I just opened fire and shot. I wasn’t
even aiming, you know, even when I’m practicing or
something I usually take my time, aim, you know, even
when I’m with my wife, I tried to show her how to use
the gun, how to prop it up, you know. But when I get
in these, these stages or something, the change, and
nobody doesn’t listen to me or what’s happening to me,
and sometime, I be mad myself. I be want to do things
to myself and things I don’t even know that I’m doing it.
You understand? Like my wife one day, I had a beard and
I was just went in the kitchen and shaved it all off. And
she says, “Whys you do that?” I says, “Just something
to do.” Shave my mustache off. I usually don’t shave my
mustache off.
(R 28:2-3)
It is essential here to draw a distinction between utterances that, while

not strictly relevant, would be perfectly appropriate under the more
relaxed expectations of ordinary conversation (as is probably true of
all the examples above), and utterances occurring in the strict truth-
telling frame of a criminal confession, in which each of the examples
above is aberrant to some degree. In each, unnecessary and irrelevant
information is supplied at considerable length (violating the maxims of
Quantity, Relevance, and Manner). VR merely seems not to know how
to “keep to the point,” or perhaps even does not understand what “the
point” is, what his interlocutors want to know, what the discourse as
a whole is about. It might be argued that VR is merely trying to throw
his questioners off by irrelevance and longwindedness. But many of his
expatiations occur after he has essentially made full confession: he has
nothing to gain from temporizing at that point. And it is questionable
whether someone with an IQ barely above the level of retardation, and
361
significant verbal and cognitive deficits, could consistently play a delib-

erate game of this kind—or even understand the advantage of playing
such a game. These examples supply evidence of other problems with
this discourse as a true “confession.” For instance, VR’s failure to adhere
to the Gricean Maxims suggests a failure to perceive the adversarial
nature of the conversation.
6. THE NONEGALITARIAN NATURE OF

THE CONFESSIONAL DISCOURSE
For a confession to have been made with informed consent, the suspect
must realize that the confessional frame itself places the confessor in jeop-
ardy, and that that risk is the interrogator’s desired outcome: it is a win-
lose situation. Since as a criminal suspect VR enters the discourse under
a disadvantage, full understanding would make clear to him the neces-
sity of keeping up the boundaries, maintaining distance and formality at
all points, giving no more information than is essential, and not expect-
ing the interlocutors to do anything to his advantage. This is the reason
why Miranda requires interrogators to offer suspects explicitly the right
to have an attorney present, and also why interrogators do their best to
discourage suspects from invoking those rights6: the presence of an attor-
ney would substantially equalize the participants. In VR’s case, the waiver
was almost too easily accomplished. In the SB confession, for example, the
Miranda statement is read rapidly and without inflection, in long phrases.
So while VR might have understood all or most of the words used, in iso-
lation, it is very possible that the warning as uttered, presented to him
in a state of agitation and exhaustion, might not have been fully compre-
hended. He answers, “Uh huh,” but it’s not clear whether this response
signifies understanding or merely compliance: “Just do what you like, fine
with me.” In the Riverside confession, after the detective has read VR his
rights, he asks: “Having these rights in mind, [Virgil], do you wish to talk
to us now?” (R 2:10). There follows a four-second pause on the tape, an
indication that VR fails to understand fully the requirement that, at this
explicit transition-relevance place, he take a turn. As the silence length-
ens, the second interrogator takes over. There is a marked stylistic shift
between Cornejo, who speaks first, and Ferguson: the former is formal
and distant, the latter spontaneous-sounding and colloquial. VR responds
6. Ainsworth (1993) discusses some of these methods.

362
inappropriately to Ferguson quite possibly because of his apparent friend-

liness, especially after Cornejo’s distance: VR may confuse the appearance
of friendliness with true “friendship,” mutuality of interests. To make
the case more strongly, Ferguson makes VR’s waiver something Ferguson
would like VR to do so that Ferguson, true friend that he is, will be enabled
to do something for VR in return.
(4) Cornejo: Having these rights in mind, [Virgil] do you wish to

talk to us now?
[4 seconds of silence]
Ferguson: Well, you can stop talking anytime you want to and
you don’t have to answer any question that you do not
want to but there’s two sides to every story. And for us to
hear your side you have to acknowledge that your rights
have been read to you and that you waive your rights.
VR: Well, what do I say? I don’t know.
(R 2:10-11)
At this point, Cornejo takes over again and makes VR’s role in the confession
into a kind of voluntary game: “Well, I’ll ask you a question. If you want to
answer it, answer it” (R 2:13). Although VR has never explicitly in so many
words indicated his willingness to give a confession, he is now locked into
a discourse frame in which he would have to aggressively perform an active
(rather than passive) violation of conversational preference rules: refuse
to carry out a directive. For someone who has been repeatedly diagnosed
as having a compliant personality (and someone with few communicative
skills), this would be particularly difficult. From VR’s subsequent behavior,
it seems clear that, rather than recognizing Cornejo’s Miranda warning as a
statement of a boundary, as the beginning of an adversarial and high-risk
interrogation frame, VR is encouraged by Ferguson’s quasi-offer of help to
see the ensuing discourse as collaborative, win-win: a safe place wherein he
can ask for advice, or openly digress about things that trouble him, in the
hope of getting help or clarification, as in examples (1)–(3) above, and his
repeated questions such as “What should I do?” “What should I say?”
7. THE PROBLEM OF CONFESSIONAL RECOGNITION

OF IDENTITY, AND THE SELF
As noted above, an essential presupposition governing the felicity of the

speech act of confession is the recognition by all participants that the
363
speaker making the confession is the same person as the one who per-
formed the action being acknowledged.7 That is, an identity relation is pre-
supposed between the two: they are, in some sense, “the same” person.
So stated, the relationship seems simple and unambiguous: either A1 = A2,
and (other things being equal) the confession is appropriate; or A1 ≠ A2, in
which case it is aberrant—both semantically meaningless and pragmati-
cally nonfunctional. The clearest cases of the latter are literally false confes-
sions, in which speakers “confess” to deeds they have not done. Such cases
are recognized as inappropriate speech acts, and have no further legal nor
moral status as confessions.
That is the simplest and most obvious problem case. It suggests that
identity relations are analyzable dichotomously: A1 = A2, or the reverse.
But in fact, both in human life and in syntactic construction, there exist
numerous intermediate possibilities: possibilities abhorred alike by lin-
guists, psychologists, and legal scholars, but nonetheless very real. Because
both our minds and (therefore) our language seem to demand strictly
delimited categories rather than continua, we either refuse to recognize
such cases, or relegate them to the margins of our analytic systems as “spe-
cial cases,” willfully refusing to recognize them as central and highly signifi-
cant. Yet these cases define us and are crucial to our language, and must be
acknowledged, and eventually accounted for, in a complete theory linking
language, mind, and the real world in which and on which they operate.
At this time, in Western culture, it is commonplace to feel that “normal”
human beings possess a unified and cohesive “self”—a persona that is felt
to be constant over time as a cohesive and rational agent, a person more or
less consistent at all times, under all conditions. These definitions seem to
us basic, self-evident, and eternal, as aspects of the human condition.
Over the last dozen years or so, postmodernist scholars in a diverse
array of fields have questioned and largely discarded the certainty and
universality of the concept of selfhood.8 This research demonstrates per-
suasively that the “self” is only a convenient organizing fiction of the
7. This definition is somewhat too narrow as stated, in that under some conditions it
is possible to “confess” appropriately to an act performed by someone with a close con-
nection to the confessor, and for whom the confessor has some degree of responsibility
or control—rather like the case of apology. For instance, a mother might confess that
her small child had damaged a neighbor’s property. But even in this case there is still a
presumptive identity relation between the confessor and the doer of the action, albeit
at one remove.
8. For example: in history, work by Davis and Starn (e.g., 1989); in anthropology,
perhaps the earliest and most influential discussion is that of Geertz (1983, ch. 3);
in discourse analysis, cf. Linde 1993; in the field of ethics, cf. the collection edited by
Johnson (1993); and in psychology, cf. Sass 1992.

364
post-Renaissance West, no more basic, universal, or eternal than (say)

Calvinist notions of predestination.
As participants in real life, however (whatever our theoretical beliefs),
we must act and believe as if we possessed selves that provide coherency
across space and time. Our language reflects that purported certainty not
only through the existence of lexical items like self, selfish, self-esteem (and
so on), but through syntactic processes that involve a notion of coherent
identity. Some devices found in English are represented in the examples of
(5) and (6)—first, in (5i–ii), “true reflexives”:
(5) i. I know myself.

ii. I stuck myself with a pin.
It might be argued that even in these “simple” cases, there are two distinct
notions of “identity” operating: in (5i) it is psychological, but in (5ii) physi-
cal. The problem only worsens with
(5) iii. I forced myself to speak.
in which the obvious identity of subject and object implied in (5i) and (5ii)
gives way to a split between the interests and perspectives of subject and
object, yet not a complete schism. The complexity increases in (5iv),
(5) iv. I dreamed that I had eaten New York.
in which there is still a notion of identity between the subjects of the two
clauses, but not the complete merger assumed in (5v):
(5) v. I know I can’t eat the whole thing.
Things get worse still, in the cases exemplified in (6):
(6) i.a. I disapprove of me.

b. myself.
(6) ii. a. In my dream, I saw me eating ice cream.
b. myself
(6) iii. a. I expected myself to be nominated.
b. Ø
In each of the sets in (6), both examples represent some sense of iden-
tity between the subject and direct object of the clause. If identity, or
365
selfhood, were a single, undifferentiable, dichotomous concept, the pos-

sibilities of choice represented in the examples of (6) would not exist.9
What we see in (6) are three options: (1) fully merged and presupposed
identity as in (6iii.b), in which the identification is so complete that there
is no need for even pronominal mention; (2) almost-complete merger,
with reflexive pronouns, in which the presence of the explicit reflexive
asserts (rather than presupposing) full identity, as in (6i.b), (6ii.b), and
(6iii.a); and (3) partial nonidentity, in the cases with the objective pro-
nouns, in which rather than total fusion, there is a sense of the subject-I
looking out at the direct object–I as distinct yet a part of itself, as in (6i.a)
and (6ii.a).
These shorthand devices of the syntactic grammar represent the sub-
tler continua of reality. It is true that we generally construe psychological
“normality” as coinciding with a sense of complete identity of personal-
ity. But psychopathology recognizes a wide range of distinct problems in
the construction of “selfhood”—a continuum that indicates how com-
plex and multiplicitous the cohesive self really is. We can set up such a
continuum:
full recognition –- “neurotic” –- multiple personality–- retrograde –- schizophrenia
of identity noncohesion disorder amnesia10
Along this continuum, VR falls somewhere between “neurotic” nonco-

hesion and multiple personality disorder: his “selves” are not disparate
9. For extensive and interesting discussion of several problems of this kind, see
A. Lakoff & Becker 1991.
10. Some clarification of the points on this scale may be helpful. The second point,
“ ‘neurotic” noncohesion,” refers to the argument in much recent psychoanalytic
writing that what brings people into therapy is a disorganized or incoherent life-
story narrative. Therapy, then, consists of restoring the narrative to coherency. In
this situation, the patient senses gaps and discrepancies among points in the narra-
tive, but recognizes him-or herself as its subject throughout. “Multiple personality
disorder” (MPD), is a condition in which, as a result of early and persistent abuse,
patients experience themselves as fragmented into multiple discrete personalities
with no continuity of memory or personality among them. Each one typically main-
tains its own continuity across time; but synchronically the patient feels divided
into multiple selves, rather than selfless. Retrograde amnesia is that rare state
beloved by writers of soap opera in which, as a result of physical or mental trauma,
memory of all past history is lost. Hence the individual has no “self ” at all, no mem-
ory providing an identity and a means of predicting and stabilizing future behavior.
Finally, in schizophrenia, especially as interpreted by Sass (1992), the individual’s
total sense of self, synchronic and diachronic, is shattered, and boundaries between
the self (physical and psychological) and other individuals and the outside world
cease to exist.

366
enough from our normal assumptions to qualify him as “psychotic,” and

still less as legally “insane”; he does not clearly fail to identify a cohesive
self (but does not clearly identify one, either). The problem (academically
if not jurisprudentially) is: If it is not fully clear to VR that VR the confes-
sor = VR the doer of the deeds confessed to, is the confession valid (since
one of its felicity conditions is disputable)?
Consider a few of many examples illustrating the problem. At first VR
strongly denies remembering any of the actions his interrogators question
him about (e.g., (7)):
(7) Cornejo: You don’t sound like [you’re upset]. I’ve accused you of
killing a little boy and you haven’t yelled at me and said
“You’re a goddamned liar,” or nothing. You just sit there as
calm as can be. I don’t understand your attitude.
VR: Well, I’ve never, you know, there’s something that I did,
I know I didn’t do it, you know.
(R 23:1-2)
But under repeated assault, VR gets flustered:
(8) Cornejo: When we’re accusing you of killing a little boy? Me

arresting you hasn’t upset you?
VR: Well, I’m in shock. I don’t know what to say. I don’t know
what to say or do.
(R 23:9-10)
After much more of the same, he begins to waver:
(9) VR: But, but if I killed somebody I am deeply, deeply sorry that I’ve
done it. By being knowing myself, I probably did kill somebody.
(R 27:8)
And he speaks similarly later on, even though by that point a more ade-
quate “confession” had already been obtained:
(10) Cornejo: Were you talking to the little girl [Virgil]?

