You are on page 1of 184


TESOL QUARTERLY Volume 26, Number 3 ❑ Autumn 1992

A ]ournal for Teachers of English to Speakers of Other Languages

and of Standard English as a Second Dialect

SANDRA SILBERSTEIN, University of Washington
Associate Editor
SANDRA McKAY, San Francisco State University
Review Editor
HEIDI RIGGENBACH, University of Washington
Brief Reports and Summaries Editor
GAIL WEINSTEIN-SHR, San Francisco State University
Research Issues Editor
GRAHAM CROOKES, University of Hawaii at Manoa
Assistant Editors
DEBORAH GREEN, University of Washington
Editorial Assistant
MAUREEN P. PHILLIPS, University of Washington
Editorial Advisory Board
Roberta G. Abraham Thom Hudson
Iowa State University University of Hawaii at Manoa
]oan G. Carson Claire Kramsch
Georgia State University University of California, Berkeley
Graham Crookes Anne Lazaraton
University of Hawaii at Manoa The Pennsylvania State University
Jim Cummins
Ontario Institute for Studies in Education Michael K. Legutke
Catherine Doughty Goethe Institute, Munich
The University of Sydney David Nunan
Miriam Eisenstein Macquarie University, Sydney
New York University
Yehia E1-Ezabi Teresa Pica
United Arab Emirates University/ University of Pennsylvania
The American University in Cairo N. S. Prabhu
Susan Gass National University of Singapore
Michigan State University Thomas Ricento
Jean Handscombe Central Michigan University
North York Board of Education, Toronto
Thomas Huckin Patricia L. Rounds
University of Utah University of Oregon
Sarah Hudelson Andrew F. Siegel
Arizona State University University of Washington
Additional Readers
Gregory Barnes, Maija Bell, Ruth Benander, Margie S. Berns, Kristine Billmeyer, Andrew D. Cohen,
Joanne Devine, William Grabe, Liz Hamp-Lyons, Donna M. Johnson, Diane Larsen-Freeman, Xiaoli Li,
Patricia A. Porter, Richard Young
Advertising arranged by Maria Minor, TESOL Central Office, Alexandria, Virginia
Typesetting, printing, and binding by Pantagraph Printing, Bloomington, Illinois
Design by Chuck Thayer Advertising, San Francisco, California


To print, select PDF page
nos. in parentheses

The Test at the Gate: Models of Literacy in Reading Assessment 433 (10-38)
Clifford Hill and Kate Parry
Writing One’s Way into Reading 463 (40-62)
Vivian Zamel
Communicative Interaction and Second Language Acquisition:
An Inuit Example 487 (64-82)
Martha Crago
Learning to Teach: Instructional Actions and Decisions
of Preservice ESL Teachers 507 (84-112)
Karen E. Johnson
Avoiding Obstacles to Student Comprehension
of Test Questions 537 (114-133)
Michal Kirschner, Carol Wexler, and Elana Spector-Cohen
L2 Tense and Time Reference 557 (134-149)
Eli Hinkel

Process and Experience in the Language Classroom 573
Michael Legutke and Howard Thomas
Reviewed by David Nunan
Developing Communicative Competence in a
Second Language
Robin C. Scarcella, Elaine S. Anderson, and Stephen D. Krashen
Reviewed by Susan Fiksdal

Discourse Analysis for Language Teachers, Michael McCarthy
(Michael Toolan) 579
Sound Advantage: A Pronunciation Book, Stacy A. Hagen and
Patricia E. Grogan (John Haberstroh)
Language Transfer: Cross-linguistic Influence in Language Learning,
Terence Odlin (Ruizhong Wang)
Understanding Computers: A Text for Developing Critical Reading,
Thinking, and Reasoning Skills, Ellen McGill and Donald DiCristoforo
(Miriam Fletcher)
Volume 26, Number 3 ❑ Autumn 1992


Bias-free Teaching as a Topic in a Course for International
Teaching Assistants 585
Janet G. Graham
Discourse Styles and Patterns of Participation on ESL Interactive Tasks 589
Christine Schuler Alvarado

Comments on Antony John Kunnan’s “DIF in Native Language
and Gender Groups in an ESL Placement Test” 595
The Limits of Biased Item Analysis
Stuart Luppescu
The Author Responds . . .
Antony John Kunnan
Research Issues
Validity and Reliability in Qualitative Research on Second Language Acquisition
and Teaching 602
Two Researchers Comment. . .
Donna M. Johnson and Muriel Saville-Troike
Another Researcher Comments . . .
Kathryn A. Davis

Information for Contributors 609

Editorial Policy
General Information for Authors
Publications Received 613
Publications Available from the TESOL Central Office 617
TESOL Membership Application 619

Editor's Note

This issue of the TESOL Quarterly is the last scheduled to be produced

using traditional typesetting. With the Winter 1992 issue, we will be
effectively moving from paper to disk: At each production stage, we will
be able to handle our copy electronically. We will continue to process
submitted manuscripts on paper, but authors of accepted manuscripts will
be asked, wherever possible, to provide an additional copy on disk.
With this change, the Quarterly will be foregoing the typesetting services
of Pantagraph Printing. During almost 15 years of association with the
journal, the staff at Pantagraph has contributed inestimably to its quality.
On behalf of the TESOL organization, I take this opportunity to thank
them and to acknowledge especially the expert counsel and professional
commitment of plant manager Lars LaBounty. Pantagraph will continue to
print our volumes.

In this Issue

Articles in this issue of the TESOL Quarterly examine language

instruction and assessment as social phenomena. The lead article critiques
approaches to reading assessment, contrasting an autonomous model of
literacy with a perspective that reflects the social dimensions of reading.
The second article argues for integrated reading and writing instruction,
stressing the transactions between readers and texts. The third focuses on
the sociocultural aspects of language acquisition and their implications for
the education of indigenous populations. Other articles address the
interaction between student actions and the instructional behaviors of
preservice ESL teachers; the interaction between test writer and test taker;
and the language/culture-based difficulties of nonnative speakers in
understanding traditional explanations of the English tense system.

• Clifford Hill and Kate Parry contrast the autonomous model of literacy
underlying traditional British and U.S. reading tests with what they call
a pragmatic model. In the former, reader, text, and reading are taken
to be autonomous entities. In the latter, readers and writers negotiate
social meaning in part on the basis of assumed social identities. Hill and
Parry note the limitations of most current reading tests, arguing for an
alternative that “reflects the social dimensions of literacy activities.”
Like Zamel, in the subsequent article, they argue that reading and
writing are inseparable.
• Vivian Zamel explores the interrelatedness of reading and writing. Like
Hill and Parry, she contrasts models of literacy—in this case a
transmission model that stresses the decoding aspects of reading with
a model that emphasizes the critical transactions between reader and
text. Zamel argues that the generative and recursive nature of writing
can help students discover that reading is not static. Rather, “reading
acts upon and gives meaning to” text; reading is “an act of composing.”
Zamel terms this discovery process “writing one’s way into reading.”
This orientation and the teaching activities it engenders help students
to see their own writing from the perspective of potential readers and
to understand reading and writing as inextricable.
• Martha Crago reports a qualitative study of first language socialization
in Inuit families. Her discussion of discourse features highlights the
incongruity between Inuit family practices and the classroom discourse
of non-Inuit second language teachers. Crago provides suggestions for
classroom activities that respect Inuit patterns of communicative
interaction. She discusses her findings in the contexts of second
language acquisition and the education of indigenous populations
• Karen Johnson explores the cognitive dimension of learning to teach,
by examining the instructional actions and decisions of preservice ESL
teachers. She suggests that some cognitive demands may be similar
across many fields, whereas others may be unique to second language
teaching. Johnson analyzes the ways in which teachers perceived and
reacted to student input. Of primary concern to these teachers were
student comprehension, student motivation, and the instructors’ desires
to maintain control over instructional management. The last of these
was served even at the cost of ignoring potentially fruitful but
unexpected student-initiated discourse. Johnson argues that teacher
preparation programs should provide avenues for apprentice teachers
to become conscious of their perceptions and responses while teaching.
• Michal Kirschner, Carol Wexler, and Elana Spector-Cohen view
reading comprehension tests as communicative interchanges between
test takers and test writers. They present preliminary test-writing
criteria designed to reflect student comprehension problems observed
in their classroom and test-writing experience. The criteria aim to


render test questions comprehensible in terms of the reader/test taker’s
pragmatic and linguistic knowledge. Application of these criteria is
facilitated by a practical checklist for evaluating and writing individual
reading-test items.
• Eli Hinkel’s study explores a source of difficulty in learning English
tenses. She finds differences between native and nonnative speakers in
their understandings of the terms typically used in ESL grammar texts
to describe English tenses. Her results suggest that native and
nonnative speakers “view time spans and their divisions and
measurements differently.” Suggestions are made for pedagogical
strategies that address this situation.
Also in this issue:
Reviews: David Nunan reviews Michael Legutke and Howard
Thomas’s Process and Experience in the Language Classroom; and
Susan Fiksdal reviews Robin C. Scarcella, Elaine S. Anderson, and
Stephen D. Krashen’s Developing Communicative Competence in a
Second Language.
Book Notices
Brief Reports and Summaries: Janet Graham describes a segment of an
ITA training course devoted to avoiding discrimination in the
classroom; Christine Alvarado reports a study of students’ discourse
styles and patterns of participation in interactive tasks.
The Forum: Stuart Luppescu’s commentary on Antony Kunnan’s
TESOL Quarterly brief report, “DIF in Native Language and Gender
Groups in an ESL Placement Test” is followed by a response by the
author. In the subsection Research Issues, Donna Johnson and Muriel
Saville-Troike, and Kathryn Davis comment on validity and reliability
in qualitative research.
Sandra Silberstein


TESOL QUARTERLY, Vol. 26, No. 3, Autumn 1992

The Test at the Gate: Models of Literacy

in Reading Assessment
Teachers College, Columbia University
Hunter College, The City University of New York

For many students throughout the world, success depends not

only on the ability to read in English but on the demonstration of
that ability through publicly recognized examinations. These
exams are based on two separate traditions, one originating in
Britain, the other in the United States; they vary considerably in
form, yet there is a fundamental similarity in the demands they
make and in the kind of reading they effectively teach. We
describe this similarity with respect to what Street (1984) calls the
autonomous model of literacy, one which assumes that text,
reader, and the skill of reading itself can be viewed as autonomous
entities. In criticizing this view, we present an alternative
approach, described as a pragmatic model of literacy, which treats
text as social instrument, readers and writers as embodying social
identities, and the skill of reading as extending beyond decoding
and use of relevant knowledge to socially structured negotiation of
meaning. We utilize these contrasting models of literacy to de-
scribe the limitations of reading tests and argue for an alternative
approach to assessment that reflects the social dimension of liter-
acy activities. Within such pragmatically oriented assessment,
reading and writing are viewed as inseparable.

