Professional Documents
Culture Documents
Editorial Address:
Linguistics (GER)
University of Antwerp (UIA)
Universiteitsplein l
B-2610 Wilrijk
Belgium
Editorial Board:
Norbert Dittmar {Free University of Berlin)
Bruce Fraser (Boston University)
John Heritage (University of California at Los Angeles)
David Holdcroft (University of Leeds)
Catherine Kerbrat-Orecchioni {University of Lyon 2)
Beatriz Lavandera (University of Buenos Aires)
Marina Sbisà (University of Trieste)
Emanuel A. Schegloff (University of California at Los Angeles)
Paul O. Takahara (Kobe City University of Foreign Studies)
Sandra Thompson (University of California at Santa Barbara)
Daniel Vanderveken (University of Quebec at Trois-Rivières)
Teun A. van Dijk {University of Amsterdam)
32
Edited by
MASAYOSHI SHIBATANI
Kobe University
SANDRA THOMPSON
University of California at Santa Barbara
CHARLES J. FILLMORE
(Photograph by Paul J. Atkins)
Preface
Masayoshi Shibatani and Sandra A. Thompson
Preface
Masayoshi Shibatani and Sandra Thompson vii
Mental-space builders:
Observations from Japanese and English conditionals
Seiko Yamaguchi Fujii 73
Index 323
A predicate construction:
Inversion in Polish1
Farrell Ackerman
University of California, San Diego
1. Introduction
This paper is largely based on the Relational Grammar analysis of Polish inversion
presented in K. Dziwirek (1991). Her presentation establishes a baseline for any adequate
alternative treatment, while the deftness of argumentation dependent on multi-stratal
assumptions constitutes a challenge to mono-stratal multi-level approaches and, hence, the
motivation for the present paper. All data is from Dziwirek (1991) unless otherwise
indicated. I would like to thank Katarzyna Dziwirek, Bozena Pruska, Mark Gawron,
Claudia Brugman and Phil Lesourd for useful discussion concerning various aspects of this
paper.
2 FARRELL ÀCKERMAN
4
That there is no necessary relation between the hypothesized centrality of lexical
heads and commitments to the nature of semantic arguments as atomic is illustrated by
the fact that there have been numerous efforts over the years to decompose semantic roles
into various properties, for example Foley and Van Valin (1984) and Jackendoff (1991)
to mention just two. As observed in Van Valin (1991) and Dowty (1991), these efforts all
share the view that such decompositions yield discrete category membership. For present
purposes, I am more concerned with the presumed locus of semantics for arguments, than
with whether the arguments can be characterized as discrete or as proto-types with
overlapping features.
5
In fact, Gawron himself extends the notion head to cover the N and A of predicative
constructions, in effect, differentiating between a functional category such as PREDICATE
and the lexical exponence of this category.
INVERSION IN POLISH 7
The basic idea underlying the present proposal for the analysis of Polish
productive inversion is this: the Polish inversion construction contains an
INVERSION PREDICATE (expressed by elements realized disjointly in
constituent structure) whose semantics entails certain properties for its SUBJ
argument, and certain of these properties determine DATIVE case marking.
As mentioned previously, I will assume that the separate pieces constitute
a unit which I will refer to as a PREDICATE. PREDICATES as, understood
here, are information units serving as the semantic head of a clause and
whose surface expression can be larger than the standard zero-level
category, but smaller than a VP (see T. Mohanan 1991 and Ackerman &
Webelhuth for discussion of the relevant notion of PREDICATE). The
PREDICATE, by hypothesis, is an entity associated with an argument
structure: all of its arguments bear a grammatical function value.
The main difference between the inversion construction and its
analogue containing a NOMINATIVE SUBJ is, on this account, located in the
Mohanan and Mohanan (1990) present evidence that Malayalam contains complex
predicates which induce the appropriate semantics for DATIVE case marking of the
SUBJ. Since these predicates consist of syntactically separate entities in constituent
structure, they resemble the Polish constructions. On their analysis these predicates are
composed lexically by combining the participating pieces.
INVERSION IN POLISH 9
Fillmore (1977) utilizes these features to explain the nuclear relations for
PREDICATES, e.g. knock vs. knock down. For a sentence such as:
He knocked on the door with his fist
he observes that though the activity requires more than a single participant,
i.e. a knocker and thing that is knocked, the entity that is active (and in the
present instance, also causal and human) is put into perspective and
functions as SUBJ. On the other hand, when the activity effectuates a
change of state in some participant, then that participant is also
perspectivalized and, following the mapping procedure presented above, it
functions as an OBJ. He says of such an activity, "... we express this with the
two word verb knock down, as in:
He knocked the door down". (1977: 77)
It should be observed that these features can be associated with phrasal
entities such as knock down on his account. In other words, Fillmore does
not identify the projector of features relevant for function assignment with
a single morphonologically integrated and syntactically atomic entity. In
addition, the arguments of this predicate are not explanatorily associated
with discrete thematic or semantic roles.
In similar spirit, but with an explicit disavowal of both case roles (e.g.
thematic roles) and saliency as criteria!, Dowty (1991) articulates a proto-
role view of the relation between arguments of a PREDICATE7 and their
grammatical function status. He suggests the following lists of proto-agent
and proto-patient properties.
Contributing properties for Proto-Agent Contributing properties for Proto-Patient
7
I am assuming that the conception of PREDICATE here is not coexistensive with
a single lexical category as on the lexical categorial approach, since, for example, Dowty
(1979) proposes lexical rules for e.g. resultative constructions such as that mentioned by
Fillmore above. In other words, there is no identification of lexical operations with an
entity that exhibits lexical integrity and is syntactically atomic.
INVERSION IN POLISH 11
Corollary 1:
If two argument of a relation have (approximately) equal numbers of entailed Proto-
Agent and Proto-Patient properties, then either or both may be lexicalized as the
subject (and similarly for objects)8
Non-Discreteness:
Proto-roles, obviously, do not classify arguments exhaustively (some arguments have
neither role) or uniquely (some arguments may share the same role) or discretely
(some arguments could qualify partially but equally for both proto-roles),
Finally, Dowty suggests that "the Proto-roles and their argument selection
principles determine hierarchies of traditional roles..." (1990: 578). The
particular hierarchy he mentions is:
AGENT > EXPERIENCER/INSTRUMENT > PATIENT > SOURCE/GOAL
In sum, Dowty articulates a view resembling the function-inducing
I interpret the reference to "both" in this passage as indicating alternity rather than
simultaneity in order to conform to uniqueness assumptions for grammatical functions.
12 FARRELL ACKERMAN
9
It would be incorrect to interpret Dowty's proposal as an effort to characterize
function-inducing semantics. Instead, as can be seen from the formulation of the Argument
Selection Principle, semantic entailments are understood as independent of the
grammatical function inventory of a predicate. The Selection Principle accounts for how
to probalistically relate entailments to functions given the co-presence of particular
entailments and particular functions. See Ackerman & Moore (1994) for discussion.
10
Just as there have been linguists who have previously argued for the independence
of case marking and grammatical function (for example, Shibatani (1977) among others,)
there have been linguists who have argued for the semantic motivation for surface case
marking (for example, Breal (1964), Wierzbicka (1980), Tsunoda (1981) among others.)
11
As often occurs in appeals to lexical semantics one enters a terminological morass:
despite the strong appearance of referring to many of the same distinctions, Wierzbicka,
Dowty and Mohanan and Mohanan, utilize different terminology for the same distinctions
and the same terms for different distinctions. I cannot attempt to reconcile or render
consistent such terminological differences here.
INVERSION IN POLISH 13
particular (class of) predicates which project this property, rather than
being construable as having undergone a change of state or location).
With respect to representational conventions for lexical items, T.
Mohanan (1991) distinguishes between several levels of representation for
the information associated with a PREDICATE: she argues for a semantic
level, an argument structure level, and a grammatical function level, among
others. In the present context these can be interpreted in the following way:
the argument level represents the valence of a PREDICATE, while the
semantic level possesses properties associated with the arguments (these
properties being, for example, Dowty type proto-role features indicated
schematically as P) and the grammatical function level represents the
functional value of arguments. This yields the following representation:
P P P P Semantic Level
Case Marking
GF Grammatical Functions
Intransitives:
SUBJ SUBJ
NOM
Transitives:
The use of the reflexive clitic in this construction is reminiscent of its use in Polish
middle constructions which generally occur without an expressed agent: in middles there
is a NOM SUBJ and, according to my informant, if the agent is expressed, it will appear
in the DATIVE case. Finally, the claim of attentuated agentivity may be evidenced by the
inability to use zeby purposive clauses with the productive inversion constructions, while
it is possible to use them with the non-inversion analogues of these constructions.
16 FARRELL ACKERMAN
to assume that there is any effect whatsoever on the OBJ argument of the
transitive base predicate. On the present proposal, therefore, the ACC
argument is an OBJ, both for the simple predicate and its inversion
analogue» Alternatively put, the present proposal predicts the retention of
the OBJ function with productive inversion, since there is nothing in the
inversion predicate that would motivate a change in function or case
marking for this argument.
In the remainder of this paper I briefly examine several 'simple
behavioral properties of the nominals in inversion constructions and
demonstrate how they can be explained in terms of the claims that the DAT
nominal is a SUBJ and that the ACC nominal is an OBJ.13
The reader should note the following with respect to OBJs in Polish: given that
passive is a reliable diagnostic of OBJECThood, it should be observed that Polish
predicates can govern several different cases. This is why it is important when formulating
the conditions for genitive of negation to specify that it operates on ACC OBJs: OBJs
which appear in other cases do not undergo this operation. Two points are relevant in
connection with this: (1) case marking for OBJs can be specified by predicates (and
motivated by semantic considerations) just as can the case marking of SUBJs and (2) the
operation of genitive of negation must refer to the combination of function and case-
marking. Both of these phenomena are expected on the present analysis since we have
argued that semantic properties can determine case marking and our representations
contain information concerning the surface case and functional status of nominals,
INVERSION IN POLISH 21
4. Conclusions
References
Ackerman, F.
1994 DATIVE SUBJ predicates in Polish. Ms,
Breal, M.
1964 Semantics: Studies in the science of meaning, Dover.
Dowty, D.
1991 "Thematic proto-roles and argument selection." Language 67:3.547-619.
Dowty, D.
1979 Word meaning and Montague grammar. Dordrecht: Reidel.
Dziwirek, K.
1991 Aspects of Polish syntax. UC San Diego doctoral dissertation.
Fillmore, C. J.
1977 "The case for case reopened." In P. Cole & J, Sadock (eds.), Grammatical
relations. New York: Academic Press.
Fillmore, C.J.
1968 "The case for case." In E. Bach & R.T. Harms (eds.), Universals in linguistic
theory, New York: Holt, Rinehart and Winston.
INVERSION IN POLISH 23
Gawron, M.
1983 Lexical representation and the semantics of complentation. UC Berkeley
doctoral dissertation.
Goldberg, A.
1992 Argument structure constructions. UC Berkeley doctoral dissertation.
Gruber, J.
1965 Studies in lexical relations. MIT doctoral dissertation.
Jackendoff, R.
1991 Semantic sttuctures. Cambridge, Mass.: MIT Press.
Jackendoff, R.
1975 "Morphological and semantic regularities in the lexicon." Language 51: 639-671.
Kubinski, W.
1987 Reflexivization in English and Polish: An arc pair grammar analysis. Niemeyer.
Mohanan, T.
1990 Arguments in Hindi. Stanford doctoral dissertation.
Shibatani, M.
1977 "Grammatical relations and surface cases." Language 53:789-809.
Spencer-Brown, G.
1972 Law of form. New York: Julian Press.
24 FARRELL ACKERMAN
Tsunoda, T.
1981 "Split case-marking patterns in verb-types and tense/aspect/mood." Linguistics
19:389-438.
Wierzbicka, A.
1988 "The meaning of a case: A study of the Polish dative." In The semantics of
grammar. Amsterdam: John Benjamins.
Wierzbicka, A.
1980 The case for surface case. Karoma Press.
Zwicky, A.
1992 Syntactic constructions. Talk given at the Milwaukee Conference on Symbolic
vs. Non-symbolic Models of Language.
The role of the example
in a frame semantics dictionary
B.T.S. Atkins
Oxford University Press
Examining the dictionary example is like running a geiger counter over the
The new lexicography sketched in this paper is based on Fillmore's work on the
semantics of risk, described in Fillmore and Atkins (1992, 1994). Any idiosyncratic
divergencies from that account simply underline the need experienced by this lexicographer
for a semanticist to light the way.
26 B.T.S. ATKINS
2
A full description of the risk frame, its schemas and frame elements, is to be found
in Fillmore and Atkins (1994).
28 B.T.S. ATKINS
3
This diagram is taken from Fillmore and Atkins (forthcoming).
A FRAME SEMANTICS DICTIONARY 29
Schema A
Schema B
Schema C
These figures are adapted from the kinds of directed graphs used in mathematical
decision theory (as discussed in Raiffa 1970).
30 B.T.S. ATKINS
It would also record that Bad [BA], whether in the form of a gerund or a
noun phrase, may be realized as the object of the preposition in the phrase
run the risk of:
He ran the risk of [BA falling down]
He ran the risk of [BA a fall],
And it would record that the syntactic realization of Possession [PO] would
be different with the phrase put at risk from take risks with:
He put [PO his own life) at risk.
He took risks with [PO his own life].
Thus the entry would set out in detail the relationship between the noun
risk and the verb risk, showing how both may be used to transmit an almost
identical message, and allowing an analysis of the speaker's (unconscious)
motivation in selecting one way rather than another of expressing the same
semantic content.
An account of the frame elements and their realizations would also
include an account of the relationships between the various multiword items
within the risk frame (take or run risks / a risk / the risk (of); (put or be) at
risk; at the risk of; at one's (own) risk, etc.), highlighting the way each idiom
or compound licenses different configurations of frame elements, and how
the whole offers a range of options allowing the expression of extremely
fine shades of meaning.
Each frame entry would of course list and discuss all the expressions in the
frame - the most obvious access point to the information, though in an
on-line dictionary not the only one. In the case of the risk frame, these
would include:riskitself, andrisky,riskily,riskiness,risker,etc., together with
the idiomatic expressions in which these words participate.
way they are used. One example will suffice: why, for instance, is the verb
risk selected to express a particular concept, rather than endanger or
jeopardize, or the phrase put at risk? The answer to this and many other
questions must be furnished by the dictionary account of the frame.
In this instance, the dictionary user would learn that when risk is used
with the meaning of 'place in danger' its object is always an expression of
the [PO] element (something or someone valued by the person involved),
while there is no such constraint on endanger, jeopardize and put at risk. I
can endanger, jeopardize and put at risk people or things that I feel nothing
for; I can only risk people or things I care about. The selection of risk to
express the concept of causing someone or something to be open to
possible harm carries with it this additional implication. The dictionary entry
would also make explicit the fact that, in these parallel uses, the [PR]
element (the central person in the frame who is vulnerable to the possible
harm) can in the context of the verb risk only be realized as the subject of
the verb, whereas with endanger or put at risk [PR] may be expressed as the
verb's object (hence you endangered me ox you put me at risk, but, with this
interpretation, not*yourisked me).
Intimately linked with the objective of the dictionary is the function of the
example: what does it bring to the entry in order to achieve its purpose?
Editors must be clear about this in the style guide of a dictionary, and
compilers must be aware of it every time they select an example. The
explicit / implicit dimension (see Atkins et al. 1986) is important here, for
an example may illustrate information given explicitly elsewhere in the
entry, or it may contain information that cannot be spelt out explicitly, for
lack of space. Often, it seems, information is more efficiently transmitted
to the user by means of an example than by an abstract statement of
linguistic facts,
The possible functions of a dictionary example include the following,
several of which it may carry out at the same time:
¤
to amplify, clarify or exemplify information given elsewhere in the entry,
for example by the definition, the part-of-speech abbreviation, the
grammar codes or other formal statement of complementation patterns,
or somewhere else in the dictionary, as for instance in verb lists outside
the main text;
¤ to give new information about one of the aspects (discussed in section
3.2) of the headword in use;
¤ to show the headword in use: some dictionaries do not specify the
aspects of word use that the example must illustrate, but simply supply
one or more example(s) of the headword used in the relevant sense,
chosen according to the lexicographer's intuition;
¤
to illustrate 'prototypical' usages of the word, or (at the other extreme)
to illustrate 'boundary' usages, to show how much the word may
encompass, both in meaning and grammar;
¤ to record every instance of the headword used with a particular
38 B.T.S. ATKINS
the example are constrained only by the fact that they must communicate
their message to all types of users. In an on-line dictionary, this need not
be too restrictive; the examples may be tagged according to prototypicality,
difficulty and other features, and the dictionary customized for each user as
it is installed, so that an individual is offered only material that he or she
can cope with.
4«, Conclusion
References
Atkins, B.T.S., J. Kegl & B. Levin
1986 "Implicit and explicit information in dictionaries." In Advances in lexicology;
Proceedings of the Second Annual Conference of the UW Centre for the New
OED, 45-63. Waterloo, Ontario: University of Waterloo.
I should like to thank Patrick Hanks and Krista Varantola for their helpful
comments on the first version of this paper,
42 B.T.S. ATKINS
1994 "Starting where the dictionaries stop: The challenge of corpus lexicography."
In B.T.S. Atkins & A. Zampolli (eds.), Computational Approaches to the
Lexicon, Oxford: Oxford University Press.
Raiffa, Howard
1970 Decision analysis: Introductory lectures on choices under uncertainty. Reading,
MA: Addison Wesley.
Shifting face from Asia to Europe1
1. Introduction
This paper could not have been written without the extensive collaboration of Seiko
Yamaguchi Fujii and Jeong-Woon Park, who supplied both syntactic and semantic
analyses. In addition, Jeong-Woon Park prepared the Korean romanization and made
important corrections to the data and their interpretation. The authors wish to thank
other advisers who gave us many examples of Chinese, Japanese, and Korean speech
practices: Dan Jun, Hea-Soog Jo, Hee Won Kang, Myn-Gyun Kwon Jeong, and Hongtu
Chen. Yoshiko Matsumoto also has made helpful suggestions. Undoubtedly we have
introduced mistakes for which these collaborators are not responsible.
The Chinese in the paper employs Mainland Pinyin, the Japanese uses the Hepburn
system, and the Korean uses the Yale romanization which gives a one-to-one correspon
dence to the Korean spelling system.
44 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
We have been in this work taking the first steps to an analysis of the
concept of 'face' based on collocations, as suggested in the recent study of
'risk' by Fillmore and Atkins (1992). However, we are far from reaching
more than a crude preliminary analysis, focussing on obvious cultural
contrasts, and for Chinese, Japanese, and Korean we used consultants be
cause at this point we do not have access to on-line text searches in those
languages.
English
There are over a dozen senses given for face in modern English dictionar
ies, including in one subset appearance or semblance, disguise, assurance,
effrontery, dignity, and prestige. The role of the physical human face in
expressing emotion affected how the word was extended to other uses in
English in reference to people. In Middle English face could refer to "(a)
facial appearance or expression, beren god face (b) facial or outward
appearance as contrasted to some other reality; two faces in a hod,
double-facedness, duplicity; (c) traits of personality, character; (d) a person
as having certain traits, a personage or person." (Kurath 357).
The current Oxford English Dictionary (643) includes as listed meanings
"outward show: assumed or factitious appearance; disguise, pretence; an
instance of this; a pretext." The OED gives a variety of historical examples
for this set of meanings. Wyclif in the 1382 translation of Corinthians: "Hem
that glorien in the face and not in the herte;" in 1568 Grafton wrote
"They...made good face and shewe to fight with the Englishe men."
The face was often referred to by Shakespeare, in vivid descriptions,
typically to insult: false-faced, black-faced, whey-faced, tallow-face,
brazen-faced, shame-faced and antic face.
"Thou damned tripe-visaged rascal, an the child I now go with do miscarry, thou wert
better thou hadst struck thy mother, thou paper-faced villain." (Doll Tearsheet)
2 King Henry IV
"The devil damn thee black, thou cream-faced loon! Where got'st thou that goose
look?" (Macbeth)
Macbeth
Current newpapers speak of "red-faced officials" in the same vein. Though
these phrases carry implications for character, they are fundamentally
physical descriptions, and do not have the abstract sense of reputation
alone which we find in the more recent usages reported in the OED.
SHIFTING FACE FROM ASIA TO EUROPE 45
In the OED, the first evidence of the sense of face in the figurative
sense of good name or repute is attested by an 1876 quotation from R.
Hart in These from the land of Sinim: "Arrangements by which China has
lost face," The collocations given by the OED involve only losing or saving
face, and the etymology is given as a translation from the Chinese for loss
of face diu lian. In these uses, one loses or saves one's own face, not
someone else's.
A search of social conversation databases revealed no instances even
among educated speakers. The usage for this term currently is educated,
and primarily appears to be diplomatic or psychological. The term is used
frequently in social science writing, as indicated by a sampling of journal
abstracts from library abstract collections. About a third of these uses are
in quotes, suggesting that both face and face-saving are still marginal, special
terms, with a foreign or technical-jargon flavor. In this context, the term has
closely related meanings: defending honor, displaying strength, maintaining
or restoring self-esteem, remedying embarrassment, and negotiating with a
strategic sensitivity to appearances. In the quotations below, italics are
added.
"Policemen frequently are called to restore order to situations in which one of the
parties is fighting 'for his honor' in front of a crowd. ... While the crowd looks on, a
citizen may only be able to save face by attacking the officers themselves. A citizen's
honor - his reputation for implacability & vengeance - is vital to him." (Muir 1980).
"The courtroom context sanctioned the violation of normal social norms and allowed
the public display of anger and indignation in order to restore face," (Brown 1990).
"Social face is the image of strength that a person wants to project in conflict... Face
may be so valued because it is based on the desire to feel respected as a group
member.... Research indicates that people are motivated to avoid appearing weak and
try to restore face in culturally defined ways." (Tjosvold 1983).
"Therapists seek to help patients 'change their stories' while allowing them to save
face," (Erickson 1984).
46 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
"A study is made of face-saving' techniques in the context of a singles dance used to
avert potentially damaging evaluations by others. Since mere attendance at a singles
dance results in some loss offace, dance management & patrons strive to offset this
aspect of disrepute" (Berk 1977).
"In such cases the granting of symbolic rewards and face-saving mechanisms to the
terrorists through the ritual of negotiation can lead to their capitulation." (Miller
1978).
French
The French expressions are perdre la face 'lose face' in the sense of lose
prestige from a serious attack on honor, dignity, or reputation, and sauver
la face 'save face' to safeguard prestige, and dignity in defeat or a setback.
Robert (1985) gives the origin as an 1850 French translation from Chinese.
Consultants regard the expressions as always concerning the agent's own
prestige, not another's, Robert's examples emphasize putting on a good
appearance:
SHIFTING FACE FROM ASIA TO EUROPE 47
"Ainsi il ne perdra pas la face (itaî. sic). Depuis le dernier coolie jusqu'au premier
mandarin, il s'agit de ne pas perdre la face, leur face de bois, mais ils y tiennent et
en effet, n'y ayant pas de principes, c'est la face qui compte,2 Henri Michaux, Un
barbare en Asie, p. 165.
"Les ouvriers désiraient retrouver leur métier et les patrons leur bureau. Il ne fallait
plus que sauver la face.3 A. Maurois, Bernard Quesnay, xviii.
"Alors, l'Allemagne, compromise malgré elle, n'a plus qu'un moyen de sauver la face:
paraître intransigeante, pour effrayer l'Europe.4 Martin du Gard, Les Thibault, t. VII,
p. 18." (Robert 1935: 352).
Borrowing
A puzzle in this history is why the Chinese expression was borrowed into
French and English at that time. In both countries there was an accelerating
fad for things and ideas Chinese which peaked in the mid-eighteenth
century, with the importation of tea, porcelain, silk, chintz, and wall-paper,
the copying of Chinese and Japanese decorative elements in Chippendale
furniture and Gobelin tapestries, and the design of Chinese-style rooms and
Chinese gardens,,
The Jesuit missionaries became interested in Chinese philosophical
traditions, particularly in Confucius, and in turn the philosophes were
attracted to the idea of a secular morality unlinked to religion (Ting 1928).
Chinese literary characters were created by European writers as a voice for
making cultural criticisms of the west (Appleton 1951). As is usual in
borrowing, assimilation to the borrowers' concerns revealed a quite
superficial acquaintance with Chinese culture, though Chinese literature was
beginning to be translated as early as the seventeenth century. Ting (1928),
in a survey of French views of China in the century before 1750, tells us
that the missionaries created a stereotype, a Chinese who was imagined on
the basis of Confucius to be soft and polite, who measured as he walked
2
Thus he won't lose face. From the last coolie to the first mandarin, it's a matter of
not losing face, their wooden face. But they stick to it, and indeed having no principles all
that counts is face.
The workers wanted to go back to their work and the bosses to their offices. All
that was needed was to save face.
Thus Germany, compromised in spite of herself, had only one way to save face: to
appear intransigeant to frighten Europe.
48 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
the length of his stride, the tilt of his head, the blinking of his eyes - in
short, an absurd exaggeration of politeness. Even when the Chinese notion
of face came to be translated it was taken up in a very restricted sense into
educated usage in the west.
2. Asian face
Face terms
As described above, in English and French the term face is mainly used in
the contexts of losing face and saving face. Unlike Indo-European languages,
Asian languages have a wide range of expressions involving the term face.
In this paper, we will focus on three Asian languages, Chinese, Japanese
and Korean. We interviewed native speakers of these three languages,
asking them the following questions:
1. Are there expressions in your language which correspond to the English
expressions 'to lose face' and 'to save face'? What are they?
2. In what contexts are they used? (e.g. who loses/saves face? under what
circumstances?)
3. Are there other expressions related to the concept of face which may not
have equivalents in English? How are they used?
5. What expressions would you use in these social contexts? (listing contexts
presented as calling for face maneuvers in Brown & Levinson, p. 65-68)
In all three languages, we found one or two terms for 'face' which have
overlapping functions. They are used to describe 'physical face/complexion',
'appearance/surface' and 'reputation/dignity'. In Chinese, the relevant term
for face is lian.5 In Japanese, there are two terms: kao or tsura. Kao is the
standard term for face, while tsura is most commonly used in colloquial
men's speech. In Korean, there are also two terms: nach and elkwul, which
are used interchangeably. An indicator of antiquity is the fact that in
Japanese and Korean indigenous terms are used in expressing the term
face, rather than terms borrowed from Chinese, of the type we will discuss
later. In addition, most of these phrases involved verbs, not adjectival
modifiers. We will discuss expressions based on these literal terms first.
No face
Certain phrases exist in all three of these languages. All have expressions
for states of extreme humiliation and shame. For example, there are expres
sions which deal with the idea of lacking face': in Mandarin, mei lian, 'lack
face', in Japanese kao ga nai, and in Korean, nach-i epsta, 'there's no face'.
In all three languages, this phrase is used in the context of T've no face' (to
see you), or 'he lacks face', because of shame. In Japanese, one might say
kimi ni awaseru kao ga nai '(there's) no face (with which) to see you'. This
might be an apology when you fail to fulfill a promise to a friend, creating
disappointment. In Chinese, the equivalent expression might be used toward
someone whose trust one has disappointed.
In Chinese, mei lian 'he has no face' said of someone else is extreme,
and is used in very serious, primarily voluntary acts, such as theft, treachery,
adultery, pregnancy prior to marriage, and failure to pass the university
entrance examination. These are losses so serious that suicide may occur.
This phrase has been in use since the 4th century B.C. in its earlier form
(Hu 1944).
Most serious in Chinese is the accusation that the person does not even
want face, which is like a curse. Unique to Chinese, the expression used is
bu yao lian 'not want face'. This is used to describe people who are
particularly cheeky or brazen, who are selfishly and shamelessly unpredict
able (Hu 1944). It can describe a man who makes inappropriate advances
to a woman. A father can use this phrase, bu yao lian 'you really don't want
your face - how shameless!' to describe a daughter who becomes pregnant
before getting married. But it is most serious when directed to superiors,
some of the figurative senses of the older term. These are not just phonological variants
but different characters.
50 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
Lose face
others" (Young 1994). A country can diu lian by failing in its obligations to
weaker nations (Hu 1944).
A common Japanese expression is kao ga tatanai, 'can't save face', (lit:
face collapses) as in the context:
kore o shinakereba kao ga tatanai 'unless I do this, my face won't stand
up',
which means that one must do something in order not to lose face. The
verb tatanai can be said of a bottle, a building, or a wave, in the sense of
their vertical dimension.
Japanese losing face' is similar to diu lian in its reference to acute
embarrassment. In all three languages face terms can be used in causative
expressions, that is to make someone lose face. In Japanese, as in Chinese,
one can lose face for someone else, that is lose their face:
kao ga tsubureru 'one's face is crushed or collapses' (intransitive like 'a
house collapses').
kao o tsubusu 'to crush someone's face' (like crushing an insect)
kao ga tsubusareru 'to have one's face crushed by someone else', (like
a house smashed by a meteorite)
kao/tsura o yogosu 'to dirty someone's face',
kao ni doro o nuru 'to paint mud on someone's face'.
These Japanese expressions are used in cases of humiliation and dis
grace. Kao ga tsubureru can be used to describe Abe's face if he recom
mends Doi for a certain job, and Doi turns out to be completely incompe
tent. A 'black sheep' of a family can be referred to as the tsura-yogoshi
'face-dirtier' of the family.
In Korean, the similar phrases are:
elkwul-ey ttongchilhata 'smear dung on someone's face'
elkwul-ey mekchilhata 'paint someone's face' (as one would paint paper
or get paint on clothes).
An expression unique to Chinese is lian fangzai nar? 'where do I put my
face?' (lit: face place-at where) said by a person who feels shame. This
same phrase can be used use for a vase or a chair, or anything movable
rather than a face. Socially, this expression is similar in contexts to wo/ni
you shenma lianmian jian ren a! 'what kind of face can I/you look at people
with!'(lit: I/you have what face to see people!), These expressions can be
52 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
Give face
We have already shown that in these languages face can be gained or lost
either by the face-owner or by someone acting in his/her behalf. The idea
of 'giving someone face' appears in these languages, but not in English or
French. In Chinese, one can say gei lian when one of a pair of two
long-standing enemies doesn't retaliate when expected. In this case, he may
claim that he 'gave (the other party) face' by not responding. A similar
expression is gei ta ge lian ba 'give him a little face', in which Chen can
attempt to persuade An to do a favor for Bai. But both of these expressions
are extremely patronizing, since they involve a concession by someone who
has the power or the right to act against someone but abstains.
There is also a negative version. If An gives Bai a present, and Bai
refuses it, then Bai has not given face to An. Also, if Chen asks An for a
favor for Bai, or for forgiveness of Bai, and An refuses, An has not given
face to Chen. If Chen asks An for a favor to himself rather than for Bai,
face is not involved in compliance, only in refusal. If An refuses the favor,
he does not give Chen face.
In Taiwan one can still say shang lian, which is'a humble way to say
'you bestowed face on me/you granted me an honor'. This expression can
be used in contexts in which a superior or colleague gives one signs of
attention, such as bestowing privileges, awards, jewelry, honors, or even
punishments. For example, the speaker would say it to a superior if he/she
attended a party the speaker held. Mainland speakers regard this phrase as
appropriate only for joking, because they consider it too obsequious or
feudal.