VR: I probably was or didn’t know what I was doing or
something like that. I don’t know.
(R 50:10-11)
367
VR’s “confessional” statements here sound (especially in their use of words

like probably) like those of someone who is trying to make sense of some-
one else’s actions or motives, rather than remembering his own. That sug-
gests that VR’s self-identification is at least problematic, if it exists at all to
the degree necessary to validate a confession. Ultimately, VR finds it easier
(especially being a “compliant” personality) to incorporate his interroga-
tor’s plausible narrative into his own “identity”: better to have a clear iden-
tity (even if it gets him into the gas chamber) than the fuzziness VR seems
to have been living with. Dubious identity for VR is a fate worse than death:
(11) Cornejo: This is the little boy right here. Handsome little boy, isn’t
he? Eleven years old. A handsome eleven-year-old little boy.
VR: I killed him?
Cornejo: You killed him [Virgil]?
VR: I killed him?
Cornejo: Uh-hum.
VR: I don’t know why I killed him.
(R 30:17-22)
While that is taken as a virtual confession by his interrogators, it is cer-

tainly arguable that it represents a failure of the requisite conditions for that
speech act, rendering it, in Austin’s terms, void. Examples have already been
given that illustrate failures in all the other significant aspects of the confes-
sional interaction: the rules of conversational behavior are frequently vio-
lated by VR—he fails to answer when an answer is mandatory; his answers
are frequently dispreferred—for instance, compliant rather than responsive.
He fails to grasp the discourse frame of confession: the power imbalance, his
jeopardy, the truth-seeking nature of the discourse. While he may technically
“understand” the language of the Miranda warning, and technically “partici-
pate” in the confessional discourse, it is far from clear that he is pragmatically
competent to understand what is happening—the subject matter of the talk,
his situation, the probable consequences. His uncertainty about identity fur-
ther casts doubt on the validity of his “confession” as a whole.
8. CONCLUSIONS AND DILEMMAS
Within the safe confines of academic interchange, the examples discussed

here are intriguing but not seriously problematic. They suggest a link which
any coherent theory of language must represent, between syntactic form,
psychological perception, and the real world both seek to capture. There

368
are fuzzinesses and uncertainties in the syntax of reflexivization in English

because the self (which that grammatical process implies or asserts) is by
no means a clear or certain construct in our psychologies or our daily inter-
(and intra-) actions. Language, as linguists love to claim, is important
because it informs and creates our physical and psychological reality: If
pragmatic analysis represents confession as complex and subtle interactive
behavior, and shows that many interactions that might be labeled confes-
sion fail to meet the pragmatic criteria for the valid performance of that act,
does that force nonlinguists—ethicists, lawmakers, and legal scholars—
to re-evaluate the rules of criminal interrogation? Does it force all of us
to re-evaluate our belief in ourselves (late twentieth century Americans)
as humane people committed to the vision of the Founders embodied in
the Bill of Rights—in this case, the Fifth Amendment protection against
self-incrimination as represented in the Miranda decision’s requirement of
informed consent? If we question the admissibility at trial of confessions
as dubious as this one (but “dubious,” rather than unambiguously coerced),
where will our good intentions leave our society? If the construct of “self-
hood” is as tenuous as data of this kind suggest, and if the admissibility
of confessions rests on this and other problematic grounds—might it not
be argued that no confession is legally admissible? Then what becomes of
criminal-trial procedure, since convictions often depend significantly if not
crucially on confessions? How can we feel comfortable about consigning
defendants to long sentences—even to death—on the basis of evidence
acquired—we must finally admit—fraudulently? Unthinkable!
So examples like VR’s confession force us to rethink the use of criminal
confessions altogether, as a great many obtained under these (quite typical)
conditions would be found to be tainted. But many convictions—perhaps
most—especially in the most serious crimes, depend on the availability
of a confession to the jury. Were this option to be lost, convictions would
surely become harder to secure. And—in America’s favorite nightmare—
criminals might go free to walk the streets. And if the government were
seen as a result to be helpless to ensure the security of the citizens, the
social contract might well be abrogated—an invitation, as the Founders
would be the first to tell us, to revolution. Intolerable!
I for one am happy that, as an academic linguist rather than a jurist,
I don’t have to make real-life Solomonic decisions of this kind. If I had to,
though, I might suggest as a first step what has already been suggested by
prominent legal scholars: that convictions, and certainly capital sentences,
never be based totally on a defendant’s confession (as is not infrequently
the current case), since confession as a speech act and a discourse genre is
notoriously subject to abuse.
369
REFERENCES
Ainsworth, Janet E. 1993. In a different register: The pragmatics of powerlessness in

police interrogation. Yale Law Journal 103(2):259–322.
Austin, J. L. 1962. How to do things with words. Oxford: Clarendon.
Davis, Natalie Zemon, and Randolph Starn. 1989. Introduction. “Memory and
Countermemory.” Special issue, Representations 26:1–6.
Geertz, Clifford. 1983. Local knowledge. New York: Basic Books.
Johnson, Barbara (ed.) 1993. Freedom and interpretation. New York: Basic Books.
Lakoff, Andrew, and Miles Becker. 1991. Me, myself and I: Spatial metaphors for the
self. Berkeley: University of California, MS.
Lakoff, Robin Tolmach. 1990. Talking power. New York: Basic Books.
Linde, Charlotte. 1993. Life stories: The creation of coherence. Oxford: Oxford
University Press.
Sacks, Harvey, Emanuel Schegloff, and Gail Jefferson. 1974. A simplest systematics
for the organization of turn-taking for conversation. Language 50:696–735.
Sass, Louis A. 1992. Madness and modernism: Insanity in the light of modern art,
literature, and thought. New York: Basic Books.
Searle, John R. 1979. Expression and meaning: Studies in the theory of speech acts.
Cambridge, UK: Cambridge University Press.

370
371
CHAPTER 16
Afterword
BY ROBIN L AKOFF
1.
There has been some commentary in the popular media about a current
tendency to premature, and postmodern, autobiography: people of ten-
der age and therefore moderate accomplishments publishing full-length
autobiographical statements whose factuality is open to dispute, if they
are not frankly fictional or “fictionalized.” In this case, the first criticism
is easily disposed of, since I am no longer of tender age. The second is
more troublesome. It is worth asking how truthful any autobiographical
statement can be, and how truthful this one actually is. All I can say is
that I think the way I describe my career as represented in these papers
is the way it happened; what I say that I meant at the time to accomplish
was indeed my intention; my current interpretations of what I was saying
are at least not strongly antithetical to the ones I would have made, had
I been asked, contemporaneously. But of course we know memory is slip-
pery. Caveat lector.
The academic autobiography is a genre with relatively few prototypes
available for imitation. Should I look to one of the heroes of my youth,
Julius Caesar, and write in the third person—suggesting disinterestedness
and thus reliability? Or represent my life as travelogue—first I went here,
then I moved there; the natives were friendly, the scenery enchanting… .
The reader may be getting impatient here: Why not just get down to work,
move through the material, leave yourself out of it like a good academic?
But these papers are my self. They were a part of me when I wrote them,
372
and they still are close to me, like children. And no one can reasonably
claim to be objective about her children.
At the same time, this is an academic retrospective. Therefore I may, or
must, omit all sorts of juicy matters, and include some rather dry ones.
Arguably I am writing not about my self, but about my work, if these can
be differentiated.
A possible compromise is to use the texts as the bases of my discussion,
focal points; but allow myself to reflect in passing not only on what the
papers in and of themselves were about, but also why I wrote them, why
I wrote them as I did, and when I did; why I said some things explicitly,
some more or less roundaboutly, and others not at all—often things that,
seen from the vantage point of the present, deserve lengthy and explicit
explication. In that sense, scholarly explication du texte meets personal
memoir. Let’s see what happens.
2.
“Once,” says the Mock Turtle in Alice’s Adventures in Wonderland, “I was a real
Turtle.” And once, reader, I was a real Linguist. In fact, before that I was even
realer: I was a classicist, more specifically, a Latinist. Like his schoolmate the
Gryphon, I studied Laughing and Grief. I have retained a soft spot in my heart
for Latin, and go back to it from time to time, on any excuse I can concoct.
But I was young back then, in the early 1960s, and I was in Cambridge,
Massachusetts, and I was hungry for answers, real truths, such as the study
of a language and its literature seemed unable to provide. At that time,
in that place, there was a compelling alternative: down the block from my
own institution, the field of linguistics was being re-created, language itself
being redefined. I could spend the morning in the tenth century bce (an
amusing enough place, but not au courant), get on the bus, and get off at
RIGHT NOW, in fact, TOMORROW. It was irresistible. At MIT, linguistics
was a combination of cutting-edge science and religious cult—a delectable
if corrupting combination for the youth. So I became a linguist, first an
Indo-Europeanist and then a transformational grammarian.
For a few years we thought, my cohorts and I, that we were following
the approved transformational program. But a problem arose because (as
is not uncommon in cults) the program itself was not completely clear,
and subject to sudden reinterpretations ex cathedra. Many of us had joined
up on the promise that the goal of transformational generative gram-
mar was to use language as a “window into the mind,” a way of rigorously
373
understanding the cognitive and social structures that make us human.

The explication of deep structure in Chomsky’s Aspects of the theory of syn-
tax (1965) and his other writings of the early to middle 1960s created con-
sternation in our minds: to the extent that the treatment was intelligible at
all (can you say “Boolean condition on analyzability,” boys and girls?), the
basic level of analysis seemed highly English-specific and purely linguistic.
How could social interaction and psychological motivation be developed
within a theory so rigidly bounded?
But we were obedient disciples and tried not to ask those questions.
Rather, we decided, often not entirely consciously, that the master’s writings
could be read so as to encompass our own agendas—he just hadn’t had the
time to do so yet. We helpful good disciples would do that for him. We did.
The results were not greeted with gratitude, or even tolerance. We were
made to realize that we were in a state of rebellion, and would be expelled
from the cult if we did not conform. We didn’t. We were.
We came to understand, by 1967 at the latest, that we were working
to develop a quite radically different model of language, not anything
that could be reconciled with what Chomsky would shortly designate as
Standard Transformational Grammar. We tried to devise our own name for
ourselves: we played with Abstract Syntax, Logical Syntax, and one or two
others before settling on Generative Semantics or GS (borrowed without
attribution from Wallace Chafe).
From the first, the premises of GS were diverse and changing (while
Chomsky kept tight rein on STG). That gave the impression of chaos, but
it was an exciting and productive chaos. It does make it difficult to define
what GS was and wasn’t, since anything I say will be disagreed with by
one or another member of the group, with perfect right. But basically the
agenda might be summarized by the following:
1. The basis of syntax is semantics and pragmatics.

2. The basic (semantic) level of syntax, Underlying/Logical Structure, is
universal.
3. The context in which the rules of syntax (transformation rules) are
stated includes information about the psychological state, and social
circumstances, of the speaker.
4. Grammar and its categories are indeterminate and multiple rather than
binary.
Members of the GS contingent tended to specialize in one or two of the

agenda items. My own area, mostly, was #3, and everything I have done
A fterword [ 373 ]
374
since I began to be a linguist can be seen as mainly a development of #3,

with incursions into the others.
3.
Though these origins can be seen in their clearest form in the first three
chapters (NB: chapter 3 includes excerpts from two papers written during
this period), written under the aegis of the GS project or during its imme-
diate aftermath, much of what I have done since then is also better under-
stood if their GS roots are made explicit. For this reason I am spending
what may seem a profligate amount of space on the development of GS and
its influence on my work. But I think it is generally true that the beliefs of
one’s formative years continue to exert a strong, if sometimes subliminal,
influence on the work of one’s maturity. Besides, the motives and findings
of GS have been misunderstood and maligned by writers who have little or
no personal knowledge of the movement, and I think it’s important to try
to set the record straight. Finally, I am spending a disproportionate amount
of time on the early papers because, as the theoretical (and theological)
premises under which they were written fade into the mists of time, they
become almost impossible to understand without a gloss.
In his “Retrospective Epilogue” to his posthumously published collec-
tion of writings, Studies in the way of words (1991), the philosopher Paul
Grice sorts the nineteen papers included in the volume into eight “strands.”
In the fashion of analytic philosophy, he doesn’t discuss his choice of ter-
minology. But I think it a profound choice, one well worth imitating; and
perhaps, in the forthright manner of a linguist, I can explain why I find the
notion of “strands” of topics useful enough to appropriate.
Strands are long and thin, like necklaces or pasta. A strand of a necklace
can stand alone, but it is in combination with others that it achieves sub-
stantiality and beauty. The strands are sometimes left separate, one on top
of the other, but often intertwined, so that the observer cannot tell one
from another. Likewise with pasta—while you can eat spaghetti strand by
strand, it is not really satisfying to do so. Tangled together, the spaghetti
picks up the sauce and other goodies, making a complex and interwoven
experience for the eater. Attractive, substantial, and tasty—that’s how
I hope you will see these papers, so I offer them as strands of a complex and
composite totality.
Something else that comes in strands is DNA. DNA differs from spa-
ghetti or necklaces in that its strands may reorganize themselves. An item
originally on one strand may link up with another, with interesting and
375
sometimes profound effects. So the strands represented in this collection

(like necklaces or pasta, as well as DNA) do not maintain their separate
identities over time. It’s true that in each case, basic passions of mine can
be traced along a twenty-five-year time line (although generally much
mutated by the end); but at the same time (like overdone spaghetti, maybe)
they fuse together, and they interchange parts and mutate, like DNA.
Thus the main subject matter of the earliest papers finds representa-
tion, if often implicit, in the later ones. The assumption that language is the
symbolic representation of the human need to express ideas and emotions,
to persuade and to achieve intimacy and distance, power, and equality, was
originally and continues to be my reason for becoming and remaining a
linguist. Based on these reflections, I can identify four major strands that
organize the fifteen papers (now chapters) here.
4.
Strand one: Theory wars: The grammaticalization of psychosocial context (chapters 1–3, 11)
The first three chapters here represent an attempt to find a way to incorpo-
rate extralinguistic information into the syntactic component of a trans-
formational generative grammar. In many ways they are characteristic of
the whole GS project, especially in its earlier stages, both in terms of con-
tent and style. They are an attempt to move from a binary, determinate,
linguistic-autonomous understanding of grammar, and especially syntax,
to something fuzzier and more complex.
Indeterminacy was a major bête noire for the Extended Standard
Theorists. As it appeared to me and still does, they were more concerned
with making a pure science of linguistics—removing it from its tawdry
humanistic origins and social-science companions—than with actually
looking at and accounting for the way real people actually use language,
often imprecisely and ambiguously. Some of my colleagues were obsessed
with fitting linguistics into a Kuhnian paradigm or achieving Popperian
falsifiability, since only thus (they argued) would linguistics be a True
Science. What they missed is that science has achieved its results and thus
its prestige by observing and generalizing carefully from natural data.
Transformational grammar, on the other hand, derived significantly from
philosophy of language, a field in which the data were produced in the mind
of the investigator, then subjected to his (just imaginably her) own analy-
sis. So it was more important to produce a formal rule, process, or theory
than to capture what happens when people speak. That focus made it easy
A fterword [ 375 ]
376
to arrive at sweeping but dubious generalizations, and to dismiss “natural-