For learners of English around the world, the acquisition of

literacy skills, despite the rapid spread of oral media, remains
paramount. Within countries that were once British colonies, an
individual’s primary experience of written language is often in
English. Even in countries with substantial bodies of written
material in an indigenous language, individuals still seek at least a
reading knowledge of English, for it is viewed as a powerful means
of acquiring the specialized knowledge on which successful partic-
ipation in the modern world is based.

Written English has become even more important by virtue of its
dominant role in formal assessment. The standardization associated
with assessment is not easily achieved with oral language; even
when speech is audiotaped to insure uniformity, the quality of
replay may vary considerably. The concern for standardization has
also led to focus on strictly controlled responses to text rather than
on actual production of writing. Reading comprehension tests are
widely used to measure an individual’s control over written English,
and they have become important gatekeepers, controlling move-
ment from lower to higher levels of education in countries as differ-
ent as Nigeria and Japan. As these tests have been developed
around the world, they generally reflect one of two major ap-
proaches to testing an older approach developed in Britain that
allows for limited use of written responses or a more recent one de-
veloped in the United States that uses a multiple-choice format.
In Britain itself, testing has been considerably modified over the
last 10-15 years, but in countries that used to be British colonies the
older tradition survives, virtually unchallenged, in school certificate
exams taken at the end of secondary school. The material in Figure 1
is taken from an English language exam for the West African
Comprehension Passage and Tasks in the British Tradition
From the School Certificate and G. C. E. (English language 1, Question 8), June 1982, Lagos,
Nigeria: West African Examinations Council. Copyright 1982 by the West African Examinations

Read the following passage carefully and answer the questions on it.
Orchids are flowers that are highly regarded and very expensive in some parts of the
world, whereas they grow wild and are barely noticed in other places. They are found in their
greatest variety and abundance in tropical climates.
There are two main categories of orchid: ground orchids, which are the common type in
temperate regions; and epiphytic orchids, which grow on trees and are usually found in the
tropics. Although these epiphytes grow on trees they do not derive nourishment from them.
The plants often have large solid swellings of the stem in which water and nutritive materials
are stored. They derive this moisture from the air by aerial roots.
More than 3000 species are in cultivation but there are many thousands of hybrids which
have been developed by cross-fertilization by horticulturalists all over the world. The orchid
has developed a prestige of its own, which is not only due to its fragile beauty, but also to its
scarcity value, and to the fact that its cultivation is somewhat specialized.
(a) Explain the word hybrid in one sentence.
(b) Give an alternative word that would fit in the place of derive.
(c) Why are orchids expensive in some parts of the world?
(d) Explain the grammatical function of the clause ‘Although these epiphytes grow on
(e) What word in the passage tells you that the roots of epiphytic orchids do not reach
the ground?


School Certificate (WASC), and it is quite typical of reading com-
prehension tests throughout the British Commonwealth.
In the United States, testing is also undergoing change, but the
traditional approach remains dominant throughout the country and
in many other parts of the world that have been influenced by U.S.
educational practices. This approach is exemplified by the material
in Figure 2, which is taken from The University of Michigan’s
Examination for the Certificate of Proficiency in English, a test used
with nonnative speakers throughout the world. We would like to
observe that the samples of test material are both well crafted
within their respective traditions.
At first sight, it is the difference between the two traditions that
is most striking. The U.S.-style test uses multiple-choice tasks; the
British-style test, open-ended tasks which vary more in the demands
they make. In effect, the British-style test requires test takers to
draw on productive skills (they must compose answers) as well as
metalinguistic knowledge (they must label linguistic units). On the
other hand, they do not have to work on eliminating alternative
interpretations suggested by distracters, as those taking the U. S.-
style test must do.
It is not surprising that those trained in one tradition initially have
difficulties when confronted by a test from the other tradition. For
example, well-educated Jamaicans, when entering a university in
the United States, are sometimes misplaced in a remedial English
course on the basis of exam performance. They complain that they
are unable to decide among all the choices provided on a multiple-
choice task. We have observed, however, that individuals who do
reasonably well on one style of test ordinarily adapt rather quickly
to the other. The reason, we suggest, is that beyond differences of
format are basic similarities, both in the kind of text used and in
what test takers are required to do with it.
If we consider the two sample passages apart from the tasks, we
cannot readily tell which one is from the U.S. tradition and which
from the British. While they deal with different subject matter—one
with botany, the other with medical technology—the way that sub-
ject matter is handled is strikingly similar. Both passages present in-
formation in a deliberately neutral way. While evaluative responses
are described, there is no indication that they are shared by the writ-
er. Nor is there any orienting material that explains matters such as
where the passage came from, who wrote it, or why it was written.
When such material is present within a text used on a test, test
makers tend to excise it or simply ignore it (Hill & Parry, 1988).
Yet another similarity can be seen in the great amount of detail in
both passages. Each begins with a general definition and then


Comprehension Passage and Tasks in the U.S. Tradition
From the Examination for the Certificate of Proficiency in English (Preliminary Test 2), 1990-
1991, Ann Arbor, MI: Testing and Certification Division, English language Institute, The
University of Michigan. Copyright 1990 by the Testing and Certification Division, English
Language Institute, The University of Michigan.

Until recently only glasses or contact lenses were available for the correction of myopia, a
vision defect in which objects can be seen distinctly only when very close to the eyes. In 1978,
a surgical technique known as radial keratotomy, also termed radial K, was touted as a
possible “cure” for nearsightedness. Since its introduction, however, radial K has been the
subject of controversy and intense scrutiny.
In this procedure a diamond blade is used to make 8 to 16 cuts that radiate out from the
center of the cornea of the eye like the spokes of a wheel. The resulting flattened curvature
of the cornea decreases myopia. While close to 50% of patients achieve 20/40 vision on
standard eye charts without glasses, the later side effects aren’t fully known. The surgery
itself involves certain risks, such as corneal perforations, which occur in about one in 600 eyes
and which require stitches. Errant incisions can cause blindness and the need for corneal
transplants; even tiny holes invite infection. Radial K patients’ most common disappointment
is that their vision isn’t perfect. Some vision is overcorrected, some is undercorrected and
sometimes the outcome is one eye each way.
The risks of radial K are real and considerable but at the same time many patients are
satisfied with the results. Some experts predict that as more operations are performed
surgeons will improve their technique and complications will be reduced; furthermore, they
say that future radial K will be automated using a computer attached to a robotic arm with
the diamond blade being replaced by a laser. Other experts counter that improving technique
won’t help much because the concept itself of cutting the cornea is a bad one, fraught with
inherent problems. Though it may become more widespread and routine in the future, at
present radial K is not the method of choice for correcting nearsightedness.
26. The term radial keratotomy refers to . . .
a. a new type of contact lens.
b. a type of corrective surgery.
c. a diamond blade used in eye surgery.
d. a technique used in corneal transplants.
27. The passage states that half the radial K patients . . .
a. have partially corrected vision.
b. have perfect vision.
c. lose vision in one eye.
d. require corneal transplants.
28. According to the passage, some experts predict that in the future radial K will . . .
a. not require the cutting of the cornea.
b. produce perfect vision.
c. use computer and laser technology.
d. become obsolete.
29. The chief disappointment of radial K patients is that . . .
a. their vision isn’t the same in both eyes.
b. they must undergo surgery.
c. their vision isn’t overcorrected.
d. they do not achieve perfect vision.
30. The author’s attitude towards the use of radial K in the future is . . .
a. very worried.
b. enthusiastic.
c. suspicious.
d. neutral.


moves rapidly into a mass of detail. This kind of text is no accident.
It has been expressly chosen, adapted, or constructed because it
provides test makers with material around which they can build
what they conceive to be valid tasks. Central to their notion of
validity is that tasks be built around odd bits of fact that readers do
not ordinarily carry around in their heads. Consider, for example,
the tasks that follow the Radial K passage. They require readers to
recycle details such as the meaning of radial keratotomy and an
anticipated use of this surgical technique. Following the Orchids
passage, four of the five tasks call for a similar recycling—the
definition of hybrid, the reason why orchids are expensive, the
nature of orchids’ roots. If there is a difference in the kind of
processing required, it is in the more obviously metalinguistic tasks
in the British-style test; in Task d, for example, candidates are
required to “explain the grammatical function” of a clause. But in
either test the essential approach to the passage is the same. In
working through the tasks, test takers must first determine what
information is called for, locate it in the passage, and then decide on
an appropriate response by comparing the exact wording of the
passage and the task.
Moreover, readers who deal consecutively with the tasks on
either test cannot move through the text in sequence. In Radial K
for example, the third task moves them into the final paragraph, but
the fourth one returns them to the second paragraph. In Orchids,
readers are forced to move around even more. Readers must begin
with paragraph 3 and then go back to paragraph 1. For the third
task, they return to paragraph 3; for the fourth and fifth, they go
back to paragraph 2. Such movement forces readers to operate at a
local level rather than a global one.
We would claim that these similarities between the U.S. and
British approaches to reading assessment are not accidental. Rather,
they reflect a common set of assumptions about what text is and
how individuals should read it. These assumptions are so deeply
embedded in the Western tradition of formal education that they
have long remained unexamined. In recent debate about the nature
of literacy, however, they have been made explicit. We here present
this debate by drawing on various fields—history, anthropology,
psychology, linguistics—and showing how dominant practices of
testing are related to a certain conception of literacy. We then
present a contrasting model of literacy that can serve as a basis for
the emerging practices of alternative assessment.


The phrase autonomous model of literacy has been used by Brian
Street (1984) to describe a set of assumptions that, he claims,
underlie Western educational practice. He identifies these assump-
tions by critically examining the work of various researchers: Goody
(1968, 1977), Goody & Watt (1968), Greenfield (1972), Hildyard &
Olson (1978), Lyons (1981), and Olson (1977). Street shows how
these researchers associate literacy with such qualities as logic,
objectivity, and rationality—qualities that are highly valued in
Western societies and that are explicitly promoted in formal educa-
tion. He claims that these associations add up to a model of literacy
which, despite its attractions, is dangerously misleading. To support
this claim, he draws on the work of anthropologists, historians, and
linguists and uses this evidence together with his own fieldwork to
construct an alternative model of literacy that he characterizes as
Street’s approach has been controversial, for he groups research-
ers who do not necessarily regard themselves as associated; and in
discussing their work, he represents them as adopting positions that
they are not always willing to acknowledge. He offers, however, a
convincing justification for this approach:
The writers I am discussing do not necessarily couch their argument in
the terms that I am adopting. But, nevertheless, I maintain that the use
of the term “model” to describe their perspective is helpful since it
draws attention to the underlying coherence and relationship of ideas
which, on the surface, might appear unconnected and haphazard. No
one practitioner necessarily adopts all of the characteristics of any one
model, but the use of the concept helps us to see what is entailed by
adopting particular positions, to fill in gaps left by untheorised
statements about literacy, and to adopt a broader perspective than is
apparent in any one writer. (Street, 1984, p. 3)
We have found the autonomous model to be a valuable heuristic
in our consideration of reading tests. This is an area that is
particularly rich in “untheorised statements about literacy”; test
makers, for example, often provide detailed reports of their
findings, but rarely, if ever, spell out their assumptions about what
they take reading to be. Those assumptions can be elucidated if we
consider how the notion of autonomy is applied in characterizing
text, literate individuals and institutions, and the skill of literacy
itself. In doing so, we will extend the discussion to material that
Street does not consider.