All of these expressions use common verbs involving possessions one
can lack, want, not want, lose, give, or bestow money, clothes, jewelry,
documents, or people. In Chinese face is thus treated as a possession to be
handled.
Japanese expressions seem to use a more architectural view of face (fall
down, stand up, raise). In Japanese, this idea is expressed as 'raising face',
kao o tateru in which someone is given a favor which leads to their
face-gain. For example, Tanaka can organize an event, but give credit to his
superior Suzuki. In this case Tanaka 'raises the face' of Suzuki. A PhD
SHIFTING FACE FROM ASIA TO EUROPE 53
student can give credit for her award-winning dissertation to her professor,
'raising' the professor's face. This is different from the instances in Chinese
in that the behavior is very humble and not patronizing at all, since it is
done by the lower person for the superior.
In Japanese, unlike English, there are both transitive and intransitive
forms of this verb. The expressions are kao ga tatsu 'one's face stands up'
and kao o tateru 'raise someone's face', which is similar to the concept of
'saving someone's face'. An example would be:
kimi no kao wo tatete ippai dake nomoo
'I'll drink one to raise up your face', i.e. 'I'll drink one for you.'
This expression might be used by a person who is reluctant to drink, but is
being pressured by his friend to do so.
In Chinese, there is an expression lou lian 'look good as a result of
receiving honor or praise' (lit: reveal one's face, let one's face appear),
which would be used in the context of a student being accepted into a
university, or winning an award. If lou lian is used with the third person, it
is considered a compliment - 'he revealed his face' by his achievements -
but if it is said to a friend, with the second person 'you showed off your
face', it would be taken to be a sarcastic remark or a joke about luck or
accident. The negative connotations of this phrase stem from the view that
conformity is desirable, and sticking out in a crowd is undesirable. When
talking about light, the phrase would be said if one leaked light during a
blackout. Expressions involving this verb suggest inappropriate or unwitting
revelation, as in revealing one's lie or revealing one's body.
One can also grow face, but only in a sarcastic context. Zhen zhang lian
'really grow face' might be said to a teenager directly in the second person,
or to a friend to criticize a third person in circumstances where the third
person showed off. The verb is not appropriate for plants, only for
abstractions like power or confidence.
Although English and French texts speak most often of face-saving, in
Chinese, Japanese and Korean, there are numerous expressions involving
the concept of losing face or lacking face, but fewer phrases about saving
face.
Thick/thin face
In addition to these phrases found in all three languages, each language has
unique phrases, Chinese has an expression based on face as a location:
lianshang bu hao kan 'doesn't look good on the face; unattractive face',
which is used when something is embarrassing, involving a minor loss of
face, This expression could be used to describe a student who cannot find
a job teaching, who instead resorts to washing dishes at a local restaurant.
The phrase bu hao kan 'doesn't look good' is a common descriptive term
for evaluating the physical looks of people, furniture, or clothes. The phrase
lianshang mo bu kai 'can't spread things out on one's face' (lit: face-location
spreads (intrans.) unsuccessfully/ not fully) using a verb frame employed for
paint or plaster, is found in contexts in which one cannot say something and
must hide one's feelings. The image involves a face one would like to shrink
or hide in shame. For example, this phrase can be used to describe a
situation in which An owes Bai money, but Bai cannot ask for his money
back, because Bai will feel embarrassed to ask for his money, so he
lianshang mobukai.
The expression lianshang you guang 'the face has lights' is used when
someone has something to be proud of, such as passing the university
SHIFTING FACE FROM ASIA TO EUROPE 55
In addition to these expressions which involve the usage of the literal term
face, there is also a large number of other words for face, many of which
have lost their original meaning and are now used interchangeably as
equivalents for 'reputation', 'dignity' and 'honor'. The terms for face used
in these expressions in Korean and Japanese are not the indigenous literal
terms kao, tsura, nach and elkwul, but forms based on the ancient Chinese
word mian which have taken on a figurative meaning. This ancient word for
face is used very commonly in Chinese, with a wide range of meaning inclu
ding some concrete meanings like surfaces, directions, and perspectives. It
no longer describes the front of the human head except in some fixed
combinations. Its primary use is for figurative meanings such as 'social
skills/ ba mian linglong, (lit: eight exquisite faces (as in jade)). It is
important to 'consider one's face', (and one's group or family's face) gu
mianzi, by the behavior appropriate to status (Hu 1944), or one can be
tactful about other people's face by not drawing attention to their
weaknesses, just as one can 'consider one's home'.
Many of the same kinds of dynamics described above for lian also
appear in phrases involving mianzi, but mildly. One can lack face, lose face
or give face using this form. Examples are mei you mianzi 'not have face'
meaning lacking the prestige to reach a particular goal, such as an
encounter with an important person. This word also is used to describe very
minor losses of face, A specialist may diu mianzi if he/she is unable to
answer a question in his/her field of specialization. One can say mianzi
shang bu hao kan 'the face doesn't look good' when there is disapproval.
One can give face to another in this version, too. Gei mianzi is an
action of An to increase Bai's prestige in front of other people by public
praise or by apparent deference to Bai's advice - i.e. making Bai look good
in public (Hu 1944). A slight is bu gei mianzi, 'not giving face'. Mianzi can
describe sinecures, given symbolically or by kicking upstairs to incompe
tents. But mianzi also implies collegiality or an 'old boys club', in the phrase
da jia you mianzi or 'we have shared face', (lit: everyone has face). A
related phrase is he wo you mianzi (lit: also I have face) 'have face with me,
help intercede for me', In these relationships, each would be embarrassed
to refuse favors for the other. The two don't share the same face or an
equal amount of the face, but the relationship is good enough for one to
SHIFTING FACE FROM ASIA TO EUROPE 57
give face to each other, and later expect a return favor. (In this same
frame, one can share a feud.)
One can also 'borrow someone's face', jie mownourende mianzi to take
advantage of another's reputation, for example in getting their reference
for a position. A reputable person in authority can ask people not to fight,
out of 'regard for my face', kan wo de mianzi (lit: look-at my face) or one's
family in the same way. But in zhaogu wo de mianzi 'be considerate of my
face9, it could be a personal situation, such as when my daughter does
something wrong but people don't spread it about since they want to
protect my face. In the third person negative, mei gu wo de mianzi 'he
lacked consideration for my face' one can refer to someone's failure to
leave some face, for instance by criticism in front of one's children. The
difference is the formal vs family relationship involved, but in both cases the
abstract mianzi is used. One can 'add face' by generosity, 'leave someone
some face' not revealing their mistakes, or 'pad someone's face' by
deference.
Unlike English and French, in these three languages, we find very few
expressions which involve the concept of saving face, the opposite of losing
when face is treated as a single whole. In Chinese wanting face for oneself,
yao mianzi, is an attempt to keep one's public image or appearance, even
if by subterfuge or excessive display. A second or third person version is lia
dian mianzi (lit: save a-bit-of face) 'keep them some face, don't make them
lose all their face,' just as one might save them space. The failure to
mention errors saves face for another person, to protect his reputation. For
the same reason, a public official might be given an empty job (Hu 1944)
to 'leave him some face'.
What is the difference between lian and mianzi since the collocations
are so similar? According to Hu (1944) the latter is more sensitive to status,
style, and connections. Hu describes the first as like integrity, and the
second as like social clout or prestige, pointing out that "Robin Hood" or
generous bandits have lian but not mianzi. To lose lian is a serious matter.
Mentsu is the Japanese equivalent of mianzi. Related expressions
include in Japanese mentsu o ushinau 'to lose face'(as one can lose time),
mentsu ni kodawaru 'to be concerned about face', mentsu no tatsu
'face-saving' (lit: face standing up), and mentsu o tamotsu 'to preserve face9
(the verb for saving money).
In addition, there are the terms mianmu in Chinese, myenmok in
Sino-Korean and menboku in Japanese, all of which are based on the same
58 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
Chinese characters for face and eye. Although mianmu is no longer used
commonly in Chinese, myenmok and menboku are both used frequently in
Korean and Japanese respectively. In Korean, there is myenmok-i epsta 'no
face'. In Japanese, there is a large variety of phrases such as menboku o
kaifukusuru 'recover face' (like recovering health), menboku ni kakeru 'com
mit one's face; pledge one's honor, put one's face on the line' (like one's
life) and menboku o tamotsu 'maintain one's honor' (like maintaining an
income), menboku o usinait 'lose face' used when one has not met expecta
tions or done something illegal or unethical.
There is also a group of phrases focusing on reputation and honor as
represented by the whole body. These are based on the Chinese characters
for body-and-face. In Chinese, these characters are read timian, and are
used in fixed phrases such as shi timian 'incur a loss of face', (using an
archaic and rare verb). Weichi timian 'maintain face, keep up appearances'
employs a verb frame for abstract nouns such as 'stability' and 'peace',
Butimian de xingwei (lit: not face REL behavior) 'dishonorable conduct' or
'undignified behavior' is a frame in which evaluative adjectives like 'elegant'
or 'conforming' can be used to replace timian de.
Similarly, in Japanese, there are expressions such as taimen o tsukurou
'put up a good face, save face; pretending or mimicking', mibunsotoo no
taimen o tamotsu 'maintain an appearance worthy of one's position' and
ikka no taimen o yogosu 'dirty/discredit the family's face'.
The related Sino-Korean word is cheymyen 'body.face'. In Korean the
usual expression for a mild loss of face uses this form. Cheymyen-ul ilhta,
'lose face' (one can also lose a job or a child) or cheymyen-i epsta 'lack
face', 'be impolite'. If a man pursues a woman over-aggressively, he will lose
face, in this sense.
Causing a face loss to another person can be expressed in this way:
cheymyen-i sonsang-toyta
body.face-NOM damage-become
'face becomes damaged'
cheymyen-ul sonsang-sikhita
body.face-ACC damage-cause
'make someone lose face'
(This is used only for abstractions like achievement or reputation.)
SHIFTING FACE FROM ASIA TO EUROPE 59
cheymyen-ul yucihata
body.face-ACC 'maintain face' (as one maintains health)
These are vivid expressions for losing one's own face or having one's face
lost by someone else. The most extreme loss would be cheymyen-ul
kkakkanaylita 'cut down someone's face', meaning that I am getting you to
lose your face (just as I could get you to reduce a price). This could be
directed at a woman who has a fight with her husband and leaves her
in-laws' house to return to her parents' house.
Using this 'abstract face' expression, one can also talk of saving face:
cheymyen-i seta 'one's face stands, is saved.' or the causative version:
cheymyen-ul seywuta
'get one's face saved.' (lit: make stand one's body/face) (this verb could
also be used for plastic surgery raising the nose)
And one can ask to have one's face saved:
cheymyen com salyecwe
'please save my face' (lit: make my face alive, as in saving fish or
animals)
A different set of expressions in Korean involves the use of the
character for body, as in the case of mangsin-ul tanghata 'experience a
body-loss; be disgraced, suffer loss' (using an adversative passive, as in
'experience a divorce' or 'suffer a deception') and mangsin-ul sikhita 'make
someone lose their body/self; disgrace someone' (using a causative verb as
in 'make someone do an errand' or 'make someone study'). In the case of
these expressions, the concept is similar to face, but clearly refers to loss of
the whole body and person.
Finally, another set of Korean expressions dealing with the idea of face
is represented not by words for the face of the person but by the terms for
the front of the house. For example, in Korean, there are expressions such
as kamwun-ul pichnayta 'make the family/house shine' which are used if a
family member brings face to many generations (e.g. winning an award,
entering a prestigious university). Thus the front of the house, like the front
of the head, represents honor.
60 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
Social usage
means "this is not very delicious, but...", upon serving tea to a guest was
considered a polite formality in China, Korea and Japan. Our consultants
emphasized that humility is an important value taught from an early age.
Parents would not report praise to children by teachers, and boasting or
showing off is strongly disapproved.
In Asian cultures, face-defensive behavior is based on avoiding behavior
that risks causing one shame (Sugiyama Lebra 1974). People must exert
self-control and rely on ritual behavior in social interactions with others.
Direct self-exposure and spontaneity are frowned upon in adults. Often
negotiations will be conducted by or mediated through a third party in
order to avoid any loss of face of the two negotiating parties. In other
situations, a person might serve as a delegate for another party, in order to
avoid direct loss of face. Writing, as opposed to speaking directly, is also a
common tactic for avoiding direct self-exposure. Another crucial strategy in
Japan is the use of anticipatory communication, in which one party
'anticipates' the other party's needs and wishes by using omoiyari, or empa
thy. In this case, the second party never has to reveal the inattentiveness of
the first party by overt speech, as their needs are constantly met by the
omoiyari of the first party. For this reason, Doi (1973) claims that a
Japanese host, unlike an American host, would never tell a guest to "help
yourself"; a Japanese host would show sensitivity in detecting what the guest
might need (e.g. drinks, food) and would provide such things without even
asking. Clancy (1986) reports that omoiyari or empathy training by mothers
of children starts in conversational interaction as early as two years of age.
Sugiyama Lebra also discusses strategies for displaying face, which she
describes as "aggressive face". The purpose of displaying "aggressive face"
is to prove one's prestige. The speaker's aggressive face can be maintained,
promoted or demoted in relation to the hearer. The most socially accept
able way of doing this is by exhibiting conspicuous generosity. This can be
done in a variety of ways, such as entertaining guests lavishly, giving gifts
and offering help. The refusal of an offer of generosity can make both the
speaker and the listener lose face. This explains why people argue over
checks in restaurants. In Chinese, this sort of competition is called zheng
mianzi 'struggling for face' (struggle to hold face up). Another example is
the escalation of gift-giving in Japanese culture. Family A will give a gift to
family B, and family B will reciprocate by giving family A an even more
expensive and elaborate gift the next time around. Consequently family A
62 SUSAN ERVIN-TRIPP, KEI NAKAMURA, AND JIANSHENG GUO
will feel pressured to find an even more exotic and unusual gift for family
B for the next occasion, and this evolves into a vicious cycle.
Another less acceptable strategy for displaying face in Japan is
self-praise, or jiman. People sometimes boast about certain things related
to themselves, such as their children. Often this is done in an extremely
subtle and hardly detectable manner, or is preceded by the phrase "jiman
ja naindesu keredo" (I don't mean to boast, but...). Excessive boasting is
considered inappropriate. Self-deprecation is valued as one is always seen
not as alone, but as part of a supporting group. One should always speak
in self-deprecatory terms which show explicitly that one has no selfish
delusions of independent grandeur. Finally, the most aggressive and least
sociable form of displaying face is arrogance. As in the case of self-praise
and boasting, this strategy is not culturally approved and is highly discour
aged, a sign of poor upbringing.
Brown and Levinson (1987) have proposed that every individual has two
specific kinds of desires: (1) negative face: the basic claim to territories,
personal preserves, rights to non-distraction (i.e. to freedom of action and
freedom from imposition) and (2) positive face: the positive consistent
self-image or 'personality' claimed by interactants. They define negative face
as "the want of every 'competent adult member' that his action be unimped
ed by others" and define positive face as "the want of of every member that
his wants be desirable to at least some others" ( 1987: 61-62). However,
Matsumoto (1988) questions the universality of the proposed constituents
of face in the light of Japanese language and culture. She claims that the
notion that all individuals desire to defend their own territory from the
encroachments of others is based on Indo-European concepts of individuals
and their rights, which "cannot be considered as basic to human relations
in Japanese culture and society" (p. 405).
According to researchers such as Nakane (1970), Doi (1973) and
Sugiyama Lebra (1976), what is central to the notion of face is not one's
own territory, but one's position in relation to others in groups and accep
tance by those others. As summarized by Matsumoto, all social interaction
is governed by "acknowledgement and maintenance of the relative position
of others, rather than preservation of an individual's proper territory"
SHIFTING FACE FROM ASIA TO EUROPE 63
Korean, we also found that in general this seemed to be true for them. The
notion that independence could be violated only exists between strangers.
While interviewing our consultants, we provided them with examples of
contexts presented by Brown and Levinson(1987: 65-68) of acts which
threaten negative face and those that threaten positive face. Examples
Brown and Levinson gave of acts which primarily threaten the addressee's
negative face, by indicating that the speaker does not intend to avoid im
peding the addressee's freedom of action include orders, requests, sugges
tions, advice, remindings, threats, warnings, offers, promises, compliments
and expressions of strong emotions toward the addressee. Examples of acts
which threaten the positive face of the addressee, by indicating that the
speaker does not care about the addressee's feelings and desires include
disapproval, criticism, accusations, complaints, insults, disagreements,
challenges, mention of inappropriate topics, boasting, non-cooperation in
an activity and use of incorrect address terms.
The speakers of our three languages reported that only some of these
contexts seemed to them to suggest problems of loss of face, despite the
centrality of this concept to social behavior. Note that this is NOT the
method Brown and Levinson used, which was to examine the occurrence
of remedies as a clue to whether face was threatened, though not which
kind of face. First, let us look at those that Brown and Levinson (1987: 65-
68) say threaten negative face because they impose on the hearer's freedom
of action.
1. "Those acts that predicate some future act A of H, and in so doing put
some pressure on H to do (or refrain from doing) the act" (B & L: 65)
(a) Making requests can lead to loss of face for a requester but only
when requests are excessive relative to social position. On the contrary,
failure to ask help of an appropriate person when you need help can make
the person who could help lose face; other people might conclude he is
stingy. That is, far from always involving a threat to the face of a hearer,
making requests can be a way of honoring someone by giving them a
chance to carry out their role - a point made vividly by Matsumoto.
Correspondingly, thanking an insider, who would normally normally give
assistance, is demeaning to their face, and distancing. "Utterances of praise
or expressions like 'thank you', 'please' or 'excuse me' work to create
distance, subtly subverting the mutual support and consideration taken for
granted among family intimates" (Young 1994.)
SHIFTING FACE FROM ASIA TO EUROPE 65
2. "Those acts that predicate some positive future act of S toward H, and
in so doing put some pressure on H to accept or reject them, and possibly
to incur a debt" (B & L: 66)
Like a request, an offer is seen by Brown and Levinson as a potential
intrusion on freedom of action. An offer has to be evaluated in the context
of its symbolism of social relationship. There are very strong social customs
surrounding offers of food, for example, which may be distinct from other
situations of offering. In some situations, offers are refused until they are
repeated several times, so that the acceptance does not imply over-
eagerness by H. The excuses offered help calibrate the intentions of both
parties. An H who persistently rejects a strong offer, for instance of a gift,
makes S lose face since the cycle of favors will keep people in the same
social circle. Rejection thus implies social rejection. In this analysis we
would expect social remedies in H's excuses to mitigate the threat to S's
positive face.
There is more agreement about positive face threats, but some of these
acts are rude, and lose face for the speaker rather than the hearer.
1. "Those that show that S has a negative evaluation of H's positive face"
(B & L: 66):
Public criticism (except in politice) is likely to be avoided or and highly
mitigated. But accusations in some contexts can involve public face or
prestige, gaining face for the accuser and losing it for the target.
"Putting thoughts vaguely or obliquely helps to mitigate threats to face and person
[...] Chinese regard one's ideas as entangled with one's identity or sense of personal
worth; an attack on one's ideas is therefore an attack on one's self, or, more
specifically, one's face. In this event, Chinese try to minimize threats to face by
hedging one's claim while looking for commonality with others. Chinese avoid a
strong stance when criticizing another's argument." (Young 1994).
2. "Those that show that S doesn't care about (or is indifferent to) H's
positive face" (B & L: 66):
(a) Expressions of violent emotions lose face for S.
(b) Boasting implies a 'thick-skinned face,' and is thus a defect in the
boaster, and not in the hearer. The boaster loses face.
SHIFTING FACE FROM ASIA TO EUROPE 67
4. Conclusions
References
Altshuler, Kenneth Z.
1984 "On the question of bisexuality." American Journal of Psychotherapy
38.484-493.
Appleton, WW
1951 A cycle of Cathay. New York: Columbia University Press.
Berk, Bernard
1977 "Face-saving at the singles dance." Social Problems 24.530-544.
Brown, Penelope
1990 "Gender, politeness, and confrontation in Tenejapa." Discourse Processes
13.123-141.
Chu, Chin-Ning
1992 Thick face, black heart. Mill Valley CA: A.M.C. Publishing.
Clancy, Patricia
1986 "The acquisition of communicative style in Japanese." In Bambi Schieffelin
& Elinor Ochs (eds.), Language acquisition and socialization across cultures.
Cambridge: Cambridge University Press, 213-250.
Doi, Takeo
1973 The anatomy of dependence. Tokyo: Kodansha.
1986 The anatomy of self: The individual versus society. Tokyo: Kodansha.
Erickson, Richard C.
1984 "Tell me a story." Individual Psychology Journal of Adlerian Theory, Research
and Practice 40.277-289.
Kirita, Katsutoshi
1984 "Tasha ni yoru kyozetsu to jisonshin." [Rejection by others and self-esteem.]
Soshioroji 29.69-86.
Matsumoto, Yoshiko
1988 "Reexamination of the universality of face: Politeness phenomena in
Japanese." Journal of Pragmatics 12.403-426.
Miller, Abraham H.
1978 "Negotiations for hostages: implications from the police experience." Ter
rorism 1.125-146.
Nakane, Chie
1970 Japanese society. Berkeley: University of California Press.
Robert, Paul
1985 Dictionnaire alphabetique et analogique de la langue française. 2nd ed. Paris:
Le Robert.
Tjosvold, Dean
1983 "Social face in conflict: A critique." International Journal of Group Tensions
13.49-64.
Young, Linda
1994 Crosstalk and culture in Sino-American communication. Cambridge: Cam
bridge University Press.
Mental-space builders:
Observations from Japanese and English
conditionals1
Seiko Yamaguchi Fujii
University of California at Berkeley
and University of Illinois at Urbana-Champaign
1. Introduction
This paper stems in part from a project on English and Japanese conditional
sentences on which I collaborated with Charles J. Fillmore in 1987-88; I have benefited
immeasurably from discussions with him; however, I am responsible for any errors or
omissions found in this paper. A slightly different version of this paper was presented at
the LACUS Forum, in Montreal, August 1992, and awarded the LACUS Presidents'
Predoctoral Prize. My thanks also go to Eve Sweetser, Gilles Fauconnier and Sandra A
Thompson for their encouragement.
74 SEIKO YAMAGUCHI FUJII
(Section 4).
(1) In France Watergate wouldn't have been a scandal.
(Fauconnier)
(2) You are so lucky. My advisor wouldn't have been so patient.
(3) A new highway would save us a lot of time.
2
Devices for expressing conditionality in Japanese include particles such as to, and
nara, verb inflections such as -tara, -(r)eba, and -vmo, and the function noun toki (time)
and baai (case) optionally followed by the case-marker, ni, and/or the topic marker, wa.
3
A more detailed discussion of the idiomatic nature of conditional constructions can
be found in Fujii (1992).
MENTAL-SPACE BUILDERS 75
In (1), (7), and (8) below, the adverbial phrase, here a prepositional phrase,
expresses the state of affairs in which the proposition contained in the rest
of the sentence holds true. These adverbial prepositional phrases are
'frame-setters': They set up a locational or situational frame of reference,
5
Fauconnier's (1985) discussion of mental spaces includes detailed analysis of
conceptual entities which appear in different mental spaces, and the mechanisms of how
each conceptual entity is mapped from one space to another.
MENTAL-SPACE BUILDERS 77
This is taken from the published English version of a Japanese original; the
Japanese sentence appears as (13) below.
78 SEIKO YAMAGUCHI FUJII
Another piece of data (11) comes from a T.V. commercial for a new
cereal, called Nutri-Grain Nuggets, which is compared with a similar brand,
Post Grape Nuts.
(11) <In a T.V. commercial advertising 'Kellogg's Nutri-Grain
Nuggets'>
<the initial remark>
(11a) Once you taste new Kellogg's Nutri-Grain Nuggets, you'll forget
about Post Grape Nuts.
<the final remark>
(11b) The taste will make you forget your old nuggets.
What is interesting about the discourse in this commercial is that it initially
uses a bi-clausal sentence with the connective ONCE (11a), to relate the
condition (tasting the new brand of cereal) and the result (forgetting about
the old brand of cereal); by constrast the concluding remark, shown in
(11b) uses a causative construction to repeat the same idea. This causative
sentence can be paraphrased as Tf you taste Nutri-Grain, you will forget
your old nuggets.' The two sentences exemplify the fact that English often
allows flexible alternation between these constructions.
Causality is an indispensable issue in the study of conditionals (cf.
Mackie 1965). The question of how and what kinds of causal meanings
figure in various conditional sentences has thus been one of the central
issues in the study of conditionals (see e.g., Comrie 1987; Sweetser 1990).
Causative constructions, of course, also express cause-effect relations. The
examination of these two formally different grammatical constructions,
namely bi-clausal conditionals and causatives, should prove a fruitful area
of inquiry, especially when we compare preferred event representation
patterns crosslinguistically (see Section 4,3 below).
The Japanese counterparts of English sentences (1), (2), (3), (8), and
(11b), discussed earlier, are given in (12) through (16) below. (12) and
(13) are the counterparts of English sentences (1) and (8), in which
prepositional phrases set up the mental spaces. (14) is the counterpart of
the counteridentical case, (2). (15) and (16) are the counterparts of the
causative cases, (3) and (11b).
(12) Huransu de dattara wootaageeto-ziken wa
France-IN copula-coND Watergate-incident-TOPic
9
The original Japanese sentence from which this sentence was adapted is given
below:
Ziyuuna kyoosoo no moto de nara sagaru hazu no kakaku mo, kisei de ganzi-garame
no zyootai dewa sagaranai.
- Tensei-jingo; Vox Populi, Vox dei, 1989, Asahi Newspaper.
MENTAL-SPACE BUILDERS 81
warurete-simau (daroo).
forget (MOD)
If (you get to) know this taste, you will forget your old nuggets.
(cf.11b) The taste will make you forget your old nuggets.
Note that all of these Japanese sentences contain conditional markers,
which are used as clause-linking morphemes (i.e., TARA in (12) and (14),
NARA in (13), and (R)EBA in (15) and (16)). General statements such as
sentence (13a) could be expressed with the topic marker WA, instead of
NARA. But it would be difficult for the other sentences to express
'conditionally' without marking the conditional adverbial adjuncts with
these conditional linkages.
I must add here that it is not simple to determine the syntactic status
of the adverbial adjuncts in these Japanese conditional sentences. For
example, do we analyze 'boku no adobaizaa dattara9 in (14) as a genuine
subordinate clause, or a constituent of the main clause? To a degree, the
answer depends on the syntactic theory which we use to describe the
various linkage types.10
But aside from the precise syntactic status of the different types of
linkage (which we need to explore in another context), the point to be
10
For example, to capture the difference in syntactic status between 'boku no
adobaizaa dattara' in (14) and 'Atarasii haiuee ga dekireba' in (15), Role and Reference
Grammar (Foley & Van Valin 1984; Van Valin 1984) might treat the former linkage at
the 'core' level while treating the latter at the 'clause' level. The hierarchical theory of
linkage in Mihami's Grammar (Minami 1964,1974) might categorize the former as a 'Type
A linkage', and the latter as a 'Type B linkage'.
82 SEIKO YAMAGUCHI FUJII
made here pertains to Japanese and English devices for building mental
spaces: Japanese, by using conditional linkage, marks the mental-space
builder (antecedent), whereas English does not have to mark the mental-
space builder with a conditional expression. This contrast in the mechanisms
for building mental spaces and expressing conditionality is systematic, and
reflects general divergences between the two languages, in particular,
regarding i) the mechanisms for expressing modality; ii) the clause-linking
mechanisms; and iii) the argument structures.
Actually, 'would' has other uses (e.g., past habituais). But they are quite marked.
12
This idea was first presented in a plenary speech entitled 'Mysteries of Conditional
Sentences in English and Japanese', by Charles J. Fillmore, in collaboration with me, at
the Annual Meeting of the ATJ (Association of Teachers of Japanese) in San Francisco,
March 1988.
MENTAL-SPACE BUILDERS 83
cf. (19') ... isioki sitai yo. (# nasaru yo. / # sitagaru yo.)
cf. (20') ... nasaru yo. /... sitagaru yo. (# osioki sitai yo.)
Sentence (18), which is the structural equivalent of the English sentence
(17), containing a full clause antecedent, can be uttered in either situation,
as in (17).13 But, as shown in (19), with the truncated antecedent Watasi
dattara (watasi, T, followed by DATTARA), only the first interpretation is
possible. The sentence (20) with the different truncated antecedent Kimi
no otoosan dattara (kimi no otoosan 'your father' followed by DATTARA)
is uttered only in the second situation.14 In either case, the mental-space
It seems that sentence (18) more typically receives the first reading while sentence
(ii) below more typically receives the second reading.
(ii) Kimi no otoosan ga watasi dattara, hippataite isioki suru yo.
you GEN father NOM I copuIa-TARA(COND), spank punish PART
(If your father were I, he would spank you.)
14
Another case in which the sentence gets only one reading is when a certain
property of the main-clause verb identifies the subject - whether it is a first person (I) or
a third person (your father) (Thanks to Yoko Hasegawa for drawing my attention to this).
For example, sentence (18) with 'sitai' only gets the first reading, whereas, with 'sitagaru'
or 'nasaru' (honorific), it only gets the second reading. Thus, notice that 'sitai' can appear
in sentence (19) but not in (20); 'sitagaru' or 'nasaru' can appear in (20) but not in (19),
as shown in (19') and (20').
MENTAL-SPACE BUILDERS 85
I am using only examples with TARA here, but other conditional markers (e.g.,
NARA) can also be used similarly.
The on-line text of this science textbook (written in Japanese, and used by junior
high school students in Japan) was made available by the Communications and
Information Processing Laboratory, Nippon Telegraph and Telephone Corporation (NTT),
Yokosuka, Japan. Sponsored by NTT, this text was used in a larger project on Japanese
clause-linkage, directed by Charles J. Fillmore. As part of the project, the original
Japanese text was translated into English by a team of bilingual speakers of Japanese and
English; first by the native speakers of Japanese, then by the native speakers of English.
Translation and editing took place through active verbal interaction, with the team
(normally four to five participants) recording various versions of the translations.
86 SEIKO YAMAGUCHI FUJII
(22A) When one heats silver dioxide, silver and oxygen are produced.
5. Conclusions
It has been noted that Japanese tends toward lower transitivity, compared with
English. This tendency has been discussed in various approaches, including analyses of
morpho-syntactic properties and case markings of intransitive and transitive verbs
(Jacobsen 1982,1985; Sugamoto 1982; Teramura 1978), and analyses of fine-graned lexical
semantics of verbs (Ikegami 1985), Ikegami (1981, 1982, 1991), among others, has also
investigated the issue of agentivity and transitivity from the point of view of event
realization patterns, which he attributes to cultural rhetorical patterns.
88 SEIKO YAMAGUCHI FUJII
References
Bowerman, Melissa
1986 "First steps in acquiring conditionals." In E.G. Traugott, A. Meulen, J.S.
Reilly & C.A. Ferguson (eds.), On conditionals. Cambridge: Cambridge
University Press, 285-308.
Comrie, Bernard
1986 "Conditionals: A typology." In E.C. Traugott, A. Meulen, J.S. Reilly & C.A.
Ferguson (eds.), On Conditionals. Cambridge: Cambridge University Press,
77-99.
Fauconnier, Gilles
1985 Mental spaces: Aspects of meaning construction in natural language.
Cambridge, MA: M.I.T. Press.