istic” examples, if any should show their ugly faces, if they didn’t conform
to the rules, as “performance errors” or special cases of one sort or another.
The context in which the applicability of syntactic rules was judged ended
at the sentence level.
(On the other hand, alternative models available at the time were strongly
anti-mentalistic and anti-interpretive. Followed literally, such theories guar-
anteed that investigators would be unable to say anything of interest about
what I wanted to know: language as a revealer of human motives.)
From the vantage point of nearly fifty years later, it is pretty easy to
discard those ideas. Certainly GS would have done better had it been
able to sneer at the pseudo-scientistic pretensions of Transformational-
Generative Grammar (TGG). We should have moved beyond syntax more
confidently; abandoned binary rule statements and determinate gram-
maticality more quickly than we did and more explicitly; realized that prag-
matics, or interactive capacity, was central to language, and syntax merely
its servant, rather than the reverse. But we had been trained by a man we
revered as transformational syntacticians, and we were loath to abandon
our principles.
The standard theorists warned us that there were dragons out there. If
we brought indeterminacy into grammar, if we introduced psychological
and social context into syntactic descriptions, our grammar would become
“too powerful”—untestable, unfalsifiable, unscientizable. We would be
reverting to the evil ways of the pre-Neogrammarians, into the murkiest
kinds of mentalism. Once we started incorporating feelings, assumptions,
and so forth into the grammar, where would we stop? Would everything
in the world, everything that a speaker might become aware of, have to
receive formal grammatical encoding? We could not, they suggested, say
when to stop. Better to just say no.
These papers are an attempt to have my cake and eat it—an attempt
to achieve precise grammatical description, yet incorporate extralinguistic
context. It was becoming clear that to ignore context was to drive meaning
out of linguistic description, to make syntax, and indeed linguistics gener-
ally, a purely formal field, to view language as merely strings of phonemes
produced without reference to speakers’ needs or proclivities. How, we
wondered, could such a field fulfill its promise to show how language was a
window into the mind?
The basic assumptions in these papers are, first, that it is necessary
to incorporate extralinguistic material into syntactic descriptions. If you
don’t, you will fail to express essential generalizations (the mortal sin of
not accounting for all, and only, the sentences of the language). Second, it
377
is possible to do so while retaining appropriate rigor and falsifiability: the

relevant criterion for inclusion is whether the extralinguistic factor has
direct linguistic correlates. Thus, for instance, the nature of the relation-
ship between participants might be reflected in the direct linguistic artifact
of honorifics; but the temperature in the room would not be, and so would
not have to be included in a complete syntactic description. These papers
are a first attempt to taxonomize, to differentiate the linguistically relevant
real world from its complement.
Stylistically, too, these papers are a challenge thrown to the formal side
of the field. Even more than today, scholarly writing then was formal and
self-important. The passive voice was used, one’s personality suppressed;
levity of any kind was frowned upon. That was especially urgent in a newly
minted field whose practitioners wished to be taken as “scientists,” with
all the seriousness that that descriptor implied. Never mind that real sci-
entists named their findings “quarks” which had “flavors” like “charm” and
“strange.” GSists like me got roundly chastised by our respectable transfor-
mational peers for our playfulness, amply demonstrated herein.
Also, these papers were conceived, written, and published during the
1960s (which really didn’t end until 1975, when the Vietnam War ended
and Nixon resigned). As children of that decade, we saw irreverence as a
form of rebellion against the status quo, our bow to sex, drugs, and rock-
and-roll, some of which figure prominently in our examples. I suppose
our style can be seen as slyly supporting our content: our fooling around
brought the playful realities of the 1960s into the sober “only-the-facts-
ma’am” world of academe. We were sneaking the real-world context into
the generally pristine ivory tower.
The last of these papers, “The way we were,” (chapter 11) was written
much later, with the advantages of hindsight. It is an attempt to justify the
ways of GS to a field that had been encouraged (by writers like Newmeyer)
to see it as an aberration, a pointless and baroque wandering off the true
path, as well as a first attempt to do what I am doing now, to make sense of
it all to me. Even its title, borrowed from a movie, is a throwback. I recog-
nize in myself as well as my colleagues a regrettable tendency to get more
serious, not to say ponderous, as we age and gain professional clout.
“The way we were” looks at linguistics from the perspective of anthro-
pology/history/sociology of science. It attempts to make sense of the way
the field worked, and works, and the way ideas within it are favored or
discarded—the convoluted connections between “truth” and politics in
academe. So it also fits into the fourth strand of papers, concerned primar-
ily with the way institutions, or rather their members, use language to cre-
ate their institutional identities.
A fterword [ 377 ]
378
5.
Strand two: The pragmatics-semantics interface: Language as creator of psychosocial

identity (chapters 4, 5, 6, 7, 12)
There is no sharp distinction between the first two strands, but merely
a shift in focus. If you join a movement advocating the incorporation of
psychosocial information into syntactic description, then you must believe
that the correct and complete identification of those psychosocial forces,
as they are represented in linguistic form, is of the utmost importance. So
even as the GS-ST (Standard Theory) wars abated and each side retreated
to its own corner, the former combatants on the GS side took their marbles
and went home: each of us returned from the battle to concentrate on the
particular aspect of the GS project that had drawn us into linguistics, and
onto the GS side, in the first place. My initial interest was psychological,
and so that occupies a large place both in the papers of strand one, and in
the first post-GS writings here. But I realized at the same time that no man
is an intrapsychic island: we create our internal selves, our belief systems,
our sense of who we are, in part on the basis of what we get from outside,
how others communicate with us and what our communicative possibili-
ties are. At the same time, these social aspects of constructed identity arise
partially out of our internal, psychic representations of the world, others,
and ourselves. And many of these considerations have explicit linguistic
representation. By recognizing these undercurrents we can carry out one
of the original missions of transformational grammar: to use language as
a window to the mind, where the “mind” is both the internal and external
selves.
Some aspects of this linguistically made selfhood are more readily con-
fined to the intrapsychic mind, others more extrinsically determined. In the
first set I would locate presupposition (both “semantic” and “pragmatic,”
if that distinction is useful or possible). In the second, I would place the
analysis of directness and politeness. But as these papers show, that neat
division fails almost at once: presupposition shapes our choice of direct or
indirect form; the decision about whether and how to speak politely to an
interlocutor is based on our assessment (“presupposition”) of the relation-
ship between us and how our culture expects us to treat each other. So
strands two and three also tend to coalesce.
This difficulty I am having in keeping these papers distinct reflects my
belief that the world and its categories are fuzzy rather than sharply deter-
minate. As I noted, later GS took explicit cognizance of indeterminacy,
diverging from the TG belief in distinct, often binary categories of many
379
kinds, a tenet of GS that tended to distress people, including some GSists.

People often fear and loathe indeterminacy, whether it be moral (“situa-
tional ethics”), political (“waffling” as deadly sin), aesthetic (consider the
furor over nonfictional fiction, or the line between fiction and [auto]biog-
raphy), or sexual (audiences used to be horrified when transvestites, trans-
gender people, or even gay persons showed up on talk shows). Scientists
are only human, so it’s not surprising that many of them are uncomfort-
able with fuzziness. Even as many hard sciences move in the direction of
acknowledging indeterminacies in the real world, adherents of less estab-
lished sciences still find it essential to cling to determinate descriptions and
processes, as if that could ensure them a place among the scientific elect.
Anyway, these papers concentrate on the role of intrapsychic processes
in the creation of language form, while the next strand is more concerned
with the interpersonal end. I classify both of these as “pragmatics,” although
arguments could certainly be made for defining the first as “semantics” and
the second as “sociolinguistics.” But again, attempts to squeeze the sub-
jects of linguistics into neat, separate little subdisciplinary boxes will yield
nothing but frustration.
In these papers, the relation among language, mind, and social inter-
change is circular: each creates and influences each other. As a way of
describing and classifying this circularity, I use the term “style,” especially
in chapter 5. Both c hapters 4 and 12 are much concerned as well with the
role of gender as part of that circle, although they treat the interface of
language and gender in somewhat different ways.
Chapter 4, “You say what you are: Acceptability and gender-related
language,” examines the connection between the way you speak and your
gender, and the way you speak and the way you are treated. In that it is a
reworking and summary of many of the arguments in Language and wom-
an’s place, published a couple of years earlier. Chapter 5, “Stylistic strategies
within a grammar of style,” is an attempt to move from the simple form–
function connection to a more complex relationship among language form,
psychological self-construction, and social position, through the concept of
communicative style.
In using the word “style,” I am thinking here especially of the way
the term is used in literary criticism: style as the sum of the expressive
choices that both make an author distinguishable from all other authors,
and impart coherency to her oeuvre. Style covers the whole linguistic
spectrum—word choice, syntactic options, pragmatic possibilities. As
I use the term here, I see style as constructing individual identity and
being based upon it, and being re-created as well as subtly changing in
each interaction with others. So we may behave quite differently talking
A fterword [ 379 ]
380
to our child or to a prospective employer, but there would be a constancy

or consistency beneath the changes. The paper also returns to the issue of
politeness first discussed in c hapter 2, and tries to suggest that it is part
of the linguistic grammar, being rule-governed like language itself and
likewise subject to diachronic change. Since it sees politeness as a socially
constructed, interpersonally relevant behavior, it moves between strands
two and three.
Chapters 6 and 7, while situated in the pragmatic exploration of the role
of language in the definition of self, move beyond it to strands three and
four: both, for instance, touch on the role of language in the making and
perpetuation of institutions, on the importance of learning the right way
to communicate if one is to achieve success within an institution, and on
the way in which institutions achieve coherence around their members’
knowing a special way to talk. That emphasis is especially clear in chapter 6,
“When talk is not cheap: Psychotherapy as conversation.” I have been par-
ticularly fascinated with two special institutional forms of language, those
of psychotherapy and the legal system. They intrigue me because in both,
the institutions achieve efficacy and satisfy the societal need for which
they were created, largely through linguistic means. Other institutions use
language, of course, for their ultimate ends. But these two more than oth-
ers work via language: interpretation is paramount in the workings of the
most influential forms of both. Both also create particularly impenetrable
forms of discourse that must be mastered by insiders and which serve to
keep outsiders at bay, excluded from meaningful critique. So the question
for psychotherapy is: How do you change people’s behavior and perception
of reality through language alone?
Chapter 7, “Some of my favorite writers are literate,” has a somewhat
different emphasis, understanding literacy as one option available to a soci-
ety for the transmission of its important knowledge. In it I suggest that our
culture may be moving past literacy into postliteracy—that the importance
we have for many centuries placed on literacy as the entrée to positions of
power and prestige may be waning; that the communicative rules of the
meta-institution are shifting, if slowly and imperceptibly, and we must
change with them.
Finally in this set, chapter 12, “Women and disability,” looks at one sub-
group of the populace and the ways in which the powerful majority con-
structs them as a lesser minority by its uses of language, and the ways in
which they are working to take back the right to make their own language
about themselves and so gain control of their own social and psychologi-
cal identities—rights traditionally denied to the non-powerful. I use the
format of a review article discussing several books written by and about
381
disabled women, a group with two strikes against their achieving social
equity. This paper links strands two and three, with reference to how
extrinsic traits (gender, abledness) affect one’s communicative options and
the meanings one makes; and strand four as well, in its concern with the
making of public meaning by individuals and institutions.
6.
Strand three: Pragmatics-cum-sociolinguistics: The grammar of interpersonal communi-

cation (chapters 1, 2, 4, 5, 6, 7, 8, 10, 12)
Language creates and is created by the self and at the same time creates
and is created by community. Even as humans are cognitive creatures (a
focal point of strand two), we are at the same time and equally importantly
social creatures, and each of these identities permits and potentiates the
other. So while strand 3 represents a change of emphasis from strand two,
it moves in many of the same directions. That connection makes sense
since often, the same linguistic devices work either as intrapsychic/cogni-
tive or as interpersonal/social, or even both at once.
While c hapters 1 and 2 were listed in the first strand, the arguments
used in the cause of theoretical persuasion were based in part on the social
functions of language, for instance the use of forms like honorifics (which
symbolize the relationships between participants) and politeness as a rea-
son for indirectness. Chapter 4 considers the gender of participants as a
significant factor in the choice of forms and their understanding. Chapter 5
also considers politeness as a function of grammar, in this case both syn-
chronically and diachronically, as well as male versus female strategies
and cross-cultural differences. In this one can see changes from earlier,
GS-oriented work like chapter 2: politeness is seen here as much as a lin-
guistically mediated social phenomenon as a social relationship with syn-
tactic consequences. Chapter 6 considers a particular form of interpersonal
communication, psychotherapy, looking at the therapeutic interchange as
a special sort of discourse in which participants engage for explicit pur-
poses; the peculiar forms of therapeutic discourse arise out of and facilitate
the functions the discourse is meant to serve. Chapter 7, like chapters 5
and 6 discussed above, is concerned in part with the social and interactive
function of literacy, versus orality. The choice of the literate channel makes
statements about the nature of the communication and the relationship
among the participants, and the currently shifting valuation of literacy will
bring changes in society.
A fterword [ 381 ]
382
Chapter 8, “Persuasive discourse and ordinary conversation, with

examples from advertising,” looks at another kind of persuasive discourse
(while all language is persuasive in intent, some types have persuasion as
their main raison d’être), advertising. This paper, too, fits into at least two
strands: it is concerned with the use of presupposition and implicature to
affect attitudes (strand 2), but mainly with the social effects of a discourse
form that is manipulative, if not coercive. Does playing that game under-
mine the basic need of a social creature—mutual trust? Where therapy,
one type of overtly persuasive discourse, requires trust on both sides to
be efficacious, advertising toys with our credulity, leaving us permanently
suspicious.
Chapter 10, “My life in court,” is my first attempt to make sense of court-
room discourse. It arose out of a prolonged term of jury service in a capital
case. As with many of the papers here, this one addresses the matter of
power and its representation via language. Who has power in court, and
how is that distinction manifested linguistically? Do the ways in which the
imbalance of power, particularly that between professionals and amateurs
(the jury), is played out affect the way justice gets done? On the other hand,
I argue that the mere simplification of technical language will not solve the
problem, and indeed in reducing the solemnity of the jury experience may
do more harm than good.
Chapter 12, as noted earlier, looks at how members of traditionally less
powerful groups find that language can keep them powerless, unless they
can make language for themselves—no easy task.
7.
Strand four: Sociolinguistics and sociology of language: The language of institutions and