Autonomy of Text
As Street points out, the writers he discusses do not themselves
use the phrase autonomous model of literacy. The word autono-
mous, however, does appear frequently, usually with reference to
writing. Goody and Watt (1968), for example, maintain that writing
is distinctive because it is, at least potentially, “an autonomous mode
of communication” (p. 40); and Ong (1982) explicates this idea
more fully:
By isolating thought on a written surface, detached from any
interlocutor, making utterance in this sense autonomous and indifferent
to attack, writing presents utterance and thought as uninvolved in all
else, somehow self-contained, complete. (p. 132)
But it is Olson who gives clearest expression to the idea that writing
is autonomous. This position is set up as fundamental in what is
probably his best known article, “From Utterance to Text: The Bias
in Language in Speech and Writing”:
My argument will be that there is a transition from utterance to text both
culturally and developmentally and that this transition can be described
as one of increasing explicitness, with language increasingly able to
stand as an unambiguous and autonomous representation of meaning.
(Olson, 1977, p. 258)
Olson does not assert that this quality of autonomy is characteristic
of all texts, but only that it can be seen most clearly in expository
prose of the British essayist tradition. He does suggest, however,
that it is the ideal for texts used in schooling:
Ideally the printed reader [i.e., a book used to teach reading] depends
on no cues other than linguistic cues; it represents no intentions other
than those represented in the text; it is addressed to no one in particular;
its author is essentially anonymous; and its meaning is precisely that
represented by the sentence meaning. (Olson, 1977, p. 276)
This view of text is clearly operative in the construction of
reading tests. Each passage in such tests, like the printed reader that
Olson describes, “depends on no cues other than linguistic cues,” as
the passages discussed earlier exemplify. Similarly, a passage in a
reading test “represents no intentions other than those represented
in the text,” or, at least, test takers are required to view it in this way.
When tasks call for a statement of “the author’s purpose” (a
common formulation, especially in a U.S.-style test), the target
response is either a recycled version of an explicit statement of
intent or a summary of the passage’s content. Again, the test
passage, like Olson’s printed reader, “is addressed to no one in
particular” and “the author is essentially anonymous.” Both Orchids


and Radial K are cases in point, and even where a passage, as in the
case of recent tests, takes the form of a letter, the accompanying
tasks are so constructed that test takers who treat it as real commu-
nication between people with defined social identities often select a
distracter (Hill & Parry, 1989). Finally, the meaning of each passage
is presumed to be, in Olson’s words, “precisely that represented by
the sentence meaning”; students must attend carefully to what
Olson and Hildyard (1980) have characterized as “the very words,”
and even where figures of speech are used, their expected
interpretation is not merely conventional but extraordinarily literal.
Taken as a whole, Olson’s description of “autonomous text”
corresponds closely to text as it is presented in reading tests.

Autonomy of Institutions and Individuals

That text is ideally autonomous is the basic premise of this model
of literacy, but we have found the word autonomous used in other
ways as well. Goody (1986), for example, claims that the autonomy
of text has much to do with the creation of autonomous institutions
within a society such as the church in medieval Europe, and he
suggests that literacy can confer autonomy on individuals as well. In
discussing the relationship between literacy and economic develop-
ment, he remarks:
If we take recent moves to expand the economies of countries of the
Third World, a certain rate of literacy is often seen as necessary to
radical change, partly from the limited standpoint of being able to read
the instructions on the seed packet, partly because of the increased
autonomy (even with regard to the seed packet) of the autodidact.
(p. 46)
At a more advanced level, Goody suggests that writing enables
individuals to integrate and develop their thinking far beyond what
would be possible in speech. His earlier book, The Domestication
of the Savage Mind (1977), is devoted almost entirely to considering
how the mind of what he calls homo legens [“reading person”] is
“domesticated” by the ability, through writing, to rearrange words
in a physical space and thus to see new ways of connecting ideas.
Ong (1982) makes even stronger claims about the radical changes
in an individual brought about by literacy:
Primary orality fosters personality structures that in certain ways are
more communal and externalized, and less introspective than those
common among literates. Oral communication unites people in groups.
Writing and reading are solitary activities that throw the psyche back on
itself. (p. 69)


Thus both Goody and Ong present homo legens as a species
whose members, through the solitary activity of reading, not only
gain access to more information, but are enabled to analyze their
own assumptions, develop new insights, and perceive themselves as
separate from the social groups to which they belong.
The assumption that literacy confers autonomy on institutions
and individuals is fundamental to traditional uses of reading tests.
As far as institutions are concerned, the major producer of tests
within the United States—the Educational Testing Service (ETS)—
insists on its status as a private, nongovernmental organization.
Outside the United States, examining bodies are often set up by
governmental legislation, but the whole point is that they should
operate, in theory at least, as independent institutions, unaffected
by the vagaries of political life. This image of such bodies as self-
contained and autonomous, is, in fact, essential to the testing
enterprise. The purpose of using an exam rather than, say,
interviews or personal recommendations is that it can be viewed as
an independent arbiter of knowledge or ability, one that is
unaffected by social interests or political preferences. The idea that
test-making institutions are not “homologous with other aspects of
the social system . . . [or] ordered . . . by the infrastructure of the
political economy” (Goody, 1986, p. 20) is crucial to their authority.
At the individual level, too, the autonomy of the reader is a basic
premise of testing. However many candidates are in an exam room,
each is considered to be alone with the text. This isolation is
reinforced by the exam procedures: Desks must be a certain dis-
tance apart, separate materials are handed to each student, no talk-
ing is allowed, and questions addressed to the proctor are to be only
about practical and administrative matters. The candidates work
alone, writing their responses on answer sheets that are identified by
their own individual numbers. When the allotted time expires, all
must stop writing; the papers are handed in; and only then is any
social exchange permitted. In short, readers, when taking a test,
must adopt the solitary habits of homo legens, for to be social in this
context is to cheat.

Autonomy of Skill
That institutions and individuals become autonomous through
literacy implies another aspect of the autonomous model, one that
we have characterized elsewhere as “autonomy of skill” (Hill &
Parry, 1989). This idea is that literacy—the ability to read and
write—can be considered separately from other factors as a
necessary, if not sufficient, cause for certain kinds of social and


cognitive development. It is characterized as a technology (Goody,
1968; Olson, 1977), the use of which may be restricted because of
limitations in other technologies (such as the manufacture of paper)
or because of particular social attitudes, but which is in itself
socially neutral. Goody (1968) demonstrates this point by suggesting
an analogy between writing and the wheel. The relationship
between literacy and the development of philosophical and
scientific thinking parallels the relationship between understanding
rotary motion and the development of wheeled transport. The
potential of the technology in question may not always be realized.
In Tibet, for example, both texts and wheels exist, but they are both
so closely associated with religion (the wheels are prayer wheels)
that the use of either for secular purposes is inhibited (Ekvall, 1964,
cited by Goody, 1968). Thus Goody concedes that social factors are
important in determining the uses to which literacy may be put, but
he does not consider reading and writing as social processes in
The presumed autonomy of the skill of literacy is closely linked to
the presumed autonomy of text. If text is considered as object rather
than action, it can then be understood as the sum of its elements
(“the very words”) rather than as a means of human communica-
tion. Reading is therefore a technical rather than a social skill—a
difficult one, however, for it involves learning not simply the
writing system, but a new way of approaching language. Olson
(1977), for example, points out that children come to school already
competent in speech, a form of language that
relies for its comprehension on a wide range of information beyond that
explicitly marked . . . [including] a shared knowledge of events and a
preferred way of interpreting them; a shared perceptual context; and
agreed-upon prosodic features and paralinguistic conventions. (p. 272)
Children must then learn to interpret written language in which this
“wide range of information” is not available, but in which there is,
by way of compensation, “an enlarged set of explicit linguistic
conventions” (p. 272). Someone who is learning to read has to
master this enlarged set and also develop the ability to discount
certain elements used in oral comprehension such as “a shared
perceptual context. ” Readers, unlike listeners, must focus
exclusively on “the sentence meaning, the meaning in the text”
(p. 273) and apply interpretive procedures to the text alone. The
ability to do this is acquired through experience with text, so
“children familiar with the use of textlike language through hearing
printed stories” (p. 276) have an obvious advantage; but once the
ability has been acquired, it should not, in principle, be affected by


the social identity of the reader, the origin of the text, or the
purposes for which the text is being read.
The belief that literacy skills can be isolated from the personal
and social characteristics of readers is basic to reading tests. The
purpose of such tests is to assess reading as an autonomous skill.
Other qualities, such as individuals’ factual knowledge, their ability
to use it in obtaining new information from text, or their capacity
for remembering and using such information for purposes other
than immediate interpretation are all considered irrelevant. Thus
test makers deliberately use a wide range of passages that
candidates are unlikely to know much about, and they try to
construct tasks that cannot be answered without reference to the
text, even by someone well informed about its content. Indeed, the
use of specialized knowledge can be a disadvantage, for it often
leads to the choice of a distracter in a multiple-choice task or the
inclusion of too much information in an open-ended one. Moreover,
test makers make no attempt to assess readers’ ability to remember
information from a text and apply it in other situations. The skill
that they test, and thereby propagate, is a circumscribed kind of
reading. It consists solely of perceiving “the very words” of the text
and drawing logical inferences from the propositions that they
A corollary to the assumption that a reading test assesses only
reading ability—and not any other knowledge or skill—is the
assumption that the results are representative of how well
individuals read in other contexts. The test takes a sample of each
person’s reading and the resulting score is used as a measure of a
more general ability. In the psychometric tradition, the scores are
not used directly; they are correlated with those of a norming
population and then commonly expressed in terms of a construct
such as grade level. These levels, once established, are widely used
as a basis not only for pedagogy but research as well. Many empir-
ical studies, for example, use test scores to place individuals into
groups defined by grade level; and once the groups are established,
individuals are then identified by terms such as high-skill and low-
skill, or even good and poor readers. Such studies generally focus on
what attributes of the individuals might account for the difference
in their levels. They do not question what the basic division of the
population means. If researchers subscribe to such reification of
grade levels, it is hardly surprising that teachers and, even more,
students accept grade levels as authoritative indicators of where
individuals stand with respect to an essential and highly valued
skill—even where, as in the case of adult ESL students, the implied
comparison with children seems quite inappropriate.