Fillmore, Charles J.
1987 "Varieties of conditional sentences." Proceedings of the Third Eastern States
Conference on Linguistics, 163-182.
Goodman, Nelson
1983 Fact, fiction, and forecast. Cambridge: Harvard University Press.
Ikegami, Yoshihiko
1981 SURU to NARU no Gengogaku [Linguistics of DOING and BECOMING].
Tokyo: Taishuukan.
Ikegami, Yoshihiko
1982 Hyoogen Koozoo no Hikaku: SURU-teki na gengo to NARU-teki na gengo
[Contrast of linguistic representation types: 'DO-language' and 'BECOME-
language']. In Tetsuya Kunihiro (ed.), Nichieigo-hikaku-kooza: Hassoo to
Hyoogen [Japanese-English comparative grammar]. Tokyo: Taishuukan, 286-
326.
Ikegami, Yoshihiko
1985 "'Activity'-'Accmplishment'-'Achievement': A language that can't say 'I
burned it, but it didn't burn' and one that can." In Adam Makkai, & Alan
K. Melby (eds.), Linguistics and philosophy: Essays in honor of rules S. Wells.
Amsterdam/Philadelphia: John Benjamins Publishing Company, 265-304.
Jacobsen, Wesley
1982 Transitivity in the Japanese verbal system. Indiana: Indiana University
Linguistics Club.
Jacobsen, Wesley
1985 "Morphosyntactic transitivity and semantic markedness." In CLS 21: Papers
from the Parasession on Causatives and Agentivity. Chicago: Chicago
Linguistics Society.
Jespersen, Otto
1940 A modern English grammar on historical principle, Vol. v.: Syntax. London:
George Allen and Unwin.
Mackie, J.L.
1965 "Cause and Conditions." American Philosophical Quarterly 2:4.245-264.
90 SEIKO YAMAGUCHI FUJII
Minami, Fujio
1964 Fukubun [Complex sentences]. Tokyo: Meijishoin.
Sugamoto, Nobuko
1982 "Transitivity and objecthood in Japanese." In Paul J. Hopper & Sandra A.
Thompson (eds.), Syntax and Semantics, Vol 15: Studies in transitivity. New
York: Academic Press.
Sweetser, Eve
1990 From etymology to pragmatics: Metaphorical and cultural aspects of semantic
structure. Cambridge: Cambridge University Press.
Teramura, Hideo
1978 'NARU' hyoogen to 'SURU hyoogen: Nichiei 'tai' hyoogen no hikaku." In
Nihongo To Nihongo-kyooiku [Japanese and Japanese Language Education].
Tokyo: Kokuritsu Kokugo Kenkyuusho.
1. General considerations
The last twenty years in linguistics has seen a great enrichment of the
conception of the lexicon. This movement, usually called Lexicalism, can be
viewed as a shaping force in all of the currently active theories. What
became evident in the years following 1970 was that a great number of the
properties of a grammar could be explained as projections of lexical
properties. Much of the variation in the grammar that was previously
handled by mappings between trees could be more perspicuously handled
as variation of lexical properties, which were then projected into the
grammar by very general principles. As a consequence, an account of the
projection of lexical properties lies at the heart of all current syntactic
theories.
The grand success of this enterprise and the great fruitfulness of this
paradigm might lead one to suppose that all of grammar can be accounted
for as the projection of lexical properties. This is clearly not the prevailing
view in contemporary syntactic theory. In the Governement Binding Theory
of Chomsky 1981, for example, one might say that the Government half
(and everything connected with the Projection Principle) deals with lexical
projection, but the Binding Theory clearly does not. Recent overhauls of
the theory of phrase-structure have put a great deal more weight on
projections of material from inflectional morphology and closed class items
such as complementizers, but the basic separation between lexically
projected and non-lexical still remains.
It is thus one of the central tasks for a linguistic theory to characterize
which properties of a grammar are projections of lexical properties and
which are not. Construction Grammar as developed in Fillmore 1987 and
92 JEAN MARK GAWRON
2. Distance expressions
2.1 A construction
There are however a few very special cases, involving NPs that describe a
property:
(13) a. A dog the size of a small pickup truck was procured.
b. A birthmark the color of an eggplant floated on his fore
head.
What is going on in (12) appears different from what is going on with the
distance nouns: There seems to be no reason to think 200 miles from Boston
is a description true of the property of being 200 miles from Boston, while
there is good reason to think the color of an eggplant is a description true
of a color property.
One more point of importance: The very same distance NPs can also
function adverbially:
(14) a. The meteorite landed a short distance from the city.
b. Mary lived twenty miles from the sea.
c. Jasha lived two day's journey from Lublin.
Again, there are other NPs that act as adverbial modifiers, but their
semantics is special:
(15) a. John shaved the way he ate aspic. Reluctantly.
b. Henryka lived on Opaczewska Street her entire stay in
Warsaw.
c. Irena left the day Miszka arrived.
The question is: What about the distance nouns links them to these other
adverbial heads?
Summing up the problem: There is a class of NPs denoting distances
which can be adnominal or adverbial modifiers. Thus, the preliminary
answer to the syntactic half of our initial question about (6) seems to be
that it is an NP whose head is miles. This answer in turn raises other
questions: Why are these NPs allowed in contexts where others are not?
What we are essentially, trying to explain here is the possibility of
certain noun-phrases acting as adjuncts. In order to take a step toward such
an account, I want to borrow an idea from HPSG. This is the idea that
adjuncts select for their heads. In HPSG, that reduces to explaining why
these noun phrases can select nominal and verbal heads to modify. I will
adopt a simplified version of this approach, trying to restrict myself to the
apparatus necessary for the problem at hand. Let us assume that lexical
entries have attributes of various sorts associated with them; among these
DISTANCE IN CONSTRUCTION GRAMMAR 97
are three that will concern us in particular, MOD, DENOTATION and MOD-
CONTENT. MOD gives a syntactic and semantic specification of what
projections of this head may modify. DENOTATION tells us what it dnots.
MOD-CONTENT tells us the semantic result of applying this modifier to this
head. The role of these three different attributes can be illustrated with an
adverbial noun like way. Let us suppose that way is a property of manners,
and way that John eats aspic a property of manners that John eats aspic.
The NP the way that John eats aspic should denote a particular manner
John eats aspic in. This is what I propose to call the denotation of the NP.
This denotation will be used the way the denotation of any NP is used in
sentences like:
(16) I like the way John eats aspic.
Now the MOD feature of the NP will telll us that it can modify verb
phrases, perhaps imposing semantic restrictions that determine which verb
phrases. The MOD-CONTENT of the NP will tell us the resulting semantics
of the verb phrase once the NP modifies it. In effect adjuncts are allowed
to carry their own semantic rules for combining with their heads.
Let us assume a Davidsonian semantics, and that the desired semantics
for the meteor landed some distance from Boston is the following:
(17) [land(e) actor(e, the-meteor) distance (e, Boston,d)]
Here e is an event variable, and d a distance variable. The relations
DISTANCE holds between two things and a distance when the things are
that distance apart. Then what the NP some distance from Boston should
specify for its three relevant features is:
changing properties of the construction which will help draw out the
differences between a lexical rule account and a bifunctional account.
The correct account needs to capture a basic fact: The complements
that can attach to distance nouns when they are ordinary NPs are different
from those in the adjunct function:
(20) a. the distance from New York to Boston
b. a city some distance from New York
c. *a city some distance from New York to Boston
d. *a city some distance to Boston
To see how this can make sense on the bifunctional account, let's propose
a noun entry that includes some details left out above:
[CAT N I
CAT V OR N 7
MOD DENOTATION π
[ DENOTATION-VAR g J
(21) CAT P CAT P
SUBCAT PFORM from, PFORM to
DENOTATION s
DENOTATION g\ J
MOD-CONTENT π A distance (g, s, d)
[DENOTATION d such that distance (g, s, d) J
I will refer to s,g, and d as the source, goal, and distance, respectively. The
idea here is that under the attribute SUBCAT, the noun contain a list of
subcaterorized-for complements, as is standard in HPSG. The crucial thing
to note is that the to-phrase fills the same argument position of the
distance-relation as the heads the NP adjunct can modify. The DENOTA
TION-VAR of the MOD and the DENOTATION of to-phrase fill the goal
argument-position of the distance relation; and both fill the source
argument-role of the distance-relation.
Given (21), the unacceptability of (20c) and (20d) will follow now if we
have some principle that precludes filling the same argument position of the
same relation twice in a single syntactic configuration. Now the competing
lexical rule account will accomplish the same by not permitting the adjunct
NP to subcategorize for a to-phrase. This is quite natural if we think of the
rule as a distant cousin of passivization: What starts out in an important
syntactic-semantic slot (the denotation role of the NP) becomes a non-
player; something else that began as oblique (the to-phrase) gets 'promoted'
100 JEAN MARK GAWRON
to be the argument of the adjunct-NP. In effect, the lexical rule just splits
the lexical entry in (21) in two. We start with:
It is worth pointing out here that what I am calling a lexical rule account
does not need to be thought of as a rule at all. If we simply think of (22)
and (23) as two distinct types licensing classes of nouns, with the relation
distance itself a type admitting further specification, then (22) and (23)
together say that there will be nouns with distance semantics with the
indicated properties. Developing this view would probably give us a more
constructional view of what is going on. In the interests of employing
familiar notions where possible, I will, however, continue to refer to a
lexical rule account. When I refer to the lexical rule account as defining a
construction, I will mean the syntax and semantics in (23).
The striking feature of the lexical rule account is that one needs to posit
non-denotational NPs (or rather, NPs that can only denote properties, like
adjectives or adverbials). The striking feature of the bifunctional account is
double-barreled semantics. One might think that tests of the denotational
status of NPs like:
(24) a. the distance from Boston
b. some distance from Boston
will help one decide between the two accounts. This turns out not to be the
case. For exammple, NPs with the definite article appear to preclude the
adjunct use, whereas those with some appear to preclude uses referring to
distances:
DISTANCE IN CONSTRUCTION GRAMMAR 101
It is a little more difficult to force adjuncts uses, but perhaps this can be done with
a special some.
Now I want to turn to characterizing the class of nouns that allow this
strange construction. In effect, we want to characterize a class of nouns that
are going to be allowed to have an adjunct semantics as well as a denota-
tional semantics. I will view this as the licensing of another meaning, even
though the account I propose does not posit a distinct lexical entry to carry
that meaning.
I propose a rule of semantic transfer or a semantic license (in the sense
of Nunberg and Zaenen 1992) of something like the following form:
Semantic License A
A noun which denotes a basic three-place relation among a SOURCE,
a GOAL and the distance EXTENT between them, and which functions
as a property of distances when saturated, can also function as a
property of GOAL»
I use the term saturated here not inn usual sense (having proposition or
term semantics), but to mean something which has gathered together all its
syntactically realizable arguments. Thus a relational noun like picture is
saturated when it has picked up its of-phrase complement. When the noun
distance has picked up its two complements as in distance from Boston to
New York, it is saturated. Since it denotes a property of distances when
saturated, it is a noun of the right sort to undergo license A. On the
bifunctional account, License A is in effect a semantic statement of the
conditions that must be fulfilled for nouns to have the sort of DENOTATION
and MOD- attributes posited in (21). On the lexical rule account, it is an
input condition on the rule. On both accounts one might well fold the
semantic conditions into the rule itself, so that a separate statement of
License A would be redundant, but it will be useful for our purposes to
have a characterization of the semantics of the nouns independently from
the construction.
But A has been stated in a very restricted fashion, so that it includes
only properties of distances. What about the following cases:
(28) a. That city is a 200 mile drive from Boston.
b. New York is a 200 mile journey from Boston.
observation, I have no detailed proposal to make. This, however, is the sticking point in
the general existential quantification invoked above in the semantic machinery:
Distinctions need to be made between those things that need to be anchored in the
discourse and those that do not.
DISTANCE IN CONSTRUCTION GRAMMAR 103
One might try to bring these nouns in under the umbrella of License A on
the basis of a metonymy account. Thus, note:
(29) The drive to Boston is 200 miles.
Here a drive too Boston seems to be being equated with a distance. A
metonymic map from travel events to the distances they cover would
account for such uses. Moreover, if travel nouns can metonymically be
mapped to distance nouns, then they will come under License A, and
adjunct uses will be licensed. The trouble with this is that travel nouns do
not fit in well everywhere that distance nouns do:
(30) a. ?John ran a 2 hour drive.
b. ?The airliner flew a two-day journey over the jungle.
Now these judgements are perfectly explicable if nouns like journey and
drive denote travel events in most contexts. But then how do we handle the
cases in (28)?
I assume that heads like drive and journey can be handled by the
following license:
Semantic License B
A noun which denotes a basic three-place relation among a SOURCE,
a GOAL and the event of moving between them and which functions as
a property of movement events when saturated, can also function as
property GOAL.
This is simply license A with movement events replacing distances. Now
under License B travel nouns still denote travel events in their adjunct uses,
and verbs that select for distances may still select against them. This is
presumably what is happening in (30).
It is License B, I suggest, that truly establishes the constructional nature
of (21) or of (23). We have now proposed that the same unusual syntactic
property extends to distinct but semantically related classes. The crucial
analogue is the extension of the ditransitive construction from events of
intended or successful transfer of objects (give, win) to verbs enabling a
transfer (permit, allow) and to verbs describing actions that will cause a
transfer at a future time (reserve). This is what Goldberg 1991b calls
constructional polysemy. What it illustrates is a construction exhibiting the
wordlike property of acquiring distinct but related meanings. Moreover the
polysemic mapping between distance and travel event is independently
motivated in other constructions, as shown in (29).
104 JEAN MARK GAWRON
3
Thanks to G. Nunberg for pointing out this question.
4
In the interests of fair play, I point out that this fact presents a problem for the
account being presented here which is not shared by the lexical rule account. Capturing
the fact that the from -phrase is obligatory in the adjunct function is not going to be easy
in a single lexical entry. Matters may not be as bad as they first appear, however. NOte
that postnominal modifiers with nothing following the head are in general, marginal: *a
man happy versus a man happy with his lot. Note also that predicative uses without the
from -phrase do appear possible:
(i) Boston is a long journey.
(ii) Denver is a two-hour walk.
DISTANCE IN CONSTRUCTION GRAMMAR 105
always filled the goal's argument position, no goal complement could ever
occur. But suppose we did allow a goal complement on the subcat list. We
could still appeal to a principle requiring that argument roles be filled only
once per structure to block:
(34) *a city some distance to New York
The question is: What would then be the status of the adjunct NP? With all
its argument positions saturated it would denote a proposition and would
be unable to function as an ordinary NP (recall that the basic purpose of
(23) was to provide a lexical entry that could only function as an adjunct.
With this semantic nature, it would not be at all surprising to find such an
NP showing up in a predicative construction with a dummy subject.
Consider:
(35) a. That New York is far from Boston is obvious to everyone.
b. It is obvious to everyone that New York is far from Boston.
Sentence(35) shows a somewhat different case of a dummy subject. If we
think of obvious as denoting a relation between an EXPERIENCER and a
PROPOSITION, then in (35b), both arguments have been supplied within the
verb phrase, and a dummy subject is the natural way to fill the requirement
of English grammar that each sentence have a subject. Now the option of
taking a dummy subject and supplying all arguments within the verb phrase
is a marked one in English grammar. Except for our path nouns, the other
cases all seem be heads taking propositional arguments. 5 It thus appears
that some stipulation would be needed to license the option for distance
nouns. But that stipulation would be a good deal less remarkable if the
dummy subjects were licensed for non-denotational nouns with all their
arguments in the VP
Consider on the other hand, the bifunctional account of (21). The point
there was to deal with adjunct and denoting uses in a single lexical entry.
But on such an account some distance from Boston to New York is just an
ordinary denoting NP Although we could still posit a lexical rule licensing
a dummy subject construction for this NP, there is no particular reason to
expect it.
5
Possibly property arguments are just as good, as in:
It costs twenty dollars to get to Scranton.
DISTANCE IN CONSTRUCTION GRAMMAR 107
6
It is striking here that the same kind of definteness restriction we saw on the
distance nouns seems to apply to the propositional nouns with dummy subjects:
?It was the fact that Bill coming.
108 JEAN MARK GAWRON
As noted in 2.1, there are several asymmetries between SOURCE and GOAL
in these constructions. The missing piece here, and perhaps the most
interesting piece of all, is the role of these notions in the semantics of
distance.
Here is what looks like another example of such an asymmetry. If the
sentences in (31) are analogous to ordinary cases of extraposition, one
would not expect either of the two prepositional phrases to be special.
Thus, the following contrast may still need to be accounted for:
(42) a. *How far from New York is it to Boston?
b. ?How far to Boston is it from New York?
3. Conclusion
We find distance heads in the following kinds of uses. In each case, the
head in question has been italicized:
(43) a. The distance from New York to Boston is 300 miles.
b. Boston is a short distance from New York. (License A)
c. Boston is a short drive from New York. (License B)
d. It is a short distance from New York to Boston. (Dummy-
Subject)
e. It is a short drive from New York to Boston. (Dummy-
Subject)
I would like to conclude by focusing on the significance of the entry of
dummy subjects into this construction, particularly in cases of travel nouns,
as in (43e). I suggested in 2.1 that occurrences of travel nouns in the
adjunct function were a case of constructional polysemy. Sentence (43e)
shows that constructions feed lexical rules. More precisely, it shows that
extending the basic meaning of the distance-noun adjunct construction to
travel nouns can license lexical entries to feed other lexical processes, here
the rule that licenses dummy subjects.
References
Chomsky, Noam
1981 Lectures on Government and Binding. Dordrecht: Foris.
DISTANCE IN CONSTRUCTION GRAMMAR 109
Fillmore, Charles
1968 "The Case for Case." In E. Bach, and R. Harms (eds.), Universals of
Linguistic Theory. New York: Holt, Rinehart, and Winston.
Goldberg, Adele
1991a "On problems with lexical rule accounts of argument structure." Cognitive
Science Conference Proceedings.
1991b "The inferent semantics of argument structure: The case of the English
ditransitive construction." Cognitive Linguistics
Morgan, Jerry
1978 "Two types of convention in indirect speech acts." In P. Cole (ed.), Syntax
and Semantics 9: Pragmatics. New York: Academic Press.
Ferenc Kiefer
Research Institute for Linguistics,
Hungarian Academy of Sciences
his account of lexical items in terms of frames. (Fillmore 1975, 1976 and
1982)
The paper begins with a brief overview of some of the earlier work on
temporal interpretation. Particular attention will be paid to the exceptions
(or counterexamples) which were noted with respect to each of these
approaches. It will also be pointed out that a number of these exceptions
have been known for some time from research in text linguistics.
The second part of the paper will be devoted to the discussion of the
types of lexical information which may be relevant for temporal interpreta
tion. It will be shown that a significant number of problematic cases can
easily be handled if this information is taken into account.
1. Perfectivity
The Hungarian verb összeturkal 'ransack' has clearly the meaning that 'the place was
left in a mess', while the verb átkutat doesn't have such an implication.
114 FERENC KIEFER
2. Tense
Kamp and Rohrer have proposed in a number of works (Kamp 1979, 1981,
Kamp & Rohrer 1983) that in French temporal structure is determined by
tense. The passé simple (replaced by the passé composé in spoken language)
is used to express temporal advancement whereas the imperfait expresses
the background or the discourse situation. The contrast between the two
tenses can clearly be demonstrated by sentence pairs such as (7)(a) and
(7)(b).
(7) (a) Quand Pierre entra, Marie téléphonait
'When Peter entered, Mary was making a phone call9
(b) Quand Pierre entra, Marie téléphona
'When Peter entered, Mary made a phone call'
In (7)(a) the first verb form is in the passé simple, the second one in the
imperfait. In (7)(b), on the other hand, both verb forms are in the passé
simple. In (7)(a) the second event expresses the discourse situation, i.e. the
background for the first event and there is no temporal advancement. In
(7)(b), on the other hand, the second event follows the first event. This
observation has led to the formulation of the following principle.
(8) If the sentence is in the passe simplé, its event follows and does
not overlap with the event of the previous sentence. But if this
sentence is in the imparfait, its event overlaps temporally with
that of the previous sentence.
This principle is obviously too strong. In the literature the following
exceptions were noted (Kamp - Rohrer 1983; Vet - Molendijk 1986).
(9) L'été de cette année là vit plusieurs changements dans la vie de
nos héros. François épousa Adèle, Jean-Louis partit pour le Brésil
et Paul s'acheta une maison à la campagne
'Last summer there were several changes in the lifes of our
heroes. Frank married Adela, John left for Brazil and Paul
bought a house in the country-side'
(10) Marie chanta et Pierre l'accompagna au piano
'Mary sang and Peter accompanied her on the piano'
(11) Nadine s'assit et lut la lettre. Elle fut étonnée de constater que la
lettre la touchait beaucoup
THE TEMPORAL INTERPRETATION OF DISCOURSE 115
'Nadine sat down and read the letter. She was surprised to find
out that the letter affected her a lot'
(12) Jean monta dans sa chambre. Pierre le suivit
'John went up to his room. Peter followed him'
The first sentence in (9) expresses the main event and the three clauses
of the second sentence represent subevents of this main event. Notice that
the subevents are unordered: They may have occurred in any order within
the time interval defined by the main event. Notice, furthermore, that the
first sentence of (9) contains the quantifier plusieurs 'several'. It would seem
that, in general, subevents of a main event do not bring about temporal
advancement.
In (10) the second clause contains the verb accompagner 'accompany'.
When you accompany a singer or a musician, you play one part of a piece
of music while they sing or play the main tune. The accompanying event is
always simultaneous with another performing event. This is equally true
when accompany means 'escort': If you accompany someone to the church,
the accompanying event is simultaneous with the going-to-the-church event.
Consequently, it is not at all surprising that (10) expresses two simultaneous
events.
The texts in (11) and (12) exemplify yet another case. Notice that the
second event in the first sentence in (11) may slightly overlap with the first
event. Nadine could have started reading the letter while still standing. The
first event of the second sentence, too, contains the passé simple, one would
thus expect that its event follows and does not overlap with the previous
event. This is, of course, not the case: Nadine might already have felt the
surprise while reading the letter. The situation is quite similar in the case
of (12): A considerable overlap between the two events is quite conceivable.
Evidently, principle (8) cannot be true. Thus Kamp and Rohrer 1983
propose to replace it by principle (13).
(13) The event introduced by the next passé simple may not be
encoded in the DRS (= Discourse Representation Structure) as
entirely preceding the event or time which at that stage of the
representation construction functions as reference point.
Dowty (1986) has argued that the Past Tense of certain verbs expresses
temporal advancement. One of his examples is repeated under (14).
(14) John arrived at 8 o'clock. He went into his room and started
reading a book. He felt warm and opened the window
The verbs which can be used in this fashion are either achievement
verbs or accomplishment verbs. The relevant principle can be found under
(15).2
(15) The Past Tense of an achievement or accomplishment verb
moves the narrative forward.
Achievement and accomplishment verbs are aspectually perfective, but
there are perfective verbs which are neither achievement nor accomplish
ment verbs. We will return to this question presently.
Dowty mentions some counterexamples to his principle (15) (Dowty
1986: 58), which are repeated under (16)-(18).3
(16) Pedro dined at Madame Gilbert's. First there was an
hors-d'oeuvre. Then the fish. After that the butler brought a glazed
chicken. The repast ended with a flaming dessert
2
It should he made clear that Dowty's formulation is different from principle (15).
His temporal discourse interpretation principle (TDIP) doesn't make any reference to
aspectual class membership. Temporal structure is inferred on the basis of the interplay
between TDIP and aspectual class membership. (Dowty 1986: 45).
3
Dowty attributes (16) to Kamp and (18) to Dry. (Dowty 1986: 58).
THE TEMPORAL INTERPRETATION OF DISCOURSE 117
(17) At the signal, everyone went to work at once. Mary searched the
room for any of the items on the list that might be there. John
went next door to do the same in Bill's apartment Susan orga
nized the rest of the players to canvass the block
(18) John knelt at the edge of the stream and washed his face and
hands. He washed slowly, feeling the welcome sensation of the icy
water on his parched skin
Example (16) is in a way parallel to (9). The first sentence represents
the main event followed by a sequence of sentences each describing a
subevent of the event mentioned in the first sentence. The subevents are
ordered: The order in which they occurred is indicated by lexical means:
First, then, after, end. Notice, however, that if we omitted from (16) the
lexical indication of temporal order, we would still interpret the subevents
in (16) in the same order. The interpretation would then be based on
everyday knowledge rather than on lexical information. We simply know
that the hors-d'oeuvre precedes the main dish and that a meal is normally
finished by a dessert.
Example (17) is not really different from (16). The first sentence
describes the main event and the subsequent events are interpreted as
subevents of the main event. The subevents are unordered: Neither lexical
nor everyday knowledge indicates any temporal order. Notice, incidentally,
that the sentence describing the main event contains the quantifier everyone.
As to (18), at first glance it is not quite clear whether it constitutes a
genuine counterexample to principle (15): Neither kneel nor wash seems to
be an accomplishment/achievement verb. However, (18) is ambiguous: It
has an imperfective and a perfective interpretation. In order to show this
the only thing one has to do is to translate (18) into a language which
makes a systematic distinction between imperfective and perfective verb
forms. Hungarian is such a language. And, indeed, kneel can be rendered
either as térdel (imperfective) or as letérdel (lit. kneel down, perfective) and
wash can be either mos (imperfective) or megmos (perfective). On the
imperfective interpretation we don't expect any temporal progression. If, on
the other hand, (18) is interpreted perfectively, the sequence of events is
expected to express temporal advancement. Contrary to this expectation,
however, the first and second sentences in (18) describe overlapping events
on both interpretations. A characteristic feature of this example is the
repetition in the second sentence of one of the predicates of the first
sentence.
118 FERENC KIEFER
Dowty goes much further in his conclusion than Kamp and Rohrer
1983. He suggests that the temporal discourse interpretation principle "must
be modified to allow that if the discourse itself conveys some implication as
to how events are to be specifically ordered, this should take priority over
(the principle) that orders reference times successively. In other word, (this
latter principle) is the 'default case' to be followed when neither time
adverbials nor entailments and implicatures of the discourse itself give clues
to the ordering of events." (Dowty 1986: 58)
Dowty's conclusion seems further be corroborated by an example due
to Ritchie 1979. Compare
(19) When they built the 39th Street bridge...
(a) a local architect drew up the plans
(b) they used the best materials
(c) they solved most of their traffic problems
Depending on the context the event described by the when-clause can
be prior to (cf. (19) - (c)), simultaneous with (cf. (19) - (b)), or posterior to
(cf. (19) - (a)) the event described in the main clause. Notice, however, that
this temporal interpretation can also be considered to be a consequence of
lexical information. Let's first have a closer look at (19)-(a). When you plan
something that you are going to make, build, or create, you decide what the
main parts of it will be and how they will be arranged. In other words, the
planning event must precede the building or creating event. As to (19)-(b),
the verb use implies simultaneity. If you use something such as a tool,
service, material, or a course of action, you do something with it in order
to do a job, achieve something, or solve a problem. On this reading the
verb use means 'to use something for doing something'. In other words, the
using event is simultaneous with the building event. Finally, to solve a
problem means to find or provide a solution to it. The solution is a
consequence of certain actions: You do something which eliminates a
problem. This means that in (19)-(c) the building event must precede the
solving event, the latter being in a way the culmination of the former.
4. Text linguistics
Research in text linguistics has long since claimed that there is a finite
number of text constituting principles which ensure the coherence of texts.
(van Dijk 1977; de Beaugrande and Dressier 1981; Longacre 1983; Sperber
THE TEMPORAL INTERPRETATION OF DISCOURSE 119
and Wilson 1986) Some of these principles have direct bearing on temporal
interpretation. They are exemplified in (20) - (26) (Isenberg 1971).
(20) Common topic
Mein Bruder hat sich einen Anzug gekauft. Peter ist die Treppe
heruntergefallen. Meine Tante hat sich den Arm gebrochen. Dies
alles erfuhr ich gestern morgen
'My brother bought a suit for himself. Peter fell from the
staircase. My aunt broke her arm. I was told all this yesterday'
(21) Specification
Gestern ist ein Unglück geschehen. Peter hat sich den Arm
gebrochen
'An accident happened yesterday. Peter broke his arm.'
(22) Causal link
Anna hat sich den Arm gebrochen. Sie ist gestern vor ihrem Haus
ausgerutscht und gefallen
Ann broke her arm. Yesterday she slipped in front of her
house and fell.'
(23) Diagnostic interpretation
Peter ist angekommen. Anna hat sein Fahrrad gesehen
Teter arrived. Ann saw his bicycle.'
(24) Pragmatic presuppositions
Der Junge ist ins Kino gegangen. Irgendjemand hat ihm Geld
gegeben
The boy went to the movies. Somebody gave him money.'
(25) Specification of the reason ('Motiv-Anknüpfung')
Hans ist hinausgegangen. Jemand hat das Fenster eingeschlagen
'Hans went out. Somebody broke the window.'
(26) Adversative contrast
Peter hat sich ein neues Auto gekauft. Sein Bruder hat dagegen
seine Wohnung neu eingerichtet
Teter bought a new car. His brother, on the other hand,
refurnished his apartment.'
Notice that in (22), (23), (24) and (25) the second event must precede
the first event, which means that these examples constitute violations of any
temporal interpretation principle which claims to be 'universal' and excludes
cases in which from among two events the earlier event is described in the
second sentence. Examples (20) and (26) express simultaneous or
overlapping events and example (21) contains one single event.
120 FERENC KIEFER
It has become a generally accepted view that events are not primitive
entities but are internally structured. (Bach 1986; Moens-Steedman 1988)
Events contrast with states, they may be momentary or durative, or may,
but need not, contain a culmination point. Both achievement and accom
plishment verbs contain a culmination point.4 More precisely, they both
exhibit the following structure (Moens-Steedman 1988: 18).
4
The taxonomy of events was first introduced in Vendler 1967. For more recent
accounts cf. Croft 1991 and Smith 1992.
THE TEMPORAL INTERPRETATION OF DISCOURSE 121
For example, the event described by They reached the top contains a
preparatory phase which consists of all the activities which precede the
achievement, the achievement itself is the culmination, and 'being on the
top' is the consequent state.
The distinction between events and states, duration and culmination gives
rise to the following taxonomy of events.
events states
momentary durative
Moens and Steedman argue (1988: 22-23) that in the case of (19)(a) the
entire culminated process of building the bridge tends to become a
culmination, which is associated with preparations for, and consequences of,
the entire business, as in (28)(a), The drawing up of the plans is then, for
reasons which have to do with knowledge of the world, situated in the
preparatory phase. In (19)(b), "in contrast, people tend to see the building
of the bridge as decomposed into a quite different preparatory process of
building, a quite different culmination of completing the bridge and some
consequences". The resulting decomposition is given in (28)(b). The use of
the best materials is, once again, situated in the preparatory phase, which
is, however, different from the preparatory phase of the previous example.
Finally, (19)(c) has the same decomposition as (19)(a), but on the basis of
everyday knowledge the event described in the main clause is situated in
the consequent state.
On Moens' and Steedman's account, then, knowledge of the world
enters into play (i) when we decompose an event into a preparatory phase,
a culmination and a consequent state and (ii) when we situate another
event with respect to this decomposition. Lexical knowledge is not called
for.