its consequences (chapters 6, 7, 8, 9, 10, 11, 13, 14, 15)
Institutions are created by individuals and small groups who unite into
large groups and so gain authority, and which are held together internally
by conventional rules and externally by the larger society’s belief in the
institution’s value. Institutional language is particularly interesting: it
must work for a large, often culturally diverse group; it must serve at once
as an internal “secret handshake,” and an external “keep off the grass” sign;
it must also work as shorthand, to express the cognitive concerns of the
institution in an economical way. It is often defensive, but as often vainglo-
rious. In short, it is an embodiment of all that makes us human, and plays
on all of the strands in these papers.
383
Hence the papers in this group have almost all been discussed under
other subheads, and need not be mentioned specifically here. One excep-
tion, chapter 8, “Doubletalk: Sexism in tech talk,” on technology and its
language, is an attempt to confront a contemporary issue fraught with
problems. When a particular form of language is connected with an espe-
cially prestigious institution, and when, further, that form of language and
that institution are, for whatever reasons, open only to certain groups in a
society, how does that affect social equity? In the case of technology, what
is the relation between the impenetrability of its discourse style and the
tendency for women to feel unwelcome? Should (can) inaccessible language
be democratized by fiat, bringing with it real-world equity? Or does lan-
guage follow reality, so that once enough women (or other disenfranchised
minorities) enter a field, the language will change with them? Or will they
merely be corrupted by internalizing the language, and with it the mores,
of the institution, turning into honorary white males?
Chapter 14, “The rhetoric of reproduction,” is an investigation of the
rhetoric of the public sphere, concerning policy on reproductive rights. The
major question here is: What is appropriate public rhetoric? In a hotly con-
tested struggle, what are the rhetorical responsibilities of each side? Is it
possible to distinguish legitimate from illegitimate persuasion?
8.
And…
I have, over the quarter-century represented by these papers, moved
from being a relatively conventional ivory- tower linguist, concerned
mainly to establish my own theory of language and linguistics, to a more
public stance, a concern with language as a vector of politics and power,
which can be used for good or ill. Many of my colleagues view this stance
with dismay, believing that a scholar must be disengaged and apolitical.
That may be an option, or even a desideratum, in some fields. But a
thinking member of a democratic society has to have feelings about the
right way for her society to go. I don’t think those feelings can or should
be compartmentalized into someone’s “political component,” brought out
only the first week in November. We are—besides being social and cogni-
tive creatures—political animals, instinctively interested in power and its
distribution, and unless we are candidates for sainthood, eager to get our
fair share of what’s going around. It would be peculiar if our work could be
insulated from those concerns, and if it even could, it would become unat-
tractively passionless.
A fterword [ 383 ]
384
It’s much better, if we’re doing work that might have any foreseeable
political consequences, to state our own biases at the outset. It is the rare
linguist these days who can rightly believe his or her work resides exclu-
sively in the ivory tower. Current political and social debates over English-
Only legislation and “Ebonics” support that argument. There are two ways
to reconcile one’s academic and political selves. One, exemplified by the dual
oeuvres of at least one eminent linguist, is the path of multiple personal-
ity: one persona publishing inaccessible linguistic theory, the other writing
impassioned political diatribes with very little direct analysis of language
per se. But if my training enables me to understand language in depth,
I should apply my skills to the analysis of language and its sociopolitical
functions. I don’t find it easy or intuitive to split myself into scholarly and
political selves, or intellectual and emotional beings. I also think—despite
a lot of conservative criticism of this position—that emotional energy does
not damage intellectual analysis, when the two are applied together to the
understanding of linguistic problems; rather, both work together to reveal
how language works, since language itself is both cognitive and emotional.
You can’t use pure logic to understand passion, and passion underlies many
of our most crucial linguistic debates and choices.
As these papers suggest, linguistics is my field because it is a universal
donor. Most of what we do as humans is done via the medium of language.
The understanding of language therefore illuminates other aspects of being
human, perhaps more than any other field. Only linguists have a way of
making sense of the human proclivity for using language to construct our-
selves, our cultures, and our societies—a task as absorbing as it is unnerv-
ing, as endless as it is immediate.
REFERENCES

Grice, H. Paul. 1991. Studies in the way of words. Cambridge, MA: Harvard
University Press.
â•‡ 385
INDEX
A morphological novelty, 195–â•‰196

Aberrant language, 64 motivational psychology, 193
Abortion origins, 193
anti-â•‰abortion sloganeering, 332 persuasive discourse and ordinary
Clinton (Bill), on, 332, 332n2 conversation, 183–â•‰201 (See also
popular support, U.S., 333 “Persuasive discourse and ordinary
rate, worldwide, 331–â•‰332 conversation, with examples from
slavery rhetoric, 340–â•‰341 advertising”)
terms, 333 power relationship, 191
“Abortion rights,” 333 pragmatic novelty, 197–â•‰198, 198n5
Abrams et al. v. United States, 319, as propaganda, 194
319–â•‰320nn10–â•‰11, 321 reproductive rights, 336–â•‰338, 338n2
Academese semantic anomaly, 196
evading responsibility, 108 subjects, 196, 196n4
as exclusionary, 212–â•‰214 subliminal, 192
male power elite, 212–â•‰215 syntactic innovation, 196
Acceptability, 82–â•‰83, 86–â•‰87. See also syntactic novelty, 195–â•‰196, 198
“You say what you are: Acceptability verbal auxiliaries, 196, 196n4
and gender-â•‰related language” words vs. images in, 192–â•‰193
judgments, 86–â•‰88 African Americans
judgments, context on, 88–â•‰89, 95 Black English, 62–â•‰63
Adorable, 90, 92 in U.S. prisons, 217, 218
Adversarial argumentation, 272–â•‰273 Ah, 91
failure of perception, 361, 362 Ainsworth, Janet E., 361n6
Advertising, 191 Alvarez-â•‰Caccamo, Celso, 148
Cooperative Principle, 199–â•‰200 Ambiguities, 107
deception, FTC on, 192–â•‰193 abortion debate, 338–â•‰341
definite article, 196 debate, 338
false, 310 Ambiguity
humor, 197–â•‰198 pragmatic, 40–â•‰42
language, 65 syntactic, 40
lexical novelty, 195, 198 American men
Manner violations, 195–â•‰198, camaraderie, 125, 126
196n4, 197n7 distance/â•‰clarity, 122, 123, 125
Manner violations, Cooperative American women, deference, 122–â•‰123,
Principle, 195–â•‰198, 196n4, 197n7 127–â•‰128, 212
386
“Anti-abortion,” 333. See also persuasive discourse, vs. addressee,

Reproduction, rhetoric of 189–190
Any, 87 religious and political gatherings, 190
Applied sociolinguistics, 309 Austin, J. L., 258, 321, 328
Appropriateness, 317 on communicative competence, 355n4
Arendt, Hannah, 206 on speech acts, 315–318, 317n3,
Ariès, Philippe, 126 355n4, 367
Artificial intelligence, language, 214 on speech acts, performatives as,
Asch, Adrienne, 302, 303, 306 317–318, 328
Aspects of the theory of syntax (Chomsky), Autobiography
37, 81, 242–253, 287, 289. See also academic, 371–372
Classical Transformational Grammar premature or postmodern, 371
(CTG); Generative semantics Autonomous semantic component, 243
classical transformational grammar, 75 Autonomous syntax, 53, 54, 88
deep structure, 74, 74n5, 106, Autonomy
242–244, 248, 249, 251, 260, bodily, 303
273, 286, 289, 373 pragmantax vs., 257–259, 257n4,
formulations, vagueness, 246–247, 259n5
372–373
McCawley on DS in, 258 B
shortcomings of, 53 Bach, Emmon, 252, 261n6
universals of form and substance, 286 Bach-Peters Paradox, 261n6
Assumptions Baker, C. L., 74n5
generally held, 38 Bateson, Gregory, 138, 138n2, 144
implicit rules of conversation, Becker, Miles, 365n9
18–19, 19n6 Behavior. See also Rule of Communicative
psychological, culture-wide, 76 Competence
Assumptions of speakers character, personality, and personal
cultural, 37–38 style, 101–102 (See also Style,
generally held, 38 personal)
on sentence use, specific contexts, culturally appropriate, 124
37–38 dualistic, 108
toward social context, 38–39 linguistic embodiment, 101
Asymmetry. See also Inequality; specific stylistic, 110
types Belenky, Mary Field, 279n9
gender-based, on Internet, 311 Beliefs about ourselves, language and, 350
power, 178 (See also Power imbalances) Bernays, Edward L., 193
As you know, 43 Bernstein, Basil, 231
Atkins, Bowman, 222 Bever, Thomas G., 107n2, 257n4
Attentiveness-oriented, 110, 113 Biber, Douglas, 203
Atwood, Margaret, 333 Binary grammaticality, 85–86
Audience Bipartite syntax, 106–107
ceremonial functions, 190 Black English, 62–63
classroom lecture, 190 Blackwell, Susan, 219
eavesdropper, 190 Bloomfieldian heresy, 257n4
interpretation, Cooperative Principle Bodily autonomy, 303
violations, 195 Bonfire of the Vanities (Wolfe), 351
participation, persuasive discourse, 195 Borkin, Ann, 60
persuasive discourse, 189–190 Boswell, James, 156–157, 170
persuasive discourse, participation, 195 Boyd, Julian, 29
[ 386 ] Index
387
Brainwashing, 186–187 deep structure, 74, 74n5, 106,

Brame, Michael K., 74n5 242–244, 248, 249, 251, 260,
Brandeis, Louis, 319 273, 286, 289, 373
Bresnan, Joan, 74n5 Extended Standard Theory, 249,
Broverman, Donald M., 100, 113 271, 275
Broverman, Inge K., 100, 113 generative semantics, 241–293
Brown, Roger, 45, 48 (See also Generative semantics (GS))
Browne, Susan E., 306 Goodman’s critique, 95
Bruderschaft, 124 influence, on students, 243
Bryant, Barbara S., 324n18 on Katz-Postal hypothesis, 260
Bucholtz, Mary, 1 Language and the mind, 243, 248
Bureaucratese, 212 on linguistic behavior
Butler, Rhett, 117 embodiment, 101
By the way, 43 sentence structure formalisms, 58
Standard Theory, 242
C Syntactic structures, 64
Camaraderie, 113f, 116, 116f–117f transformational grammar, 58, 102
advice books and strategies, 127 “Choosing life,” 337
friendship, 124 Churchill, Winston, 170
males, American, 125, 126 Civil rights movement, 302
males, modern shift to, 125–126, 128 Clarity, 113–115, 113f, 116f–117f
narcissism, 128 American men, 122, 123, 125
Camaraderie-Deference axis, 117–118, 119 politeness and, 42–43
Cameron, Deborah, 134, 179, 221 Clarity/distance. See Distance/clarity
Can, speaker certainty, 28–30 Clarkson, Frank E., 100, 113
Capitalization, 162–163 Classical Transformational Grammar
Capital punishment, 341 (CTG), 242–249
Carroll, John B., 203 Aspects theory, 242–243 (See also
Carter, Jimmy, 171 Aspects of the theory of syntax
Cartesian linguistics (Chomsky), 243 (Chomsky))
Catford, J. C., 71n3 attraction to and rise of, 244
Certaintives, 28–30 autonomous semantic
Certainty, speaker, 28–30 component, 243
Challenged, 300 dichotomy of linguists,
Chaplinsky v. New Hampshire, 320–321, mathematicians vs. humanists,
320n12 247–249, 247n3
Character, 101–102 formulations, vagueness, 246–247
Charming, 90, 92 fragments, 244
Children, on speaking–writing Postal’s paradigm argument, 244–246
relationship, 148–149 as sentence-level grammar, 246
Choice (reproductive rights) Classroom
anti-abortion sloganeering, 337 lecture, 190
choices for, 345 small-group or collaborative talk, 134
real, 346–347 Clinchy, Blythe McVicker, 279n9
Chomsky, Noam. See also Classical Code, restricted, 231
Transformational Grammar (CTG) Cohen v. California, 322–323,
Aspects of the theory of syntax, 242–253 322nn13–15
(See also Aspects of the theory of Coherency, 103, 110
syntax (Chomsky)) Coleman, Linda, 183, 199n6
Cartesian linguistics, 243 Collaborative talk, classroom, 134
Index [ 387 ]
388
Color-discrimination vocabulary, truth-seeking nature, 358–361

women, 90, 92 what confessor needs to know,
Comic strips, 166–171 357–358, 357n5
Command tag-form, 20 Connors, Debra, 306
Commercials, 187. See also Advertising Consent
power relationship, 191 engineering of, 192–193, 200
spontaneity, 164–165 informed, criminal confession,
Communication. See also specific types 316, 355
frequency, 204 Conservatism, women’s language, 91, 92
of ideas, linguistic means on, 38–39 Consistency, 103, 110
language rules, 105–106 Constantives, 317, 317n3
nonlinguistic rules, 105–106, 105n1 Contextual categorizations, 74–75, 75n6
Communications Decency Act, 311 Contextually linked linguistic
Communicative competence, 4, phenomena, 9–10
112–113, 113f Continuum vs. dichotomy, 263–265,
definition, 353 265f, 265n7
full, demonstration of, 356 Conversation. See also specific types
Virgil Reilly case, 354n4, 355–356 nature of, 137–138
Communities, speech openings and closings, 189
individual relationship to, 350 stable form, 189
of law enforcement and criminals, 351 Conversational implicature, 48, 49, 63
Competence Conversationalization, 134–135
communicative, 4, 112–113, 113f, 353, Conversational Logic (Grice), 191
354n4, 355–356 courtroom discourse, 230–231
pragmatic, criminal confession and, Conversational Maxims (Grice), 100,
353–368 (See also Pragmatic 113, 180, 190–192, 194, 358, 361
competence, criminal confession and) persuasive discourses, 180
Computer Conversational rules. See Rules of
home, 147–148 conversation
literacy, 147–150 Conwell, Chic, 206
Conduct vs. expression, 321–322, Co-occurrence, 103–105, 111–112
328–329, 329f Cooke, Michael, 221–222
Cohen v. California, 322–323, Cook-Gumperz, Jenny, 148, 149
322nn13–15 Cook-Gumperz, John, 148
Confession Cooperative Principle
definition, 363, 363n7 advertising, 199
true, 349–350, 352 courtroom discourse, 230–231
Confession, criminal, 353–368. See also familiarity, 195
Pragmatic competence, criminal ordinary conversation, 230
confession and persuasive discourses, 180, 194, 199
communicative competence, Cooperative Principle violations
355–356, 355n4 audience interpretation, 195
confessional discourse, nonegalitarian Manner, 195–198 (See also Manner
nature, 361–362, 361n6 violations, advertising)
felicity, 362–363 Cory, Donald Webster, 206n2
informed consent, 316, 355 Coulthard, Malcolm, 221
Miranda v. Arizona, 355, 355n3, 361, Courtesy manuals, 126
362, 367 Courtroom
recognition of identity and self, in justice, 217–218
362–367, 363nn7–8, 365nn9–10 speech in, admissibility of, 219
[ 388 ] Index
389
Courtroom discourse, 225–234. See also true, 121