To sum up, the autonomous model of literacy suggests not only
that text is autonomous, but that it confers autonomy on both
institutions and individuals. It also treats reading as an autonomous
skill that is independent of other factors and transferable across all
kinds of texts. We find such assumptions consistently reflected in
the practice of reading assessment; yet, like Street, we find many
objections to the model not only from our own experience as
teachers, but from recent theoretical work in various disciplines, on
both literacy in general and reading in particular. In the following
section, we summarize Street’s position and then outline our own,
adding to his anthropological and historical material evidence from
the field of linguistics, particularly as it deals with social and
psychological processes.


Before presenting what we call the pragmatic model of literacy,
we will summarize Street’s own ideological model in relation to the
three aspects of the autonomous model that we distinguished
above. First, Street denies the autonomy of text; he cites both his
own work and that of J. Parry (1982) in India to show, on the one
hand, how text often depends on oral mediation, and, on the other,
how variable its interpretation may be. Second, Street asserts that
literacy is far from being a universal means to autonomy for its
practitioners, whether in the social or cognitive sense. He concedes
that particular literacy practices may bring economic advantages—
as in the use of Arabic script for commercial purposes in Iran—but
he also cites historical research (Clanchy, 1979; Graff, 1979) that
counters the commonly held notion that literacy skills necessarily
bring social and economic benefits; he further points out that if
being able to read and write has any cognitive effects, it is only in
specific operations and as a function of particular practices in
particular scripts (Cole & Scribner, 1981). Third, Street insists that
literacy cannot be considered a neutral or independent technology.
He cites Heath’s work with different communities in South Carolina
(1983) to show the extent to which literacy is embedded in ideology
and social practice. He concludes that it is such embedding that
determines the nature and therefore the consequences of literacy.
As linguists rather than anthropologists, our perspective differs
somewhat from Street’s. Whereas he is interested in literacy in
general and how it is used in different societies, we are more
concerned with particular texts—those used on reading tests—and
how they are interpreted by particular readers of various language


backgrounds. This perspective leads us to conclusions that in some
respects differ from his, and so we distinguish our own model of
literacy by describing it as pragmatic rather than ideological. The
choice of word is deliberate. Pragmatics is that area of linguistics
concerned with “the use of language in communication” (Richards,
Platt, & Weber, 1985, p. 225). Hitherto it has largely dealt with oral
interaction, but obviously our concern is with how communication
takes place through text.

Text as Communication
To begin, then, we would like to establish our own view of text.
We have shown that those who subscribe to the autonomous model
tend to view text as object, emphasizing certain physical properties
of writing: (a) given its means of production (e. g., making marks on
paper), it can be easily separated from its producer; (b) given ap-
propriate technology (e.g., printing on a durable substance), it can
be relatively permanent; and (c) given its visual form (as opposed to
auditory), it can be easily retrieved. In effect, as language becomes,
in a purely physical sense, more separable, durable, and retrievable,
it acquires a greater independence from the context in which it is
produced. To paraphrase Olson (1977), the meaning comes to
reside in the text itself.
From our vantage point, however, this focus on physical
properties has obscured the essentially social character of text, a
point that becomes apparent when we consider the various
activities involved in its production, storage, and retrieval. First, in
producing text, even the most solitary writer, by the very fact of
using words, draws on communal property, the system of commu-
nication that de Saussure has described as langue. This system, as
de Saussure points out, is inherently social:
It [langue] is the social side of speech, outside the individual who can
never create nor modify it by himself; it exists only by virtue of a sort of
contract signed by the members of a community. (1966, p. 14)
Thus, in selecting words, writers do not simply refer to the world
they are writing about; they associate themselves with particular
communities of language users. Our own selection in writing this
particular text is a case in point. We had to decide, for example,
whether to use ideological or pragmatic in describing an alternative
model of literacy and are well aware that the former links us with
anthropology, the latter with linguistics; and, of course, we are
locating ourselves in relation to current debates about literacy by
using expressions such as autonomous model. The very act of


producing text involves continuous evocation of other text makers,
and so text itself can best be understood as social gesture.
Second, the act of preserving text necessarily involves social
judgment. Right from the beginning, when individual writers save
what they have written instead of tossing it in the wastepaper
basket, they are deciding that it has some worth. They are
envisioning it as a vehicle of communication, even if its only readers
are to be the selves that they eventually become. When text is not
simply preserved but reproduced, this ascription of communicative
potential—and hence social value—is even more apparent. Within
the manuscript tradition of medieval Europe, a text would be
copied only when it was viewed as having intrinsic worth for future
generations; in the case of present-day publication, a book is printed
only when publishers anticipate sufficient response to justify their
effort and expenditure.
Third, when a text is retrieved, it can only be understood in a
social context, one that necessarily differs from the one in which it
originated. It still draws on the social property, language, but then
that property will have changed, however slightly, as will the world,
especially the social world, to which it refers (in fact, change within
language and the world is mutually reinforcing). Further, the
understanding of any text is affected by the literacy practices within
the particular community to which the reader belongs. Clanchy
(1979) shows how these practices change over time; Besnier (1986),
Conklin (1949), J. Parry (1982), and Street (1984) document widely
differing uses of writing in cultural settings around the world; and
Heath (1983) shows how three communities living within the same
geographical area and drawing on the same language vary in their
attitudes toward text. Finally, a particular reader approaching a
particular text must have some reason for doing so, even though it
may remain unconscious and unarticulated; and this, too, affects
how the text is understood. We have all experienced how much a
text can change as our approach to it varies: Jane Eyre when read on
holiday is hardly the same as when studied in school.
What is the significance of all this in relation to reading tests?
From the vantage point of production, a reading test is particularly
complex because of its declared intention of representing many
different kinds of text. There are thus several passages, each of
which is embedded in a textual world of its own. Test makers often
extract these passages from larger texts, editing them in minor ways
to ensure that they function independently. But as Hill and Larsen
(1983) illustrate in their analysis of a passage taken from Lewis
Carroll’s Through the Looking Glass, these efforts are seldom
successful. The skilled writer has woven text so carefully that any


effort to take out a piece inevitably results in a number of loose
threads—and it is just these threads that the reader may use to
weave a larger story. Other passages, intended to represent
particular genres such as children’s stories, are written specifically
for test purposes. In these instances, however, the role of the writer
is not what it seems to be, for close examination of the passages
often reveals a discrepancy between the ostensible intention—to tell
a story or convey information—and the actual one—to present a
text that will discriminate between readers.
A further level of complexity derives from the new context that is
constructed for the passages. Each passage is necessarily accompa-
nied by tasks whose presence often has a distorting effect. This ef-
fect is particularly evident in the case of specifically written pas-
sages in which certain peculiarities of text structure seem to result
from the very process of constructing tasks. Such distortion can also
be present in the case of excerpted text where a particular word or
phrase may be altered in the light of the tasks to be constructed.
Moreover, various passages, each filtered through its own set of
tasks, are then grouped together to form a larger whole; and with-
in this whole the various passage-task configurations inevitably—
though they are not supposed to—influence each other. Beyond the
individual test, there is a further context—the larger world of
reading tests itself comprising a textual tradition closely associated
with formal schooling. Within this tradition, text is viewed as
autonomous, and so a passage, no matter how brief it may be, must
be approached as if it were an independent vehicle of meaning.
The manner in which reading tests are preserved and reproduced
has a particularly marked effect on their status. In order to maintain
face validity, the physical form of a particular test varies little. The
same typeface is used from year to year and the layout of headers,
passages, and tasks is relatively stable. Moreover, the printing of
tests is managed in great secrecy, and the distribution of the finished
product is carefully controlled. In many countries, the tests are sent
out in sealed envelopes to be opened only in the presence of the
candidates; at times, tests are even kept at the police station until the
day of the exam.
Thus, when test takers come to retrieve a text included in a test,
they are doing so at several removes from the original. Moreover,
the context in which they approach it can be quite powerful. In
many instances, students’ future careers depend on their responses,
and we have noted how the solemnity of the occasion is reinforced
by exam procedures. Indeed, it is our claim that reading tests, as
commonly used around the world, demonstrate not so much that


text is autonomous as that its significance, like that of any other
social gesture, depends largely on context.

Communication Between Readers and Writers

The autonomous model of literacy does not simply emphasize the
isolation of text but the isolation of those who use it. Goody (1986)
claims that various institutions develop a separate identity by virtue
of writing, and he, along with others, suggests that individuals do so
as well. The claims for institutional autonomy have been cogently
addressed by Street, so we will restrict ourselves to the claims made
for individuals. In examining these claims, we can begin with a
familiar picture of the solitary writer:
Writing is a solipsistic operation. I am writing a book which I hope will
be read by hundreds of thousands of people, so I must be isolated from
everyone. While writing the present book, I have left word that I am
“out” for hours and days—so that no one, including persons who will
presumably read the book, can interrupt my solitude. (Ong, 1982,
p. 101)
As a writer, however, Ong is hardly alone, for other people are
intruding on his thought all the time. To begin with, the room in
which he works is undoubtedly lined with books, and the book on
which he is working is full of references to them. In a real sense, he
is keeping company with the many people whose work he cites
(and, for that matter, many others whose talk has influenced his
thinking). Moreover, Ong is keenly aware of his readers, as
indicated by his own reference to “hundreds of thousands of
people.” Of course they are not physically present, and so the writer
must imagine them. As Ong (1982) himself puts it,
the writer must set up a role in which absent and often unknown readers
can cast themselves. Even in writing to a close friend I have to
fictionalize a mood for him, to which he is expected to conform. (p. 102)
From our perspective, the fact that the writer is addressing peo-
ple who are not in a position to interject words or gestures does not
make writing any less social. The audience may be invented but it
still plays a powerful role in determining what words the writer
actually puts on the page; and it is, of course, these words that allow
us to infer what kind of audience the writer is constructing. The
same can be said of oral interaction. Speakers, too, must socially
construct their listeners, even though they may be physically pres-
ent; and by virtue of what speakers say, we are able to ascertain
what kinds of social identities they ascribe to those they are talking


These identities are often at odds with externally observable ones;
but then listeners—and, of course, readers, too—are able to reject
an ascribed identity. As readers, we are all familiar with refusing to
play the role that the writer has set up for us, much as in
conversation we let others know that we do not agree with what
they have said. Indeed, the interactive dimensions of talk are much
more present in reading than is commonly recognized (Bakhtin,
1981). The fact that writers cannot immediately talk back does not
prevent our challenging what they have put on the page. We may be
even more prone to challenge their words simply because they have
been granted the power to control the discourse. It is much like the
resistance we feel when someone monopolizes conversation.
In characterizing this fundamental reciprocity of oral and written
communication, we draw on the model that Widdowson (1978)
presents in Teaching Language as Communication. In this work he
rejects the traditional dichotomies between, on the one hand,
speaking and listening and, on the other, reading and writing. As he
points out, such dichotomies do not take account of the
interdependence of productive and receptive activities. With
regard to oral communication he writes:
What is said . . . is dependent on an understanding of what else has been
said in the interaction. . . . Speaking . . . is part of a reciprocal exchange
in which both reception and production play a part. (pp. 58-59)
Widdowson’s model of oral communication is displayed in Figure 3:

Widdowson’s Model of Oral Communication

As can be seen, he introduces talking as a generic term for this

reciprocal exchange, distinguishing it from its productive and
receptive aspects, saying and listening, and from their external
manifestations, speaking and hearing. He uses the term interpreting
in parentheses at the top of the display to indicate “a psychological
process, which, unlike talking, is not realized as actual social
activity” (p. 64). This notion of reciprocal exchange is an important
corrective to models of communication based on discrete skills, for
it is clear that to engage in talk is to listen as well as to speak.