The above account suggests the following interpretation principle:
(29) From among two events ei and ej
(i) ej precedes ej if it is placed into the preparatory phase
of ej;
(ii) ei follows ej if it is placed into the consequent state of
e j;
(iii) ei is simultaneous with ej if it can neither be placed into
the preparatory phase, nor into the subsequent state of
e j.
If one allows for more than one decomposition, as Moens and
Steedman do, then simultaneity can also be arrived at in different ways, i.e.
both ei and ej can fall either into the preparatory phase or into the
consequent state.
THE TEMPORAL INTERPRETATION OF DISCOURSE 123
In this section we are going to inquire into the role of lexical information
in the temporal interpretation of discourse in more detail. It would seem
that the relevant lexical information may basically be of two kinds. One type
of lexical information is directly related to temporal structure, i.e. temporal
ity is encoded in the lexicon. In this case the predicate contains an event
variable as one of its arguments. Since events presuppose temporality, there
must be a temporal relationship between the event expressed by the
predicate and the one expressed by its argument. This relationship, too,
must be part of lexical representation.5 In the second case lexical informa
tion is only indirectly related to temporal interpretation. In some cases we
9
This is also implicit in Fillmore's description of scenes and frames. For example, the
elements in a prototypic commercial event scene are the buyer, the seller, the money that
changes hands, and the goods that change hands. (Fillmore 1976: 102-103) Consequently,
x bought y from z, z sold y to x, and x gave z money for y express essentially the same event
and cannot therefore be used to advance the narrative.
124 FERENC KIEFER
The verb accompany temporally relates the two events with each other.
That is, we have to add (32)(b) to the lexical characterization of accompa
ny:
(32) (b) SIMULTANEOUS(ei,ej)
ej = piano-playing event, ej = singing event
6
The lexical representations to be proposed here are utterly oversimplified. They are
process semantics-like representations and thus similar to the ones proposed in Miller &
Johnson-Laird 1976.
THE TEMPORAL INTERPRETATION OF DISCOURSE 125
(b) The predicate 'to draw up a plan' plan (Ex. (19)(a))
When you draw up a plan for something that you are going to make,
build, or create, you decide what the main parts of it will be and how they
will be arranged. The planning event presupposes two participants: An
agent and the object to be planned. Let us denote the predicate 'to draw
up a plan' by PLAN, which thus contains two arguments of which the second
argument is an event argument.
(33) (a) PLAN(x,y) and y = BUILD(u,v)
And, once again, the temporal relationship between the planning event
ei and the building event ej must be made part of the lexical representation.
(33) (b) PRIOR-TO(ei,ej)
gone to, and where they are now staying. Compare They followed after me
and He followed them to Venice. In principle, with verbs of movement both
interpretations are possible. Notice, however, that if the first event is not
yet completed, only the first interpretation is available. E.g. Bill got up and
made for the stairs. Mary followed. On one of its readings the verb follow
thus means 'to move behind in the same direction'. As to temporal
structure note that one of the arguments of follow is an event variable. That
is,
(35) FOLLOW(x,y) and y = MOVE(u,v)
which means that 'x follows the event y which consists in u's moving toward
v'. Since the lexical characterization of follow contains an event variable,
according to the above generalization the temporal relationship of the two
events (the moving event and the following event) must be made part of
the lexical representation.
(36) (a) OVERLAP(ei,ej)
e j = following event, ej = moving event
Condition- ej is not as yet completed
(b) POSTERIOR-TO^ej) otherwise
(39) (b) corresponds to the default case, therefore only (39)(a) has to
be stipulated
7. Conclusion
In the present paper I have argued that lexical information can be used to
explain a great number of exceptions to the principles proposed in the
THE TEMPORAL INTERPRETATION OF DISCOURSE 129
References
Bach, Emmon
1986 "The algebra of events". Linguistics and Philosophy 9.5-16.
Croft, William
130 FERENC KIEFER
Dowty, David R.
1986 "The effects of aspectual class on the temporal structure of discourse:
Semantics or pragmatics?" Linguistics and Philosophy 9.37-61.
Fillmore, Charles
1975 "An alternative to checklist theories of meaning". Proceedings of the First
Annual Meeting of the Berkeley Linguistic Society, 123-131.
1982 Frame semantics. In The Linguistic Society of Korea (ed.), Linguistics in the
Morning Calm. Seoul: Hanshin, 111-138.
Hopper, Paul
1979 "Aspect and foregrounding in discourse." In T. Givón (ed), Syntax and
Semantics Vol 12: Discourse and Syntax. New York: Academic Press, 213-241.
Isenberg, Horst
1971 "Überlegungen zur Texttheorie." In Jens Ihwe (ed). Literaturwissenschaft
und Linguistik. Frankfurt: Athenäum, 150-172.
Kamp, Hans
1979 "Events, instants and temporal reference." In Rainer Bäuerle, Urs Egli,
Arnim von Stechow (eds), Semantics from Different Points of View. Berlin:
de Gruyter, 376-417.
Lascarides, Alex
1990 "Knowledge, causality and temporal representation." Edinburgh: The
University of Edinburgh, Human Communication Research Centre.
THE TEMPORAL INTERPRETATION OF DISCOURSE 131
Longacre, Robert
1983 The Grammar of Discourse. New York and London: Plenum.
Pete, István
1983 "Az igeszemlélet, a cselekvés megvalosulásának foka, a cselekvés módja és
minősége a magyar nyelvben" (Aspect, resultativity and aktionsart in
Hungarian'). Magyar Nyelv 79.137-149.
Ritchie, George D.
1979 "Temporal clauses in English." Theoretical Linguistics 6.87-115.
Smith, Carlota S.
1992 The Parameter of Aspect. Dordrecht: Kluwer Academic Publishers.
Vendler, Zeno
1967 Linguistics in Philosophy. Ithaca: Cornell University Press.
Verkuyl, Henk J.
1972 On the Compositional Nature of the Aspects. Dordrecht: D.Reidel.
Wacha, Balâzs
1978 "Aspect in predicate constructions (in Hungarian)." Doctoral Dissertation,
The University of Budapest.
Reflections on metaphor and grammar
George Lakoff
The University of California at Berkeley
Chuck Fillmore walked into my office one day and asked coyly if I had ever
thought about the sentence,
- From my office, I can see the bay
The sentence, he observed, raised the issue of how it is possible to
semantically unify the adverbial phrase "from my office", which designates
a source, with "I can see the bay" which is about perception. The answer,
he observed, had to be a metaphorical one, in which seeing involves the
metaphorical motion of the gaze from my eyes to the bay, as in the
sentence,
- From my office, I can see all the way to the bay.
Chuck was aware, of course, that Johnson and I, in Metaphors We Live
By, had pointed out the general conceptual metaphor SEEING IS TOUCHING;
EYES ARE LIMBS, which as as instances cases like:
My eyes picked out every detail of the pattern. I ran my eyes over the
walls. I couldn't take my eyes off her. His eyes are glued to the tv. My
gaze is out over the bay. Their eyes met.
In this conceptual metaphor, there is a mapping from the source domain
of TOUCHING to the target domain of SEEING. The word "gaze" designates the
metaphorical Visual limb' that reaches out, just was the word "eyes" can
designate such visual limbs. The metaphor maps the following source
domain knowledge onto target domain knowledge:
Source: Limbs can be directed.
Target: Vision can be directed.
134 GEORGE LAKOFF
from the tactile domain, and apply them to the auditory and olfactory
domains:
- From the top of the mountain, I can pick up broadcasts from Moscow
on my radio.
- From my deck, I can pick out the sounds of construction on campus.
- Through the walls, I can pick out the aroma of the saffron in the
bouillabaisse being cooked in the next apartment.
- Through the piles of leaves that had fallen since the storm last night,
the dogs picked up the scent of the escaped prisoner.
So far, so good. But there were further implicit questions raised by the
sentences Fillmore gave me. The SEEING IS TOUCHING metaphor maps the
directability of limbs onto the directability of vision. Hearing and smell are
not directed in the way vision is. Sounds and smells are perceivable from
all directions at once. Could a single, more general metaphorical mapping
map directability for vision but not map directability for smell?
As one might expect with Chuck, his implicit question presupposed even
more implicit information. At the time, he and I were both teaching
seminars on inheritance in construction grammar. We differed in our
theories of inheritance. He claimed that inheritance could not have
overrides, and I, on the basis of my there-construction research (Lakoff
1987, case study 3), claimed that overrides were necessary. I had also
claimed that the Invariance Principle (Lakoff 1990) was a form of
inheritance (the source domain inferences are inherited by the target
domain) to which target domain overrides could apply. The directional
inferences about touch are mapped in the vision case, but not in the cases
of hearing and smell. Can this be due to target domain overrides of the
Invariance Principle?
Again the answer is yes. Suppose we assume that the more general
metaphor is of the following form:
PERCEPTION IS TOUCH
PERCEPTUAL ORGANS ARE LIMBS
and the knowledge mapped is:
Source: Limbs can be directed.
Target: Perceptual organs can be directed.
Metaphorical experiences
But the implications do not stop there. There are implications as well for
the classical theory of semantic roles first suggested by Fillmore himself.
Consider Fillmore's Experiencer role. In analyses dating back to the 196Q's,
Fillmore has insisted on distinguishing the Experiencer from semantic roles
like Agent, Patient, Source, and Goal. An Experiencer is someone who
experiences perception, emotion, or cognition, no matter what grammatical
role that person plays in a given sentence. Examples of Experiencers
include the italicized first-person pronouns in the following sentences:
- I can see the bay.
- That idea just came to me.
- John looks sick to me.
- The idea struck me as silly.
- I was letting off steam.
- I'm picking up Radio Moscow.
What I would like to suggest is that the Experiencer role is a very general
role, and that it does not strictly contrast with Agent, Patient, Source, and
Goal. Rather, conceptual metaphors can impose all of those roles on the
REFLECTIONS ON METAPHOR AND GRAMMAR 141
experiencer role, with the result that the metaphors, for the most part,
determine the grammatical relations borne by the experiencers.
Let us begin by returning to the sentence,
- I can see the bay from my office.
As Fillmore observed, "I can see the bay" must be metaphorically under
stood in order for semantic unification to take place with the source phrase
"from my office". But what are the details of that unification? What, for
example, is the role played by "the bay"? In the metaphor, my gaze is
extending to the bay. That is, the bay is the metaphorical goal. And, indeed,
it is marked as such in sentences like:
- From my office I can see all the way to the bay.
Here, "the bay" which is the Content of the perception in "I can see the
bay" is also the metaphorical goal, and is grammatically marked as such in
the Path+Goal phrase "all the way to the bay".
Now consider:
- That idea just came to me out of the blue.
"Me" would be merely an Experiencer on Fillmore's semantic role analysis,
since it indicates the person experiencing cognition. But "me" is marked
grammatically as a goal, which is just what it is metaphorically. The general
conceptual metaphor involved here is the CONDUIT metaphor, according to
which ideas are objects that can be sent and received. "Out of the blue" is
a metaphorical source phrase, and "That idea" is not just the Content of the
cognitive experience, but is also the metaphorical Theme that moves to
"me." The grammar of the sentence is a reflection of the metaphor. That
is, it has the grammar of a literal Theme-Goal-Source sentence, like the
literal "The dog came to me out of the kennel." To put it another way, the
sentence has source domain syntax. This contrasts with "I can see the bay,"
which has target domain syntax.
Now consider
- John looks sick to me from the pictures.
This sentence is about the acquisition of information via perception; that
is, "me" is an Experiencer of perception and cognition. Again we have a
case of metaphor, in which "from the pictures" is a source phrase; the
pictures are the metaphorical source of both perception and information.
"Me" is the metaphorical goal, and is expressed grammatically as a goal.
142 GEORGE LAKOFF
But, whereas "came" in the previous example is taken from the source
domain, in this sentence "look" is taken from the target domain. That is, the
grammar (TO X FROM Y) is taken from the source domain of the metaphor,
while the lexical item is taken from the target domain.
In the above examples, there are Experiencers which are metaphorical
Goals, and have the syntax of Goals, Now let us turn to a case where an
Experiencer is a metaphorical Patient and has the syntax of a Patient.
- The idea struck me out of the blue,
Again, we have the CONDUIT metaphor, with an idea that is conceptualized
as an object that comes from a source "out of the blue" to me, not just
reaching me as a goal but striking me. Thus, "me" is not merely a Goal, but
moreover, a Patient that is affected by being struck. The verb "struck" is
from the source domain, as is the syntax, in which "me" is direct object,
which is the natural grammatical relation for a Patient to have.
We have just seen that Experiencers can be metaphorical Goals and
Patients, with the corresponding syntax. They can also be metaphorical
Agents. Consider:
- I was letting off steam very carefully.
- I'm picking up Radio Moscow with some difficulty.
In both these sentences "I" is an Experiencer, an Experiencer of emotion
in the first sentence and an Experiencer of perception in the second. But
in both cases "I" is a metaphorical Agent, taking two of the grammatical
correlates that go with Agents but not with literal Experiencers: A
progressive auxiliary and a manner adverbial » "very carefully" and "with
some difficulty" select for agentive verbs. "Let off steam" and "pick up" are
both agentive predicates in the source domain.
In sum, Experiencers can play at least the metaphorical roles of Goal,
Patient, and Agent, and can have the corresponding source domain syntax
associated with those roles. The syntactic results could also be achieved by
arbitrary linking rules or arbitrary lexical constraints. But the conceptual
metaphors are needed independently, and since they are there to be used,
the most parsimonious solution is to make reference to them and avoid
linking rules that would only do the same work as the independently needed
metaphors.
The moral: Metaphor enters a grammar in at least three ways. (1)
Metaphor may be needed for semantic unification. (2) The clause grammar
may be a reflection of metaphorical semantic roles. And (3), a metaphorical
REFLECTIONS ON METAPHOR AND GRAMMAR 143
lexical item may be used. Thus, linking rules that specify how semantic roles
are paired with grammatical relations can often be dispensed with when we
take into account metaphorical semantic roles characterized by indepen
dently existing metaphors.
References
Lakoff, George
1987 Women, Fire, and Dangerous Things. Chicago: University of Chicago Press.
D E N N I S T H E M E N A C E / H a n k Ketcham
Knud Lambrecht
University of Texas, Austin1
1. Introduction
1
Earlier versions of this paper were presented at the Second Colloquium on
Romance Linguistics at UT Austin, at the Department of Modern Languages and
Linguistics at Cornell University, and at the UC Berkeley Linguistics Colloquium. I am
indebted to John Aske, Suzanne Fleischman, Mirjam Fried, Adele Goldberg, Laura
Michaelis, Sue Schmerling, Eve Sweetser, and Robert Van Valin for valuable comments
on various versions of the manuscripts.
146 KNUD LAMBRECHT
2
I will ignore in this paper intonational differences coding semantic or pragmatic
distinctions other than the focus structure of a sentence, such as the marking of speech-act
differences or of speakers' attitudes towards propositions. See Chapter 5 of Lambrecht
forthcoming for justification of this procedure.
150 KNUD LAMBRECHT
3
In the Gontext of (2) the subject noun foot is likely to receive what is often called
a 'contrastive accent'. This accent, called 'topic-establishment accent' in Lambrecht (1994,
Chapter 5.4) is functionally different from the focus accent and is irrelevant for the
argument at hand.
4
For an explanation of the projection mechanism whereby a single subject accent
signals a focus which extends over an entire proposition see Lambrecht 1992.
THE PRAGMATICS OF CASE 151
5
I am using the term 'subject' here in the traditional sense, in which it encompasses
postverbal NPs like un piede in (3). Whether such NPs have certain object properties or
may be considered 'objects' at some more abstract level of analysis is not at issue here.
152 KNUD LAMBRECHT
with the pronoun you. Therefore the linguistic expressions which refer to
the speaker in the answer sentences, i.e. the possessive determiner my in
English, the dative pronoun mi in Italian, and the determiner mon as well
as the subject and object pronouns je and me in French, are TOPIC
expressions; However, since these expressions do not code the primary
argument of the proposition, their topic status does not affect the focus-
articulation type of the sentence. The fundamental criterion for distinguish
ing the predicate-focus and the sentence-focus articulations is that in
sentences of type (2) the SUBJECT ARGUMENT of the underlying proposi
tion has a TOPIC relation to the proposition while in sentences of type (3)
the subject has a NON-TOPIC (i.e. FOCUS) relation to the proposition.
In the ARGUMENT-FOCUS sentences in (4), the focus domain is
'narrow", i.e. restricted to the designatum of the accented NP argument.
This pragmatic feature distinguishes the argument-focus articulation both
from the predicate-focus and the sentence-focus articulations. The context
question in (4) evokes a discourse situation in which the fact that the
speaker feels pain is assumed to be known to the addressee. The informa
tion conveyed by the answer consists merely in identifying the body part in
which the pain is felt.6 The argument-focus articulation corresponds to the
well-known 'focus-presupposition' type discussed e.g. in Chomsky (1972)
and Jackendoff (1972: Ch.6), in which the focus constituent corresponds to
a variable in a pragmatically presupposed open proposition. The argument-
focus articulation is only indirectly relevant to the issues presented in this
paper, which concern mainly the contrast between the predicate-focus and
the sentence-focus articulations. I will therefore not discuss it here at any
length.
Notice that the English sentence in (4a) is formally identical to the
sentence-focus structure in (3a). Given that the difference between sentence
focus and argument focus is a universal information-structural distinction
which is grammatically reflected in other languages (e.g. French or Italian),
this identity of form in English is to be seen as a case of focus-structure
homophony. This is confirmed by the fact, to be illustrated later on, that the
argument-focus type enjoys greater distributional freedom than the
sentence-focus type in English. All sentence-focus structures involving
6
In Lambrecht (1994, Chapter 5) I refer to sentences with argument-focus
articulation as 'identificational sentences', contrasting with 'topic-comment sentences'
(predicate-focus) and 'event-reporting' sentences (sentence focus).
THE PRAGMATICS OF CASE 153
7
I will avoid the term 'thematic role' because in information-structure research the
term 'theme' is traditionally understood as referring to a pragmatic rather than semantic
relation. Somewhat reluctantly, I will go with the trend and sometimes use the term 'theta
role', in the sense in which it is used by Fillmore & Kay.
154 KNUD LAMBRECHT
buyer is the agent and the seller the source, while with sell the buyer is the
goal and the seller the agent. But with both verbs, it is the agent role that
gets matched with the grammatical role of subject in a sentence.
Let us look at the participant roles associated with the predicator hurt,
The semantic frame of this predicator involves at least the roles listed in
(5), which I will first designate with frame-specific names:
(5) (a) a sentient being feeling the pain, or SUFFERER
(b) a body part where the pain is felt, or LOCUS OF PAIN
(c) a (non-volitional) stimulus causing the pain, or CAUSE OF
PAIN.
(d) a (volitional) agent inflicting the pain, or INFLICTOR.
(e) an instrument with which an agent inflicts pain, or INSTRU
MENT.
One or more (but rarely all) of the roles in a semantic frame will be
expressed in syntactic argument positions in a sentence when the verb is
called upon to describe a certain state of affairs, or SCENE, as Fillmore calls
it. In (5), roles (c) and (d) are mutually exclusive in a given scene. Pain is
caused either by a volitional agent or a non-sentient stimulus, but not by
both at the same time. Moreover (e) conceptually presupposes (d). I will
be concerned mostly with roles (a), (b), and (c) in this paper, i.e. most
examples to be discussed are to be understood as involving non-volitional
stimuli.
The frame-specific role names in (5) translate into the following theta
roles: The sufferer is an EXPERIENCER (EXP), the locus of pain a
LOCATION (LOC), the cause of pain a STIMULUS (STI), the inflictor an
AGENT (AGT), and the pain-causing instrument an INSTRUMENT (INS).
Notice that the mapping from frame-specific to theta roles is not an
automatic translation procedure but depends on what Fillmore & Kay call
the "notional parsing of the event" in a given scene. Since I am dealing in
this paper with a limited set of predicators, requiring few theta-frame
generalizations, the distinction between frame-specific roles and theta roles
is relatively unimportant for my argument. I will therefore allow myself
relative terminological freedom with respect to this otherwise important
distinction.
There is notorious and vast disagreement among scholars about the
THE PRAGMATICS OF CASE 155
nature and number of semantic roles in linguistic theory.8 Fortunately,
there is much less disagreement concerning the relative ranking of semantic
roles with respect to each other. In the following discussion I will be mainly
concerned with questions of relative ranking, therefore the controversy over
the nature and number of semantic roles has no major consequence for the
present analysis.
Since Gruber (1965) and especially Fillmore (1968) it has been
generally recognized that the syntactic arguments associated with a
predicator in a sentence may be assigned different semantic roles (or,
alternatively, that the semantic roles associated with a predicator may be
mapped with different syntactic arguments), depending on the scene
depicted by the predicator in a given sentence. Consider, for example, the
difference in the semantic role of the single argument of the predicate hurt
in (6) and in (7) (examples from Fillmore 1988b). The theta role of the
nominal argument is indicated in angled brackets after the sentence:
(6) My foot hurts. <LOC>
(7) This shoe hurts. <STI>
In (6) the subject NP indicates the body part where the pain is felt, or the
location of the pain, whereas in (7) it indicates the entity which causes the
pain, or the stimulus.
As Fillmore has shown (1968, 1988b), in the case of predicates with
more than one argument the various possible mappings from semantic roles
to syntactic positions are at least in part regulated by general principles,
which can be expressed in the form of SEMANTIC ROLE HIERARCHIES.
These role hierarchies determine which of the two or more semantic roles
associated with a predicate will appear in which syntactic position, in
particular which one will be selected for the role of subject. For example,
if in a given scene the predicate hurt is associated with two arguments, as
in (8)
(8) This shoe hurts my foot <STI - LOC>, <*LOC - STI>
the subject argument can only be the pain-causing stimulus, and the locus
of the pain must be the object. If the order is reversed, i.e. if the NP my
foot is selected as the subject and the NP this shoe as the object, the subject
will still be interpreted as the cause of the pain and the object as the entity
For a useful recent overview over the debate see Dowty (1991).
156 KNUD LAMBRECHT
porte s'est ouverte, where the verb of the intransitive sentence must have
reflexive morphology (cf. *La porte a ouvert).9
Returning to the English predicate hurt, we notice that while in (6),
(7), and (8) the semantic roles of the locus and/or the cause of pain are
overtly expressed as arguments of the predicate, the role of the EXPERIEN-
CER of the pain, which is a necessary participant role at the conceptual
level (it is the person, not the foot, that feels the pain), does not appear
as an overt argument of hurt. Rather the experiencer role is either
expressed indirectly, via the possessive NP determiner my (example (6) and
(8)), or it is omitted altogether, as in (7). As we will see, this is a feature
whereby English markedly differs from French (and other Romance
languages), where such a separate experiencer argument plays an important
syntactic as well as pragmatic role,10
While certain restrictions are placed upon the coding of the experiencer
as an object argument with predicates like hurt, it is possible, at least in
certain dialects or with a certain semantic specialization, to use the
experiencer as the unique argument, as in (14):
(14) 1 hurt (ail over). / I'm hurting. <EXP>
(14) is comparable to (13), where a patient argument could appear as a
subject, provided that no other argument was competing with it for this
position. Again, English is different in this respect from other languages,
As we will see in Section 4, in French it is necessary, in order to express
the meaning of (14), to choose a lexical predicate different from the one
9
This restriction is not absolute in French. For example under certain semantic
conditions the transitive casser 'to break' may be used intransitively, as in Le verre a cassé
'the glass broke'.
10
Among the English speakers I have consulted on the matter, several rejected
sentences like (i) (a) and (b) below, in which the experiencer is expressed as a separate
pronominal object argument, although some accepted corresponding versions involving
progressive aspect such as (i) (c) and (d):
(i) a. ?My foot hurts me. <LOC - EXP>
b. ?This shoe hurts me. <STI - EXP>
c. My foot is hurting me more every day. <LOC - EXP>
d. My shoes are hurting me. <STI - EXP>
Be that as it may, the general preference in English is not to express the experiencer in
the form of a separate argument, i.e. speakers tend to say (6) or (7) rather than (i) (a) or
(b).
THE PRAGMATICS OF CASE 159
Thus far we have looked at the relationship between semantic roles and
syntactic functions in English. In accordance with the goal of this paper, I
will now add a third dimension, by asking how the syntactic and semantic
levels relate to the level of INFORMATION STRUCTURE, in particular to the
pragmatic roles of TOPIC and FOCUS.
Looking first at the relationship between SEMANTIC and PRAGMATIC
roles, without reference to syntax, we notice that the two are in principle
independent of each other. The independence of semantic and pragmatic
roles is an obvious consequence of the fact that information structure has
to do with the USE of SENTENCES, rather than with the MEANING of
PROPOSITIONS. Information-structure distinctions are necessarily manifested
at the level of the UTTERANCE, i.e. at the level where propositional
contents are communicated via sentences with prosodic and morphosyn-
tactic structure. Since, as we have seen above, the same semantic role can
appear in different syntactic functions, and since it is syntactic positions that
express pragmatic roles, semantic roles per se cannot be pragmatically
constrained. Any pragmatically determined mapping constraints intervene
necessarily at the point where semantic roles are manifested as syntactic
constituents in a sentence.
Let us then consider the relationship between PRAGMATIC and
SYNTACTIC roles. Here we observe a revealing parallel with the relation
ship between syntax and semantics. As noted, English allows both agent
like and patient-like roles, including that of patient or highly affected
experiencer. (Foley & Van Valin's undergoer), to appear as subjects,
without concomitant morphosyntactic or lexical changes (see e.g. (13) and
(14)). In the same fashion, English allows in principle both the topic
relation and the focus relation to be mapped with the syntactic role of
subject, without concomitant changes in syntax or morphology. In this
respect, English differs from many other languages, including Italian and
French (see the contrasts illustrated in (2), (3), and (4) above). In English
the same unmarked SV(O) sequence may be used to express different
focus articulations, although, as we will see, the sentence-focus articulation
is semantically more constrained than the predicate-focus and the
argument-focus articulations. Differences in focus articulations are typically
160 KNUD LAMBRECHT
11
For discussion of a particular type of constructionally coded focus articulation in
English see Lambrecht 1988c.
THE PRAGMATICS OF CASE 161
(19a) is rather unnatural, compared to (14), in which the subject is a first person
pronoun subject. The lexical NP is needed for the present argument because a correspond
ing sentence-focus structure containing an accented pronominal subject would be
unacceptaable for independent reasons (see below). In (19b), the question mark indicates
a certain difficulty in assigning a semantic role to the subject argument. According to
DeLancey 1991, the subject that slap takes on the properties of the agent in that sentence.
162 KNUD LAMBRECHT
sentence.14
Given the restrictions against the co-mapping of focus and subject and
the absence of such restrictions in the case of topics, we may posit as a
general information-structure principle, for English at least, that the topic
argument OUTRANKS the focus argument for subject selection. This general
ranking difference between topic and focus arguments may be expressed
in the form of a PRAGMATIC ROLE HIERARCHY:
(21) THE PRAGMATIC ROLE HIERARCHY: topic > focus
Not having analyzed the matter in sufficient depth, I leave open the question of
whether the mapping of TOPIC and OBJECT is cross-linguistically constrained in the
same way as that of focus and subject. For strong statistical evidence that topics are rare
as lexical object arguments in French see Lambrecht (1986b: Chapter 6).
THE PRAGMATICS OF CASE 165
See Lambrecht 1988b for further discussion and for a proposed explanation of the
asymmetry between lexical and pronominal arguments with respect to the expression of
focus structure.
166 KNUD LAMBRECHT
not depend on the type of predicate alone. But neither does it entirely
depend on the semantic role of the subject argument. The latter conclusion
is confirmed by the fact that the availability of a sentence-focus reading for
a sentence with a given semantic structure often depends on the scene
which it is used to evoke. For example, the slightly varied version of the
unacceptable (20b) in (20b')
(20b') The SLAP hurt him!
has a possible sentence-focus reading in a context where a nurse is testing
the pain sensitivity of a patient emerging from anesthesia. 17 We must
conclude, then, that the pragmatic subject-selection hierarchy does not
directly match any natural semantic hierarchy such as e.g. Foley & Van
Valin's hierarchy in (12). It does not seem possible to define a cut-off point
on a semantic role hierarchy beyond which semantic roles can no longer
serve as the subject arguments of sentence-focus structures.
In spite of examples such as (22) and (23), it is nevertheless possible,
I think, to establish an inherent connection (i.e. one that goes beyond a
mere discourse tendency) between sentence-focus subject and non-agentive
semantic role. In order to see this connection it is necessary to take into
account an important fact concerning the interpretation of argument roles,
which in my opinion has not received sufficient attention: The construal of
the role of an argument may be influenced by discourse factors, i.e. by
factors other than the inherent lexical meaning of the predicate.
The discourse factor I have in mind is well-illustrated in examples (22)
and (23). Although in (22) the subject John designates the person who
makes the phone call, and although the activity of making a phone call
undoubtedly involves a willful agent, it is intuitively clear that the individual
making the phone call is not primarily CONCEPTUALIZED in (22) as an
agent on the level of the discourse. This intuition correlates with the fact
that (22) could not serve as an answer to the question What did John do?,
i.e. that the subject NP John has the pragmatic status of a NON-TOPIC in the
discourse. On the level of information structure, the event described by (22)
is not PREDICATED of the referent of the subject expression John in the
given context, i.e. the communicative purpose of the utterance is not to
convey information ABOUT this referent. The function of (22) is not to
17
This fact was pointed out to me by Laura Michaelis. For analogous examples and
discussion see Fuchs (1980) and Faber (1987).
THE PRAGMATICS OF CASE 167
inform the addressee that the caller has the property of having made a
phone call but rather to announce an event, in which the caller is involved
as a necessary participant. This fact necessarily leads to non-agentive
construal of the argument in the given context. Similarly, the reply in (23)
is not primarily to be understood as describing an action done by the
husband, but as announcing an event which happened to the topic referent
'Mary'. Pragmatically, the action of leaving Mary is not predicated of the
individual designated by the subject NP her husband. Rather (23) is a
predication about the topic 'Mary', which is expressed in the unaccented
object pronoun her.18
The above notion of a relationship between semantic roles and
information function is by no means obvious. (It has been received
skeptically by colleagues to whom I have presented it). How can the
semantic interpretation of the subject role of a predicate, say to call (on the
phone), be influenced by the degree of topicality or salience of the caller in
the discourse? As a first step towards making this notion less objectionable
I would like to remind the reader that theta roles themselves - as opposed
to the frame-specific roles of individual predicators - represent schematiza-
tions of events which apply to large classes of predicates in relative
independence of their specific lexical meanings. And pragmatic roles may
be seen to correlate with such abstract theta roles rather than with frame-
specific roles.
But the real explanation for the interpretive phenomenon at hand is
found, I believe, in the notion of CONSTRUCTIONAL MEANING, as
developed in the framework of CONSTRUCTION GRAMMAR (Fillmore
1988a, 1988b; Fillmore, Kay, & O'Connor 1988; Fillmore & Kay 1993), in
particular in recent work by Goldberg (1992):19 Goldberg argues,
convincingly I think, that the meaning of grammatical constructions cannot
be seen as a mere projection of the meanings of the lexical items they
contain, as postulated e.g. in the 'Projection Principle' of the Government
18
The idea that agentive arguments are not interpreted as fully agentive when used
in certain "detopicalizing" constructions is already discussed in Kirsner (1973) in a
comparison of active and corresponding 'pseudo-passive* versions of the same sentence in
Dutch.