“My life in court” women, American, 122–123,
conversation analysis, 230 127–128, 212
Cooperative Principle and Deference–Camaraderie axis, 117–118, 119
conversational logic, 230–231 Definite article, advertising, 196
data, obtaining, 219–220 Description/prescription
interrogation, triadic, 221 distinction, 94–95
legal discourse vs. ordinary Deutero learning, 138
conversation, 220 Dibbell, Julian, 312
minority groups, 222 Differently abled, 300
for non-native speakers, 223 Disability, women and, 295–308, 302,
power inequalities, 221–222 378–381
rhetorical questions, 228 abortion and “right-to-life,” 303–304
ritual and spontaneity, 231–232 bodily autonomy, 303
speech acts, 229–230 dichotomization, 307
sympathy elicitation, 232–233 vs. disabled man, 302
tag questions, 229 double “pedestal,” 303
trial as discourse situation, 227–229 hidden disabilities, 296, 306–307
“women’s language,” 222 ignoring, 299–300
Criminal confession. See Confession, impossible dilemma, 296
criminal linguistic problems, 296–297
Cripple, 300–301 naming and what person becomes,
Cronkite, Walter, 117 295–296
Crouch, Isabel, 179, 221 needs and desires, 304–305
Culture, linguistic relevance, 9–10 parents’ conditioning of, 296
cummings, e. e., 163n2 speaking for themselves, 307
terminology and power of
D naming, 300
Davis, Mike, 219 wholeness, human, 308
Davis, Natalie Zemon, 363n8 words alter reality, 298
Declaration, tag question on, 21, 21n8 Disability-rights movement, 302
Declarative statement, tag questions Disabled
and tag imperatives on, 21–22, category definition, 301–302
21n8, 26 as euphemism, 301
Deep structure (DS), 74, 74n5, 106, ignoring, 299–300
242–244, 248, 249, 251, 260, 273, passive particles and victimization, 301
286, 289, 373 terminology and power of naming,
Defamation law, 310 300–301
opinion vs. factual statement Disabled man, 302
(Milkovich v. Lorain Journal Company Discourse. See also specific types
et al.), 326–327, 326n24 function, linguistic form and, 353
paraphrase, 325–326, general principles, 137
325–326nn22–23 Discourse-learning, 138
responsibility of meaning, 323–324, Discourse types. See also specific types
327–328 cultural preferences, 187
Deference, 113f, 115–116, 116f–117f taxonomy, 184
euphemisms, 116 Displacement, 343, 343n5
expressions, 68 Distance, 113f, 114–115, 116f–117f
hysteric, 118, 119, 127–128 Europe, 13th century, 126
Japanese, 120, 122, 128–129 Germans, 120, 123, 124
Index [ 389 ]
390
Distance (Cont.) English, nonstandard, 61–62. See also

men, American, 122, 123, 125 specific types and topics
men, technical language codes, 212 Ervin-Tripp, Susan M., 203
Distance/clarity, 117 Etiquette manuals, 126–127
men, American, 122, 123, 125 Euphemisms
obsessive-compulsive symptoms, deference, 116
111, 127 disabled, 301
Divine, 90, 92 politeness, 46–47
Doonesbury (Trudeau), 166 vocabulary, 206
“Doubletalk: Sexism in tech talk,” women’s use, 91–92
209–215, 382–383 Explanatory notions, 3
humanities, technologizing, 211–215 Expositives, 317
impersonality, 203–204 Extended Standard Theory (EST), 236,
intelligibility, 209–210 242, 249, 271, 275, 375
introduction, 203–207 masculine nature, 278–279n9
language codes, 210–211 Extralinguistic factors. See also specific
language use as technology, 209 factors and topics
masculinity, 204 contextual and societal concepts, 13,
obfuscation, 210 15–16, 26, 31
speech similarity, social factors, English modals, 3
204–205 grammaticality/acceptability, 89
vocabulary, 209 implicit rules of interpretation, 106
Dougherty, Ray, 271 Japanese particles, 22
Dramatic performances, conversations, linguistic theory, 71
189–190 nonlinguistic rules, 105–106, 105n1
Dubitatives, 75–76 psychology and sociology, 82
I guess, 27–28 relations, inter-individual, 12
Japanese, 76 women’s language, 89–90
verb/particle endings, 27–28
Dubois, Betty Lou, 179, 221 F
Dyadic conversation, 227, 355 Factual statement, vs. opinion (Milkovich
v. Lorain Journal Company et al.),
E 326–327, 326n24
Eades, Diana, 221, 222 Fairclough, Norman, 134
Eavesdropper, 190 False advertising, 310
Eckert, Penelope, 205 Fanshel, David, 131, 133
Egalitarianism, 187 Felicity, 317, 318, 355
Eisenhower, Dwight, 170 Feminization, collaborative or small-
Eissler, Kurt, 271 group talk, classroom, 134
Either/or distinction, 85–86 Fields of discourse, 133
Ekman, Paul, 105n1 Fighting words, 310, 320–323, 328
Ellipses, 163–164 Fillmore, Charles, 75n6
Ellison v. Brady, 324, 324n18, 325 Fine, Michelle, 303, 306
Eloquence, 157 First Amendment, 319. See also Free
Email, on conversation, 148 speech, right to
Emotion Abrams et al. v. United States, 319,
expression, women’s language, 90–92 319–320nn10–11, 321
satisfaction, psychotherapeutic “clear and present danger,” 320–321
conversation, 140 definition, 310
Engineering of consent, 192–193, 200 dichotomization, 327
[ 390 ] Index
391
“fighting words,” 310, 320–323, 328 on psychotherapy, 139, 139n3

flag burning, 323, 323n16 on women as disabled, 304, 304n2
freedom in, 320, 320n12 Friendliness, politeness, 44
harmfulness, words on others, 312 Friendly conversation, reciprocal, 143
illocutionary force, 319, Full pragmatic competence, 355
319–320nn10–11, 321, 323 Fuzziness, hierarchical, 87
interpretation of utterance, Fuzzy categories concept, 55
responsibility for, 320 Fuzzy rules, 55
language as action, 310–311, 312
language crime vs., 310–311 G
marketplace of ideas, 321 Gable, Clark, 117
misogynistic, harassing language, 311 Garfinkel, Harold, 226
pornography transmission, Gazdar, Gerald, 265, 270n8
computer, 311 Geertz, Clifford, 363n8
Schenck v. United States, 319, Gell-Mann, Murray, 205n1
319n8, 327 Gender. See also specific genders
speech-act theory vs., 315–329 and topics
speech and intent to action, 310, on conversation, 133
319, 319–320nn10–11, 322–323, interlocutors’, English, 4–5
322nn13–15 interlocutors’, Japanese, 4–5
speech harms to others, 312 male privilege, 4
wartime speech, 319, 319n8, positioning, 83
320–321 styles, 83
First name use, 125, 125n7 tech talk and, 204
Ford, Marguerite, 48 Generative grammar/syntax, 3
Forethought, written communication, rules, success, 109
151, 153–157 shortcomings, 53–54
Formalism Generative semantics (GS), 4–5, 74,
limitations and constrictions, 235–238
59–60, 59n1 Classical Transformational Grammar
as linguistics proper, 59–60 and, 242–249 (See also Classical
overemphasis and overdependence, Transformational Grammar (CTG))
53–54, 59 early premises, 373–374
sentence structure, 58–59 feminine nature, 278–279n9
syntax, women’s, 55 methods, 237–238
usefulness vs. blind dependence, 59n1 origins, 373–374
Freedom, in First Amendment, values, 237
320, 320n12 Generative semantics (GS), history,
Free speech, right to, 310. See also First 241–293, 378
Amendment basic, 242–243n2, 242–249, 247n3
civil libertarian ideology, 311 conclusion, 290–293, 292n13
extreme stance, 311 failure, why and if, 286–290
First Amendment clause, 319, 320 personal in theoretico-political,
freedom in, 320, 320n12 275–283, 278–281nn9–11
harmfulness, words on others, 312 pragmatics on, 237–238
harms to others, 312 in transformational grammar,
virtual rape, 312 recent, 238
Freud, Sigmund universality and relativity, 283–286,
on ego as rider controlling horse, 144 284n12, 286f
on motivational psychology, 193 valediction, 236–237
Index [ 391 ]
392
Generative semantics (GS), theory, Green, Georgia M., 60

249–275 on children recognization books’
base component, 249–256, 255f styles, 148–149
continuum vs. dichotomy, 263–265, on classical transformational
265f, 265n7 grammar, 60
Katz-Postal hypothesis, 259–262, 261n6 on generative semantics, 237–238
lexicon, 249–252 Greenberg, Joseph, 284
pragmantax vs. autonomy, 257–259, Grice, H. Paul, 4, 18, 137n1, 221
257n4, 259n5 Conversational Logic, 191, 355, 355n4
rift, basis, 266–275, 270n8, 271–272n Conversational Maxims, 100, 113, 180,
transformations, power, 256–257 190–192, 194, 358, 361
VP nonexistence, 252–253 Conversational Maxims, persuasive
word order, 253–256, 255f discourses, 180
Germans on rules of conversation, 42–43,
Bruderschaft, 124 63, 350
distance, 120, 123, 124 Studies in the way of words, 374
particle use, appropriate, 8 theory of implicature, 48, 49, 63, 259
Gilligan, Carol, 279n9 Griswold v. Connecticut, 343, 343n6
Gilman, Albert, 45 Guess, 21n8
Gleitman, Lila, 87 Gumperz, John, 207, 221
Goffman, Erving, 305
Goldberger, Nancy Rule, 279n9 H
Goldflam, Russell, 223 Haley, Jay, 141, 141n4, 144
Golly, 19–20 Hall, Kira, 1
Good human being, cultural concept, 120 Hammer, Muriel, 203
Goodman, Paul, 95 Handicapped, 300
Goodwin, Charles, 218 Harlan, Justice, 322
Goodwin, Marjorie Harness, 218 Healthy human being, male stereotype,
Gordon, David 120, 129
ambiguity, context and, 10n1 Hedges, 45–46
on conversational implicature, 48, 63 social utility, 122n5
normal conversation, women’s use, 91
assumptions, 18–19 Herring, Susan, 311
on speech-act theory, 259 Hierarchical fuzziness, 87
Gothic fiction, 157 Hierarchical grammaticality, 37
Graddol, David, 135 Holmes, Oliver Wendell, 319
Grammar Honorifics, 8, 10–11, 16
elements, 184 (See also specific types definition, 13
and theories) Japanese, 10–11
rule applicability, context, 4–5 Howe, Florence, 306
of style, 101–129 (See also “Stylistic Humanities
strategies within a grammar of style”) definition, U.S. Constitution, 340
Grammaticality, 5 technologizing, 211–215
binary, 85–86 Humility markers, 12–13
hierarchical, 37 Humor, in advertising, 197–198
Grammaticality/acceptability Hymes, Dell, 284
distinction, 87–89 Hysterical style, 102, 110–112
Gravity’s Rainbow (Pynchon), 57–58 deference, 118, 119, 127–128
Greavu, Elena, 70 neurotic, 118
Greek, Classical, particle use, 8 women, 119–120, 119n4, 127–128
[ 392 ] Index
393
I Informed consent, criminal confession,

Identity, 364–365 316, 355
in confession, speaker–performer, 357 Initial capitals, comic strips, 169
in confession, subject–direct object of Intellectual persuasion, 191–192
clause, 364–365 Intelligibility, technical language, 209–210
confessional recognition, self, and, Interactional styles, 133
362–367, 363nn7–8, 365nn9–10 Interjections, English, 71, 71n3
language as creator of, 378–381 Internal-state signaling devices, 67–68
(See also Pragmatics; specific topics) Internet, gender-based asymmetry, 311
in psychological “normality,” 365 Interpersonal communication, grammar
psychopathology continuum, of, 381–382. See also specific topics
365–366, 365n10 Interpretation. See also “My life in court”
Identity marking, 204–205 context and, 4
Ideology, 83 Interpretive semantics, 4
I guess, 27–28, 90, 118 Interrogation. See also “My life in court”;
Illocutionary force, 310 Pragmatic competence, criminal
Abrams et al. v. United States, 319, confession and
319–320nn10–11, 321 Rhode Island v. Innis, 324–325,
definition, 318 324–325nn19–21
First Amendment, 319, Intonation
319–320nn10–11, 321, 323 semantic-pragmatic functions, 70–71
theoretical interest, 318, 318n5 tag statements and questions, 21n7
I mean, 48, 62, 91 verbs of thinking, 21n8
Imperatives women’s language, 91
on declarative statements, 21–22 Italics, 160–162
politeness, 16–18, 17n5 comic strips, 169
tag-forms, 20 ironic detachment, 169
Impersonal expressions I thought, 28
politeness, 45 I wonder, 90
tech talk, 203–204
Implicature J
conversational, 48, 49, 63 Jackendoff, Ray, 247n3
Grice’s theory, 48, 49, 63, 259 James, Deborah, 60
rules, 191 James, Henry, 65
Implicit rules, linguistic, 109 Japanese
Implicit rules of conversation deference, 120, 122, 128–129
assumptions, 18–19, 19n6 dubitatives, 76
violations, 19–22, 21nn7–8 honorifics, 8, 10–11
Indeterminacy, 339, 373, 375 ne, 21n7, 22–24, 22n9, 26, 27n11
fear and loathing of, 291 o-, 12
in generative semantics, 378–379 particle use, 8, 22–27, 22n9, 24n10
in grammar, 376 politeness, 68
hard sciences on, 291 -san, 11, 12
Indirections, social utility, 122n5 women’s vernacular literature, 215
Inequality, 187. See also Asymmetry; yo, 22, 22n9, 24–27, 24n10, 27n11
Power imbalances; specific Jargon, technical, 209, 210–211. See also
topics Vocabulary, social factors
language and, 311 Jefferson, Gail, 137n1, 355n4
Informal conversation, rules, 140 Jesperson, Otto, 91
Informalization, 134–135 Johanson, Donald C., 271, 271–272n
Index [ 393 ]
394
Johnson, Barbara, 363n8 on hedges, 45