Widdowson sets up a parallel diagram for written communica-
tion, which is shown in Figure 4:

Widdowson's Model of Written Communication

As can be seen, he uses the terms composing and comprehending to

parallel speaking and hearing, the terms writing and reading to
parallel saying and listening, and the term corresponding to parallel
talking. Widdowson justifies his use of a more specific term such as
corresponding on the following grounds:
Whereas it is reasonable to think of saying and listening as reciprocal
aspects of the one basic activity of talking, reading and writing cannot so
readily be considered as reciprocal activities in quite the same
sense. . . . It is true that we do have written as well as spoken
interactions, as in the case of an exchange of correspondence, and
indeed correspondence might be considered as the larger-scale version
of talking in the written mode. But there is a vast amount of written
discourse that does not take the form of an exchange. Usually, what is
written does not directly depend on a previous reading activity and a
particular act of writing is not necessarily prompted by a particular act
of reading. (p. 61)
From a strictly empirical point of view, Widdowson’s focus on
correspondence can be justified, for he is concerned only with what
might be called external reciprocity (i.e., that which takes place
between externally identifiable participants in written communica-
tion). He could also have presented oral communication in a more
restricted way, for there are different kinds of talk that do not
involve an empirically verifiable exchange (e. g., someone talking on
a radio show where the listeners do not call in). To capture this
restriction, he could have perhaps chosen a term such as conversing
rather than talking.
Given our own view of communication, we are interested in
moving beyond the strictly empirical and expanding Widdowson’s
model to include what may be called internal reciprocity, which we
posit as fundamental to all communication. In explaining this
notion, we would like to return to a point made above. In externally


observable interaction—whether conversation or correspondence—
language users can be viewed not simply as addressing externally
identifiable language receivers but rather as ascribing an identity to
these receivers by virtue of the discourse they construct: the choice
of particular vocabulary and syntactic patterns, the amount of
background knowledge assumed, and so on. By the same token, it
is by means of this discourse that language users take on an identity
for themselves. We can refer to these linguistically constituted
identities as an ascribed identity and an assumed identity, respec-
In characterizing internal reciprocity, we can shift our perspec-
tive from the productive aspect to the receptive, in which case we
can describe individual readers as assuming an identity and
ascribing one to the writer. This contrast between production and
reception is displayed in Figure 5

Production and Reception in Written Communication

1 In both production and reception, the arrow runs from the writer to the reader simply to
indicate temporal sequence: A writer must write before a reader can read.

Indeed, it is only as readers actively assume an identity that they

are able to ascribe an appropriate identity to the writer. This
shifting of assumed and ascribed identities is not, however, the
whole story on internal reciprocity. There is a further displacement
that takes place within both the writer and reader, for each engages
in socially patterned behavior modeled on talk in the external
world, namely, taking on the role of the other in order to negotiate
meaning. Indeed, such displacement may be viewed as potentially
more radical in literate as opposed to oral activity. A writer, in
principle, has ample time in which to anticipate, even rehearse, the
reader’s response.


By the same token, we can view readers as involved in a similar
displacement. By exploring the role that the writer assumes, they
gain a sense of whom the text was written for. Hence they
continuously pose questions, whether consciously or not, about the
writer’s choice of words (Why code rather than style?) or decision
to include particular material (Why is Bernstein brought in here?),
and it is by means of this probing that they ultimately establish how
reliable the writer is. As we suggested earlier, such exploration may
lead readers to modify, even reject, the role that the writer has
ascribed to them. Their negotiation of the writer’s communicative
stance is, in principle, continuous and thus can be paralleled to what
the writer does in revision. In effect, reading, like writing, is an
active negotiation of socially constituted meaning, which can be
represented in the following way:

Internal Reciprocity in Written Communication


As indicated by Figure 6, the larger communication between

writer and reader takes place only as each internalizes the social
reciprocity with which they communicate in the external world.
Such internalization is facilitated by the social character of
language, the means by which both oral and written communication
are achieved: The very structure of language calls forth social
Given this richer notion of reciprocal exchange, we would revise
Widdowson’s model of written communication in the following


A Revision of Widdowson's Model of Written Communication

We have made a number of changes in the model. To begin with,

we have reversed the position of the terms in the bottom line to
reflect our sense of how they are ordinarily understood. Writing and
reading now describe externally observable activities (parallel to
speaking and hearing in the model of oral communication);
composing and comprehending describe the internal aspects of
reciprocity (parallel to saying and listening). More importantly, we
have coined the term texting to take the place of corresponding.
This term parallels talking, thereby signaling that social reciprocity
is fundamental to both written and oral communication.
There is one further change that we have made in the model. We
have removed (INTERPRETING) from the top of the display,
since we prefer not to separate the psychological from the social (as
indicated by our characterization of internal reciprocity as
inherently social). Widdowson (1978) distinguishes interpreting
from the social processes of talking or corresponding because it is
for him a “non-reciprocal skill . . . the psychological process of
understanding” (pp. 65-66). For us, interpreting cannot be
separated from talking or texting. As Vygotsky (1962) observed,
interpretation, when viewed developmentally, is best understood as
incorporating social exchange into individual thought. It is the social
character of all individual thought that underlies our own approach
to written communication. Any literate activity, no matter how
solitary it may be, necessarily embodies reciprocity. TO write is to
be a reader and to read is to be a writer.
What is the application of this revised model to reading tests? We
should first observe that within a testing situation the reciprocity
can be viewed as both external and internal. Test takers do, after all,
produce externally observable responses to the writing presented
by test makers. In Widdowson’s terms, they are involved in
corresponding. Their response is, of course, limited. On a machine-
scored multiple-choice test it consists only of bubbles shaded in on
an answer sheet, and even on a test with more open-ended
questions, the answers are necessarily brief. Nevertheless, these
responses constitute an important act of communication, one that


revolves around the question of whether the test takers, from the
test makers’ perspective, know how to read.
As to internal reciprocity, the question naturally arises as to the
linguistically constituted identities represented in the test material.
As we have pointed out, this material ordinarily represents many
writers, for test makers include samples of different kinds of writing
in the interests of validity; their notion is that a test should represent
the various kinds of text to which the target population is exposed.
Hence in a test that includes many passages, the writers may be
quite diverse. On the other hand, they are not ordinarily named, and
few readers can recognize who they are from the passage. This
anonymity is deliberate, for test makers, in the interests of fair play,
seek out passages that are likely to be unfamiliar to those taking the
test. Further, the passages come as a set, printed in the same format
and framed by the same kinds of tasks, so that they look like the
product of a single hand. In effect, the work of the various writers,
and the manifold identities that they assume in writing, is mediated
through the more encompassing identity of the test maker; and it is
with the latter that test takers must engage in reciprocity if they are
to be successful. In our own research (Hill, 1992), we have
discovered that inexperienced test takers are often led astray by
focusing unduly on the individual writers themselves.
But then who are the test makers and what kind of identity do
they assume in constructing a test? In many parts of the world, their
actual identity cannot be known, since it is, for reasons of security,
shrouded in secrecy. Test takers cannot thus resort to external clues
in ascribing an identity; they must rather attend closely to what
constitutes test discourse—the various passage-task configurations
that the test makers have constructed; and from these they must
infer the presence of homo legens, a figure that draws only on
knowledge normatively associated with the words on the page. It is
not easy, however, for inexperienced test takers to make this
inference, since their own view of what constitutes normative
knowledge does not necessarily correspond to the test makers’; and
such lack of correspondence is particularly likely to occur when the
test takers are reading in a foreign language.
Homo legens, as represented by the test, can be quite an
intimidating figure, for test takers know that they must produce
responses already designated as correct. Since these responses have
little to do with the way that many test takers ordinarily read, they
often become unsure about how to respond. In our own research,
we have discovered that many become fearful that the questions
have been set up to trick them. As one test taker put it, “I can
understand what I read, but not the questions; it’s like they’re there


to trip me up.” Hence the test maker comes to embody an alien
authority: He—and we feel homo legens is best described as a
“he’’—is one who possesses peculiar powers with respect to how
text is to be comprehended. In short, he is the kind of institutional-
ized figure identified by Foucault (1977) in his provocative essay
“What Is an Author?”
It is not simply that test makers assume the identity of homo
legens; they ascribe this same identity to anyone who can negotiate
tests successfully. Test makers do acknowledge various sociocultu-
ral identities in so far as they avoid using passages that are obviously
offensive to particular groups. They do not, however, use any
means of evaluating how sociocultural norms of language, thought,
and experience are reflected in how test takers respond; they do
not, for example, draw on the process-oriented interviews or think-
aloud protocols that are now common in reader-response research.
Rather they assume that anyone who selects the target response,
whether shading a bubble or writing the appropriate word or
phrase, has responded as homo legens. In short, the test makers use
a highly attenuated external exchange as evidence that a particular
kind of internal reciprocity has taken place.