19
See also Michaelis 1993, especially Chapter 2. Preliminary remarks concerning the
importance of constructional meaning for the information-structure interpretation of
sentences are found at the end of Lambrecht 1987b.
168 KNUD LAMBRECHT
and Binding framework. To use one of Goldberg's examples, it is difficult
to see how the meaning and form of a transitive sentence like Sam sneezed
the napkin off the table can be computed on the basis of the lexical meaning
of the verb sneeze alone, since sneeze is an intransitive verb (cf. *John
sneezed the napkin). Rather the interpretation of such a sentence is
determined by the combined meanings of the lexical items it contains and
of the grammatical construction instantiated by the sentence, in this case
the verb phrase construction [V NP PP], where V is a verb of motion and
PP a directional phrase.
I believe that the notion of constructional meaning can be applied
fruitfully to the issue of the pragmatic construal of the semantic role of
sentence-focus subjects. Such an application requires that we first define the
basic constructional meaning of sentence-focus structures.20 As I have
argued elsewhere (Lambrecht 1987b and 1988b), while certain unaccusative
predicates (like arrive, disappear, die etc.) crosslinguistically favor sentence-
focus construal by their lexical meaning, the criterion determining which
predicates are permissible in sentence-focus constructions is not primarily
lexical but PRAGMATIC, having to do with the ability of the predicate to be
used in a PRESENTATIONAL function. I term a sentence 'presentational' if
its communicative purpose is not primarily to depict a referent as playing
a role in some state, event, or action, but to introduce the referent into a
discourse or else to report an event which involves the referent not as a
topic but merely as a necessary participant (in the latter case it is the event
involving the referent which is 'presented'). The presentational discourse
function corresponds on the semantic level to the earlier-mentioned
THETIC judgment type.21
One of the strongest pieces of evidence in favor of the presentational
analysis of sentence-focus constructions lies in the fact, first observed by
Fuchs (1980), that the subjects of such sentences can never be PRONOUNS
i.e. cannot have referents which are assumed by a speaker to be already at
20
The present argument presupposes, of course, that subject-accented sentences of
the kind analyzed here are indeed grammatical constructions, i.e. that their semantico-
pragmatic structure is not entirely predictable from their form, and vice-versa. That this
is indeed the case is demonstrated in detail in Lambrecht 1992 and in Chapter 5 of
Lambrecht (1994).
21
Sasse (1987) refers to the two kinds of presentational sentences as 'entity-central'
and 'event-central thetic sentences.
THE PRAGMATICS OF CASE 169
the forefront of the addressee's mind (cf. e.g. the contrast between My
FEET hurt and #THEY hurt.) The behavior of pronominal subjects in
sentence-focus constructions is thus the exact opposite of that of pronomi
nal objects. The discourse status of a subject referent which can appropri
ately be coded with a pronoun, i.e. which is taken to be presently under
discussion, clashes with the inherent presentational function of the sentence-
focus structure, resulting in pragmatic unacceptability of the sentence.22
Apparent exceptions to the rule that sentence focus subjects must have
'new', i.e. not yet topical, referents can be accounted for in terms of the
notion of 'constructional accommodation' (Lambrecht 1994), whereby a
pragmatic feature which is conventionally attached to a grammatical
construction - such as the 'new' requirement for the referent of the subject
in a sentence-focus construction - is inherited by an expression appearing
in the construction even if the actual pragmatic status of the expression in
the discourse is not that normally required by the construction. In a
presentational construction the pragmatic feature of 'newness' or 'unexpect
edness' which is conventionally attached to the construction can be
inherited by an expression appearing in it even when the referent of the
expression is 'objectively' not new in the particular discourse context in
which the construction is used (for further discussion see Lambrecht 1987b).
I would like to claim, then, that the non-agentive construal of certain
agentive sentence-focus subjects, such as those in (22) or (23), is a result of
the basic presentational meaning inherent in the sentence-focus CON
STRUCTION. Verbs with agentive subjects take on the function of
presentational predicates when they head such constructions. It is of course
necessary, for the agentive subject of an intransitive sentence to be
pragmatically construed as non-agentive, that the basic meaning of the
predicate be amenable to the presentational function of the sentence-focus
construction. For example, the reason the verb call can be used as a
presentational predicate has no doubt to do with our knowledge of the
phone-call scenario: Announcing an unsuccessful phone call is a conven
tional way of introducing the caller into the consciousness of the addressee.
The above argument concerning non-agentive construal of agentive
22
An apparent exception to the constraint against pronominal subjects in sentence-
focus structures is the use of a heavily stressed pronoun as a 'euphemism' for a proper
name, as in When I drove by Dad's house, I could tell that SHE was there, where 'she' is
Dad's scandalous mistress (example provided by Sue Schmerling).
170 KNUD LAMBRECHT
3.3. Summary
expressed in the same clause, the cause of pain wins out (exx. (8) and (9)).
The locus of pain appears as a direct object (even though semantically it
designates a place). If the cause of pain and/or the locus of pain are
expressed in a clause, the role of the experiencer tends to remain unex
pressed or else to be combined with that of the possessor coded by the
possessive determiner within the NP expressing the locus-of-pain or the
cause-of-pain argument. If the predicate has no other roles associated with
it in the sentence, the experiencer role may appear as subject, without the
concomitant lexical, morphological, or syntactic changes in the verb which
are required in other languages (ex. (14)).23
As for the relationship between PRAGMATIC and GRAMMATICAL roles,
we observe that in principle both the topic relation and the focus relation
may be mapped onto the role of subject, without concomitant changes in
constituent order. However, just as with semantic roles, the pragmatic
mapping possibilities are constrained by a hierarchy: The co-mapping of
subject and focus is much more restricted than the co-mapping of subject
and topic. Subjects with strongly agentive semantic roles are normally not
compatible with the focus relation. Moreover, focus subjects cannot cooccur
with lexical objects (cf. (20)). There is thus a relationship between the
semantic and the pragmatic role hierarchies, even though the two cannot
be conflated. The existence of such a relationship is confirmed by the fact
that the construal of the semantic role of an argument may be influenced
by the communicative function of the sentence in context. In sentence-focus
constructions, the semantic role of any agentive subjectNP is neutralized or
backgrounded and the proposition is given a presentational or event-
reporting interpretation.
In the English examples discussed, the vehicle of the pragmatic
information is not the syntactic but the prosodic structure of the sentence.
The general typological picture emerging from these facts (and from many
others not discussed in this paper) is that English is a language in which
SVO constituent order can be used for a variety of semantic and pragmatic
purposes and in which the canonical sentence-initial subject NP position is
23
The statement that the mapping of experiencer and subject is contingent upon the
absence of any other argument is not meant as a general statement about English, cf. e.g.
the existence of such sentences as I feel pain in my foot, in which an experiencer/subject
cooccurs with a location/object. Moreover, as pointed out to me by Sue Schmerling, in
sentence (14), in which the subject is an experiencer, the phrase all over may be seen as
expressing a location argument.
172 KNUD LAMBRECHT
24
That in (24a) both mon pied and le pied are in principle possible has to do with the
grammar of inalienable possession in French. This fact is not directly relevant for the
argument that follows. It is mentioned for completeness and in order to avoid misunder
standings due to the use of one rather than the other construction.
THE PRAGMATICS OF CASE 173
(27) @ Les pieds (*me) font mal à ces ch. <STI - (EXP) - LOC>,
<*LOC-(EXP)- STI>
25
The experiencer argument may remain unexpressed in French in cases of
'Indefinite Null Complementation' (Fillmore 1986), i.e. if the referent is not specific, as
e.g. in Ca fait mal 'that hurts',
174 KNUD LAMBRECHT
the feet (to-me) make pain at these shoes 'my feet hurt these
shoes'
(If the experiencer argument is present in (27), the sentence is not only
pragmatically anomalous but uninterpretable; hence the star in front of me.)
The similarity between French and English with respect to the ranking of
the two semantic roles is predicted by the subject selection hierarchies in
(10) and (12), which are assumed to be universal (see Van Valin 1993).
Alongside this similarity in the respective ranking of the stimulus and
the location there are again important differences between French faire mal
and English hurt. With faire mal, the locus of pain does not appear as a
direct object argument, as my foot does in the English (8). Rather, it must
be expressed in the form of the locative prepositional phrase au pied.
meaning 'at the foot'. This makes semantic sense, since the locus of the
pain is a body part and since locations tend to be expressed by adpositions
across languages (ci English She hit me on the head). And as in (25), we
observe in (26) that the experiencer of the pain must be coded as a
separate pronominal argument. The alternative version of (26) in (26'),
which lacks this argument, is as unacceptable as (25') (in a non-metaphori
cal reading, as indicated by the semantic roles in angled brackets):
(26') *Ces chaussures font mal à mes pieds. <STI - LOC>
these shoes make pain at my feet 'these shoes hurt my feet'
Finally, unlike with English hurt (ex. (14) and discussion), the experiencer
can never appear as the subject of faire mal.
This leads us naturally to the alternative lexical expression avoir mal.
The semantic frame of this predicate involves only two roles: Experiencer
and location. The experiencer is necessarily expressed as a subject, and the
location can only appear as a locative prepositional phrase. The predicate
avoir mal thus replaces faire mal if the experiencer argument is to appear
as subject. The locative argument of avoir mal is optional, as (28b) shows:
(28) a. J'ai mal au pied, 'my foot hurts' (=25b) <EXP - L O O
b. J'ai mal. 'I feel pain/I hurt' <EXP>
c. *Mon pied a mal 'my foot hurts' <LOC>
Sentence (28c) is ungrammatical (in the intended non-metaphorical
reading) because only an experiencer can be the subject of avoir mal
Moreover, since the non-subject argument of avoir mal can only be the
locus of pain, a sentence such as (29) can only have a facetious reading:
THE PRAGMATICS OF CASE 175
26
The same effect is possible (but not necessary) in English with the gloss My shoes
hurt, in particular with the possessive my instead of the demonstrative these, and with
sentence-focus articulation. The effect is exploited in the Dennis-the-Menace cartoon
reproduced at the beginning of this paper.
27
One (apparent) exception to the equation between focus and non-subject in French
is the presentational [il V-unacc NP] construction, illustrated in (i)
(i) Il est venu trois femmes. 'There came three women'.
it is come three women
in which a focal lexical subject FOLLOWS the verb. In such constructions the subject
properties are distributed over the non-preverbal pronoun il (which can be neither a topic
nor a focus expression) and the postverbal lexical subject constituent.
28
In spoken French, predicate-focus structures involving lexical topic NPs typically
require a 'detachment' (or 'dislocation') construction, in which a bound preverbal pronoun
cooccurs with the lexical NP (see Barnes 1985; Lambrecht 1987a), while in the written
standard the canonical SV(O) structure is preferred. For reasons of simplicity, I will use
the canonical versions in the example sentences below.
176 KNUD LAMBRECHT
The exact nature of the violation in (31a) is unclear to me. Given the oxytonic
rhythm of French, one might consider (31a) an instance of phonological (prosodic) ill-
formedness rather than a violation of an information structure constraint. However (31a)
is not, strictly speaking, ill-formed. Leftward 'accent movement' does occur occasionally
in French, especially in metalinguistic corrections of previous statements.
THE PRAGMATICS OF CASE 177
necessarily appears as a non-subject argument, a sentence containing this
structure allows the focus accent to fall on a postverbal NP, i.e. in its
unmarked position. The resulting sentence is given in (32):
(32) J'ai mal au PIED (=24b) <EXP/SUBJ/TOP - LOC/OBJ/
FOO
I have pain at-the foot 'My FOOT hurts'
Moreover, since the experiencer argument of avoir mal is necessarily the
subject and since, as I have emphasized, the experiencer (i.e. the owner of
the body part) is pragmatically a topic in both contexts, it becomes possible
for the subject and the topic to be combined in preverbal position.
Notice that formally the resulting sentence (32) does not have sentence-
focus but predicate-focus articulation, given that its subject is a topic.
French thus expresses here via a predicate-focus construction a pragmatical
ly structured proposition which English and Italian express via sentence-
focus constructions. This is a noteworthy fact. It shows that it is not always
possible to predict from a given discourse context (such as the one assumed
in the model example (3)) which focus articulation will be used in a given
language to express a given propositional content. What IS predictable from
the utterance context, and what remains constant across languages, is that
a certain NP will be marked as a focus rather than a topic constituent (in
our case the NP designating the speaker's foot). But whether this focus
constituent will appear as a subject, resulting in a sentence-focus structure,
or as an object, resulting in the unmarked predicate-focus structure,
depends on the valence of the predicate provided by the lexicon of the
language. It is important to understand that the term 'sentence-focus
structure' designates a GRAMMATICAL CATEGORY, whose occurrence,
though entirely motivated by pragmatic considerations, is contingent upon
the structural and semantic properties of the sentence in which it is
instantiated. Information structure MOTIVATES grammatical form, it does
not DETERMINE it.30
The lexical alternation involving the predicates faire mal and avoir mal
and its relevance for the informational structuring of the French sentence
30
The observation that a given pragmatically structured PROPOSITION may be
formally expressed either by a sentence-focus or by a predicate-focus STRUCTURE makes
it necessary to abandon the notion, which I indirectly defended in Lambrecht 1987b, that
there is a necessary correlation between the thetic judgment type and the sentence-focus
structure.
178 KNUD LAMBRECHT
31
As the English glosses 'He has long hair' and 'His hair is long' show, the contrast
between the predicates have and be in English allows in principle for the same pragmatic
contrasts as that between French avoir and être. For the analysis of the pragmatic function
of have in English cf. Lambrecht 1988c.
32
Following Foley and Van Valin 1948: 48, I assume that the semantic role of the
subject of 'have' or 'avoir' is locative.
180 KNUD LAMBRECHT
MATICAL CONSTRUCTION in the sense of the framework of Construction
Grammar, i.e. a complex grammatical unit with semantic, pragmatic, and
syntactic properties which are not predictable from other properties of the
grammar. In this construction, the first clause, e.g. J'ai mon pied 'I have my
foot' in (36), is semantically non-compositional in that the proposition it
expresses becomes meaningful only in combination with the proposition
expressed in the following relative clause. Both clauses present themselves
formally as topic-comment structures, with the pronominal subjects as the
topics; yet the combination of the two has the specific pragmatic function
of expressing a single proposition with sentence-focus articulation.33
Let us now consider the proposition These shoes hurt9, in which the
subject argument is the cause of pain. If the sentence expressing this
proposition is to be construed pragmatically as conveying information about
the topic 'these shoes', it will have the form of the predicate-focus structure
in (37):
(37) Ces chaussures me font MAL. <STI/SUBJ/TOP - EXP/OBJ/
TOP>
these shoes to-me make pain These shoes HURT'
(37) is pragmatically and syntactically well-formed. The cause-of-pain
argument (the shoes), which is a topic, appears as the subject NP ces
chaussures; the experiencer, which, as in (31), is a secondary topic, appears
again as the atonic pronominal object me.
But how can the proposition underlying (37) be articulated in a
discourse context in which the NP representing the cause of pain is to be
in focus, as in the English sentence-focus structure in (18) {These SHOES
hurt)? Both (38a) and (38b) are unacceptable, for the same reasons the
parallel structures in (31) were unacceptable:
(38) a. #Ces CHAUSSURES me font mal.
b. *Me font mal ces CHAUSSURES.
In this case, the lexicon provides no alternative structure (as it did in the
case of the ill-formed sentences in (31)) which would allow the experiencer
to become a preverbal subject-topic and the cause of pain a postverbal
33
As far as I can tell, (36) is not only semantically but also pragmatically equivalent
to (32). Such cases of free variation are not the norm, since the occurrence of lexical
alternatives such as (32) hinges on the existence of particular predicates in the lexicon.
THE PRAGMATICS OF CASE 181
4.3. Summary
but syntax. -French however does not easily fit into the binary typology
proposed by Thompson (1978), according to which languages may be
divided into two major types: Those with 'grammatically controlled word
order' (say, English) and those with 'pragmatically controlled word order'
(say, Latin or Russian). In French, constituent order is grammatically
controlled as strongly as in English. Nevertheless French does not express
focus-structure alternations prosodically but via syntax, by using a variety of
pragmatically specialized grammatical constructions.
5. Conclusion
34
As the reader will have noticed, I have not addressed in this paper the question of
the criteria according to which a syntactic constituent may be said to be the subject of its
clause, in particular I have not addressed the issue of ergative languages. The validity of
my crosslinguistic generalization remains to be investigated for such languages.
184 KNUD LAMBRECHT
implications for the view of the overall organization of grammar and of the
role played by information structure in the grammatical system. The fact
that information structure may motivate PROSODIC STRUCTURE is well
known. Sentence-focus structures involving accented subjects, such as those
found in English, have been discussed in the literature at least since
Bolinger 1954. A particularly abundant discussion has arisen from the
stimulating observations in Schmerling 1976 (in particular her Chapter 5).
The approach to focus structure sketched in the present paper (and
discussed in more detail in Lambrecht 1988b, 1992, and 1994) is one in
which subject accentuation is seen as an information-structure device for
indicating the marked mapping of a pragmatic and a syntactic role.
The influence of information structure on sentence form is not limited
to prosody. Like other pragmatically motivated grammatical constructions,
such as e.g. the often-discussed English it-cleft and wh-cleft constructions
(and their equivalents in other languages), the French avoir-cleft construc
tion demonstrates the far-reaching consequences information structure may
have for the SYNTACTIC organization of sentences. Since such constructions
have the specific effect of changing the mapping relation between syntactic
and pragmatic roles found in the corresponding canonical sentences, they
may be referred to as PRAGMATIC-RELATION-CHANGING constructions.
These are in many ways comparable to such constructions as passive,
dative shift, tough-movement, etc., which are often referred to in generative
grammar as 'grammatical-relation-changing' constructions, a term which
primarily evokes their semantic and syntactic properties. However such
constructions may equally well be seen as permitting alternative mappings
of syntactic and pragmatic roles.
Finally, the data analyzed in this paper show that information structure
may manifest itself in certain aspects of the LEXICAL SELECTION process.
Crucial in this respect is the contrast between the structures in (30) and
(32) {Le pied me fait mal - J'ai mal au pied). If the mental process of
sentence formation only involved procedures for relating syntactic structure,
semantic structure (or logical form) and phonetic form, it would be difficult
to understand why alternative lexical structures like faire mal and avoir mal
should EXIST in a language and why a speaker would ever use one rather
than the other. The contrast between (31) and (32) shows that the
selection of a predicating expression from the lexicon is sometimes
determined not only by the contribution of the expression to the meaning
of the sentence but also by information-structure constraints on the
186 KNUD LAMBRECHT
syntactic form of the sentence in which the predicate is used to express this
meaning. There is a point where lexical structure, grammatical structure,
and information structure meet.
In selecting a predicate to express a proposition, a speaker must take
into account rules determining which of the roles in the semantic frame of
the predicate is to be mapped onto which syntactic position in the sentence.
But the speaker must also take into account pragmatic rules of information
structure determining which pragmatic role is to be associated with which
syntactic position, in accordance with her assumptions about the mental
state of the addressee. These pragmatic rules may in turn have conse
quences for the formal organization of the sentence (focus accent on the
subject vs. on the predicate, preverbal vs. postverbal position of the subject,
use of a special grammatical construction, choice of a different predicate,
etc.).
Facts such as those described in this paper suggest that a complete
explanation of the grammatical structure of a sentence requires not only
a theory of the relationship between semantics and syntax on the one hand,
and between syntax and prosody on the other, but a theory of the four-
termed relation between semantics, syntax, prosody, and information
structure. This paper has suggested one way in which this relation may be
described.
References
Barnes, Betsy
1985 The pragmatics of left detachment in spoken standard French. Amsterdam:
John Benjamins.
Bolinger, Dwight
1954 "English prosodic stress and Spanish sentence order." Hispania 37.152-156.
Burzio, Luigi
THE PRAGMATICS OF CASE 187
1981 Intransitive verbs and Italian auxiliaries. MIT dissertation.
Chomsky, Noam
1972 "Deep structure, surface structure, and semantic interpretation." In Studies
on semantics in generative grammar. The Hague: Mouton, 62-119.
Comrie, Bernard
1981 Language universals and linguistic typology. Chicago: The University of
Chicago Press.
DeLancey, Scott
1991 "Event construal and case role assignment." BLS 17.338-353.
Dowty, David
1991 "Thematic proto-roles and argument selection. Language 67. 547-619.
Faber, David
1987 ''The accentuation of intransitive sentences in English. " Journal of Linguistics
23.341-358.
Fillmore, Charles J.
1968 "The case for case." In E. Bach & R. Harms (eds), Universals in linguistic
theory. New York: Holt, Rinehart & Winston, 1-88.
Fuchs, Anna
1980 "Accented subjects in 'all-new' utterances." In G. Brettschneider & Chr.
Lehmann (eds.), Wege zur Universalienforschung. (Festschrift für Hansjakob
Seiler). Tübingen: Gunter Narr Verlag.
188 KNUD LAMBRECHT
Goldberg, Adele
1992 Argument structure constructions. PhD dissertation, University of
California, Berkeley.
Gruber, Jeffrey
1965 Studies in lexical relations. PhD dissertation, MIT.
Hawkins, John
1981 "The semantic diversity of basic grammatical relations in English and
German." Linguistische Berichte 75.1-25.
Höhle, Tilman
1982 "Explikationen für "normale Betonung" und "normale Wortstellung"." In W.
Abraham (ed.), Satzglieder in Deutschen. Tübingen: Gunter Narr Verlag,
75454.
Jackendoff, Ray
1972 Semantic interpretation in generative grammar. Cambridge: MIT Press.
Kirsner, Robert S.
1973 "Natural focus and agentive interpretation: On the semantics of Dutch
expletive er." Stanford Occasional Papers in Linguistics 3.101-13.
Kuno, Susumo
1972 "Functional sentence perspective: A case study from Japanese and English."
Linguistic Inquiry 3.299-320.
Kuroda, S.-Y.
1972 "The categorical and the thetic judgment. Evidence from Japanese syntax."
Foundations of Language 9.153-185.
Lambrecht, Knud
1984 "A pragmatic constraint on lexical subjects in spoken French." CLS
12.239-256.
1986b. Topic, focus, and the grammar of spoken French. Unpublished PhD
dissertation, University of California, Berkeley.
1987a. "On the status of SVO sentences in French discourse." In R. Tomlin (ed.),
Coherence and grounding in discourse. Amsterdam: John Benjamins, 217-262.
1988b. "When subjects behave like objects: A markedness analysis of sentence focus
constructions across languages." Expanded version of a paper read at the
1987 LSA Meeting.
1988c. "There was a farmer had a dog. Syntactic amalgams revisited." BLS
14.319-339
1994 Information structure and sentence form: A theory of topic, focus, and the
mental representations of discourse referents. Cambridge: Cambridge
University Press.
Lyons, John
1977 Semantics, 2 vols. Cambridge: Cambridge University Press.
Mathesius, Vilém
1928 (1964) "On linguistic characterology with illustrations from modern English."
Actes du Premier Congrès international de linguistes à la Haye. Reprinted
in J. Vachek (ed.), 1964. A Prague School Reader in Linguistics. Blooming
ton: Indiana University Press.
Michaelis, Laura A.
1993 Toward a grammar of aspect: The case of the English perfect construction.
PhD dissertation, University of California, Berkeley.
190 KNUD LAMBRECHT
Perlmutter, David
1978 "Impersonal passives and the unaccusative hypothesis." BLS 4.157-189.
Reinhart, Tanya
1982 Pragmatics and linguistics: An analysis of sentence topics. Distributed by the
Indiana University Linguistics Club.
Sasse, Hans-Jürgen
1987 "The thetic/categorical distinction revisited." Linguistics 25.511-580.
Schmerling, Susan F.
1976 Aspects of English sentence stress. Austin: University of Texas Press.
Thompson, Sandra A.
1978 "Modern English from a typological point of view: Some implications of the
function of word order." Linguistische Berichte 54.19-35.
1993 "A synopsis of Role and Reference Grammar." In Van Valin (ed.), Advances
in Role and Reference Grammar. Amsterdam: Benjamins Publishing
Company.
Walters, John R.
1979 "Focus in Aghem: A study of its formal correlates and typology." Southern
California Occasional Papers in Linguistics 7.137-197.
Wandruszka, Ulrich
1982 Studien zur italienischen Wortstellung. Tübingen: Gunter Narr Verlag.
Ancestor-descendant and cultural-linguistic
relativity
Charles N. Li
University of California, Santa Barbara
1. Introduction
condemns such a parent. The importance of filial piety in the culture has
given rise to ancestor worship, which has the characteristics of a religious
ritual. Each traditional Chinese family would typically display an altar at a
prominent spot in the house to commemorate the ancestors. On certain
days of the year, incense and candles would be burned and food would be
placed on the altar. Hence, an ancestor even after death commands a great
deal of power over his/her living descendants. In short, we can describe the
ancestor-descendant relation in China as an asymmetrical dimension of
ethical value with a positive direction toward the ancestor. The ancestor,
dead or alive, is always worthier than the descendant. The descendant is
forever beholden to the ancestor because it is the ancestor who gave life to
the descendant, and it is within the realm of social legitimacy for an
ancestor to take life away from a descendant.
The expressions of the ancestor-descendant relation in the Chinese
language reflect the asymmetry of cultural values. First, the temporal aspect
of the relation is by and large stated in terms of spatial metaphors.1
Among the pairs of spatial terms denoting height and positions: gao 'high' -
di low', qian 'front' - hou 'back', shang 'up' - xia 'down', the positive
member of each pair is used to express 'ancestor', and the negative member
of each pair is used to express 'descendant'. Consider the following data:
(1) (a.) gao yi bèi
high one generation
'One generation older'
1
There are also a few terms which employ temporal expressions, e.g. xian ren 'early
person = ancestor', wan ren 'late person = descendant'.
LINGUISTIC RELATIVITY 193
(c.) shi - dì
teacher - younger brother
'a junior colleague trained by the same master'
(d.) shi - zŭ
teacher - ancestor
'founding master'
(e.) tu - dî
disciple - youth
'disciple'
(f.) tu - sŭn
disciple - paternal grandson
'disciple of a disciple'
In spite of the obvious biological correlation between old age and the
loss of physical prowess, the opposite correspondence is upheld in Chinese
folklore and novels on martial arts, i.e. the older a master of martial arts,
the greater the physical prowess he possesses.
To the Chinese mind, then, age is not only a repository of authority,
knowledge and wisdom, but also a source of physical prowess and agility.
The correlation of this mind set with the cultural practices and linguistic
expressions in China serves as an example of a Whorfian hypothesis - a
connection between language, culture, and mind. What is interesting about
LINGUISTIC RELATIVITY 195
this example, however, are the questions it poses: Does the Confucian ethic
still dominate the cultures of the various Chinese populations scattered
across the world? If the Confucian ethic is no longer upheld and the
Chinese language retains the positive-negative polarity in its expressions of
ancestor-descendant relation, what can one say about the Whorfian
hypothesis? Answers to those questions will shed light on the fundamental
nature of language.
The first and most obvious caveat for any Whorfian hypothesis is that there
isn't any necessary and sufficient condition for establishing correlations
between language, culture and mind. Even if one substantiates an
association between a certain cultural practice, X, with a certain language
pattern, Y, it does not mean that the existence of X necessarily implies the
existence of Y, or vice versa. In case of the ancestor-descendant relation,
the correlation which I have described does not imply that if the Confucian
ethic has been adopted by a society, the language of that society will
necessarily reflect its people's perception of age, ancestor and descendant.
Conversely the correlation which holds in the Chinese civilization does not
imply that if a language exhibits the Chinese pattern of expressions
concerning ancestor and descendant, the speakers of that language will
necessarily subscribe to the Confucian ethic.
To illustrate the former case, I will cite a Mongolian language, Baonan,
spoken in the eastern edge of Inner Asia in China. For centuries, the
Baonan people have completely adopted the Confucian ethic,2 and
borrowed a number of kinship terms from Chinese through contact with the
2
In a field trip to the Baonan villages, I was struck by the rigidity of the unequal
relation between parent and offspring. All power rests in the hands of parents, especially
the father, regardless of the age of the offspring. There is no question that nowadays, the
Baonan adheres much more strictly to such Confucian ethical norms as filial piety and
parental authority than the Chinese does. At one point I asked my Baonan language
consultant, whose son as an adult had already taken over the primary responsibility of
cultivating the family farm, what his son would do if he had ordered his son to follow a
particular farming practice that was clearly inappropriate. He responded with
laughter,"Well, that's unlikely. But if it did happen, my son would always do as I say.
Whether I am right or wrong would be irrelevant." His son confirmed this assertion of
parental authority.
196 CHARLES N. LI
Chinese-speaking Hui ethnic group in Inner Asia. But their language does
not show a positive direction toward 'ancestor' along the ancestor-descen
dant axis, and it does not have terminology of the type cited in (l)-(3). In
fact, Baonan does not have expressions for 'ancestor' and 'descendant'. The
Baonan word for parents is borrowed from a Chinese expression, xian-ren,
which literally means 'early people' in Chinese, and the Baonan term for
offspring is an indigenous word, rarla.
As an illustration of the latter case where the language exhibits a
consistent pattern of bias in favor of ancestor in the ancestor-descendant
dimension, but the people do not adhere to the Confucian ethic, I will cite
the transformation in contemporary China.
Over the past four decades the various political movements promoted
by Mao and his cohorts in China severely discredited the validity of the
Confucian ethic. Ancestor worship was prohibited. During the Cultural
Revolution and other political campaigns, rebelling against authority to the
point of denouncing one's own parents as counter-revolutionaries was
encouraged. The dismantling of ancestors as an exalted caste was also
enhanced by the policy of one child per family which has been strictly
enforced in urban areas. One of the consequences of having only one child
is that many Chinese parents have become such doting and indulging
parents that Chinese children have been described as "lording over their
parents". In other words there appears to be a shift from filial piety toward
parents to parental devotion to offspring. In addition, contact with western
civilization has resulted in significant changes of attitudes regarding the
ancestor-descendant relation, especially among Chinese populations in
Hong Kong, Singapore and other places.3 I would not suggest that the
Confucian ethic with respect to the ancestor-descendant relation is no
longer valid in Chinese civilization, but there is no doubt that it has moved
towards Western framework of values. Yet the Chinese language has not
changed or abandoned the terminological skewing shown in (l)-(3).
We may conclude that correlations between cultural principles,
linguistic patterns, and human perception do exist, but they are not
predictable. In other words, a language may or may not reflect a particular
pattern of social cognition and culture of its speakers. The reason for this
unpredictability lies in the nature of language and language change which
3
One example is that ancestor worship is no longer practiced in most families in
urban China, Hong Kong and Singapore.
LINGUISTIC RELATIVITY 197
4. Conclusion
6
The so-called formalism in modern linguistics is no more than a set of notations
with a few simple conventions. Notations and conventions can be useful for the scholar
to present his/her analyses more succinctly and rigorously. As a former student of
mathematics and physics, I have always found the use of the term 'formalism' in linguistics
somewhat pretentious.
LINGUISTIC RELATIVITY 199
References
Clark, Herbert
1973 "Space, time semantics, and the child." In T.E. Moore (ed.), Cognitive
Development and the Acquisition of Language. New York: Academic Press,
28-63.
Crick, Francis
1988 What mad pursuit: A personal view of scientific discovery. New York: Basic
Books, Inc.