Johnson, Samuel, 156, 170 on performance/competence
Joyce, James, 170 distinction, 77
on speech-act theory, 259
K “Toward generative semantics,” 246
Kairys, David, 319 Langacker, Ronald, 256
Katz, Jerrold J., 257n4, 259–262, 261n6 Langendoen, D. Terence, 107n2
Katz-Postal hypothesis, 259–262, 261n6 Language. See also specific topics
Kay, Paul, 199n6 as action, 310–311, 312
Keller, Evelyn Fox, 279n9 beliefs about ourselves and, 350
Kennedy, Edward, 170–171 creating reality, 317n3, 368
Kennedy, John F., 170–171 creation of, intrapsychic
Kennedy, Robert, 170–171 processes, 379
Kilpatrick, James J., 326n23 as creator of psychosocial identity,
Kinda, 118 378–381 (See also specific topics)
King, Rodney, beating and trials, inequality and, 311
218–219, 218n2 misogynistic, harassing, 311
Kitagawa, Chisato, 22n9 sociology of, 382–383 (See also specific
Klein, Ewald, 265, 270n8 topics)
Klima, Edward, 87 universal functioning, 316
Knoblauch, Hubert, 148 Language and the mind (Chomsky),
Kubler-Ross, Elisabeth, 274–275 243, 248
Kuhn, Thomas, 73n4 Language and woman’s place (Lakoff, R.),
paradigm, 245 1–2, 55, 81–83, 179, 379
on science, 270 Language and woman’s place: Text and
on theoretical disputes, 237, 237n6 commentaries, 1
Language codes, 210–211
L male, distancing, 212
Labov, William male technical, on women and
on conversational logic rules and minorities, 212–213
class/ethnicity, 69 “Language in context,” 3–36,
on judgments of own speech 375–377, 381
patterns, 90 contextually linked linguistic
on Martha’s Vineyard young peoples’ phenomena, 9
vowels, 205 culture and linguistic relevance, 9–10
phonological innovation, 91 dubitatives, verb/particle
on sociolinguistics, early, 4 endings, 27–28
Therapeutic Discourse, 131, 133 honorifics, 8, 10–13, 16, 16n4
on women’s linguistic innovation, 91 humility markers, 12–13
Laing, R. D., 64 imperatives, 16–18, 17n5
Lakoff, Andrew, 365n9 implicit rules of conversation,
Lakoff, George, 10n1 assumptions, 18–19, 19n6
on adjectives, 252 implicit rules of conversation,
on assumptions, rules of violations, 19–22, 21nn7–8
conversation, 18 McConnell-Ginet on, 3–6
on conversational implicature, 48, 63 modals, English, 3, 11–15, 14n3
on generative semantics, 74 modals, English, paraphrases (must/
on governed rules and major and have to), 30–31
minor rules, 256 modals, English, verbs of perception
as GSist, 247n3 (can), 28–30
[ 394 ] Index
395
modals, politeness conventions, Like, 48, 62, 92

11–15, 14n3 Linde, Charlotte, 363n8
ne, 21n7, 22–24, 22n9, 26, 27n11 Linguistics. See also specific topics
origins, 3 adversarial argumentation,
particle use, speaker gender, 26 272–273
performative verb, explicit definition, 315
presence, 25–26 description and prediction, 315–316
politeness conventions, 11–16, 12n2 as functioning, 317
presuppositions, 16 as interpretive discipline, 315
statements vs. questions vs. orders, at M.I.T. (Chomsky era), 372 (See also
10, 10n1 “The way we were”)
verb endings/forms, 8 scientific method and,
will-deletion (-insertion), 30 269–272, 270n8
yo, 22, 22n9, 24–27, 24n10, 27n11 as social science, 315
Language-particular grammatics, 4 “Linguistic theory and the real world,”
Larkin, Don, 30–31, 67n2 67–78, 375–377. See also specific
Lasch, Christopher, 128 topics
Last name use, 125, 125n7 introduction, 53–56
Law Literacy
as interpretive discipline, 315 children’s, textuality, 149
language, 225–234 (See also “My life diminishing, 152–153, 171
in court”) vs. nonliteracy, judgment,
pragmatics, 315–329 (See also 151–153, 171
“Pragmatics and the law: Speech Literature language, 65
act theory confronts the First Luborsky, Lester, 143n6
Amendment”) Luborsky, Lise, 143n6
as social science, 315
Lawler, J., 60 M
Law practitioners, inequalities, 217 MacGraw, Ali, 160
Leakey, Mary, 271, 271–272n Mailer, Norman, 157
Leave-taking, 140 Mairs, Nancy, 305, 306
Lecture Malcolm, Janet, 325–326,
audience, 190 325–326nn22–23
Grice’s maxims, 191 Manner, 195
nonreciprocal, 185 courtroom discourse, 231
power relationship, 191 Manner violations, advertising,
psychotherapeutic conversation, 141 195–198, 197n7
Lees, Robert B., 272 definite article, 196
Legalese, 231–232. See also Courtroom humor, 197–198
discourse lexical novelty, 195, 198
Lexical novelty, 195, 198 morphological novelty, 195–196
Lexicon, generative semantics theory, persuasive discourse and ordinary
249–252 conversation, 183–201
Libel pragmatic novelty, 197–198,
definition, 323 198n5
paraphrase, 325–326, semantic anomaly, 196
325–326nn22–23 subjects, 196, 196n4
“Life” syntactic innovation, 196
choosing, 337 syntactic novelty, 195–196, 198
hypocrisies of, 341–342, 341n verbal auxiliaries, 196, 196n4
Index [ 395 ]
396
Manner violations, Cooperative Principle Minority groups

advertising, 195–198, 196n4, 197n7 courtroom discourse, 222
humor, 197 language use, 61–63
Manner violations, political rhetoric, 198 Miranda v. Arizona, 355, 355n3, 361,
Mansfield, Michael, 220 362, 367
Marketplace of ideas, 321 Misogyny, defense of, 311
Masson, Jeffrey Moussaieff, 325–326, Mitford, Jessica, 206
325–326nn22–23 Modals, English, 3, 11–15
Masson v. New Yorker Magazine, 325–326, may, 11, 13, 15–16
325–326nn22–23 may/be allowed, 31
May, 11, 13, 15–16 must, 11, 13–15, 14n3, 16n4
May/be allowed, 31 paraphrases (must/have to), 30–31
McAlinden, Fiona, 179, 221 should, 11, 13, 15
McCawley, James D., 247n3, 251 speaker certainty, 28–30
on Aspects DS, 258 speaker responsibility, 30–31
on VSO, 253–254 verbs of perception (can), 28–30
McFadden, Cyra, 144 will/be to, 31
McLuhan, Marshall, 154, 172 Monroe, Marilyn, 117–118
Meaning, responsibility of Morgan, Jerry L.
defamation law, 323–324, 327–328 on generative semantics, 236,
sexual harassment (Ellison v. Brady), 237nn3–4, 287
324, 324n18 on generative semantics collapse,
Meiklejohn, Alexander, 321 236, 287
Memory, short-term, nonliterate on selectional restrictions
cultures, 171–172 violations, 88
Men Morphological novelty, 195–196
formalism perspective, 278–279 Morrow, Lance, 125n6
gender privilege, 4 Motivational psychology, 193
healthy human being stereotype, Multiple personality disorder, 365n10
120, 129 Must, 11, 13–15, 14n3, 16n4
language, vs. women’s Must/have to, 30–31
language, 89–90 “My life in court,” 225–234, 382–383.
personal style, 122, 123 See also Courtroom discourse
personal style, German vs. American, author’s account, court duty and case,
123–125 225–227
speaker, context, 95–96 conversation analysis, 230
technical language codes, 212–213 Cooperative Principle and
Men, American conversational logic, 230–231
camaraderie, 125, 126 introduction, 217–223
clarity, 122, 123, 125 ritual and spontaneity, 231–232
distance/clarity, 122, 123, 125 speech acts, 229–230
Metacommunication, 144 sympathy elicitation, 231–232
Metaphor, psychotherapeutic trial as discourse situation, 227–229
conversation, 144
Michaels, Sarah, 150 N
Milkovich v. Lorain Journal Company Naming
et al., 326–327, 326n24 power, 300–302
Miller, Roy A., 70 on what person becomes, 295–296
Milne, A. A., 162–163, 163n2 NARAL Pro-Choice America, 333
Milroy, Lesley, 203 Narcissism, camaraderie, 128
[ 396 ] Index
397
Narcissistic characters, 128 Noonan, Peggy, 340

National Abortion Rights Action League Normal language, vs. women’s
(NARAL), 333 language, 119
Ne, 21n7, 22–24, 22n9, 26, 27n11 Novelty
nee, 21n7 lexical, 195, 198
Neologism, 195, 198 morphological, 195–196
Neural, 205 ordinary conversation, 194–195
Neuronal network, 205 persuasive discourse, 189, 192
Neurotic hysterical style, 118, 365n10 pragmatic, 197–198, 198n5
Neurotic obsessive-compulsive style, syntactic, 195–196, 198
118, 128
Neurotic styles (neuroses), 102, 118–120, O
119n4, 128–129 o-, 12, 16, 16n4
definition, 120 O’Barr, William, 222
women vs. men, 120 Obfuscation, 210
Neutral language, 83 Obscurantism, cultivating, 214
Newman, Edward, 91 Obsessive-compulsive style, 110
Newmeyer, Frederick, 270n8 distance/clarity, 111, 127
on formalism as male perspective, masculine behavior, 118–120, 119n4
278–279 neurotic, 118, 128
on generative semantics, 236, 241–242 O’Leary, Kathy, 179, 221
on generative semantics, Olson, David R., 172–173
salesmanship, 289 O’Malley, Michael H., 67n2
on generative semantics, stylistic O’Meara, John, 237n2
traits, 281n11 Opinion vs. factual statement (Milkovich
on generative semantics as fuzzy, 259n5 v. Lorain Journal Company et al.),
Nicknames, intimacy and number 326–327, 326n24
of, 48 Oral communication
Nineteen eighty-four (Orwell), 339 capitalization, 162–163
Nixon, Richard, 106, 158, 171 comic strips, 166–171
Noncohesion, 365n10 italics, 161–162
Nondirectness, women’s language, primacy, 171–173
90–92, 95–96 quotation marks, 159–160
Nonegalitarian, 187 strategies, 151
Nonfluencies, 163–165 vs. written communication, as
Nonfree goods spontaneity vs. forethought, 151,
generosity, 50 153–157, 170
politeness, 44–45 Ordinary conversation
Nonliteracy definition, 177
vs. literacy, judgment, 151–153 emotional satisfaction, 140
new, 173–175 Gricean maxims, 191
short-term memory, 171–172 Gricean maxims, violating, 194
storytelling, 172 informal, 178
Nonreciprocal conversation, 132–133, leave-taking, 140
138, 187 novelty, 194–195
lecture, 185 permissible discourse, 140
psychotherapeutic, 132–134, 141–142 persuasion vs., 185
unilateral, 188 power imbalances, 178
Nonreciprocity, persuasive discourse, 185 in psychotherapy, 131–132, 139–140
Nonstandard English, 61–62 Rapport rules, 199
Index [ 397 ]
398
Ordinary conversation (Cont.) Personality, 101–102

Rapport vs. Cooperative Principle, Personal style, 101–105. See also Style,
199–200 personal
rationality, acknowledging, 140 Persuasion, 186
reciprocal, 132–133, 138, 143, 178, Persuasive discourse
185–188 changes over time, 192
ritual and custom, 178–179 Cooperative Principle, 199–200
as spontaneous, 179 definition, 186
vignette, 177–178 intellectual, 191–192
Orwell, George, 34, 339 “Persuasive discourse and ordinary
conversation, with examples from
P advertising,” 183–201, 382–383
Paralinguistic behavior, 68–71, 106 advertising, 191
nonlinguistic rules, 105–106, 105n1 audience (vs. addressee), 189–190
Paraphrase, 108, 325–326, audience participation, 195
325–326nn22–23 brainwashing, 186–187
modals (must/have to), 30–31 commercials, 187
Partee, Barbara, 261 Cooperative Principle, 180, 194, 199–200
Partial-Birth Abortion Act, 331–332 deeper purpose, of discourse, 184
Particle endings, dubitatives, 27–28, 27n11 definition, 186
Particle use Dorothy Sayers, 183–184
appropriate, 8 vs. egalitarianism, 187
ne and yo (Japanese), 8, 21n7, 22–27, emotional persuasion, 192
22n9, 24n10, 27n11 engineering of consent, 192–193, 200
speaker gender, 26 Gricean maxims, 180, 190–192, 194
Passive particles, victimization and, 301 Gricean maxims, violating, 194–195
Passives, politeness, 45 humor, 197–198
Pathological language, 64 intellectual persuasion, 191–192
Pauses, vocalized, 163–164 introduction, 177–181
Performance/competence distinction, Manner violations, 195–198,
77, 85 196n4, 197n7
Performance utterances, hedges, 45–46, means of persuasion, 191–192
91, 122n5 motivational psychology, 193
Performative/constative split, 317, 317n3, nonreciprocity, 185, 187–188
329, 329f nonverbal means, 192–193
Performatives, 317, 317n3 novelty, 189, 192, 194–197 (See also
felicity, 317, 318 specific types)
Performative speech acts, 310, 311, persuasion vs. ordinary
317–318, 328 conversation, 185
creating reality, 317n3, 368 political propaganda, 187
Performative verb, 318 political rhetoric, 191
explicit presence, 25–26 power imbalance, 187–188
superficial structure, 256 power relationship, 190–191
Perlocutionary force, 310, 311, 318, 323 propaganda, 191, 193–194, 193n4
Permissible discourse Quality violations, 198–199
ordinary conversation, 140 Rapport, 199
psychotherapy, 141 spontaneity, 188–189
Person types, 186
anti-abortion sloganeering, 339, 339n3 words vs. images, power and
definition, U.S. Constitution, 340 regulation, 192–193
[ 398 ] Index
399
Persuasive discourses hierarchical grammaticality, 37