Literacy Skills and Commmunication

Our representation of text, and of those who use it, obviously
entails a view of reading that differs radically from that implied by
the autonomous model of literacy. As already indicated, this model
presents literacy as a technology and reading as a technical skill. It
is a complex skill, as everyone acknowledges. It involves familiarity,
obviously, with the language of the text as well as with the writing
system so that the reader can reconstruct words, phrases, and
sentences from a series of discrete symbols. Moreover, it requires
knowledge of a particular set of conventions for making meaning
explicit. According to Olson, not only is a wider range of
vocabulary typically used in text than in speech but words are used
in more precise ways; similarly, the syntactic resources of the
language are exploited more fully for the purposes of rhetorical
emphasis, especially given the substantial loss of paralinguistic cues
in writing.
This view of reading as a technical, albeit complex, skill does not
account for a rather obvious fact. Even the most proficient readers
are incapable of interpreting some texts—even though those texts
are in their own language. We ourselves, for instance, as linguists,
can identify and characterize in some detail all the constituent
elements in an English text on, say, elementary particle theory in


physics, but we have little confidence in our ability to learn much
from such a text. We simply do not have enough knowledge of the
relevant theory to make sense out of the words on the page, even if
we look them up in a dictionary. In short, much of the meaning that
is said to be present “in” such a text is there only by virtue of the
highly specialized bodies of knowledge it assumes—and the skill of
reading becomes largely a matter of having that knowledge and
being able to apply it.
In addition, when we view reading as an act of communication,
it becomes clear that more is involved than decoding the words of
the text and applying appropriate background knowledge. Readers
must also draw on the communicative skills that we have described
under the notion of internal reciprocity. As they work with a text,
they must not only ascribe an identity to the writer but assume one
for themselves; and they must then work with these identities—
revising them and reconstituting them in new ways—as they
continuously negotiate meaning. As we have claimed, such social
reciprocity is modeled on what we do as we negotiate meaning
orally. In all our interactions, we are concerned, whether
consciously or not, with such matters as how trustworthy our
interactant is, why s/he is making this point, and what we should do
in response. These interactive concerns are fundamental to
negotiating text as well as talk; and the answers to the questions they
raise vary considerably from one culture to another.
We have thus distinguished three kinds of skill that are necessary
to reading. First is knowing a writing system and understanding the
linguistic forms used in a text; second is possessing appropriate
background knowledge and knowing how to apply it; and third is
being able to engage in a reciprocal exchange that is appropriate to
the text being read. Reading tests, as they exist at present, are
designed to assess primarily the first of these three skills. Readers
who do well on these tests can decode accurately. Moreover, they
can perceive formal relationships between words, especially
between those in the passage and those in the tasks; test takers are
often asked to recognize or supply a word or phrase that parallels,
or opposes, one used in the passage—consider Tasks b and e, for
example, in Orchids in Figure 1. Successful test takers can also
manipulate the syntax; they can, for example, express the relations
between subject and object in alternate ways; Task 29 in Radial K in
Figure 2 calls on just this ability.
As to the second kind of skill, reading tests are not designed to
assess whether individuals possess appropriate background knowl-
edge. On the contrary, test makers claim, following the tenets of the


autonomous model, that the information given in any passage is
sufficient for handling all the tasks. From our vantage point, such a
claim is misleading, for no text can be altogether clear about
everything. Test makers attempt to combat textual indeterminacy
by using passages in which only commonplace knowledge is
assumed. The problem is that the test maker’s view of what is
commonplace can diverge quite sharply from the test taker’s; and
such divergence is exacerbated when a test is administered to
people from a wide range of ethnocultural backgrounds who are
dealing with a second or foreign language. Consider, once again,
the Orchids passage, which assumes various pieces of information
that would be useful in responding to Task c: “Why are orchids
expensive in some parts of the world?” Among potentially useful
pieces of information are that (a) the expensive kinds of orchids are
grown in the tropics, (b) temperate regions are where orchids are
expensive, and (c) in these regions people often pay high prices for
a scarce entity, even though it lacks practical use or longevity. To
students raised in the tropics where the orchid is so abundant and
where plants are valued primarily for their practical use, these
assumptions are far from obvious and many are unable to respond
appropriately to this task; see K. Parry, 1986, for a study of Nigerian
students responding to this test material.
As for the third kind of skill, the traditional tests that we have
described make no attempt to assess the ability to engage in
different kinds of reciprocal exchange. Given the character of the
passages used and the particular focus of the tasks, only one kind of
exchange is called for, namely, the response of homo legens to “the
very words” of the text. The implicit message is that this limited
exchange is all that reading is; variation in literacy practices is
ignored not only in different cultures but even in dominant forms of
Anglophone culture, where readers and writers engage in a great
variety of reciprocal exchanges. More recently developed tests, it is
true, do attempt to be more communicative, in that they include
material, such as letters and advertisements, in which social
reciprocity is overtly signaled—the 1987 edition of the Test of Adult
Basic Education (TABE) is an example. In analyzing the tasks based
on such passages, however, we have found that they call for the
same kind of limited response as do those in the older style of tests.
The effect is unfortunate, for when such pragmatically oriented
material is coupled with autonomously oriented tasks, the resulting
discourse can be quite misleading, and the tests can engender even
greater confusion than the ones they are replacing (Hill & Parry,


Throughout our discussion of the pragmatic model, we have
emphasized the social character of literacy. Text is viewed as social
instrument, readers and writers as embodying social identities, and
literacy skills as involving social interaction. When traditional tests
of reading comprehension are considered in this framework, their
limitations become immediately apparent. Based on a strictly
autonomous model of literacy, they ignore the social dimension of
written communication. For individuals who have the opportunity
to develop literacy skills beyond formal education, these effects are
not necessarily serious. They can learn to treat testing as a
distinctive genre that has little or no influence on their experience of
text more generally. But when individuals depend on formal
education for the development of literacy, particularly their
English-based literacy, the tests can have a pernicious influence if
they are allowed to play a dominant role, as they often do, in the
classroom. Reading, and particularly reading in English, comes to
be viewed as a narrowly circumscribed activity in which personal
knowledge must be continually suppressed for fear of making an
inappropriate response. Readers are encouraged to monitor closely
what they know about textually mediated worlds and to hold such
knowledge apart from that strictly assumed by the texts themselves;
as we have pointed out, this assumed knowledge is often alien to
their own sociocultural norms of language, thought and experience.
It is ironic that readers are encouraged to hold separate the very
knowledge which is crucial to their effective engagement with text,
for there is clear evidence that people learn best when integrating
what they already know with what they are reading (Bransford,
Franks, Vye, & Sherwood, 1989).
It is because of these problems that the current development of
alternative approaches to assessing literacy skills has taken on such
importance. In an edited volume, Hill and Parry (in press)
characterize more fully these problems of autonomously oriented
testing as well as alternative approaches to assessment based on a
pragmatic model of literacy. Here we do not have space to deal
with alternative assessment but can suggest what its basic goals are
to document the range of literacy activities in which individuals
engage, the repertoire of abilities they draw on, and how effectively
they use these abilities in carrying out various activities. These are
demanding goals that call for methods of assessment that are
ongoing and comprehensive. Such methods are being developed,
particularly within the field of adult education where reliance on
formal models of schooling is less dominant. Many such programs


still use testing but in a more limited way. A test may be given as
part of an intake interview to provide baseline information about
students: for example, how well they can decode, use an index, or
interpret a chart. Moreover, such testing is often performance
oriented. Students are asked to do literacy tasks embedded in
everyday life, such as writing a check or filling out a bureaucratic
The alternative approach to assessment extends well beyond test-
ing, as exemplified by the increasing use of portfolio documenta-
tion, which is generally associated with writing; students typically
are asked to present a selection of written work. To our way of
thinking, portfolios, given the reciprocal nature of reading and
writing, show students as readers as well. Indeed, their reading can
be used as the primary means of structuring their portfolios. They
can make notes on textbook selections, keep journals in response to
short stories and novels, and write essays about community issues
on which they have conducted research. In this sense, portfolios can
reflect external reciprocity—what students write is directly linked
to what they read. More important, portfolios can provide evidence
of internal reciprocity, for they help us to understand to what
degree students can ascribe appropriate identities to writers and
take on appropriate ones for themselves.
Perhaps the most important benefit of alternative assessment is its
emphasis on students becoming more responsible for evaluating
their own work. Indeed, we would claim that the most basic goal of
any assessment should be to help students internalize socially
constituted norms that they can use in such evaluation. As students
internalize these norms, they become increasingly aware of
themselves as members of a literate community who can use text to
learn. As a Nigerian student once ended a letter to one of us, “my
reading has led me into many strange worlds.”

We would like to acknowledge the help of many students in the Program in
Applied Linguistics at Teachers College, Columbia University; here we have space
to mention only Eric Larsen, Patrizia Magni, Kathie McClure, and Joye Smith.


Clifford Hill is the Arthur I. Gates Professor of Language and Education at
Teachers College, Columbia University, where he directs the Program in Applied
Linguistics. He has conducted research on oral culture in West Africa, African
American language and communication, the linguistic representation of space and
time, and varying conceptions of literacy in relation to education.
Kate Parry is Associate Professor in the Department of English at Hunter College
of The City University of New York and previously taught ESL in Uganda,
England, and Nigeria. Her research interests include cultural variation in reading
strategies, vocabulary acquisition in a second language, the development of a
pedagogical grammar, and the social history of English.

Bakhtin, M. (1981). The dialogic imagination. (M. Holquist, Ed.;
C. Emerson & M. Holquist, Trans.) Austin: University of Texas Press.
Besnier, N. (1986). Spoken and written registers in a restricted-literacy
setting. Unpublished doctoral dissertation. University of Southern
California, Los Angeles.
Bransford, J., Franks, J., Vye, N., & Sherwood, R. (1989). New
approaches to instruction: Because wisdom can’t be told. In
S. Vosniadou & A. Ortony (Eds.), Similarity and analogical reasoning
(pp. 470-497). New York: Cambridge University Press.
Clanchy, M. (1979). From memory to written record: 1066-1307.
Cambridge, MA: Harvard University Press.
Cole, M., & Scribner, S. (1981). The psychology of literacy. Cambridge,
MA: Harvard University Press.
Conklin, H. (1949). Bamboo literacy on Mindoro. Pacific Discovery, 2,
de Saussure, F. (1966). Course in general linguistics (W. Baskin, Trans.).
New York: McGraw-Hill.
Ekvall, R. B. (1964). Religious observance in Tibet. Chicago: Chicago
University Press. -
Foucault, M. (1977). What is an author? In D. F. Bouchard (Ed.),
Language, countermemory practice: Selected essays and interviews by
Michel Foucault. (D. F. Bouchard & S. Simon, Trans.) Ithaca, NY:
Cornell University Press.
Goody, J. (Ed.). (1968). Literacy in traditional societies. C a m b r i d g e :
Cambridge University Press.
Goody, J. (1977). The domestication of the savage mind. C a m b r i d g e :
Cambridge University Press.
Goody, J. (1986). The logic of writing and the organization of society.
Cambridge: Cambridge University Press.
Goody, J., & Watt, I. (1968). The consequences of literacy. In J. Goody
(Ed.), Literacy in traditional societies. Cambridge: Cambridge
University Press.