Fillmore, Charles J.
1971 Lectures on deixis. University of California, Santa Cruz, Summer Program
in Linguistics. Manuscript.
1972 "How to know whether you are coming or going." In K. Hylgaard-Jensen
(ed.), Linguistik 1971. Frankfurt: Athenäum, 369-379.
1982 "Toward a descriptive framework for spatial deixis." In Robert G. Garvella
& Wofgang Klein (eds.), Speech, place and action. Chichester: John Wiley
and Sons Ltd, 31-59.
Givón, Talmy
1979 On understanding grammar. New York: Academic Press.
Rosch, Eleanor
1977 "Human categorization." In N. Warren (ed.). Advances in cross-cultural
psychology, Vol. 1. New York: Academic Press.
Saussure, Ferdinand de
1916 Cours de linguistique generale. Paris: Payot. Translated by Wade Baskin as:
Course in general linguistics. New York: McGraw-Hill, 1959.
200 CHARLES N. Lï
Toshio Ohori
University of Tokyo at Komaba
1
The present paper originates in part from Ohori (1990), and the question
formulated in (1) was originally presented by Charles J. Fillmore. As such, I consider it
is fitting to dedicate this study to him, with deep gratitude for his guidance in our
linguistic research.
Romanization of examples is phonemic, basically following Kunrei-system. Long
vowels are represented by doubling, e.g. sensee 'teacher'. Interlinears are given according
to the following conventions: (i) function words are glossed in capitals (linkage markers
are treated as primitives, so they are also in capitals); (ii) when a single word in one
language corresponds to more than one word in the other, dots are used instead of spaces,
e.g. ryuugaku-'study.abroad'; (iii) elements that are not grammatically realized in the
original sentence, e.g. subject NPs, are put in the parentheses in the English translation.
Abbreviations for function words are as follows: ACC(usative), BEN(efactive),
COMP(lementizer), DAT(ive), EMPH(atic), NEG(ative), NOM(inative),
NZ(=nominalizer), PERF(ective), POL(ite), PRE(fix), PRED(ication), PRT( = Particle),
STAT(ive), TOP(ic), VOL(itional). When case markers serve non-canonical functions, they
are glossed PRT.
202 TOSHIO OHORI
2
The data for this paper come from both introspection and casual observation of
everyday talk, reflecting the preliminary nature of the study.
REMARKS ON SUSPENDED CLAUSES 203
elliptical answers.3 For both of these uses, virtually any linking device
allows suspension. The criterion relevant to the present discussion is
whether a given linking device can be used in an utterance that is
self-initiated and yet structurally incomplete.
3
Examples (i) and (ii) represent echo question and elliptical answer, respectively:
(i) A: Moo sukosi nomu-to, ...
more little drink-TO
'If (I) drink a little more, ...'
B: Moo sukosi nomu-to?
more little drink-TO
'If (you) drink a little more, then ... [what will happen]?'
(ii) A: Doo suru-to fairu-ga utuse-ru no?
how do-TO file-NOM transfer-can PRT
'How can (I) transfer a file?'
B: Koosite kurikku-sita mama aikon-o ugokasu-to, ...
thus click-do whle icon-A€C move-TO
'Move the icon while (you) click on it like this,
then...[you can transfer the file]'
Note that the sequential marker TO, which is not used in the suspended construction, is
possible in the above sentences.
REMARKS ON SUSPENDED CLAUSES 205
4
Here I refrain from using cumbersome non-sexist pronominal reference such as s/he,
his/her, and the like, in favor of clarity of prose. Besides, this example is a real quote
anyway, the teacher being myself.
208 TOSHIO OHORI
reduction of the main clause thus takes place due to its pragmatic
redundancy, without involving any sort of idioms as in the constructions
discussed in 2.1. In NONI-rnarked suspended clauses, because only the
semantic relation of counterexpectation is realized without the consequent
statement, their pragmatic effect involves regret or very mild blaming tone.
By now, it seems reasonable to consider a clause-linking particle as having
its own discourse function in the suspended construction.
The mechanism of inference that gives rise to this kind of suspension
is by no means simple. Here, NONI no longer codes any logical relation
between the linked clauses, but marks some aspect of the participant's
involvement in the conversational context. The pragmatic effects of such
suspended clauses are not easy to describe, but, generally speaking, the
effect of NÖNI may be rendered approximately, 'Alright, but the due
consequence has not followed in reality, you know'. In (13), the 'due
consequence' is the teacher's giving a copy of his book to his student. In
this example, there is only a feeling of weak regret in the teacher ( = B)'s
utterance. In some cases, however, concessivity or counterexpectation
encoded by NONI can give rise to a strong blaming tone. See also the
following example:
(15) Konnani hituyoo.na-noni.
this.much necessary-NONI
'(You) never know how much (I) need (you)', lit. Though (I)
need (you) this much, ...' or '(I) need (you) this much, but ...'
Here, what is to be inferred is not actually the consequent clause, but the
state of affairs that give rise to the unhappy situation, so a more exact
translation would be 'Although I need you so much, the due consequence
does not follow, which I regret'. In this case, an appropriate rendering of
NONI would be 'Alright, but I'm/you're not getting what I/you ought to'. As
a result, one of the main pragmatic effects of NONI is to urge the hearer to
think of the cause of misfortune.
In addition to NONI, there is another concessive marker, namely KEDO,
which is also frequently used in the suspended construction:
(16) Undookai-tte.ie.ba watasi are-o omoidasi-masu-kedo ne.
athletic.meet-TOP I that-ACC recall-POL-KEDO PRT
'Athletic meet? I recall that (incident).'
KEDO is a shortened form of KEREDOMO (the former is limited to the
spoken register). The pragmatic effect of KEDO in this context may be
REMARKS ON SUSPENDED CLAUSES 209
That's right, but that's not the whole story' with a warning tone. Here the
challenged presupposition can be any aspect of what is being talked about,
for example, its positive/negative evaluation. Example (16) is taken from a
DJ's talk about athletic meets. To take part in them is usually supposed to
be fun. But the DJ comes to think of her own experience and says 'I recall
that incident, but...', the assumption being 'right, many people have fun in
athletic meets, but I don't totally agree on that point'. After this, she starts
to talk about her miserable experience, which however is funny to others.5
By now the paths for the functional extension from linkage markers to
pragmatic particles are schematized as follows:
(17) X-NONI, Y (Though X, Y') = > X-NONI, ø (Though X, the due
consequence hasn't followed,' or 'X, but I regret that the
adversative condition holds.')
5
Itani (1992) recently discusses the utterance-final use of KEDO (together with
NODE), focusing on its 'softening' and hence 'polite' effect. One example is (romanization
and the gloss are slightly regularized):
(iii) Otya-ga hairi-masi-ta-kedo
tea-NOM ready-POL-PAST-KEDO
Tea is ready.'
By using KEDO, 'but', the speaker gives the hearer an option of either having or not
having the tea (in which case the omitted clause would be something like '...but would you
like to have it?'). I do not consider that the suspended version is always 'polite', but the
feeling of reservation as I described above in 2.2 may give rise to indirectness, which is an
important element of politeness.
210 TOSHIO OHORI
Here the speaker (=the broadcaster) is not saying that the runner really
ought to steal the next base. He is just wondering,with certain detachment,
how it would be if that happens and KEDO is used to express the 'prepare
for more' part of (18). In addition to NONI and KEDO, GA also participates
in this construction, though less frequently (when suspended, its function is
closer to KEDO).
Next, markers for reason, especially KARA, are frequently used in the
suspended construction. See the following examples:
(20) A: Kyoo hima?
today free
'Are (you) free today [i.e. this evening]?'
B: Ee, demo, tukare-teru kara.
yes but tired-STAT KARA
'Yes, but (I)'m tired, so...[I can't make it]'
[teru is a reduced and colloquial form of te.iru which is a stative
marker]
(21) Nani it-te.mo iu koto kika-nai-n.da-kara.
what say-TEMO say thing listen-NEG-PRED-KARA
'Whatever (I) say (s/he) won't listen to (me), so...'
In both of these examples, reduction takes place because the message of
the consequent clause is inferable. But they do differ with respect to the
significance of the suspended clause. In (20), since A wants to know
whether B can go out with him (for dinner maybe), B's giving an excuse
means a rejection of A's invitation. To this extent, the unexpressed message
of rejection is central to the communicative goal. In (21), on the other
hand, the reason clause is suspended even though the clause that follows
it is not identifiable. Here, the speaker is not giving any excuse, but is
complaining about somebody's stubbornness. Since KARA is a marker of
reason, a possible non-suspended version of (21) might be:
(22) Nani-o it-temo iu koto kika-nai-n.da-kara komaru yo.
what-ACC say-TEMO say thing listen-NEG-PRED-KARA troubled
PRT
'Whatever (I) say (s/he) won't listen to (me), so (I)'m troubled.'
The speaker's complaining tone comes from the meaning of KARA and the
inference it induces. In this example, the consequent part, that the speaker
is troubled, is inferred only indirectly. The speaker utters a statement, which
REMARKS ON SUSPENDED CLAUSES 211
seems to provide a reason for some condition, but this consequent condition
is not expressed. Then, assuming that the speaker is talking about
something relevant to his/her present condition, the hearer makes an
inference that the KARA-marked clause expresses the speaker's current
concern. In this sense, the main clause in (22), '(I)'m troubled', is only one
possibility. A maximally broad interpretation of the KARA-marked
suspended clause is that the expressed statement has a strong concern on
the speaker and possibly the hearer. In an extreme case, KARA 'so' solely
marks the speaker's urge for the hearer's involvement, as in the following
example:
(23) Watasi nante motto taihen-dat-ta-n.da kara ne.
1sg EMPH more troubled-PRED-PAST-PRED KARA PRT
'I was even more troubled,' or 'I had an even tougher time, so...'
The pragmatic effect of suspension in this case may be paraphrased as 'I
had an even tougher time, and so you know what'. An expected response
would be, for example, 'Oh yeah, that's really bad'. Thus the hearer does
not simply recover the omitted part, but actively responds to the hearer by
showing sympathy.
Pragmatically, the suspension of the reason clause has the following
mechanism:
(24) X-KARA, Y ('X, so Y') = > X-KARA, ø ('X, so you know what,'
or 'X, so it concerns me/you')
NODE is also used with the same function, though less frequently. With
NODE, the speaker-hearer's concern seems a little weaker, partly because
it is associated with the formal register and is less forceful.
An important point to be noted in this connection is that linkage
markers that are not specialized for reason, such as SI and TE, can only
have a reason reading when they are suspended. Thus the Si-marked clause,
which by itself codes a very loose linkage as in (25), can be suspended when
it is put into an explanation- or excuse-frame, as in (26):
(25) Kinoo-wa eega-o mi-ta-si, kyoo-wa iti.niti hon-o yon-de.i-ta.
yesterday-TOP movie-ACC see-PAST-SI today-TOP all.day
book-ACC read-STAT-PAST
'(I) went to see a movie yesterday and was reading a book all
day today.'
212 TOSHIO OHORI
23. By now the relevant conditions for suspended clauses that have been
discussed so far are summarized as follows:
(30) A suspended clause occurs most frequently when:
(a) it is in the spoken register, especially conversation, and
(b) it is part of an idiomatic expression and can code the
meaning of the whole without the rest, or
(c) it is marked by a (pseudo-)logical connective, and the
intended message is either contextually inferable or convention
alized.
Condition (a) is self explanatory. Some linkage types, for example the
linkage by the 'gerundive' ending (ren'yoo-kee) of verbals, do not occur in
the suspended construction because they are mostly limited to the written
or formal register. Condition (b) refers to such idioms as V-BA-ii (suggestion
or permission), and V-TE-'thanks'. As discussed in 2.1, a clause is suspended
when the linkage marker can stand for the whole idiom in the communica
tive context. Condition (c) refers to cases in which the main clause is
omitted because it is inferable, and the pragmatic effect, for example regret
or blaming, is conventionalized, as we have seen in 2.2. This is the case with
NONI. Another marker of counterexpectation, KEDO, adds a reservational
nuance, implying that there is room for further consideration. The reason
marker KARA indicates the speaker's concern and consequently expresses
an urge for the hearer's involvement, usually sympathy. What is common
to these constructions is that they occur when the (pseudo-)logical conse
quence is contextually inferable. Markers for concession and reason are
typically inference-intensive (i.e. they put various sorts of constraints on
interpretation), and with these markers, the hearer tries to infer what is
coming after the dependent clause more actively than otherwise. This
means that, when faced with suspended clauses, the hearer must infer what
may come after them. In doing so, relevant aspects of the context are
picked up and generalized. When they are sufficiently generalized, they
become conventionalized part of the construction.
214 TOSHIO OHORI
3.2. Here, I would like to discuss the nature of suspended clauses from a
broader perspective.
First, suspended clauses can be aptly considered as constructions in
their own right, in the sense of Fillmore (e.g. 1988; also cf. Fillmore, Kay,
and O'Connor 1988). That is, suspended clauses have their own discourse
functions that are not manifest in a non-suspended version, and their
well-formedness conditions must be specified as a set of features that
pertain to particular lexico-grammatical configurations. Notice that the
pragmatic effects of suspended clauses we have examined so far are only
partially reducible to the properties of clause-linking devices themselves.
For example, the blaming or regretting tone of NONI, or the urge for the
hearer's involvement with KARA should best be described as a property of
the suspended construction, for such effects are found only when suspended
(also cf. the case of SI and TE in (26)-(27)). At the same time, it does not
seem to be reasonable to reduce everything to pragmatic inferences and
make no reference to structural properties. To do so would fail to recognize
the highly conventionalized nature of the pragmatic effects associated with
suspended constructions. The foregoing analysis thus seems to support the
view that the notion of construction is important and indeed indispensable
for understanding the structure of language.
Second, one important issue to be put on the agenda is the explication
of the pragmatic mechanism involved in the suspension of dependent
clauses. Since reinforcement of pragmatic inferences and their conventional
ization are crucial aspects of grammaticization, a closer look at the
inferential processes associated with suspended clauses will be necessary.
In this connection, suspended clauses seem to raise an interesting issue for
the pragmatic theory of connectives. In Blakemore (1987), it is argued that
the function of connectives is to put semantic constraints on the way an
utterance is interpreted. On the one hand, suspended clauses can be seen
as demonstrating the usefulness of the theory, because it says that
connectives guide the hearer's inference and this function may not change
whether the dependent clause is suspended or not. On the other hand, the
'semantic constraint' theory runs into a trouble when faced with the fact
REMARKS ON SUSPENDED CLAUSES 217
References
Blakemore, Diane
1987 Semantic constraints on relevance. Oxford: Blackwell.
Blass, Regina
1990 Relevance relations in discourse: A study with special reference to Sissala.
Cambridge: Cambridge University Press.
Chafe, Wallace L.
1980 "The deployment of consciousness in the production of a narrative". In
Wallace L. Chafe (ed.), The pear stories: Cognitive, cultural, and linguistic
aspects of narrative production. Norwood, NJ: Ablex, 9-50.
218 TOSHIO OHORI
Fillmore, Charles J.
1988 "The mechanisms of 'Construction Grammar'". Berkeley Linguistic Society
14.35-55.
Haiman, John
1988 "Inconsequentials in Hua and the typology of clauses". In John Haiman &
Sandra A. Thompson (eds.), Clause combining in grammar and discourse.
Amsterdam: John Benjamins, 49-69.
Itani, Reiko
1992 "Japanese conjunction kedo ('but') in utterance-final use: A relevance-based
analysis". English Linguistics 9.265-283.
Ohori, Toshio
1990 Clause linkage in Japanese: A synopsis. Special Field Examination, The
University of California, Berkeley. To be distributed by the Indiana
University Linguistics Club.
Pragmaticization of meaning in some
sentence-final particles in Japanese
Shigeko Okamoto
California State University, Fresno
1. Introduction1
The complementizers (COMPs) no, koto, and to/tte (tte is a colloquial form
of to) in Japanese appear at the end of a complement clause, as illustrated
in (l)-(3): 2
1
An earlier version of this paper was presented at the 66th Annual Meeting of the
Linguistic Society of America (1992). I would like to thank Charles Fillmore, Orin
Gensler, Kiyoko Hirose Ohara, Yoshiko Matsumoto, Masayoshi Shibatani for their
valuable comments and discussions.
The following abbreviations are used in this paper: AUX (auxiliary verb), COMP
(complementizer), COP (copula), DMT (diminutive suffix), GEN (genitive case marker),
IMP (imperative mood marker), NEG (negative auxiliary), OM (object marker), PASS
(passive voice), PPX (polite prefix), PRG (progressive aspect marker), PRT (particle), PST
(past tense marker), SEP (sentence-final particle), SM (subject marker), TM (topic
marker), and TQ (tag question marker).
220 SHIGEKO OKAMOTO
3
Although it is beyond the scope of the present paper, the actual distribution of the
complementizers no, koto, and to/tte is much more complicated than the generalizations
given here and requires further study.
PRAGMATICIZATION OF MEANING 221
sentence-final particles.) Employing the notion of grammatical construction
advanced by Fillmore and others (Fillmore 1985, 1988; Fillmore, Kay, and
O'Connor 1987),4 I will argue that the sentence-types S no/koto/to/tte (e.g.
(4) and (5)) are not context-dependent elliptical expressions, but indepen
dent grammatical constructions used for specific purposes, and that they are
not complements, but main clauses, in which these SFPs serve as modality
indicators. These main clause constructions are, however, closely related to
complement clause constructions. It is then argued that the usages of these
SFPs are instances of reanalysis of morphological functions through
pragmaticization of meaning (Traugott 1988, 1989; Traugott and Konig
1991).5
The paper is organized as follows: Section 2 outlines the semantic-
pragmatic properties of no, koto, to, and tte as SFPs. Section 3 compares
these final morphemes and the corresponding COMPs with regard to
morphosyntactic and semantic-pragmatic properties, and argues for the
status of the fomer as SFPs. Section 4 discusses the implications of the
analysis given in section 3.
2.1. No
4
Other studies that advocate the view of grammar based on the notion of
grammatical construction and other releant notions (e.g. formulas, preferred clauses,
preferred argument structures) include Du Bois (1985), Lakoff (1987), Lambrecht (1984,
1988), and Hopper (1987). Okamoto (e.g. 1993, 1994) analyzes a number of instances of
grammatical constructions in Japanese.
5
See also Heine et al. (1991) and Matsumoto (1988).
Kuno and Teramura discuss only no da, an expression related to no. I discuss the
difference between no da and no in section 3.1.
222 SHIGEKO OKAMOTO
2.2. Koto
There are two usages of koto as an SFP: (a) to express exclamation (Martin
1975: 945) and (b) to give an order or direction. The former usage is
PRAGMATICIZATION OF MEANING 223
2.3. To
2.4. Tte
7
Tte here is a contracted form of to yuu no wa - i.e. COMP (to) + say (yuu) +
pronoun (no) + TM (wa)), which translates into 'what is called (mountain).'
PRAGMATICIZATION OF MEANING 227
Though interesting, it is beyond the scope of the present study to investigate the
difference between tte and soo da.
228 SHIGEKO OKAMOTO
3.1. No
As outlined in section 2.1., the SFP no is typically used to indicate that the
utterance is an explanatory comment about a certain situation in the
discourse context. One might argue that this use of no, as in (8a), is an
abbreviation of no da, as in (8b):
(8a) Montoree ni Uta no.
to go PST
'He went to Monterey.'
PRAGMATICIZATION OF MEANING 229
(8c):
o
The no da construction has been studied extensively. Maynard (1992) provides a
good review of previous studies on no da.
230 SHIGEKO OKAMOTO
3.2. Koto
10
I assume that the role of no for objectification is relevant at the cognitive level. As
discussed, syntactically no da as a whole functions as an auxiliary verb.
considered paraphrases.
With regard to the issue of inferability, ellipsis is, in general, used when
the meaning is assumed to be inferable from the context without an explicit
mention. For example, in (28), the elliptical utterance is based on the
inferability of the intended meaning:
(28) Kiyoki ippyoo o.
honest one vote OM
'(Cast) an honest vote.'
In the case of the exclamatory koto, however, the criterion of inferability
does not seem to be met. For example, a beginning student of Japanese,
who knows only the 'basic' use of koto as a COMP, would have difficulty
interpreting the utterance O-niwa ga utukusii koto 'How pretty the garden
is!', particularly in light of the fact that koto occurs without a case marker.
Even in an appropriate context, a listener might wonder about the presence
of koto or be left with many possibilities of interpretation: e.g., koto ga
wakatta '(I) realized that', koto o kiita '(I) heard that'.
Based on these observations, we can conclude that koto in question is
not a COMP, but an SFP used to express exclamation in a feminine manner.
In other words, the exclamatory S koto is not an elliptical expression.
Rather, it should be considered a grammatical construction used as a main
clause, in which koto serves as a specific modality marker. (Note, however,
that the SFP koto reflects to some extent the function of the corresponding
COMP, or nominalizer, in that the situaion being described is an object of
amazement (see section 4).)
However, (4a) and (11b) are explicit performative sentences; they are
authoritative, hence stronger orders than (4) and (11). These performative
sentences are formal and not commonly used, whereas directives such as
(4) and (11) are used in ordinary situations. The directive koto might also
be thought equivalent to the auxiliary koto da.13 However, koto da is used
to make a suggestion or recommendation. Thus, in (31), for example, the
use of koto, not koto da, is appropriate, because the teacher is giving an
order, not making a suggestion:
(31) (a teacher to students in class)
marker. 14 (Note, however, that the directive koto reflects to some extent
the function of the corresponding COMP, or nominalizer, in that the
proposition is a rule/duty or an object of abidance (see section 4).)
3 3 . To
14
Interestingly, Korean kes has a directive usage similar to koto. Ransom (1988: 373)
points out that the COMP kes is used interchangeably with the imperative mood marker,
and it could evolve into a mood marker.
PRAGMATICIZATION OF MEANING 237
15
It is unclear to me why da is required in the use of to for defiance.
238 SHIGEKO OKAMOTO
3.4. Tte
The choice of a COMP is largely controlled by the verb, while the choice of a
modal auxiliary depends on the speaker's subjective judgment about the situation being
described.
PRAGMATICIZATION OF MEANING 243
17
Note that in the case of the SFP no, it is related to the COMP via the auxiliary no
da (see section 3.1.).
244 SHIGEKO OKAMOTO
final particles in general (as well as other linguistic devices): A final particle
"imparts some additional hint of the speaker's attitude toward what he is
saying" (Martin 1975: 914) without explicitly articulating it. Although the
processes of reanalysis are quite different, the shift of S no/koto/to/tte from
complements to main clause constructions may be also compared to the
functional reanalysis of clauses involved in the grammaticization of
epistemic parentheticals in English (e.g. I think) discussed by Thompson
and Mulac (1991) - the process in which the head element becomes an
epistemic phrase or a dependent element, while the 'complement' becomes
the main element.
In this study, I have demonstrated how the constructional view of
grammar enables us to consider subtle but significant differences in the
semantic-pragmatic properties of certain formally similar linguistic
expressions as well as the contiguous nature of semantics and pragmatics
that underlies the usages of these expressions.
References
Cook, Haruko M.
1990 "An indexical account of the Japanese sentence-final particle no." Discourse
Processes 13.401-439.
Du Bois, John W.
1985 "Competing motivations." In J. Haiman (ed.), Iconicity in syntax. Amsterdam:
John Benjamins Publishing Company, 343-365.
Fillmore, Charles J.
1985 "Syntactic intrusions and the notion of grammatical construction." BLS
11.73-86.
Josephs, Lewis S.
1976 "Complementation." In M. Shibatani (ed.), Syntax and semantics 5: Japanese
generative grammar. New York: Academic Press, 307-369.
Kuno, Susumu
1973 The structure of the Japanese language. Cambridge: MIT Press.
Lakoff, George
1987 Women, fire, and dangerous things. Chicago: University of Chicago Press.
Lambrecht, Knud
1984 "Formulaicity, frame semantics, and pragmatics in German binominal
expressions." Language 60.753-796.
Martin, Samuel E.
1975 A reference grammar of Japanese. New Haven: Yale University Press.
Matsumoto, Yo
1988 "From bound grammatical markers to free discourse markers: History of
some Japanese connectives." BLS 14.340-351.
Maynard, Senko K.
1984 "Functions of to and koto-o in speech and thought representation in
Japanese written discourse." Lingua 64.1-24.
1992 "Cognitive and pragmatic messages of a syntactic choice: The case of the
Japanese commentary predicate n(o) da." Text 12.563-613.
McCawley, Noriko A.
1976 "Another look at no, koto, and to: Epistemology and complementizer choice
in Japanese." In J. Hinds & I. Howard (eds), Problems in Japanese syntax
and semantics. Tokyo: Kaitakusha, 178-212.
Okamoto, Shigeko
1993 "Nominal repetitive constructions in Japanese: The 'tautology' controversy
revisited." Journal of Pragmatics 20.433-466.
246 SHIGEKO OKAMOTO
Okamoto, Shigeko
1994 "Augmentative verbal repetitive constructions in Japanese."
Cognitive Linguistics 5.381-404.
Ransom, Evelyn N.
1988 "The grammaticalization of complementizers." BLS 14364-373.
Teramura, Hideo
1984 Nihongo no shintakusu to imi II. (Japanese syntax and semantics II.) Tokyo:
Kuroshio Shuppan.
Traugott, Elizabeth C.
1988 "Pragmatic strengthening and grammaticalization." BLS 14.406-416.
Jan-Ola Östman
University of Helsinki
0. Preamble
1
My first serious encounter with deixis came a long time ago in Charles Fillmore's
lectures on semantics at UC-Berkeley. In fact, the present paper has grown out of an
essay I once wrote for Chuck, and which he - faithful to his habit - commented on
extensively. In addition, I would like to thank Erik Andersson and Farrell Ackerman for
248 JAN-OLA ÖSTMAN
1. Introduction
comments on very early versions, and the audience at the 1990 International Pragmatics
Conference in Barcelona for their response to a brief presentation of the major ideas in
this paper - I particularly remember that Jens Allwood, Dan Slobin and Auli Hakulinen
gave me very useful suggestions. I am also very grateful to Sandra Thompson for her
valuable comments on the penultimate version of this paper.
DEIXIS IN FINNISH 249
Table 2.
Although there have been a great many discussions about the semantics of
2
Capital-letters are used in suffixes to indicate variations in accordance with the
rules for vowel harmony in Finnish; e.g. -A- stands for -a- or -ä- depending on the
characteristics of the stem.
250 JAN-OLA ÖSTMAN
Inner vs. Outer case marking in Finnish in general (for a recent, cognitive
approach, ef. Leino (1989)), the distinction has received hardly any
systematic attention with respect to its manifestation in the deictic system.
One fairly typical, almost intuitive, kind of suggestion is represented by
Lehtinen's (1967) sugggestion that the size of the area referred to is
crucial: If the area referred to is relatively small, as in (1),
(1) Tässä on makuuhuone.
here-INESSIVE is bedroom
'Here's the bedroom'
Inner cases are used, whereas Outer cases are used if the area is of a
relatively larger size; cf. (2).
(2) Sataako täällä paljon?
rain-QUESTION here-ADESSIVE much
'Does it rain a lot here ( = in this area) V
This explanation is not, however, uncontroversial. In fact, I would
rather support the view that the Inner vs. Outer distinction is directly tied
to the role of Local case marking in general in Finnish. Inner cases refer to
a Bounded location of some sort; in the terminology of Lakoff and Johnson
(1980), the space with respect to which something is metaphorically viewed
as a container. The Outer, cases, in contrast, are used in reference to
Unbounded areas. These are issues on which Talmy in particular has done
invaluable research (cf. e.g. Talmy 1978).
The behavior of the postposition päin is particularly revealing for a
proper interpretation of the function of the Inner-Outer distinction. Päin is
a lexical hedge (G. Lakoff 1972) which refers to a very unspecified area,
something like 'hereabouts' when it is added to a demonstrative which
indicates proximity:
(3) Hän asuu täälläpäin.
He/She lives at-here-PÄIN
'He (or She) lives around here somewhere.'
This postposition can only be added to locations expressed with Outer case
marking. It is not, for instance, possible to say (4), with the demonstrative
adverb in the Inessive case.
(4) *Hän asuu tässäpäin.
If we know the specific location where somebody lives, it is feasible to view
DEIXIS IN FINNISH 251
3
As another case, consider the function of the indefinite pronoun jossain
'somewhere'. When used with an Outer-case demonstrative, as in Täällä (ulkona) jossain,
'Here (outside) somewhere', it indicates an unspecified, Unbounded area in general;
whereas when it occurs with a demonstrative adverb in an Inner case, the unspecified
target area is seen as located within some Bounded area: Tässä (laatikossa) jossain, 'Here
(inside this box) somewhere'.
4
Cf. also Fillmore's (1975: 18) reference to Robert Austerlitz.
252 JAN-OLA ÖSTMAN
5
There are, of course, also other factors influencing the particular lexicalized form
a language codifies. As Fillmore (1977) showed very early on, and returned to even more
in detail in his discussions on Construction Grammar (1988; cf. also Fillmore, Kay and
O'Connor 1988), lexical items very often have constructional properties, in particular, they
have to be seen in relation to their opposite or converse terms. Cf. on land vs. at sea, but
on the ground vs. in the air.
Thus, in Finnish, the reason why we sometimes say lahdessa and sometimes lahdella
('in the bay') is only partly due to the Boundedness properties of the bay. In particular,
the expressions partake as terms in the following paradigms:
(i) lahdella vs. rannalla vs. merellä
in-the-bay on-the-shore at-sea
vs.
(ii) lahdessa vs. meressä vs. ilmassa
in-the-bay in-the-sea in-the-air
(In addition, there is the town Lahti, which takes the Inessive form Lahdessa, 'in Lahti',
rather than Lahdella.)
I am here restricting my comments to apply only to what the evidence from the
function of demonstrative adverbs indicates.
DEIXIS IN FINNISH 253
7
Lehtinen (1967), for instance, takes the large area vs. small area distinction to be
the basic distinction. However, she herself admits that it does not apply to the distinction
between siellä and siinä. The suggested distinction in terms of Boundedness, however,
covers all three terms in what I will shortly deal with under the heading of the Proximity
system.
254 JAN-OLA OSTMAN
8
Cf. Östman (1992).
9
Another distinction that has been suggested (cf. e.g. Lehtinen 1967) to explain the
semantic difference between the use of Inner and Outer cases is in terms of
circumlocutions like 'over here, hereabouts' for Outer cases, and 'right here' for Inner
case markings. I can to some extent go along with the 'right here' vs. 'hereabouts'
distinction, as should be obvious from my discussion above. However, the rendering 'over
here' creates additional problems.
In particular, I would say that 'over here' should be seen as a translation of the
semantics of the Inner, rather than of the Outer, case system. Talmy (Fillmore, p.c.) has
argued that whereas here and there designate areas with a possible border-line between
them, over here and over there rather indicate points with (a considerable amount of)
space between them ~ i.e. between the speaker/designated object and the addressee. Thus,
since the areas referred to by tässä, '(in) here', and tuossa, '(in) there', are Bounded, it is
easy to conceive of them as having space between them.
DEIXIS IN FINNISH 255
new information.
In order to decide among these three alternatives, extended text
analyses need to be made. But even without such in-depth analyses, I
suspect that it might not be possible to weigh the factors appropriately.