Cooperative Principle, 180, 194 Japanese vs. Americans, 68
definition, 177 Make A feel good–be friendly,
ever-present, 179 44, 48–51
Gricean maxims, 180 options, 44
power imbalances, 178 passives and impersonal
Phil Donahue show, 167n3, 188n1 expressions, 45
Phonological innovation, women, 91 please, 40–41
Pickpockets’ vocabulary, 206 pragmatic ambiguity, 40–42
Pitch, women’s language, 91 pragmatic content, on
Plaintext: Essays by Nancy Maris (Mairs), acceptability, 39–42
305, 306 pragmatic presuppositions, 38
Planned Parenthood, 333–334, 346 real world linguistics, 63–64
Please, 12n2, 40–41 rude utterances, 50–51
“Pluralism in linguistics,” 57–66, 375–377 Rules of Conversation, 42–44, 49
Chomsky, 53–54 Rules of Politeness, 44–51
formalisms, 58–60, 59n1 Rules of Pragmatic Competence, 42
growth, 65–66 subcultural systems, 69
informal work, 59–61 syntactic ambiguity, 40
introduction, 53–56 syntactic phenomena, 63
literature, advertising, and Politeness conventions, 11–16, 12n2
propaganda, 64–65 honorifics, 8, 10–13, 16, 16n4
minority language use, 61–63 imperatives, 16–18, 17n5
pathological and aberrant modals, 11–15, 14n3
language, 64 Political correctness, 323
politeness and its difficulties, 63–64 Political discourse
real world linguistics, 61 Manner violations, 198
sentence structure formalisms, 58–59 Quality violations, 198
transformational grammar, 58 Political propaganda, 187
women and, 66 Political rhetoric, 191
Poetry, 174 Manner violations, 198
Polgar, Sylvia K., 203 Portnoy, Stephanie, 203
Police procedures Postal, Paul M.
deaths during arrests and custody, 217 on crossing quantifiers, 262
inequalities, 217–218 CTG paradigm argument, 244–246
Politeness, 37–56, 68, 375–377, 381–382 on deep structure, 273
actions, 49 Katz-Postal hypothesis,
assumptions of speakers, 37–39 259–262, 261n6
clarity, 42–43 Power imbalances, 178, 187–188
communication of ideas, linguistic courtroom, 221–222
means on, 38–39 persuasive discourses, 178
conversational implicature, 49 Power relationship, 191
cultural differences, 49–50 advertising and commercials, 191
cultural systems, 68–69 lecture, 191
deferential expressions, 68 ordinary conversation, 178
Don’t impose, 44–51 psychotherapeutic conversation, 191
euphemisms, 46–47 women’s language, 55
friendliness, 44 Pragmantax vs. autonomy, 257–259,
Give options, 44, 45–51 257n4, 259n5
hedges, 45–46 Pragmatic ambiguity, 40–42
Index [ 399 ]
400
Pragmatic competence, criminal “fighting words,” 310, 320–323, 328

confession and, 353–368 First Amendment, 319–320, 319n8, 327
communicative competence, 354n4, First Amendment, Schenck v. United
355–356 States, 319, 319n8, 327
conclusions and dilemmas, 367–368 illocutionary force, 310, 318, 318n5
confession, criminal, truth-seeking interrogation, 324–325,
nature, 358–361 324–325nn19–21
confession, criminal, what confessor introduction, 309–313
needs to know, 357–358, 357n5 language, universal functioning, 316
confessional discourse, nonegalitarian linguistics vs. law, 315–316
nature, 361–362, 361n6 marketplace of ideas, 321
confessional recognition of identity Masson v. New Yorker Magazine,
and self, problem, 362–367, 325–326, 325–326nn22–23
363nn7–8, 365nn9–10 Milkovich v. Lorain Journal Company
facts of case, 354–355, 354–355nn1–3 et al., 326–327, 326n24
introduction, 349–352 opinion vs. factual statement,
language, beliefs about ourselves, 350 326–327, 326n24
Miranda v. Arizona, 355, 355n3, 361, performative/constative split, 317,
362, 367 317n3, 329, 329f
speech communities, 350–351 pragmatics, 318–319, 318n6
true confession, 349–350, 352 “reasonable man” standard, 324
truth-seeking discourse genre, responsibility, for communicative
357, 357n5 behavior, 323, 323n17
Pragmatic Competence, Rules of, 42 responsibility of meaning, defamation
Pragmatic content, on law, 323–324, 327–328
acceptability, 39–41 responsibility of meaning, sexual
Pragmatic novelty, 197–198, 198n5 harassment, 324, 324n18, 325
Pragmatics, 4–5, 38–41, 379 Rhode Island v. Innis, 324–325,
acceptability, context, 39–42 324–325nn19–21
definition and content, 318–319, Schenck v. United States, 319, 319n8, 327
318n6, 353 speech-act theory, 316–318, 317n3
linguistic form and discourse speech and intent to action, 310,
function, 353 322–323, 322nn13–15
universal information categories, 55 theory in linguistics vs. law,
“Pragmatics and the law: Speech 316, 316n2
act theory confronts the First Pragmatics-cum-sociolinguistics,
Amendment,” 315–329, 382–383 381–382. See also specific topics
“clear and present danger,” 320–321 Pragmatics–semantics interface,
Cohen v. California, 322–323, 378–381. See also specific topics
322nn13–15 Predicate (VP), generative semantics
conduct vs. expression, 321–322, theory, 252–253
328–329, 329f Pressley, Sue Anne, 351
defamation law, paraphrase, 325–326, Presuppositions, 16, 18
325–326nn22–23 pragmatic, 38
dichotomization, 327 Primary-process thought, 193
dichotomous terms, on appellate court Prison inequalities, 217, 217n1
decisions, 321–322 “Private” face-to-face conversation, 134
dichotomous terms, on jury “Pro-choice,” 333, 333n3
decisions, 321 “Pro-life,” 333, 333n3, 341–342
Ellison v. Brady, 324, 324n18, 325 as absolutist, 341–342
[ 400 ] Index
401
capital punishment and, 341 Quantity, courtroom discourse, 231

contraception and sex education vs., 342 Quark, 205n1
fear and, 342–345, 343n6 Question. See also Tag questions
Iraq War vs., 341, 341n vs. declarative, women’s use, 91
violence and, 342 intonation, 21n7
Propaganda, 191 tag-form, 5, 20–21, 21n7
advertising as, 194 Quotation marks, 158–160, 159n1
definition, 193n3
etymology, forms, and meanings, R
193–194, 193n4 Rampton, M. Benjamin, 205
language, 65 Rape, virtual, 312
political, 187 Rapport-oriented, 110–111
reproductive rights, 335–336, 336n1 Rapport rules, 199
Prose, 174 changing, 125–127
Psychotherapeutic conversation Rationality, acknowledging, 140
metaphor, 144 Reagan, Ronald, 266
nonreciprocal and bilateral, 188 Reality, language creating, 317n3, 368
persuasiveness, 187 Really, 19–20
power relationship, 191 Real world, linguistic theory and, 67–78.
reciprocity, superficial, 186 See also “Linguistic theory and the
semipersuasive, 178, 180 real world”
Psychotherapeutic discourse. See applied linguists, 71–72, 76–77
also “When talk is not cheap: contextual categorizations, 74–75, 75n6
Psychotherapy as conversation” cultural differences, 68–69
empirical analyses of, 131 cultural differences, psychological
fields of discourse, 133 assumptions, 76
heterogeneity, 133 deep structure, assumptions, 74
interactional styles, 133 (See also Deep structure (DS))
Lakoff’s intuitive approach, 131–132 dubitatives, 75–76
metacommunication, 144 generative semantics, 74
nonreciprocal, 132–134, 141–142 internal-state signaling devices, 67–68
ordinary conversation, 131–132, intonation, semantic-pragmatic
139–140 functions, 70–71
on other conversation forms, linguistic theory vs. applied linguistics,
132–134, 139 71, 77–78
“private” face-to-face conversation, paralinguistic behavior, 68–71
made public, 134 performance/competence
reciprocal, 132–133 distinction, 77
rules, learning, 143 politeness and formality
as struggle for power, 141n4 devices, 68–69
symbiotic relation, 132 sociological information, 68
Pullum, Geoffrey, 238 subcultural differences, 69
Pynchon, Thomas, 65 theorists’ isolation, 73
The Crying of Lot 49, 168 theory vs. real world, 71–72
Gravity’s Rainbow, 57–58 transformational grammar vs., 73–74
well and pitch-patterns, 70
Q Real world linguistics, 61–65
Quality, 180 literature, advertising, and
Quality violations, 198–199 propaganda, 64–65
advertising, 198 minority language use, 61–63
Index [ 401 ]
402
Real world linguistics (Cont.) Restricted code, 231

pathological and aberrant language, 64 Retrograde amnesia, 365n10
politeness and its difficulties, 63–64 “Review essay: Women and disability,”
“Reasonable man” standard, 324 299–308, 378–380, 382. See also
Reciprocal conversation, 138, 188 Disability, women and
bilateral, 188 introduction, 295–298
definition, 188 naming and what person becomes,
dyadic, 227 295–296
egalitarian, 185 terminology and power of naming,
friends, 143 300–301
implications, 187 Rhetorical questions, courtroom
ordinary conversation, 132–133, 143, discourse, 228
178, 185–186 Rhode Island v. Innis, 324–325,
psychotherapeutic conversation, 324–325nn19–21
132–133, 186 Right-to-life. See Reproduction, rhetoric of
Recontextualizing, 144–145 Ritual, spontaneity and, 231–232
Repetitions, 163–164 Rodney King, beating and trials,
Reproduction, rhetoric of, 331–352, 218–219, 218n2
382–383 Roe v. Wade, 331–333, 343–344
abortion rate, worldwide, 331–332 Roosevelt, Franklin D., 170
“abortion rights,” 333 Rosenkrantz, Paul S., 100, 113
“anti-abortion,” 333 Ross, John Robert (Haj)
anti-abortion sloganeering, 332 on auxiliaries as main verbs, 264
Center for Reproductive Rights, 333 GF version of DS mobile, 254, 255f
choice, a real, 346–347 islands of, 256
choices for “choice,” 345 Linguist’s National Anthem, 263
introduction, 331–334 on squishes, 259n5, 263, 318n4
“life,” hypocrisies of, 341–342, 341n on superficial structure of
modest proposal, 345–346 performative verb, 256
National Abortion Rights Action on verbs of thinking, 21n8
League, 333 Rule of Communicative Competence,
Planned Parenthood, 333–334, 346 112–113, 113f
“pro-choice,” 333, 333n3 between-modality style, 116, 116f
“pro-life,” 333, 333n3 camaraderie, 113f, 116, 116f–117f,
reproductive freedom, 333 124–126, 128
Roe v. Wade, 331–333, 343–344 clarity, 111, 113–115, 113f, 116f–117f, 127
semantic flash vs. honest talk, Clarity–Distance axis, 117
335–336, 336n1 deference, 113f, 115–116, 116f–117f,
semantic obfuscation, 338–341, 339n3 118–122, 127–129
sloganeering and slick ads, Deference–Camaraderie axis,
336–338, 338n2 117–118, 119
slogans, fear behind, 342–345, 343n6 distance, 113f, 114–115, 116f–117f,
Reproductive freedom, 333 120, 121–126
Responsibility style combinations, 116–117, 117f
for communicative behavior, 323, 323n17 Rules of Conversation, 42–44, 355
speaker, 30–31 implicit, assumptions, 18–19, 19n6
Responsibility of meaning implicit, violations, 19–22, 21nn7–8
defamation law, 323–324, 327–328 informal conversation, 140
sexual harassment (Ellison v. Brady), persuasion, 192
324, 324n18, 325 politeness, 42–44, 49
[ 402 ] Index
403
Rules of Politeness, 44–51 Semantics, 379

Don’t impose, 44–51 Semantics–syntax interface, 3–4
Give options, 44, 45–51 Semipersuasive discourse, 178, 180
Make A feel good–be friendly, Sentence structure, formalisms, 58–59
44, 48–51 Sexual harassment, responsibility of
Rules of Pragmatic Competence, 42 meaning (Ellison v. Brady), 324,
Rules of Rapport, 194 324n18, 325
ordinary conversation, 199–200 Shakespeare, 156
Rycroft, Charles, 343n5 Shapiro, David
attentiveness continuum, 110, 113
S hysterical style, 102, 110–112,
Sacks, Harvey, 137n1, 203, 355n4 118–120, 119n4, 127, 128
Sadock, Jerrold, 236, 287 obsessive-compulsive style, 110, 111,
Salzinger, Kurt, 203 118–120, 127
-san, 11, 12 on personal style, 100–103
Sapir-Whorf hypothesis, 298 Shepard, Alicia C., 333n3
Sass, Louis A., 363n8, 365n10 Sherzer, Joel, 179
Saxton, Marsha, 306 Short-term memory, nonliterate
Sayers, Dorothy L., 183–184 cultures, 171–172
Schafer, Roy, 122n5, 144, 144n7 Should, 11, 13, 15
Schegloff, Emanuel, 137n1, 221, 355n4 #ShoutYourAbortion, 334
Schenck v. United States, 319, 319n8, 327 Siegler, David, 179
Schizophrenia, 365n10 Siegler, Robert, 179
Schizophrenics’ language, 63 Sinclair, John, 221
poetry and, 105 Singer, Barton, 143n6
Science Slander, 323
doing vs. playing, 270n8 Slick ads, reproductive rights,
social sciences and linguistics as, 336–338, 338n2
269–272 Slogans, reproductive rights,
Searle, John R., 221, 316, 355n4 336–338, 338n2
Secondary-process thought, 193 fear behind, 342–345, 343n6
“Secret handshake,” 211 Small-group talk, classroom, 134
Self-distancing, rhetorical, 351–352 So, 90
Self(hood) Soap opera, spontaneity and
nature and reality, 353 nonfluencies, 164
normal humans, 363 Social factors, vocabulary, 204–206.
Self(hood), concept See also Vocabulary, social
construct, dubious, 368 factors
identity, 364 Social network, 204
language, 364 Social sciences
real life, 364 adversarial argumentation, 272–273
scholars’ questioning, 363–364, linguistics and law as, 315
363n8 as science, 269–272, 270n8, 271
Semantic anomaly, advertising use, 196 Sociolinguistics, 4–5, 379, 382–383
Semantic competence theories, 3 applied, 309
Semantic flash vs. honest talk, Sociology of language, 382–383. See also
335–336, 336n1 specific topics
Semantic obfuscation, reproductive Socio-syntax, 53
rights, 338–341, 339n3 Some, 87
Semantic representations, 4 courtroom discourse, 228
Index [ 403 ]
404
“Some of my favorite writers are literate: reciprocity and bilaterality, 189