Gaff, H. (1979). The literacy myth: Literacy and social structure in the
nineteenth-century city. New York: Academic Press.
Greenfield, P. (1972). Oral or written language: The consequences for
cognitive development in Africa, U.S. and England. Language and
Speech, 15, 169-178.
Heath, S. B. (1983). Ways with words. Cambridge: Cambridge University
Hildyard, A., & Olson, D. (1978). Literacy and the specialisation of
language. Unpublished manuscript, University of Toronto, Ontario
Institute for Studies in Education.
Hill, C. A. (1992). Testing and assessment: An ecological approach.
Inaugural lecture for the Arthur I. Gates Chair in Language and
Education. New York: Teachers College, Columbia University.
Hill, C. A., & Larsen, E. (1983). What reading tests call for and what
children do. Washington, DC: National Institute of Education.
Hill, C. A., & Parry, K. (1988). Reading assignment: Autonomous and
pragmatic models of literacy (LC Report 88-2). New York: Teachers
College, Columbia University, Literacy Center.
Hill, C. A., & Parry, K. (1989). Autonomous and pragmatic models of
literacy: Reading assessment in adult education. Linguistics and
Education, 1, 233-283.
Hill, C. A., & Parry, K. (Eds.). (in press). Testing and assessment:
International perspectives on English literacy. Harlow, England:
Lyons, J. (1981). Language and linguistics: An introduction. Cambridge:
Cambridge University Press.
Olson, D. R. (1977). From utterance to text: The bias of language in
speech and writing. Harvard Educational Review, 47 (3), 257-281.
Olson, D. R., & Hildyard, A. (1980). Literacy and the comprehension of
literal meaning. Paper presented at the Conference on the Development
and Use of Writing Systems, Bielefeld, Germany.
Ong, W. J. (1982). Orality and literacy: The technologizing of the word.
New York: Methuen.
Parry, J. (1982). Popular attitudes towards Hindu religious texts.
Unpublished manuscript.
Parry, K. (1986). Readers in context: A study of northern Nigerian students
and school certificate texts. Unpublished doctoral dissertation, Teachers
College, Columbia University.
Richards, J., Platt, J., & Weber, H. (1985). Longman dictionary of applied
linguistics. Harlow, England: Longman.
Street, B. V. (1984). Literacy in theory and practice. C a m b r i d g e :
Cambridge University Press.
Vygotsky, L. (1962). Thought and language. Cambridge, MA: MIT Press.
Widdowson, H. (1978). Teaching language as communication. Oxford:
Oxford University Press.


TESOL QUARTERLY, VoL 26, No. 3, Autumn 1992

Writing One’s Way into Reading

University of Massachusetts at Boston

Despite research and theory to the contrary, approaches to the

teaching of reading continue to reflect a transmission model of
reading, focused on the retrieval of information from a text. This
model prevents students from experiencing reading as an active,
exploratory process, one that involves the making of meaning. It
thus denies them their transactions with a text and the realization
that reading involves such transactions. In order to give students
experiences with reading that demonstrate the ways in which
readers engage, contribute to, and make connections with texts,
writing needs to be fully integrated with reading. Writing, because
of its heuristic, generative, and recursive nature, allows students to
write their way into reading and to discover that reading shares
much in common with writing, that reading, too, is an act of

It has become commonplace to characterize the act of writing as

a meaning making, purposeful, evolving, recursive, dialogic, tenta-
tive, fluid, exploratory process. Recent research and theory in
reading have shown us that these terms can be applied as well to the
act of reading, for like writing, reading is characterized by active
engagement through which meaning is created. But although there
have been ongoing attempts to incorporate the implications of
writing research and theory into the classroom, the teaching of
reading seems to be less responsive to what we know about the
process of reading, perhaps because it is difficult to reconcile the
kinds of outward performance and demonstrations required by
schooling with the internal experience of reading.


The way reading gets taught (and evaluated) in schools tends to
keep hidden from students the sense making and exploration that
makes reading possible and that, in turn, reading makes possible.
What is practiced in the guise of reading suggests to students that

reading is a receptive, and static process, rather than an active, par-
ticipatory one involving the dynamic contributions of a reader.
Shannon’s (1989) historical analysis and detailed description of
reading instruction point to the ways in which the teaching of read-
ing has been translated into a technology requiring students to fol-
low a set of prescribed and uniform procedures which are discon-
nected from students’ lives and subjectivities, and over which stu-
dents have little control. Rose’s (1989) critique of educational prac-
tices applies particularly well to reading instruction. Teaching, he
simplifies the dynamic tension between student and text and reduces the
psychological and social dimensions of instruction. The student’s
personal history recedes as the what of the classroom is valorized over
the how. Thus it is that the encounter of student and text is often
portrayed . . . as a transmission. Information, wisdom, virtue will pass
from the book to the student. (p. 235)
Reading is often reduced to the act of finding a particular idea, as
if this idea resides fixed and absolute in a text. As Bartholomae
(1986) has pointed out, “the language of reading instruction . . . is
loaded with images of mastery and control. . . . A reader finds a
controlling idea and follows it” (p. 96). This approach is most
evident in the display or test questions which appear in reading texts
(for both native speakers and ESL students), questions which
immediately follow the assigned passage and which not only call for
a predetermined answer rather than the interpretation of the
student reader but which may, in fact, keep students from
understanding the text: “Since traditional comprehension checks
generally focus on myriad text details, many students learn to
answer not by understanding the text well, but by looking
progressively throughout the text, following the questions as they
go” (Barnett, 1989, p. 134). It is not surprising, therefore, that
students conclude that the very “purpose of reading text is to
answer the questions that follow it” and that answering these
questions correctly sigifies that they have understood what the text
means (Belanoff, 1987, p. 102). Given the ritualized activities that
characterize reading instruction and the specification of correct
answers upon which it is based, students come to believe that
“reading is the attempt to memorize text which someone else selects
so that you can reproduce factual information when questioned’
(Shannon, 1989, p. 96). Meaning thus becomes understood as
something contained in a text or something that exists “out there,”
rather than something that results when a reader finds a way to
make the representation of meaning possible. In the case of ESL


students, the problem is compounded; the emphasis on determining
and identifying the meaning in a text quite naturally gives way to
the ineffective strategy of reading word-for-word, bilingual
dictionary in hand. Some textbooks reinforce this strategy by
assigning numerous exercises on vocabulary; however, even when
textbooks try to dissuade students from the notion that reading
proceeds in this way, by pushing students to time themselves as they
read and providing them with exercises in skimming and scanning,
the emphasis on the retrieval of information predominates.
By being taught according to the reductive assumption that
reading is a matter of identifying and retrieving a set of ideas that
reside in and are transmitted by a text, a set of ideas that all readers
can agree upon, as implied by Hirsch’s (1987) “cultural literacy,”
readers are denied both their own transactions with a text and the
critical understanding that reading involves such a transaction
(Bartholomae & Petrosky, 1986a; Freire, 1983; Rosenblatt, 1978).
Furthermore, this denial is often enacted quite literally. Given a
fragmented approach to curricula, whereby language and literacy
processes are artificially separated from one another, students are
prevented from writing about their reading because it is believed
that their writing “confounds reading” (Petrosky, 1982, p. 19). But
teachers, too, are kept from interacting with the text in any
meaningful way, for the technology of reading instruction, with its
mandated textbooks, directives, guidebooks, and tests, limits the
teacher’s role to one of monitoring and managing students, thereby
undermining a teacher’s autonomy, experience, knowledge, and
literate behavior (Shannon, 1989). Thus, a transmission model of
reading characterizes the experiences even of teachers involved in
reading these texts, and this model is passed on to the student.
Yet another context in which such a model of reading keeps
teachers from interacting with texts in a genuine way is when they
read students’ papers, looking for predetermined kinds of texts, for
certain sets of ideas, and for particular textual features. White’s
(1984) discussion of the relationship between theories of reading
and the ways in which teachers interpret and respond to student
writing makes explicit how the expectation of ideal texts prevents
teachers from genuinely reading students’ texts. And Tobin (1991)
argues that despite teachers’ recognition that interpretations,
reconstructions, and misreadings are unavoidable, “in practice,
most of us cling to the notion that our readings of student essays are
somehow objective,” as if these texts have an “objective reality”
(p. 336). Although this issue is only indirectly related to my concern
here, I mention it nevertheless, for we need to be aware that our
responses to students’ papers, because they both illustrate and


model a kind of reading, are yet another, though more subtle, lesson
we teach our students about what it means to read.
Given such an approach to instruction, how do students come to
view reading? Bartholomae and Petrosky (1986a) describe the
lessons that students learn about reading as a result of their
classroom experiences. These students assume that in order to read,
one must understand and remember all that has been read and that
their inability to do so suggests that they are not good readers. They
are focused on a concern for getting the definitive meaning. They
come to believe that the difficulties, ambiguities, and confusions
they are wrestling with are necessarily a sign of a problem that
resides in them, and that because what resonated for them did not
match with the answers expected, they cannot read. They come to
understand that what they do seem to get is not important, for
frequently it is the very associations that make the text come alive
and stay with them that are left unacknowledged. They become
afraid, they cease to take risks. Some of this is reflected in the
following comment written by a Japanese graduate student who
was assigned to work with a native speaker of English:l
I have a student who is taking Japanese as a foreign language since last
semester. He is having trouble learning the new language. While his co-
students mastered basic writing and reading skills, he is unable to do so
yet. His last tutor decided he has a learning disability. He himself
admitted that he was categorized with a learning disability and waived
to taking the SAT for that reason. I asked what that meant, and he said
that he had to read something twice before he fully comprehends it. I
thought perhaps I also suffer from this learning disability, because I
sometimes need to read something a few times to understand it well.
It should be noted that these lessons about reading also teach
students lessons about writing. It is no wonder that, as a result of
these kinds of experiences with reading, students come to see
writing as a matter of (re)presenting the right set of ideas or the
correct interpretation; that students distrust and dismiss their own
written attempts, trying to determine instead what the instructor
wants; that they view their struggles with writing as a reflection of
their inability to write; that they develop a deep-seated fear of
writing. The irony, of course, is that the more difficulty students
seem to experience, “the less reading and the more nonsense drills
we typically arrange for them to do” (Smith, 1983, p. 5). SO a
cyclical process gets underway whereby students who have little
1 The student texts that appear throughout this article were written by students attending my
ESL composition courses and graduate courses on the teaching of writing.


insight into what reading means are excluded from the possibility of
making this discovery.