The Boundedness criterion is a grammar-internal factor — with possible
cognitive repercussions; the Givenness and Specificity criteria are more
pragmatic and interactionally determined. I suspect the criteria work
together (cf. also the discussion in sections 5 and 6 below).
I have preferred to single out Boundedness, because it is a
grammatical feature referred to extensively in Finnish grammar anyway. If
the distinction between Outer and Inner cases in Finnish in general is seen
as an aspectual distinction, where, say, the reason for the use of the
Inessive case in kyyneleissä 'all in tears', is due to the idea that the
designated state of affairs (that of being all in tears) is being characterized
as completed, perfected, an ultimate state, and fulfillment of such a state
of affairs, then the Boundedness feature can be seen merely as a particular
case of a more general aspectual distinction in terms of (im)perfectivity or
(in)completeness.
Whereas the Location/Direction parameter is fairly unproblematic, the
explication of the Inner-Outer, Boundedness parameter is not as
straightforward. Although the problems that arise might at first seem
specific to Finnish grammar, they have important implications both for
deixis and for the semantics of case marking in languages in general.
Figure 1.
10
Key:
D demonstrative term expressing:
Dl 'closeness to speaker' [Sp. Proximal]
D2 'closeness to addressee' [Addr. Proximal]
Dx 'at a distance from speaker' [Medial/Distal]
Dsd 'short distance from speaker' [Medial]
Did 'long distance from speaker' [Distal]
D-v 'non-visible' [- Visible]
DEIXIS IN FINNISH 257
types, and given detailed treatment by scholars (cf. e.g. Levinson 1983, ch.
2), as we move from Place, Time, and Person deixis to Discourse and
Social, not to mention Affective deixis, we also gradually (and qualitatively)
move away from a prototypical deictic situation as defined in terms of
physics correlates. For instance, and in particular, the use of this vs. that
affectively, to express one's positive or negative attitude, respectively, as in
(14) and (15)
(14) This Bush isn't such a bad guy after all.
(15) That Bush really made a fool of himself again yesterday.
is often seen as a spatial metaphor, as a metaphorical extension of the
Proximity-Distance distinction in the demonstrative pronouns in English:
this is proximal, thus positive; that is distal, thus more negative. (Cf. further
R. Lakoff 1974; and Collinson 1937.)
My purpose is not to argue against a description in terms of
metaphorical extension with respect to the use of this and that (in 14-15)
in English; in fact, the suggestion seems highly plausible. Rather, I want to
argue that not all languages work in the same way, and that other types of
deixis — even Affective deixis — might have to be considered (cognitively)
'basic' (and not 'metaphorical extensions') in other languages, and thus in
language in general.
In order to show this convincingly, I need to make a brief digression of
a more theoretical nature.
4. A theoretical challenge
11
In this connection, cf. R. Lakoff (1989: 966): "... 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 dubious advantages of modeling our own
theory and method on those of an obsolescent field."
12
Still, for those of us who go for both cognitively and socially realistic descriptions
of phenomena in language, who in some way or other work on Yngve's (1986) assumption
that all things do ultimately hang together, a sane Bloomfieldian-kind-of scepticism is
often only healthy.
DEIXIS IN FINNISH 259
13
An aspect which will not be dealt with in this paper, but which can be deduced
from Östman (1992) would be the status of physics (as a science very closely connected
to mathematics) vis-à-vis linguistics, which has at least in some quarters attempted to
break free from the bonds of mathematics during the last decades.
260 JAN-OLA ÖSTMAN
(When referring to these as lexemes, the abbreviations TÄÄ, TUO, and SIE,
respectively, are used in the ensuing discussion.)
The whole system of demonstrative adverbs in Finnish is given in table
3; the demonstrative pronouns 'this', and 'that', as well as the temporal
adverbs, 'now' and 'then', are included for ease of reference.14
tämä tuo se
Table 3.
[In = Inessive; Ad = Adessive; El = Elative; 111 = Illative;
All = Allative; Ab = Ablative]
14
Only those temporal adverbs that are formed from the same base form as the
demonstrative adverbs are given; for instance, the default word for 'now' is nyt.
15
The Inessive and Elative forms of the SIE-term are exceptional, but functionally
their use is comparable to what would be regular forms.
DEIXIS IN FINNISH 261
success. It is obvious, for instance, that neither the SIE nor the TUO term
is used in Finnish as an equivalent to the English yonder, as we can see
from example (16).
(16) A: Mitä tuossa laatikossa on? (pointing to a box on the kitchen
table, as A and B are in the kitchen making dinner)
'What is in that [=TUO] box?'
B: Siinä on käärme.
There's a snake (in) there [=SIE].'
Instead of starting by going through other people's suggestions in
detail, and presenting counterexamples to their characterizations of the
TÄÄ-TUO-SIE distinction, I will proceed by taking into account what I have
said about other possible ways of looking at deictic systems in languages,
and criticize alternative analyses of the Finnish 'Proximity' system as I go
along. In particular, I want to suggest what an alternative analysis of the
Finnish demonstrative system might look like, an analysis which is not
primarily based on physical aspects like location.
The greatest problems with the Finnish system are centered around the
interpretation of the third term in the system, the SlE-forms. At least in an
initial analysis we come close to understanding the TÄÄ and TUO terms if
we keep to the characterization in figure 1. But the third term does not fit
in neatly. There are at least three alternative strategies to follow in this
situation:
Strategy 1 would simply say that the general scheme is valid, and if there
are uses of, in this case, the third term that do not fit, such
uses are due to metaphorical extensions/interpretations of
what the third term in a three-term system typically stands for.
Strategy 2 would say that the proximity distinction is not a major or basic
one in Finnish at all. The primary distinction is in terms of
something else, and only as a consequence of this other
distinction can the uses of TÄÄ, TUO, and SIE be talked about
in proximity terms.
Strategy 3 would say that the first two terms make up a basic, simple
proximity system - just as in English; but that the third term
is either
(3a) completely outside the system that the other two terms
262 JAN-OLA ÖSTMAN
a loose tendency. TUO can very well be used of something close to the
addressee, especially if the speaker and addressee are further apart than a
couple of feet:
(19) Tuossa sinun vieressäsihän veitsi on!
there-INESS your beside-INESS:2sg-RELEV knife is
T h e r e [=TUO] next to you is where the knife is!'
Secondly, the idea of SIE indicating Givenness of referent is not as
straightforward as one would like it to be. (And in addition, Givenness is
of course inherently linked up with Discourse deixis.16) Wiik gives the
following example:
(20) Tuolla näet tuomiokirkon. Siellä on kaunis alttaritaukh
T h e r e [=TUO] you see the Dome. There's a beautiful altar
painting there [=SIE]/
However, I can perfectly well introduce the topic of the Dome with a
sentence like (21).
(21) Siellä tuomiokirkossa on kaunis alttaritaukh
T h e r e [-SIE] in the Dome is a beautiful altar painting.' 17
And thirdly, it would surely be more satisfying if we were able to
stipulate one basic function for all the uses of the deictic SIE. In Wiik's
analysis we would have to say that there are in fact two homonymous
terms: SIE1 and SIE2. Here my stand is a cognitive-linguistic one in line with
the notion that languages tend to adhere to a basic one-meaning-one-form
principle (cf. Bolinger 1977).
16
It may indeed be worth while investigating to what extent the SIE term has a basic
Discourse deictic function in present-day Finnish. The use of se (which is both the third
person singular neuter pronoun, and the third-term demonstrative pronoun) for cohesive
purposes in discourse is well-known.
One can of course here argue that the difference is that in (21) the deictic SIE-
term is used attributively, and that the new referent is thus introduced in the same phrase.
Still, the difference between SIE and TUO in this respect is that both can occur
attributively, but if they occur independently TUO is accompanied by some ostensive
gesture, whereas SIE is not. Thus in the first sentence of example (20), TUO can be said
to be used attributively to the necessary gesture of pointing, and SIE in the second
sentence has a Discourse deictic function. The difference between New and Given is thus,
it seems to me, secondary, and dependent on the presence of an ostensive gesture.
DEIXIS IN FINNISH 265
Strategy 2 invites the stipulation of a completely different 'basic'
parameter behind the semantics and pragmatics of the demonstrative
system in Finnish, a parameter which would be capable of accounting for
the data more homogeneously than the Proximity distinction.
Such a parameter — I propose — can be set up in terms of Affective
deixis. As I have mentioned in section 4 and above, a distinction on Spatial,
Temporal, and Personal grounds cannot a priori be taken as more basic
than one on Discourse or Social grounds, and I would like to suggest that
Affective deixis, or Emotional deixis as it is sometimes called, is also a
potentially primary way of perceiving and thus codifying the world of deixis.
According to this suggestion, the SIE-term is the neutral one, the TÄÄ-
term is speaker-oriented, and the TUO-term is non-speaker oriented.18
The distinction is in terms of what scholars refer to as Involvement, Affect,
or Empathy: The TÄÄ-term is used for referents the speaker has empathy
with, or is involved with. Thus, as a consequence, it will also be used for
things proximal to the speaker. The TUO-term is used for cases where the
speaker does not have empathy with the referent, or where s/he even has
antipathy towards the referent — thus, as a consequence it is used of distal
things, and of referents that are new, because they are unknown. The SIE-
term, finally, is basically used when empathy, involvement, and affect is not
at issue in the communicative situation.
Before going through this suggestion in some detail, I want to make
the general comment that I see my own approach as complementary to
Larjavaara's very elaborate one. Mine is an approach that tries to get at
the (unconscious) cognitive categorization that Finnish speakers may be
involved in as they make decisions of what demonstrative to use.
Furthermore, I take for granted that since the three terms TÄÄ, TUO, and
SIE do pattern together, we should try to find some common denominator,
some prototype definition of what semantico-functional parameter the
TÄÄ-TUO-SIE system in Finnish is based on.
One kind of counterargument against the suggestion of Affect and
Empathy would be to say that, since the other two subsystems in the
18
A somewhat similar distinction might be appropriate for German hier - dort - da,
with da as the neutral term (cf. e.g. Ehrich 1982: 62); and for Japanese sore - kore - are,
with are being the neutral term (cf. Fillmore 1982: 54). Compare also the -va, -sa, -ny
distinction in Yuma, where according to Halpern (1946) the third term indicates location
unspecified'. Perhaps also the Spanish ahi can be seen as a neutral term in the sense I
have in mind here.
266 JAN-OLA ÖSTMAN
Finnish deictic system have to do with space, why should not the third
subsystem have this feature in common with the other two? However, as I
have shown in the discussion above, it is not at all clear that both of the
other two subsystems have to do with space as a basic parameter. It is true
that the Location/Direction subsystem definitely has to do with physical
space, but the Inner-Outer system proved to be based on an aspectual
distinction. (And we even found - cf. examples (12) and (13) - that
Boundedness, too, is relative to the addressee's orientation.) That is, space
is not as such of primary importance in all aspects of the Finnish
system.19
In the spirit of the generative argumentation of the 60s, I will now give
a number of arguments for, and instantiations of, the distinction I have
suggested. I will pay most attention to the SIE-term, since its use has been
notoriously difficult to pin down. Also, the SIE-term is crucial in that if we
accept the challenge from cognitive linguistics of finding some unifying
parameter with whose help we can explain the TÄÄ, TUO, and SIE terms
and relate them to one another, then focussing on the functions of the TÄÄ
and TUO terms might become counterproductive: In both the figure-1
approach and in my alternative, the TÄÄ-term will be linked to the
speaker, and the TUO-term to the addressee, or to some form of distance
from the speaker. Against this background, it will be predictably difficult to
say which of the suggested interpretations is more basic and primary than
the other. However, if the SIE-term can be shown to fit one system better
than another, we can obviously decide in favor of the system it better fits.
As we have already seen, the use of SIE is not easily amenable to the kind
of model presented in figure 1. The points below attempt to show the
feasibility of the alternative model. The first two arguments establish the
SIE term as more 'neutral' than the other two. The rest of the arguments
address the issue of Affective deixis more directly.
1. First of all, although Finnish does not usually make use of any
grammaticalized deictic form like English existential there in sentences like
There's a snake in the garden, on the few occasions that Finns do need such
This being the case, we could even be so bold as to use this counterargument for
our benefit: Finnish makes one distinction in terms of space, and another in terms of
Boundedness. There is thus no a priori reason why the third distinction should be in
terms of space. The default expectation in this situation would be that the distinction is
not based on location.
DEIXIS IN FINNISH 267
a grammatically neutral term, siellä or siinä (i.e. SIE) are the ones that are
given this neutral status.20
The inherent neutrality of the SIE-term also shows up in that it is the
term that is used neutrally without necessary reference to any particular
location. Thus, if I say (22),
20
We can here note that the grammatical neutrality of German da, Japanese are, and
Finnish SIE is motivated in a sense that English there is not.
21
Expressions more similar to English 'here and there', with the ordinary
conjunction ja 'and', have recently begun to occur; thus, we may find siellä ja täällä
instead of the more 'archaic' siellä sun täällä or siellä täällä,
268 JAN-OLA ÖSTMAN
22
Kalliokoski (1986) has argued that the order of elements in freezes of the siellä
täällä kind is one indication that the ME-First principle cannot be upheld cross-
linguistically. Alternatively, I would say that the reason Finnish does not follow the
principle is not because the western-culture based principle as such is wrong, but only
that speakers of Finnish, and no doubt of other languages too, do not necessarily
categorize the world in the same terms (of physics) as speakers of Indo-European, western
languages have grown up to do.
23
In general, Finnish seems to be more apt to use what I have called the YOU-
Importance principle (Östman 1981a), rather than the ME-First principle. Thus, in
Finnish one typically says Marja ja minä 'Marja and me', and sinä ja minä 'you and me'.
The opposite order of the terms is almost unthinkable.
24
Neither suggestion explains why täällä and not tuolla is chosen as the second term
in siellä täällä, 'here and there'. My suggestion is that since both SIE and TUO, according
to almost all analyses, are viewed as distal in relation to the TÄÄ term (but cf. below, for
additional explanations), having two distal terms to stand for an expression that means 'a
little bit all over', would exclude the area in close vicinity to the speaker. Thus, since siellä
has to be chosen because of its neutrality, täällä is the only, and obvious, choice as the
second term.
25
I understand, though, that täällä tuolla is indeed an expression that in the very
recent years has started to creep into spoken Finnish - probably due to influence from
the format of equivalent expressions in English. (Kari Pitkänen, p.c.) I have also found an
instance in a newspaper of the expression tuolloin tällöin 'now and then' (or rather, 'then
and now'), but I presume this is word play rather than a general tendency.
DEIXIS IN FINNISH 269
3. Thirdly, when these terms are used in an emotional deictic manner
(what I have called Affective deixis in this paper; cf. R. Lakoff 1974;
Collinson 1937; Lyons 1977), TUO is the most negatively loaded:
(24) Tuo Vayrynen on sitten maailman suurin älykääpiö.
That [TUO] Väyrynen is really the world's biggest pighead.'
TÄÄ is more positively loaded:
(25) Tä(m)ä Fillmore on sitten fiksu tyyppi.
This [TÄÄ] Fillmore is a really sharp guy.'
Note also that se is the third person neuter singular pronoun, 'it', and in many
dialects of Finnish - and very frequently in informal speech - it is often used without
derogatory implication of people.
The distinction that Fillmore (1975: 71) mentions for English between 'this' as being
used when only the speaker knows what is referred to, and 'that' when both speaker and
addressee know, seem not to work for Finnish; cf.
(i) Tapasin eilen poliisin Espalla; no, se/tää/tuo kaveri/kundi/tyyppi oli ihan
umpikännissä.
'I met a policeman on the Espa boulevard yesterday; well, this/?that guy was
really stoned.' '
(ii) Muistatko (sitä/tätä/tuota) kaveria, joka myi meille lippuja matsiin? No,
se/tää/tuo kundi kertoi minulle, että...
'Do you remember the guy who sold us tickets to the game? Well that/?this guy
told me that...'
I am not completely convinced that this holds as a tendency even in English, unless
reinterpreted in terms of Affective deixis.
21
I am here following a suggestion made by Auli Hakulinen (p.c.).
270 JAN-OLA ÖSTMAN
presupposes a shared experience, a shared perspective on things.28 TÄÄ
(as in (25)) involves a process of the speaker 'placing' the addressee next
to him/her and imposing his/her perspective on the addressee in order for
both of them eventually to end up sharing the speaker's perspective. Thus,
as a consequence, TÄÄ will be used to introduce something the
addressee/audience is not familiar with from before, as in Sit tää koira
yhtäkkiä ilmestyy ja ... Then this [TÄÄ] dog suddenly appears and ...'
whereas some other specifier has to be used if the speaker is not interested
in pulling in the addressee into his/her story: Sit yks' koira ilmestyi ja ...
Then a [='one'] dog appeared and ...'. In a Cognitive-Affective
perspective, if a referent is viewed as important enough by the speaker for
min/her to pull in the addressee, the speaker (possibly inadvertently)
displays a potentially positive attitude to this referent. Dialogically
speaking, TUO (as in (24)) would primarily focus on the speaker's relation
to the object referred to. Here, at the same time as the speaker (almost
accidentally) pulls in the addressee, the speaker's negative attitude towards
the referent under discussion is clearly displayed. In this way, the function
of Affective deixis can be supported interactionally.29
28
Cf. Fillmore's (1982: 54) discussion of are in Japanese.
29
The primacy of cognition vs. interaction is a notorious chicken-and-egg issue. In
this paper whose main empirical task is to find the parameter according to which the
demonstrative adverb system in Finnish is codified, I have preferred to treat them as
cooperating.
DEIXIS IN FINNISH 271
5. Fifthly, why is SIE often felt to be used for old information?30 Old
information is common to both speaker and addressee. Thus, the neutral
SIE is obviously the default choice. If any of the speakers are particularly
involved with the matter talked about, this can be negotiated in other ways
- explicitly or implicitly. If implicitly (cf. Östman 1986), then TUO is indeed
one alternative (Bl, B2, and B3 are alternative responses to A):
(27) A: Tää Jörn osaa sitten olla herttainen.
This [=TÄÄ] Jörn really knows how to be sweet.'
Bl: Kuules, tuota Donneria et kyllä tuo tähän taloon yöksi!
Tisten, don't you even think of bringing that [=TUO] Donner
into this house for the night.'
B2: Sinähän taidat olla vähän rakastunut siihen Donneriin!
'You seem to be a little bit in love with that [SIE] Donner.'
B3: Olkoonkin herttainen, kunhan muistat, että tää söpöliini, se on
mun!
'Call him sweet if you like, as long as you remember that this
[TÄÄ] cutie is mine.'
6. Sixthly, with the suggested interpretation of the function of the third
term, SIE, we can also see why one might not so readily want to use TUO
when the speaker and addressee are close together (cf. Wiik's suggestion
above):
(i) TUO is non-speaker oriented (thus TÄÄ could just as well be used);
and,
(ii) SIE, being neutral with respect to Proximity, can thus be used in all
situations that do not explicitly require TÀÂ or TUO.
7. Seventhly, we can now also, referring back to example (20) discussed
earlier, explain why we do not so readily use siellä to accompany an
ostensive act (cf. also fn. 17). The reason is the neutral character of siellä:
If one points to something, one makes a marked choice of indicating one
phenomenon over and above all others.
30
Saukkonen's (1967) findings rather suggest that se is neutral with respect to Given
and New. Cf. also fh. 16.
272 JAN-OLA ÖSTMAN
These arguments will have to do for the time being.31 The important
point to note is that the parameter of Affective deixis that is needed in
order to adequately describe the Finnish demonstrative-adverb system is
not derived from the place-time parameter, but is an independent, basic
parameter. And since this is so for Finnish, it is most likely a parameter to
be taken seriously for debris in general.
As we know, our views of place and time are based on, and filtered
through, our perception. And if as a corollary to this, we can in addition
say that time and place a priori do not have any more justified status in the
world according to physics than does any other product of our perception,
then there is no reason to assume that place and time are the only
concepts with cognitive status which partake as determinants of
demonstrative-adverb systems.
5, Further implications
Having a member of a system as the default member - the way SIE can be
seen as a default member in the demonstrative system — is nothing
extraordinary in Finnish grammar. As I pointed out in Östman (1981b)
there seems to be a clear tendency to use the nominative case as a default
case where it should not really appear according the the rules of Finnish
grammar. This happens in cases where there is either just, say, one NP in
the sentence, or where there for other reasons can be no possibility of
misunderstanding. Thus, the impersonal construction (sometimes
erroneously called a 'passive') takes a nominative form as object case
where one would expect an accusative (cf. Östman 1981b), Also, in
imperative sentences, not only do objects occur in the nominative, as in
(28) Ota auto!
take-IMP:2sg car-NOM:sg
Take the car.5
31
For instance, it is not inconceivable that the uses of sitä and t(u)ota as pragmatic
particles could also be explained with reference to the discussion in this paper. Basically,
t(u)ota is a hesitational device; hesitation marks indeterminacy, which might be connected
with negative attitudes. Sitä has been extensively discussed by Hakulinen (1975); perhaps
my suggestion of its euphemistic use for politeness purposes might come in handy here.
These are, obviously, fairly speculative suggestions and matters that need to be discussed
more fully.
DEIXIS IN FINNISH 273
Cf.
(29) Otan auton.
take-PRES:lsg car-ACC:sg
T m taking the car.'
but also adverbial phrases in imperative constructions can take the default
nominative case instead of the accusative; cf.
(30) Juokse reippaasti koko matka!
run-IMP:2sg briskly whole journey-NOM:sg
'Run briskly all the way.'
6. Conclusion
32
I am grateful to Mauri Ronimus for lengthy discussions on these issues.
DEIXIS IN FINNISH 275
References
Bolinger, Dwight L.
1977 Meaning and form. London & New York: Longman.
Collinson, W.E.
1937 Indication. (Language Monographs 17.) Linguistic Society of America.
Ehrich, Veronika
1982 "Da and the system of spatial deixis in German". In Jürgen Weissenborn &
Wolfgang Klein (eds.), Here and there: Cross-linguistic studies on deixis and
demonstration. Amsterdam & Philadelphia: John Benjamins, 43-63.
DuBois, John W.
1985 "Competing motivations". In John Haiman (ed.), Iconicity in syntax.
Amsterdam & Philadelphia: John Benjamins, 343-365.
Fillmore, Charles J.
1975 Santa Cruz lectures on deixis 1971. Bloomington, IN: Indiana University
Linguistics Club.
276 JAN-OLA ÖSTMAN
Hakulinen, Auli
1975 "Finnish sitä - An instance of the interplay between syntax and
pragmatics". Scandinavian Conference of Linguistics 2.147-163.
Halpern, A.M.
1946 "Yuma III: Grammatical processes and the noun". UAL 12.204-212.
Itkonen, Terho
1966 "Tutkimus suomen asyndetonista". [Investigating asyndeton in Finnish]
Virittäjä 70.
Kalliokoski, Jyrki
1986 "Rinnastusfraasin jäsenten järjestyksestä". [On the order of units in
coordinate phrases] Kieli 1.73-102.
Karlsson, Fred
1982 Finnish grammar. Porvoo, Helsinki, Juva: WSOY.
Lakoff, George
1972 "Hedges: A study in meaning criteria and the logic of fuzzy concepts". CLS
8.183-228.
Lakoff, Robin
1974 "Remarks on this and that". Berkeley Studies in Syntax and Semantics, XVII.
Berkeley, CA: Department of Linguistics, UC-Berkeley.
1989 "The way we were; or, The real actual truth about generative semantics: A
memoir". Journal of Pragmatics 13:6.939-988.
Larjavaara, Matti
1985 "Suomen demonstratiivisysteemin rakenne". [On the structure of the
demonstrative system in Finnish] Sananjalka 27.15-31.
Lehtinen, M.
1967 Basic course in Finnish. Indiana University Press.
Leino, Pentti
1989 "Paikallissijat ja suhdesääntö: Kognitiivisen kieliopin näkökulma". [Local
cases and the relation rule: An approach from cognitive grammar] Virittäjä
93:2.161-219.
Levinson, Stephen C.
1983 Pragmatics. Cambridge: Cambridge University Press.
Lyons, John
1977 Semantics. Cambridge: Cambridge University Press.
Östman, Jan-Ola
1981a "You know: A discourse-functional approach". Amsterdam & Philadelphia:
John Benjamins.
Saukkonen, Pauli
1967 "Persoonapronominien hänise, he:ne distinktiivi oppositio". [The distinctive
opposition between the personal pronouns hän he:ne] Virittäjä 71:3.286-
292.
Talmy, Leonard
1978 "The relation of grammar to cognition - A synopsis". TINLAP 2.
Verschueren, Jef
1987 Pragmatics as a theory of linguistic adaptation. (IPrA Working Document
1). Antwerp: International Pragmatics Association.
Wiik, Kalevi
1975 "Ruotsinkielisten tuolla-siellä -vaikeudesta". [On the difficulties that
Swedish speakers have with tuolla-siellä] Kielitieteellisiä lastuja 1.36-37.
Turku: Linguistic Association of Finland.
Yngve, Victor H.
1986 "To be a scientist". LACUS Forum 13.1-21.
Frame semantics and the lexicon:
Nouns and verbs in the body frame
1. Introduction
Fillmore (1975, 1976a) introduced the FRAME idea into linguistics in part
as a response to what has been called "checklist theories of meaning". In
that approach to semantics, the meaning of a linguistic form is represented
in terms of a checklist of conditions that have to be satisfied in order for
the form to be appropriately or truthfully used. In FRAME SEMANTICS,
word meaning is characterized in terms of experience-based schematizations
of the speaker's world - i.e. frames. It is held that understanding any
element in a frame requires access to and understanding of the whole
structure. Later on, Fillmore (1978) characterized the frame as the most
central and powerful kind of domain structure, thus alluding to a
frame-based organization of the lexicon. In a frame-based organization of
the lexicon, it is the frame which provides the conceptual underpinnings for
related senses of a single word and semantically related words. This would
necessarily include all categories of words, including nouns, verbs, adjec
tives, adverbs, prepositions, and conjucntions, as well as phrases and
expressions.1
In my own work (Petruck 1986, forthcoming), I have attempted to
validate and extend the frame semantics approach to the study of meaning
- word meaning, word structure, and semantic structure in the lexicon - by
demonstrating its applicability to one area of the lexicon in one language,
1
Of course, Fillmore's other writings on frame semantics (1977, 1982, 1985, 1992)
have influenced my work on the subject as well.
280 MIRIAM R.L. PETRUCK
First, in section 2.1,I present a set of compound noun data to illustrate the
phenomenon with Hebrew body part terms; a subset of the compound noun
data is discussed in detail. Next, in section 2.2.1, I cover a set of complex
nouns, more specifically, names for clothing and clothing-like items, which
are derived .from body part terms. Then, in section 2.2.2, I turn to a set of
denominal body part verbs whose parent nouns are instruments. With each
set of data, I claim that knowledge and experience of the body and its parts
2
There are other instances of body part terms used as measure words in Hebrew
including zeret - 'pinky', 'span' and 'ama - 'forearm', 'cubit'. In English we find foot and
finger.
282 MIRIAM R.L. PETRUCK
are integral to the creation or formation of the compound or complex
forms.
Some extensions of body part terms are far from their source - that is, the
extended sense does not refer to something literally so close to the body.
For example, in English we have head of the department, body of work,
mouth of the river, eye of a needle, and many more. Data of this sort
illustrate the non-corporeal sense of polysemous body part terms and
indicate the domains with which the body part domain interacts. A
complete specification of the body frame would have to include all such
domains and an explanation of the interaction.
One way of illustrating this sort of extension in Hebrew is with
compound nouns, smixut forms, where the first of the two nouns is a body
part term. Compounding is a process available in the language for the
creation of new words; in particular, compounding is a naming device
(Zimmer 1971; Downing 1977). Objects and entities or parts of objects and
entities named in this way are given in Tables la and lb. The claim is that
compound forms of this sort are linked to body part terms in the body
frame because it is experience of the body and knowledge about the body
that has been drawn upon for their formation and understanding.
The data show that there are a number of characteristics of a body part
which can serve as the basis of the extension, including function, size, shape,
and position. Other factors such as configuration and orientation may need
to be considered as well. For example, pi ha-me'ara - 'mouth of the cave'
and pi ha-be'er - 'mouth of the well' are based on the shape of the mouth
in a particular configuration. Forms such as pney ha-binyan (face of the
building) - 'surface of the building' and pney ha-yam (face of the sea) -
'surface of the sea' show the need for different orientations: The former
assumes the canonical vertical orientation of human beings and the latter
assumes a horizontal orientation. In some cases, more than one factor plays
a role in the extension. For instance, in ragley ha-ŝulxan (legs of the table)
- 'table legs', while the function of the legs may be viewed as the major
factor, the extension obtains in part because of their shape.
FRAME SEMANTICS AND THE LEXICON 283
FUNCTION
SIZE3
SHAPE
3
As a body part word guf has two senses - it refers to the whole body or a large part
of the body, the trunk. It is the latter sense which figures into the meaning of the
compounds listed here.
4
Theuse of yad - 'arm' for 'monument' shows up in other constructions which are
not smixut forms - e.g. yad la-banim (arm to-the-sons) - 'Monument to the Sons', a
monument to fallen soldiers, and yad va-šem (arm and-name) - 'Yad VaShem', a memorial
monument to the martyrs of the Holocaust.
284 MIRIAM R.L, PETRUCK
POSITION
5
katef ha-har - shoulder of the mountain is listed in parentheses because it is a
military topographic term, not necessarily used by the ordinary speaker in everyday
conversation.
FRAME SEMANTICS AND THE LEXICON 285
physical experience (Lakoff and Johnson 1980a, 1980b) and offer further
justification for the experience-based schematization of word meaning
(Fillmore 1982, 1985; Lakoff 1987). If words acquire new senses through
speakers' experience of and interaction with the world, then it is most
appropriate to invoke experience in the characterization of word meaning.
A closer examination of this phenomenon is warranted in cases where
the semantic motivation for the extension is based on similarity of position.
The data show that roS - 'head' is used for the top part of a variety of
objects and entities; panim - 'face' names the outside surface of different
things; safa - 'lip' can be used for an edge; gav - 'back' for the back of other
things, lev - 'heart', 'ecem -'bone', and beten - 'stomach' name different
kinds of inside parts of other entities. Extensions based on similarity of
position are particularly interesting. When parts of objects are named in
this way, we see that there are entitites and objects which can be viewed as
structured in ways which parallel the structuring of the human body and, in
fact, it is relative position that matters. Some examples from the present
data are given in Table 2.6 The metaphors obtain because of a perceived
similarity between the structure of the body and the structure of some other
object, along with the ability to conceptualize one object in terms of
another.
The clearest illustration of this is har - 'mountain' where a schematiza
tion of the human body has been applied to a mountain to name its parts.
Thus, we findrošha-har (head of the mountain) - '(the) summit', gav ha-har
- 'back of the mountain', katef ha-har - 'shoulder of the mountain', which
is a military/topographic term, and ragley ha-har (feet of the mountain) -
'foot of the mountain'. Note that this schematization assumes the canonical
upright and forward-facing orientation of the human body (Clark 1973).
Note also that har - 'mountain', which has no inherent front/back orient
ation, has been schematized as facing the speaker, as evidenced by the use
of gov - 'back' for the distant part. Thus, knowledge of the way the body is
carried and oriented also figures into the way entities structured in terms
of the human body are named.