The mingling of oral and literate ritual and, 231–232
strategies in written communication,” soap operas, 164
151–175, 378–383 Squishes, 259n5, 263, 318n4
capitalization, 162–163 Standard Theory (Chomsky), 242. See
comic strip, 166–171 also Classical Transformational
introduction, 147–150 Grammar (CTG)
italics, 160–162 Starn Randolph, 363n8
literacy, diminishing, 152–153, 171 Statement
literacy vs. nonliteracy, judgment, intonation, 21n7
151–153, 171 tag-form, 20–21, 21n7
nonfluencies, 163–165 Statement-and-plus question tag-form,
nonliteracy, new, 173–175 21, 21n8
oral modes, primacy, 171–173 Stereotypes
quotation marks, 158–160, 159n1 study, 121
spontaneity vs. forethought, 151, women, 120–121
153–157 Stern, Nanci, 306
transferring spoken discourse to Stevenson, Adlai, 170
writing, 157–158 Stigma (Goffman), 305
written vs. oral communication, 151–153 Stockwell, Robert, 281n11, 284n12
Speaking–writing relationship, children Storytelling, 149
on, 148–149 nonliterate cultures, 172
Special effect language, 64–65 Strands, Lakoff’s work, 374–375. See also
Speech. See also First Amendment specific topics
unprotected, 310 pragmatics-cum-sociolinguistics,
Speech acts. See also specific types 381–382
definition, 317 pragmatics–semantics interface,
felicity, 355 378–381
Speech-act theory theory wars, 375–377
Austin in, 315–318, 317n3 Stubbs, Michael, 311
vs. First Amendment, 315 Studies in the way of words (Grice), 374
free speech vs. language as action, Style, personal, 101–105
310–311, 312 attentiveness, 110, 113
illocutionary force, 310, 318, 318n5, coherency, 103–104, 110
319, 319–320nn10–11, 321, 323 consistency, 103, 104, 110
Lakoff (R.) on, 318 co-occurrence, 103–105, 111–112
origins and basic claims, 316–317 men, 122, 123
performative/constative split, 317, men, German vs. American, 123–125
317n3, 329, 329f rapport, 110–111
performative speech acts, 310, 311, 328 Shapiro on, 102–103
pragmatics, 318–319 women, 120–123
Speech communities women vs. men, 120–123
individual relationship, 350 Styles of gender, 83
law enforcement and criminals, 351 “Stylistic strategies within a grammar of
Speech similarity, social factors, 204–205 style,” 101–129, 378–381
Spoken discourse, to writing, 157–158 ambiguities, 107
Spontaneity attentiveness continuum, 110, 113
commercials, 164–165 behaviors, character, personality,
oral communication, 151, 153–157, 170 personal style, 101–102
ordinary conversation vs. lecture, 188 behaviors, culturally appropriate, 124
[ 404 ] Index
405
bipartite syntax, 106–107 Surface structures (SS), 242–243

camaraderie, 113f, 116, 116f–117f, Sutherland, Edwin H., 206
124–126, 128 Swann, Joan, 135
clarity, 111, 113–115, 113f, Symbiosis, psychotherapeutic
116f–117f, 127 conversation, 132, 141
coherency and consistency, Sympathy, courtroom elicitation, 232–233
103–104, 110 Syntactic ambiguity, 40
co-occurrence, 103–105, 111–112 Syntactic innovation, 196
deference, 113f, 115–116, 116f–117f, Syntactic novelty, 195–196, 198
118–122, 127–129 Syntactic structures (Chomsky), 64
distance, 113f, 114–115, 116f–117f, Szasz, Thomas, 142n5
120, 121–126
dualistic behavior, 108 T
grammar of style, 109–110 Tag command, 20
healthy woman paradox, 120–121, 129 Tag-forms. See also specific types
hysterical style, 102, 110–112, 118–120, intonation, 21n7
119n4, 127, 128 Tag imperatives, 20
intention and execution match, 110 on declarative statement, 21–22,
introduction, 99–100 21n8, 26
men, American, 122–125, 126 politeness, 63
men, German, 123–125 Tag questions, 5, 20–21, 21n7
narcissistic characters, 128 context, 221–222
neuroses, 102, 118–120, 119n4, 128–129 courtroom discourse, 221–222, 229
nonlinguistic rules, 105–106, 105n1 on declarative statements, 21–22,
obsessive-compulsive style, 110, 111, 21n8, 26
118–120, 119n4, 127 politeness, 63
paraphrases, 108 women’s vs. men’s use, 179
personal style, 101–103 Tag-statement-and-plus question, 21, 21n8
psychic economy, 105 Tag statements, 20–21, 21n7
rapport, 110–111 Talking power: The politics of language
rapport rules, changing, 125–127 (Lakoff, R.), 179, 180
Rule of Communicative Competence, Tannen, Deborah, 180
112–113, 113f Tarule, Jill Mattuck, 279n9
Shapiro’s taxonomies, 110–111 Technical jargon (terminology), 209,
stereotype vs. cultural style, 123–124 210–211. See also Vocabulary,
stereotype vs. gender style, 122–123 social factors
style changes, for both genders, 128–129 Technical writing, 203–215. See also
style changes, men, 125–126, 128 “Doubletalk: Sexism in tech talk”
style changes, women, 128 badge of pride, 207
stylistic behavior, 110 doctors, 207
target strategies, culture and gender, emphatics and hedges, 204
121–122 vs. euphemisms, 46–47
transformational rules, 107, 107n2 gender and, 204
women, language, 102 impersonality, 203–204
women, language, vs. children’s masculinity, 204
language, 108 vocabulary, 204–206
Subjects, advertising, 196, 196n4 for women, 207
Subliminal advertising, 192 Technologization, of language, 209–215
Such, 90 Texas v. Johnson, 323, 323n16
Suppose, 21n8 Text, iconicity, 149
Index [ 405 ]
406
Textuality, 149 Transformational grammar (TG), 54

The Crying of Lot 49 (Pynchon), 168 binary grammaticality, 85–86
The Handmaid’s Tale (Atwood), 333 formalisms and generalizations, 58–59
The Life of Samuel Johnson, LL.D. inadequacies, 60
(Boswell), 156–157 missions, original, 378
“The logic of politeness; or, Minding your origins, 375–376
P’s and Q’s,” 37–51, 375–377, 381. vs. real world, 73–74
See also Politeness Transformational rules (T), 107, 107n2,
introduction, 33–35 242–243
Theoretico-political, personal in, Transformations, power of, generative
275–283, 278–281nn9–11 semantics, 256–257
Theory. See also specific theories Tribe, Laurence H., 315
in law, 316 Trudeau, Gary, Doonesbury, 166
in linguistics, 316 True confession, 349–350, 352
Theory wars, 375–377. See also specific “True confessions? Pragmatic
topics competence and criminal
Therapeutic Discourse (Labov and confession,” 353–368, 382–383.
Fanshel), 131, 133 See also Pragmatic competence,
“The rhetoric of reproduction,” 331–352, criminal confession and
382–383. See also Reproduction, introduction, 349–352
rhetoric of Truth-seeking discourse, 357, 357n5
The Right Stuff (Wolfe), 169–170 criminal confession, 357, 358–361
“The way we were; or; The real (See also Confession, criminal)
actual truth about generative Tu, 47
semantics: A memoir,” 241–293, Tucker, Karla Faye, 351–352
375–377, 382–383. See also
Generative semantics U
Classical Transformational Grammar, Um, 91
242–249 (See also Classical Universal information categories, 55
Transformational Grammar (CTG)) Uyeno, Tazuko, 16n4, 21n7, 22n9, 27n11
generative semantics theory, 249–275
(See also Generative semantics (GS), V
theory) Verbal auxiliaries, advertising,
introduction, 235–238 196, 196n4
methods, 237–238 Verbs
values, 237 endings, 8
Think, 21n8 endings, dubitatives, 27–28, 27n11
Thorne, James, 29 speaker responsibility for claim, 8
Thought, primary- vs. Verbs, performatives, 318
secondary-process, 193 explicit presence, 25–26
Tiersma, Peter, 323–324, 327 superficial structure, 256
“Toward generative semantics” Verbs of perception, modals (can), 28–30
(Lakoff, G.), 246 Verbs of thinking
Transcripts, 157–158 on declarative statements, 21n8
Transference, 144–145 intonation, 21n8
Transformational Generative Vocabulary, social factors, 204–206
Grammar (TGG) euphemisms, 206
clarity, lack, 372–373 exclusionary, 211
lure, 372 high-school cliques, 205
pseudoscientist pretensions, 376 in-group, 206
[ 406 ] Index
407
language codes, 210–211 Why, 19–20, 22

Martha’s Vineyard, young peoples’ Will/be to, 31
vowels, 205 Will-deletion (-insertion), 29–30
new, 205, 205n1 Winnie-the-Pooh (Milne), 162–163, 163n2
pickpockets, 206 With the power of each breath: A disabled
secrecy, 204 women’s anthology (Browne), 306
“secret handshake,” 211 With wings: An anthology of literature by
semantic use, 210 and about women with disabilities
solidarity and cohesion, 204 (Saxton and Howe), 306
speech similarity, 204–205 Wolfe, Tom, 169–170
unitary, 211 Bonfire of the Vanities, 351
writing similarity, 205 The Right Stuff, 169–170
Vous, 45 Women
VP, nonexistence, generative semantics healthy, paradox, 120–121, 129
theory, 252–253 inequality, 344
linguistics careers, 66
W pluralism in linguistics, 66
Walker, Anne Graffam, 219 stereotypes, 120–121
Wallace, George, 167 Women, American, deference, 122–123,
We, authorial, 45 127–128, 212
Well, 19–20, 22 Women’s language, 82–83, 212
pitch-patterns, 70 academia, 97
West, Lindy, 334 acceptability, 94
“When talk is not cheap: Psychotherapy characteristics, 90–93
as conversation,” 137–145, 378–383 vs. children’s language, 108
conversation, nature, 137–138 color-discrimination
discourse, general principles, 137 vocabulary, 90, 92
discourse-learning, 138 conservatism, 91, 92
emotional satisfaction, 140 courtroom discourse, 222
introduction, 131–135 as deviation from norm, 94–97
learning psychotherapy emotional expression, 90–92
conversation, 138 evading responsibility, 108
lecture, 141 expectations and
metacommunication, 144 prescriptivism, 94–95
metaphor, 144 grammar of style, 102 (See also
nonreciprocal conversation, 132–134, “Stylistic strategies within a
138, 141–142 grammar of style”)
ordinary conversation, 131–132, 139–140 italics, 162
ordinary conversation, series of linguistic conservatism, 91
interviews as, 142–143 vs. men’s language, 89–90
other conversation forms, nondirectness, 90–92, 95–96
psychotherapeutic discourse on, vs. normal language, 119
132–134, 139 syntactic rules and, 93–94
vs. permissible discourse, 140–141 Women’s movement, 302
reciprocal conversation, 132–133, 138 Women’s style, 118–119
rules, learning, 143 Women with disabilities: Essays in
spontaneity, denial, 142 psychology, culture, and politics
symbiosis, 141 (Fine and Asch), 302
transference and recontextualizing, Word order, generative semantics theory,
144–145 253–256, 255f
Index [ 407 ]
408
Words alter reality, 298 “You say what you are: Acceptability and

Writing similarity, social factors, 205 gender-related language,” 85–98,
Written communication 378–381
capitalization, 162–163 acceptability judgments, 86–88
comic strips, 166–171 acceptability judgments, context,
forethought, 151, 153–157 88–89, 95
intuitive writers, 154 description/prescription
italics, 160–162 distinction, 94–95
oral and literate strategies, 151–175 either/or distinction, 85–86
(See also “Some of my favorite grammaticality, context, 88
writers are literate: The mingling grammaticality/acceptability
of oral and literate strategies in distinction, 87–89
written communication”) grammaticality judgments, 86
vs. oral communication, 151–153 introduction, 81–83
vs. oral communication, as male vs. female speaker, context, 95–96
spontaneity vs. forethought, norm, 94
153–157 performance/competence
quotation marks, 158–159, 159n1 distinction, 85
from spoken discourse to, 157–158 subculture languages, 92
strategies, 151 women’s language, 90–93
women’s language, academia, 97
Y women’s language, acceptability, 94
Y’know, 48, 62, 91 women’s language, deviation from
Yo, 22, 22n9, 24–27, 24n10, 27n11 norm, 94–97
particle use context, speaker women’s language, expectations and
gender, 26 prescriptivism, 94–95
speaker gender on, 24n10
Yoshida, Kazuhiko, 22n9, 24n10 Z
Your, 197, 197n5 Zadeh, Lofti A., 55
[ 408 ] Index
409
410
411
412

Robin Tolmach Lakoff - Context Counts. Papers On Language, Gender, and Power (2017)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Robin Tolmach Lakoff - Context Counts. Papers On Language, Gender, and Power (2017)

Uploaded by

Copyright:

Available Formats

i

Edited by Laurel A. Sutton

Published in the United States of America by Oxford University Press

© Oxford University Press 2017

All rights reserved. No part of this publication may be reproduced, stored in

You must not circulate this work in any other form

Library of Congress Cataloging-​in-​Publication Data

Illa ipsa loquitor

Susan Blackwell, Department of Language, Literature and Communication,

Laurel A. Sutton, Department of Linguistics, University of California,

T his book did not exist, so it became necessary to invent it.

Bucholtz, Mary. 2004. Introduction. Language and woman’s place: Text and

Introduction to “Language in context”

W hen “Language in context” first appeared in 1972, it broke much new

efficacy) of utterances—​that is, sentences uttered in particular contexts.

the use of certain particles in Japanese provides “implicit personal infor-

Traditional transformational grammar attempts to define the conditions on the applica-

W hen studying exotic languages, the speaker of English often runs

The problems I am referring to will of course be immediately recogniz-

(i) Particles, like doch in German, or ge in Classical Greek, or zo in

problems confronting such diverse areas as theoretical linguistics and the

(a) Contextually linked linguistic phenomena are probably identifiable,

We may distinguish certain aspects of context from others. Some have

L anguage in context [ 9 ]

(1) You must have some of this cake.

L anguage in context [ 11 ]

already been noted, the effect of such bogus translations is generally

(4) Have some of this cake—​yecch.

My claim is that a sentence like (1) is a much closer translation of (4) than

L anguage in context [ 13 ]

of an obligation. In a sentence like (1), the most natural assumption is

Any of these assumptions might be primary in a given instance. In non-​

Theoretically, then, a sentence like (1) should be triply ambiguous, and two

(8) Have some of this delicious cake.

L anguage in context [ 15 ]

one’s own possessions—​work the same way in determining appropriate-

(9) Have some of this ‘o-​cake’.4

(13) Come in, won’t you?

L anguage in context [ 17 ]

In a normal conversation, the participants will make the following assump-

Rule I. What is being communicated is true.

L anguage in context [ 19 ]

(18) Leave, won’t you?

than demanding agreement or acquiescence from the hearer (as is true in a

( a) I say that Fritz is a Zoroastrian.

L anguage in context [ 21 ]

operates in this way, as a “softened” version of (21). We already know that

Ne may be appended to declaratives, imperatives, and interrogatives:

In all three cases, as the interpretations indicate, a normally obligatory rule of

L anguage in context [ 23 ]

be true even if the participants are in fact social equals, as a “hum-

This last is possible only as an incredulous echo-​question based on a prior

As is evident from the foregoing, the behavior of yo with questions is

(28) A: Is Agnew a liberal?

In both these cases, depending on context, B’s response may be construed

L anguage in context [ 25 ]

situation inapplicable), just as it employed a specific transformational rule,

(a) Kore-​wa anata-​no hon da wa yo ne.

If yo demands compliance with the rules of conversation, and ne allows relaxation of

(c) This is your book, isn’t it?

L anguage in context [ 27 ]

have seen—​demand the addressee’s belief as he ordinarily would. The best

(30) John is in Antarctica

Library of Congress Cataloging-in-Publication Data

efficacy) of utterances—that is, sentences uttered in particular contexts.

(4) Have some of this cake—yecch.

Any of these assumptions might be primary in a given instance. In non-

one’s own possessions—work the same way in determining appropriate-

(9) Have some of this ‘o-cake’.4

This last is possible only as an incredulous echo-question based on a prior

(a) Kore-wa anata-no hon da wa yo ne.

have seen—demand the addressee’s belief as he ordinarily would. The best

place. This phenomenon has been referred to as “will-deletion,” though per-

(44) *My girl must be home by midnight—I think it’s idiotic.

of pragmatic rules, dictating whether an utterance is pragmatically well-

(16) Nixon is an arch-conservative.

(18) (a) When the natives of Whango-whango want to

and, even while there, imagining and beginning to create a performance-