6
Except for lešon yam (tongue of the sea) - 'fjord', listed in parentheses, all of the
extensions in Table 2 are based on similarity of position.
286 MIRIAM R.L. PETRUCK
har - mountain
hinyan - building
'ir - city
yam ° sea
7
Infact ros ha-'ir is not a part of 'ir in the same way that, for example, ros ha-har is
part of har, although roS ha-'ir may well be "attached", emotionally or psychologically to
'ir.
288 MIRIAM R.L. PETRUCK
concrete domains. More generally, the compound noun data are impor
tant because they exemplify an extremely basic level of metaphor which
serves as the basis of further extensions. The conceptual underpinnings of
these extensions are provided by and structured in the body frame. As we
have seen, different characteristics can serve as the basis of an extension
and other factors such as orientation and configuration need to be
considered. Characterizing these compound nouns in terms of the body
frame makes it possible to incorporate all of this information in the
description of their meanings.
Now let us turn to complex nouns and verbs derived from body part terms.
The formal apparatus of Hebrew derivational morphology is such that the
creation of vocabulary from native material is a relatively simple and
straight-forward matter. In principle, once the consonantal structure of an
existing word is determined it can be used in any of the miškalim, nominal
patterns, or hinyanim, verbal conjugations. For the present discussion, a new
word is created as a result of the body part term having undergone one of
the derivational processes in the language. Two kinds of examples - derived
nouns based on body part terms and derived verbs - are presented.9
8
InPetruck (1988), there is a discussion of mappings from the body to the domain
of location, providing a detailed example of mapping from a concrete domain to an
abstract domain.
9
I have limited the discussion in this paper to nouns and verbs. It should be noted
that there are also adjectives, adverbs, prepositions and conjunctions derived from body-
part terms in Hebrew.
10
Petruck (1987) covers other groups of complex nouns derived from body part terms
including names for diseases, (e.g. damernet - 'hepatitis' from dam - 'blood'), instruments
(e.g. yadit - 'handle' from yad - 'arm/hand', and abstract nouns (e.g. resit - beginning from
roš - 'head').
FRAME SEMANTICS AND THE LEXICON 289
yad (spoon of arm) - 'hand', named a part of the body.12 Today, the
monolexeme kaf only means 'spoon'; kfafa is a cover for the hand, not for
a spoon, cavaron - 'collar' and karsolit - 'gaiter' are straight-forward in that
they name clothing worn on or near the body part from whose name the
term is derived; in both cases the garment is worn around the body part.
Like hitna - lining', gufiya - 'undershirt'and xaziya - 'bra' are based on
polysemous body part terms. With both guf - 'body, trunk' and xaze - 'chest,
breasts', the two senses of each word can be characterized as whole and
part of that whole. In each case, the derived form is based on the "part"
sense of the word. In principle, gufiya could have meant 'wetsuit' and xaziya
could have meant 'undershirt', motniya - 'vest' (worn on the chest) doesn't
seem to have anything to do with moten - 'waist', although a typical vest
reaches to the waist.13 Although micxiya - 'visor' is worn on the forehead,
its purpose is to protect the eyes, ktefiya - 'shoulder strap' seems straight
forward as it names the part of a garment worn on the shoulder. However,
shoulder pads and epaulettes are also worn on the shoulder and the term
ktefiya does not refer to either of these items. Note that kotfot - 'epaulettes'
is heard in the army.
While they are not really articles of clothing, 'ecba'on - 'thimble' and
'ozniya - 'earphone' are also things which are worn - 'ecba'on is worn for
protection; it covers the body part all around. Given the size and shape of
a thimble, it seems possible to analyze 'ecba'on comparable to cavaron -
'collar', where -on is one of the clothing word suffixes, 'ozniya is different
because it goes inside the body part, in contrast with, say, gufiya and xaziya.
Clothing terms are linked to body part terms in the body frame because
it is knowledge about the body - the physical characteristics of its parts -
and what is done with the body - that it is customary to cover the body in
a particular way - that is drawn upon in the formation or creation of such
words. The data demonstrate this in a profound way: In Hebrew names for
clothing and parts of clothing are derived from body part terms. I am not
saying that all clothing terms have to be morphologically related to body
part terms. Rather, the fact that there are such clothing terms provides
support for the connection to the body frame.
14
Petruck (1987) also covers other groups of complex verbs derived from body part
terms including those where the body part is an affected object (e.g. 'akav - 'follow' from
'akev - 'heel') or a product (e.g. hištin - 'urinate' from Men - 'urine').
Apparent deviations from these patterns are a result of predictable changes. Thus,
for example, the second vowel in yida, whose underlying consonantal root is YDH, is a
rather than the expected e. Although the glottal fricative is not realized in contemporary
Hebrew pronunciation, there are still surface phonetic manifestations of its underlying
existence. The underlying form /yideh/ is realized as [yida] because in word-final position,
a glottal lowers the preceding vowel.
292 MIRIAM R.L. PETRUCK
and the work to which the instrument can be put, which is indicated in the
denominal verb. Some of the verbs designate actions which are the body
part's anatomical function. Examples of this are ziyen - 'fuck' and 'if'ef -
'blink'. Other verbs designate actions that are more conscious, deliberate or
purposeful than their regular/typical function. For example, 'iyen - 'look
into' is not just to use the eye for seeing, but rather to see consciously, with
a purpose in mind. Similarly, yida - 'throw', which is restricted in use,
involves the use of the hand for an action with some force, throwing, and
not for example, giving, grasping, or handling. Likewise, with hicbia - 'raise
finger, vote' and hilšin - 'tell on', the actions are purposeful. A finger is
raised to get attention, as for example, in a class; the purpose is especially
clear in the 'vote' sense of the word. The tongue is used for talk, but telling
on (someone) is talk with a specific purpose and seems to include a value
judgement. The distinction can best be seen with 'izen - 'balance' and he
he'ezin - 'listen'. The verb 'izen reveals folk knowledge of the anatomical
connection between the inner ear and a person's sense balance, the inner
ear being the instrument which controls balance. 16 For the person whose
ear is doing the balancing, this act is involuntary. In contrast, listening is a
deliberate and purposeful act. In addition, these verbs differ from each
other in that izen designates an action physically outside of the body.
naxar - 'snore', 'inpef - 'talk nasally', and rigel - 'spy' are particularly
interesting because the action designated doesn't really involve the body
part. It is not the nostrils (or even the nose) that "snores", but rather the
soft palate. The vibrations of the soft palate, which can also create some
movement of the nostrils, cause the sound that we associate with snoring.
Physiologically, talking nasally only involves the nose in a negative way - i.e.
air doesn't pass through the nose, usually because the nasal passages are
blocked. The verbs suggest that people believe that the body part
mentioned is the instrument performing the action, more specifically, that
the body part is the point of origin of the action. And it is that belief which
16
For a detailed discussion of the bodily basis of balance, see Johnson 1987: 74-96.
The physiological and anatomical relationship between the inner ear and balance is known
by ordinary speakers. Indeed it was a linguistically sophisticated native speaker of Hebrew
who pointed out the relationship between 'ozen - 'ear' and the verb 'izen - 'balance'.
However, it should be noted that the histories of these two words show that they are from
different sources, which becomes apparent when considering the comparable terms in
Arabic. Whereas the Arabic body part term 'u un - 'ear' is cognate with the Hebrew term
'ozen, the Arabic verb wazana - 'weigh' is not cognate with the Hebrew verb 'izen.
FRAME SEMANTICS AND THE LEXICON 293
is lexicalized in the verb. Although the verb rigel - 'spy' incorporates the
notion of going on foot, spying involves using one's eyes much more so than
one's legs or feet. In this case, the history of the word is important.
Consider the Biblical phrase le-ragel 'et ha'arec (to-spy OBJ the-land) - 'to
spy the Land', where the spying actually included more physical involvement
of legs/feet.
girger - 'gargle', related to gargeret - 'Adam's apple' seems to incorporate
the fact that the Adam's apple is the visible external manifestation of the
throat which is an internal part. In addition, for a person who has a
protruding Adam's apple, it is possible to see its vibration while gargling.
Finally, with pana - 'turn', which is clearly related to partim - 'face', the verb
incorporates the idea that turning one's face counts as turning. If someone
says Turn to me, turning one's face toward them is appropriate. It is
possible to comply with the request by turning your face without turning
your body, but not vice versa.
These denominal verbs are linked to body part terms in the body frame
because their formation and understanding depend upon knowledge and
beliefs about the body - its physical structure, how it functions, and how
speakers experience it altogether. Again, not all verbs invovlving an action
of a body part are derived from a body part term. However, the existence
of these denominal body part verbs provide support for the connection to
the body frame.
3. Concluding remarks
We have seen that body part terms are used in the naming of (1) that
which is conceived of in terms of the body, as in the compound nouns used
to designate parts of objects or entities outside the body and (2) that which
is close to the body - either actual physical contact, as with clothing words,
or physical actions that parts of the body perform, as with denominal verbs
whose parents are instruments. The claim is that these words are linked to
the body frame not simply because of the obvious lexical or morphological
relationship between the compound or complex forms and the body part
terms from which they are derived. They are linked to the body frame
because it is knowledge and experience of the body and its parts which
provide the conceptual underpinnings for their formation and understand
ing.
294 MIRIAM R.L. PETRUCK
References
Andersen, Elaine S.
1978 "Lexical universals of body-part terminology". In J.H. Greenberg, CA.
Ferguson, & E.A. Moravcsik, (eds.), Universals of human language, Vol. 3
(Word Structure). Stanford: Stanford University Press, 335-368.
Clark, Herbert
1973 "Space, time, semantics, and the child". In T. Moore (ed.), Cognitive
development and the acquisition of language. New York: Academic Press, 27-
63.
Downing, Pamela
1977 "On the creation and use of English compound nouns". Language
FRAME SEMANTICS AND THE LEXICON 295
53:4.810-842.
Fillmore, Charles J.
1975 "An alternative to checklist theories of meaning". Proceedings of the First
Annual Meeting of the Berkeley Linguistics Society, 123-131.
1976a "Frame semantics and the nature of language". Annals of the NY Academy
of Sciences 280.20-32.
1976b "The need for frame semantics within linguistics." Ms. Typescript of lecture
given at Computational Linguistics (COLING) meeting. Ottawa.
1977 "Topics in lexical semantics". In Roger Cole (ed.), Current issues in linguistic
theory. Bloomington: Indiana University Press, 76-138.
1982 "Frame semantics". In Linguistics in the morning calm ed. by The Linguistic
Society of Korean. Soeul: Hanshin, 111-137.
Jakobson, Roman
296 MIRIAM R.L. PETRUCK
1971 "Two aspects of language and two types of aphasic disturbances". In Roman
Jakobson & Morris Halle, Fundamentals of language. The Hague: Mouton,
69-96.
Johnson, Mark
1987 The body in the mind. Chicago: University of Chicago Press.
Lakoff, George
1987 Women, fire and dangerous things. Chicago: University of Chicago Press.
forthcoming
Frame semantics and structure in the lexicon: The case of Hebrew body part
terminology.
Sweetser, Eve E.
1986 "Polysemy vs. abstraction: Mutually exclusive or complementary?" Proceed
ings of the Twelfth Annual Meeting of the Berkeley Linguistics Society, 28-538.
Traugott, Elizabeth
1982 "From propositional to textual and expressive meanings: Some semantic-
pragmatic aspects of grammaticalization". In Winfred P. Lehmann & Yakov
Malkiel (eds.), Perspectives on historical linguistics. Amsterdam: John
FRAME SEMANTICS AND THE LEXICON 297
Benjamins, 245-271.
Zimmer, Karl E.
1971 "Some general observations about nominal compounds". Working Papers on
Language Universals 8:C1-21.
The conceptual basis
of
performativity
Jef Verschueren
Belgian National Fund for Scientific Research
and
University of Antwerp (IPrA Research Center)
1. Introduction.
1
It is not only for this reason that the present paper is suitable for insertion in a
Festschrift in honor of Charles Fillmore. In fact it could and should have been written as
early as 1982 when I presented the basic ideas as a lecture in a course entitled
"Metapragmatics: Or, the lexicalization of linguistic action" which I was co-teaching with
Charles Fillmore (at his suggestion), John Gumperz, and John Searle at the University of
California at Berkeley. No doubt, the present formulation owes a lot to discussions with
my co-teachers and the participants in this course, as well as with more or less recent
audiences in Amsterdam, Budapest, Duisburg, Ljubljana, and Vienna (where I was invited
to talk about the subject by, respectively, Rob Grootendorst and Lachlan Mackenzie,
Ferenc Kiefer, René Dirven, Igor Žagar, and Wolfgang Dressier). Special thanks are due
to John Searle whose "How performatives work", the general drift of which I violently
disagree with, made me decide that I should commit my thoughts to paper, and to
Masayoshi Shibatani and Sandra Thompson, whose decision to compile a Festschrift made
300 JEF VERSCHUEREN
me do so. In its final stages, this paper has benefited from comments by Sandra
Thompson, Jan Blommaert, Gino Eelen, and Michael Meeuwis.
2
The extensive philosophical and linguistic literature on performativity, heavily
concentrated in the 1970's, includes: Åqvist 1972, Bach 1975, Bach & Harnish 1979,
Cornulier 1975, 1980, Davis 1964, Fauconnier 1979, Ginet 1979, Grewendorf 1979,
Grodzinski 1980, Heal 1974, Hedenius 1963, Lemmon 1962, McCawley 1977, Récanati
1980, 1981, Sadock 1974, Urmson 1977, Walker 1969, Warnock 1973, Yanovsky &
Maclaran 1980, Zaslawsky 1979, Zuber 1981. For more references, see Nuyts &
Verschueren 1987.
3
See Keck & Stubbs 1984, Hinrichs 1985, Leeuwen-Turnovcová 1986, and Žagar 1988.
4
"Ich habe in meiner eben erwähnten Polemik das Wort 'effectivum' in einem etwas
verschiedenen Sinne gebraucht, nämlich von einem Praesens, das die Handlung nicht nur
bezeichnet, sondern eben durch das Aussprechen des betreffenden Verbums zugleich vollzieht:'
(Italics correspond to the translated portion of the quote in the main body of the text; all
translations in this article are mine.)
THE CONCEPTUAL BASIS OF PERFORMATIVITY 301
2. The phenomenon
form in (9), a third person singular form in (10), and the passive in (11)
through (13):
(8) I am asking you to do this for me Henry, I am asking you to do it
for me and Cynthia and the children. (Searle 1989)
(9) We pledge our lives, our fortunes and our sacred honor. (Searle
1989)
(10) Le porteur déclare être majeur. (Cornulier 1980)
(The holder [of this ticket] declares that he/she is over 18 years
old.)
(11) Passengers are hereby advised that all flights to Phoenix have been
cancelled. (Searle 1989)
(12) Passengers are warned not to lean out of the window.
(13) You are dismissed.
Of these examples, (8) is the most disputable case of a performative: it
does not even contain the content of the request referred to with 'asking,5
which indicates that the act is a description of the request after it has been
made by other (linguistic) means, rather than the making of a request itself.
This does not mean that the present continuous cannot carry a
performative meaning; but if it does, at least repetition of a foregoing act
seems to be involved. Thus (14):
(14) I'm ordering you to get out!
is likely to occur after the 'primary' performative "Get out!" has failed to
yield the desired result.
A second class of non-central cases of performatives contains verbal
modifiers of various kinds:
(15) Yd guess there were twenty of them.
(16) I would say that this is completely wrong. (McCawley 1977)
(17) I would estimate/reply/etc. that ... (McCawley 1977)
(18) I will offer you the following alternative: ... (McCawley 1977)
(19) I will (simply) answer you that ... (McCawley 1977)
(20) Yd like to ask you to change your plans.
THE CONCEPTUAL BASIS OF PERFORMATIVITY 303
3. The problem
Eine Ausnahme macht das Slovenische; es scheint in demselben nämlich bei den
imperfektiven Verben das Gefühl für das Unvollendete der Handlung zu lebhaft zu
sein, als dass man ohne weiteres sich ebenfalls so ausdrücken dürfte. [...] es ist aber
schwer zu begreifen warum das logisch richtige gegenüber dem logisch falschen ein
Verderbniss sein sollte." (1903: 555)
[By uttering the words 'I thank, promise, pledge, offer my services,' the corresponding
acts are performed. Uttering the word may occupy a certain, even if very brief, time
span, but the speaker does not think of that and it also does not matter. The act is
to be regarded as momentaneous. It is strange though, that this momentaneous act
is in most Slavic languages performed by uttering the praesens of an imperfective
verb. (...)
An exception is Slovenian; it seems as if in this language the connotation of the
incompletedness of the action associated with the imperfective verb is too prominent
for this form of expression to be used as such. [...] but it is difficult to understand
why the logically correct should be a deviation vis-à-vis the logically false."
In other words: explicit performatives can only be viewed as momentaneous
acts, without duration; therefore it seems illogical that in most Slavic
languages, with few exceptions, imperfective verbs are used.
Replacing the term praesens effectivum with Koinzidenzfall5 many
5
"Diesen Fall, in dem die Handlung, von der gesprochen wird, eben im Aussprechen
der betr. Form besteht, habe ich den Koinzidenzfall genannt, weil in ihm Sprechen und
Handeln tatsächlich identisch sind, nicht aber weil beides zeitlich zusammenfiele, denn das
letztere ist ja doch natürlich bei jedem aktuellen Präsens der Fall." (Koschmieder 1935:
288)
[I have called this case, in which the action of which one speaks consists in the
uttering of the corresponding form, Koinzidenzfall because speaking and acting are really
identical, not just because they coincide in time, since that is of course the case with every
'actual present.']
THE CONCEPTUAL BASIS OF PERFORMATIVITY 305
[Why is it that in the Koinzidenz the perfective can refer to the present? The
coincidence of speaking and acting is only in the present imaginable, since in the past
or the future it is only possible to speak about something. But when I want to express
in the present that the uttering of my words 'is' at the same time the action, then I
express the 'entering' of this action into the present, and I can do that in the present,
since the uttering of the word in question is itself the realization of the action. In this
case - but not in any other case - the continuation and the entrance of an action are
opposed to each other in the present, just as they always are in the past and the
future. If the Koinzidenz must designate 'entrance' into the present, then the
perfcctive should be the norm. Therefore, what needs explaining is the imperfective
which is used in several languages.]
cannot fry an egg by saying "I fry an egg." In more philosophical terms: how
do we explain the existence of a class of sentences the meaning of which is
such that we can use them to perform the act named by the main verb,
simply by uttering the sentence (which literally says that we are performing
the act in question)? How do we explain the existence of a class of verbs
which lends itself to such usages? How do we explain that many other verbs
which look semantically similar, cannot be used in the same way (so that,
e.g., I cannot boast by saying "I boast that ..." or insinuate by saying "I
insinuate that...")? Is there anything in the meaning of the verbs that makes
the difference?
In our own terms: does the phenomenon of performativity have a
conceptual basis? Are the observed facts merely determined by convention?
Or are there any deeper pragmatic (i.e. cognitive and/or social) constraints
involved? Before venturing upon a response to these questions, we shall
first scrutinize Searle's account.
adjourned/' and succeed in my declaration, then I make it the case that what I said
is true; similarly with "I order you to leave the room." But it is important to
emphasize, contrary to the hypothesis that I considered earlier, that the truth of the
statement derives from the declarational character of the utterance and not
conversely. In the case of performative utterances, the assertion is derived from the
declaration and not the declaration from the assertion." (pp. 553-554)
Though 'the truth of the statement' may indeed derive from the
declarational character of the utterance, this does not warrant the step
towards a claim as to a derivation of the assertive nature of performatives
itself. Leaving this unwarranted argumentational step aside, the conclusion
is as counterintuitive as the idea that all performatives would be indirect
speech acts. It is counterintuitive in view of the function of sentence types:
declarative sentence types are usually associated with the force of an
assertion as their basic function; therefore, viewing the assertive nature of
a declarative sentence as derived in the case of performatives stretches the
imagination to an unacceptable extent.
A third problem is associated with the notion of self-reference which is
central to Searle's account. After restating the problem as follows
(repeating the denial of ambiguity),
"How can the literal meaning of an ordinary indicative sentence encode the actual
performance of an action named by the main verb? And how can the literal meaning
both encode the performative and the assertive meaning without being ambiguous?
It is not enough to say that in the one case the speaker intends the utterance as a
performative and in the other as an assertion. The question is: how could one and
the same literal meaning accommodate both intentions?" (p. 551)
he formulates the main elements of his explanation:
"The apparatus necessary for answering these questions includes at least the following
three elements:
Second, we need to recognize the existence of a class of verbs which contain the
notion of intention as part of their meaning. To say that a person performed the
act named by the verb implies that he or she did it intentionally, that if it wasn't
intentional, then the agent didn't do it under that description. Illocutionary
verbs characteristically have this feature. I cannot, e.g., promise unintentionally.
If I didn't intend it as a promise, then it wasn't a promise.
[When the action about which one speaks consists in uttering the corresponding
form, then we are clearly dealing with verba dicendi though of course in the widest
possible sense of the word, i.e. verbs which describe actions that can be performed
by speaking or which may symbolize them.]
Clearly, what needs explaining is the behavior of a specific type of speech
act verbs, in their capacity as speech act verbs and not in their capacity as
intentional verbs (though they all are intentional verbs). This shift of focus
is, incidentally, symptomatic of the overdependence of Searlean speech act
theory on the notion of intention in general (see Rosaldo 1982; Du Bois
1987). The superfluous widening of the focus is further aggravated by the
(not unrelated) introduction of an equally superfluous distinction: The one
between self-referentiality and executiveness. It is hard to imagine what a
completely self-referential utterance could be if not an instance of the type
of act it describes, and therefore at the same time executive in Searle's
sense.
If we confine the discussion to properties of utterances containing
speech act verbs, and if we start from the thesis that a completely self-
310 JEF VERSCHUEREN
whitewash (as opposed to excuse), slander (as opposed to accuse), but also
to lie and to boast or brag, all require an evaluative distance between A and
D. Reason: as was already anticipated in relation to to lie, the social
acceptability of the described acts is low, so that negative value judgments
on the part of Sd are associated with the choice of the LAVs listed. D-
condition: Sd's negative value judgment. This D-condition makes it
impossible for a speaker to describe his or her own act as an instance of
threatening, slandering, lying or boasting, without passing a value judgment
on it and thereby undercutting the act while it is performed. This is the
classic case of what Vendler (1976) called 'illocutionary suicide' in an article
which contained most of the ingredients needed for the type of explanation
I am trying to formulate for the phenomenon of performativity. That this
is exactly what is happening can best be clarified with reference to forms
of usage in which the negative value judgment is cancelled. Consider (38)
and (39):
(38) I boast myself a patriot
(39) I promise I'll kill you
Sentence (38) was taken from a British dictionary6; since patriotism is
usually evaluated positively, the content of the sentence cancels any
negative associations the verb to boast normally has. Therefore, the D-
condition is overruled, and to boast can be used performatively. Similarly
to promise in (39) does not carry the negative connotations which to threaten
is burdened with, so that if to promise is used as a euphemism one can
perfectly threaten performatively.
Finally, to convince, annoy, persuade, amuse even require a temporal
distance between A and D. Reason: the communicative effect of A must
already have taken place before A can be described as having convinced,
annoyed, persuaded, amused someone. D-condition: Sd's assessment of the
communicative effect of A. It goes without saying that A an D cannot
coincide if they must be consecutive.
6
The fact that this sentence was found in a dictionary, does not make it acceptable for
all speakers of English. In fact, a self-reflective use of to boast is no doubt extremely
exceptional - which underscores the general point we have made about this verb.
THE CONCEPTUAL BASIS OF PERFORMATIVITY 315
Figure 1
316 JEF VERSCHUEREN
6. Back to 1903
7
On the contrary, Žagar (1990) argues about Slovenian (the language in which,
according to Škrabec, a perfective form of the performative would be possible) that only
the imperfective can express a real performative, whereas the use of a perfective expresses
only an intention or a readiness to perform an act Thus the imperfective Obljubljam
would mean "I promise", but the perfective Obljubim means "I am about to promise, I am
willing to promise,"
THE CONCEPTUAL BASIS OF PERFORMATIVITY 317
between D-conditions that are 'reflective' and those that are not). This
gradability, in turn, may not be unrelated to the preference for a specific
form of expression. But a much more detailed investigation is called for to
make those issues tangible.8
References
Åqvist, Lennart
1972 . "Performatives and verifiability by the use of language". Filosofiska Studier
14.
Austin, John
1962 How to do things with words. Oxford: Oxford University Press.
Bach, Kent
1975 "Performatives are statements too." Philosophical Studies 28.229-36.
Cornulier, Benoît de
1975 "La notion d'auto-interprétation." Etudes de Linguistique Appliquée 19.52-
82.
1980 Meaning Detachment (Pragmatics & Beyond 1:7). Amsterdam: John
Benjamins.
Davis, Steven
1964 "'I know' as an explicit performative." Theoria 30.157-65.
Du Bois, John
1987 "Meaning without intention: Lessons from divination." IPrA Papers in
Pragmatics 1:2.80-122.
8
It should be stressed that the descriptive model for linguistic action verbs, as handled
in this article, makes use of only a few research angles provided by the more general
theory of pragmatics with reference to which it was claimed that it has to be situated. The
same theoretical perspective also provides the means for further, more detailed,
investigations of actual patterns of usage.
THE CONCEPTUAL BASIS OF PERFORMATIVITY 319
Fauconnier, Gilles
1979 "Comment contrôler la vérité: Remarques illustrées par des assertions
dangereuses et pernicieuses en tout genre." Actes de la Recherche en
Sciences Sociales 25.3-22.
Fillmore, Charles
1971 "Verbs of judging". In CJ. Fillmore & D.T. Langendoen (eds.), Studies in
Linguistic Semantics. New York: Holt, Rinehart & Winston, 272-289.
Ginet, Carl
1979 "Performativity." Linguistics and Philosophy 3:2.245-65.
Grewendorf, Günther
1979 Explicit performatives and statements. Journal of Pragmatics 3:5.431-45.
Grodzirński, Eugeniusz
Performative utterances. Wroclaw: Ossolineum.
Heal, Jane
"Explicit performative utterances and statements." The Philosophical Quarterly
24.106-21.
Hedenius, Ingemar
1963 "Performatives." Theoria 29.115-36.
Hinrichs, Uwe
1985 "Der Koinzidenzfall in den Balkansprachen." Zeitschrift ßr Balkanologie
1985.136-51.
Koschmieder, Erwin
1935 "Zu den Grundfragen der Aspekttheorie." Indogermanische Forschungen
53.280-300.
Lemmon, E.J.
1962 "On sentences verifiable by their use." Analysis 22:4.86-89.
320 JEF VERSCHUEREN
McCawley, James D.
1977 "Remarks on the lexicography of performative verbs." In A. Rogers et al.
(eds.), Proceedings of the Texas conference on performatives, presuppositions
and implicatures. Arlington, Virginia: Center for Applied Linguistics, 13-25.
Récanati, François
1980 "Some remarks on explicit performatives, indirect speech acts, locutionary
meaning and truth-value." In John Searle, Ferenc Kiefer & Manfred
Bierwisch (eds.), Speech act theory and pragmatics. Dordrecht: Reidel, 205-
20.
Rosaldo, Michelle
1982 " "The things we do with words: Ilongot speech acts and speech act theory
in philosophy." Language in Society 11.203-237.
Sadock, Jerrold M.
1974 Toward a linguistic theory of speech acts. New York: Academic Press.
Searle, John R.
Urmson, J.O.
1977 "Performative utterances." Mid-West Studies in Philosophy 2.120-27.
Vendier, Zeno
1976 "Illocutionary suicide." In Alfred F. MacKay & Daniel D. Merrill (eds.),
Issues in the philosophy of language (Proceedings of the 1972 Oberlin
Colloquium in Philosophy). New Haven: Yale University Press, 135-45.
THE CONCEPTUAL BASIS OF PERFORMATIVITY 321
Verschueren, Jef
1980 On speech act verbs. Amsterdam/Philadelphia: John Benjamins.
Warnock. G.J.
1973 . "Some types of performative utterance." In Essays on J.L. Austin, by Sir I.
Berlin et al, 69-89. Oxford: Oxford University Press, 1973.
Žagar, Igor 2 .
1988 "Aspect et performativité en Slovène." Acta Linguistica Hungarica
38:1/4.275-287.
Zaslawsky, Denis
1979 "Pronoms personnels, performatifs et actes de langage." Langue Française
42.48-59.
Zuber, Ryszard
1981 "Marche modali e performativi espliciti." Materiali Filosofici 4:5.111-20.
Index
Ehrich: 265, 275 Foley: 6, 7, 23, 81, 88, 156, 157, 159,164,
Ellipsis: 83-85 166, 179, 187
Emotive: 127 Formality: 236, 237
Empathy: 265 Frame: 279-294
English: 44-46, 73-90, 145-190 — semantics: 25-42, 279-294
Middle - : 44 Body - : 279-294
Erickson: 45, 69 Theta - : 153
Etymology: 26 Francis: 67, 70
Event: 120-123, 128 French: 46-47, 114-115, 145-190
Exclamation: 222, 231 Fried: 145
Excusing: 314 Fuchs: 151, 166, 168, 187
Experiencer: 106, 154, 156, 158, 173 Fujii: 74, 88
Metaphorical — : 140-142 Functions:
Explicitness vs. implicitness: 37 Grammatical — : 16-21
Extent: 102
Gawron: 2, 6, 9, 12, 23, 109
Faber: 151, 166, 187 Gaze: 133
Face: 43-71 Gensler: 219
—saving: 45 German: 265, 267
Give - : 52-53 Ginet: 300, 318
Lose - : 50-52 Givenness: 254, 255, 262, 263, 264
Negative — : 62, 64-66 Givón: 149, 188, 197, 199
No - : 49 Gloss: 26
Positive - : 62, 66-68 Goal: 30, 31, 102, 103, 104, 107, 108,
Restoring — : 67 140, 142
Thin/thick - : 53-54 Goldberg: 7, 23, 92, 103, 109, 145, 167,
Fauconnier: 73, 76, 77, 88, 300, 319 168, 188
Fillmore: 2, 7, 8, 9, 10, 12, 22, 23, 25, 27, Goodman: 78, 88
28, 41, 42, 44, 69, 73, 75, 78, 82, 83, Government binding theory: 91, 167, 168
85, 88, 91, 93, 104, 109, 111, 112, Grammar: 133-143
123, 130, 133, 136, 137, 140, 141, Case - : 247
153, 154, 155, 156, 157, 167, 173, Construction — : 7, 91-109, 167,
187, 191, 198, 199, 201, 216, 218, 180, 252
219, 221, 234, 244, 247, 251, 252, Head-driven phrase structure — : 94
254, 256, 265, 269, 270, 275, 276, Lexical functional — : 7
279, 285, 295, 299, 319 Relational — : 1,. 15
Finnish: 247-275 Role and reference — : 7
Fleischman: 145 Grewendorf: 300, 319
Focus: 145, 147-153, 159, 164 Grodzinski: 300, 319
— articulations: 148 Grootendorst: 299
— types: 146 Gruber: 7, 23, 155, 188
Argument - : 149, 150, 152 Grumbling: 313
Predicate - : 149, 150, 175 Gumperz: 299
Sentence - : 149, 150, 161
Unmarked — position: 163 Haiman: 217, 218
326 ESSAYS IN SEMANTICS AND PRAGMATICS