Reading, as we have come to understand it, in fact, has little to do
with what we ask our students to do. The growing body of literature
in reading theory and research in both L1 (see, e.g., Newkirk, 1986;
Peterson, 1986; Smith, 1971) and L2 (see Barnett, 1989; Carrell,
Devine, & Eskey, 1988; Grabe, 1991; Mikulecky, 1990; Wallace,
1988) has underlined the importance of recognizing that reading has
as much to do with what the reader brings to the text and how the
reader interacts with the text as with the text itself. The following
excerpts represent the ways in which the process of reading has
been characterized by writers whose work is situated in a range of
disciplines but whose conceptualizations of reading reveal a
remarkably similar perspective:
There is no such thing as a generic reader or a generic literary
work. . . . The reading of any work of literature is, of necessity, an
individual and unique occurrence involving the mind and emotions of
some particular reader. (Rosenblatt, 1938, p. 34)
The understanding of a text has not begun at all as long as the text
remains mute. But a text can begin to speak. When it does begin to
speak, however, it does not simply speak its word, always the same, in
lifeless rigidity, but gives ever new answers to the person who questions
it and poses ever new questions to him [sic] who answers it. To
understand a text is to come to understand oneself in a kind of dialogue.
(Gadamer, 1976, p. 57)
At the center are the readers’ responses . . . to the meanings they make
and re-make as they read. (Atwell, 1987, p. 154)
Reading is usually thought to be a matter of discerning what is there,
but . . . it is a matter of knowing how to produce what can thereafter be
said to be there. Interpretation is not the act of construing but the act of
constructing. Interpreters do not decode [readings]; they make them.
(Fish, 1987, p. 226)
There is no meaning on the page until a reader decides there is. (Tierney
& Pearson, 1983, pp. 568-569)
Much of the meaning understood from a text is really not actually in the
text, per se, but in the reader. (Carrell & Eisterhold, 1988, p. 79)
That is the terrible, wonderful power of reading: the texts we create in
our own minds while we read—or just after we read—become part of


the life we believe we lived. . . . [A]ll reading is autobiographical.
(Murray, 1991, p. 74)
A piece of writing must be understood as a product of a person or
persons . . . taking its meaning from the interpretive gestures of
individual readers. (Scholes, 1982, p. 16)
Reading always involves critical perception, interpretation and
rewriting what is read. (Freire, 1983, p. 11)
Reading, then, if it is to represent engaged and meaning-making
activity, must allow for the ways in which readers contribute to and
make connections with the text. Writing provides a unique
opportunity for discovering and exploring these contributions and
connections, for it allows the reader to dialogue with a text and find
a particular way into it. Explicating the work of a number of
reading theorists, Petrosky (1982) explains, “the only way to
demonstrate comprehension is through extended discourse where
readers become writers who articulate their understandings of and
connections to the text in their responses” (p. 24).
Despite recent attempts to make connections between reading
and writing, these efforts are such that we presume that reading is
what makes it possible for us to write rather than the other way
around. Note, for example, that whereas writing textbooks typically
include reading material upon which writing assignments are based,
it is rare for reading textbooks to include assignments that call for
extended, open-ended written responses. Furthermore, when
written assignments are included in reading textbooks, they
invariably appear at the ends of units or chapters, thus reinforcing
the notion that writing is done as a final activity after the text has
been read, analyzed, worked through, rather than used as a means
for understanding the text. We seem to assume a static and
unidirectional effect for reading on writing, believing that exposure
to reading texts provides us models, that reading provides so-called
comprehensible input which, if acquired, will later be displayed in
the writing produced, that reading provides ideas that can be used
as a basis for writing one’s own text. Thus, reading continues to be
viewed as necessarily preceding writing, to offer a paradigm to
internalize, to act as a stimulant for writing, or to provide subject
matter to write about. With any of these situations we assume that
if students read, they will become adept at putting their thoughts on
paper. Reading and writing are thus not fully integrated, and
reading controls the writing. (Arguing for balanced integration,
Silberstein, in press, cautions against a similar danger: having writ-
ing instruction dominate reading, “using reading [only] as grist for
a writing mill.”)


Calkins (1986), whose work is primarily in children’s acquisition
of literacy, describes how such a perspective influences the ways in
which writing is now being introduced in reading classrooms.
Under the rubric of reading-writing connections, writing is being squeezed,
stretched, and distorted to service existing reading programs. . . . In the
name of reading-writing connections, writing is being treated as an elab-
orate ditto. (p. 234)
In other words, writing is being inserted into the reading program
and is typically viewed as either reinforcing reading or taking away
from reading time. Calkins questions whether these writing activi-
ties make a significant contribution to reading because they fail to
take into account the genuine connections readers make as they
read, because they don’t explore the reading and writing relation-
ship provocatively enough. These writing activities keep hidden
from developing readers/writers that one can find one’s reactions
and responses to texts by reflecting on them through writing.
That writing, in fact, contributes to the development of reading
has been demonstrated in elementary school classrooms in which
writing becomes an opportunity to generate and explore meaning.
In these classrooms, children with limited literacy and English
language proficiency, even before they have learned to read (even
before some of them can express themselves orally in English),
write stories which become their first reading texts (Edelsky, 1982;
Hudelson, 1984). This initiation into literacy provides children with
the opportunities to develop and extend their understanding about
text. It allows them to test out their “growing understanding of
storiness, of wordness, of how one keeps ideas apart in writ-
ing . . . of how one uses writing to mean” (Harste, Woodward, &
Burke, 1984, p. 218). Writing, because it requires these beginning
readers/writers to make decisions about purpose, sequence, and
language, because it helps them to understand how and why texts
are written, gives these learners insights into the goals, constraints,
and concerns of authors, insights which they apply to their reading.
Writing, because it helps learners understand that “everything they
read is writing,” therefore needs to be recognized as “the
foundation of reading,” and the “most basic way to learn about
reading” (Hansen, 1987, p. 178).
Like the children in these “writing to read” classrooms, older
students who have had little experience with reading or who have a
limited understanding of what reading means can learn how print
comes to represent meaning through writing. Adults, for example,
who are in the process of acquiring both English language and
literacy, can benefit from creating their own texts; these written


pieces, because they can provide a means through which students
record their own experiences and consider their own realities, have
greater resonance for them than those they are typically assigned to
read and thus have the potential for becoming the basis for literacy
development (Jones, 1991; Rigg, 1981; Rigg, 1991). Voices: New
Writers for New Readers, a quarterly journal written by and for
adults acquiring literacy (both native and nonnative speakers of
English), is a powerful demonstration of how an appreciation for
literacy begins with writing. In the case of ESL college composition,
writing has been shown to affect the reading of literature when the
writing assigned is fully integrated with and provides strategies for
reflecting about literary texts (Spack, 1985). But even at more
advanced levels, writing has much to contribute to reading. The
following excerpt, written by a graduate student who was reflecting
on his own development as a reader, demonstrates how applying
writing strategies to his reading contributed to his development as a
Until sometime after I began majoring in English, I was a poor reader.
I treated each word on the page as a separate entity. Reading this way
always caused me to become easily distracted. I’d restart continually (I’d
read “Call me Ishmael” over and over), lose my concentration, etc. Once
it had become clear to me by way of learning about the writing process,
I applied a sort of free-writing approach to my reading. I forced my
eyes to speed up, to move forward as a pen is commanded to during
freewriting. I took in chunks of sentences at a time in the same way that
phrases rolled off my pen when I was writing without having time to
think about them. And slowly, without fully realizing it at the time, I
began predicting where the author was going so that I didn’t need to
attend to every word in order to get the meaning.
Becoming conscious of an approach to writing that could be
brought to bear on his reading had a critical impact on this student’s
process of reading. But there are more profound ways in which
writing teaches reading. Because the process of writing shares much
in common with the process of learning, it gives rise to the
generation and reconceptualization of ideas that may not have been
possible otherwise (Emig, 1977; Odell, 1980). The heuristic nature
of writing allows one to discover and consider one’s stance, one’s
interpretation, one’s immediate reactions to a text. Moreover, it
makes these responses to a text overt, concrete, and tangible.
Making students conscious of their own reactions to texts gives these
readers the sense that experienced readers have when they read.
Though we may not always respond in writing to our texts, when
we underline portions of texts, mark them up, stop and verbalize
our reactions, or scribble marginal comments, we are using


interpretive strategies that give us insight into our meaning making.
What these actions represent is the complex intersection of who we
are, what we know, and how we make sense.
By reading like a writer writes, students come to see that to
understand a text is to come to understand oneself in a kind of
dialogue, that understanding is initiated by a response, and that
understanding means not so much taking away from a text as giving
to the text what is not there. Writing in this way makes reading an
activity of finding and making connections, of figuring out what
speaks to the reader and why. Writing, because it gives rise to our
own ways of probing and working with texts, is thus a way to
construct, to compose the reading. Note the following account,
written by a graduate student, which reveals the essential role that
writing plays in composing her reading:
I never fully understand what I read until I write about it. This goes
for school subjects, letters, and novels that I read in my spare time. . . .
Every once in a great while when I read a novel I will write about it
at great length in my journal. I am not sure whether it was the initial
interest or subsequent writing about it but these novels are all among my
favorites. For the most part I developed an intense interest with the main
characters and the way the author developed them. In a sense I removed
the characters from the author’s domain into my own world. How would
this character deal with a situation in my own life?. . . . By writing about
these characters I find out a tremendous amount about myself. I think
mostly because by putting words on paper my voice appears. By
discovering that voice, I have a better understanding of the work itself.
We read everyday, but how much do we maintain. I read the Sunday
papers this week, and I have a vague understanding of what is
happening in the world but the only articles I vividly remember are the
articles I wrote my mother about. . . .
Reading sparks something in the mind like a nighttime of fireflies.
Associations both emotional and intelligent speed across the brain and
we have a hard time capturing them. Groups of letters, words and
paragraphs affect us with sound, meter, and meaning. We continue to
read because we like the fireflies. The light and movement mesmerize
us. When we try to focus on one firefly, it is elusive and difficult to catch.
But by capturing it, and getting close to it we have a better
understanding of what motivates it. We can only capture those sparks of
imagination by putting them down on paper so they can be read and
reread, and by rereading those thoughts we are usually moved to write
One cannot write without reading for as we write we read the words
that pour, or dribble, onto the page. One can read without writing, but
understanding is much deeper if response to that reading is done in


In addition to giving us insight into the generative nature of
reading, because writing has the inherent capacity to let us see our
thoughts and to discover in that seeing alternatives and new
perspectives, it allows us to experience reading as an evolving and
speculative process. Writing invites us to entertain our initial
responses, to offer tentative reactions, but then allows us to go back
into the text and to revise these original readings, Thus, it helps us
to understand that reading changes as we bring new responses to it,
that reading is open to revision, that reading is work in progress.
(See, e.g., Langer, 1990, whose study of students’ reading processes
revealed the “changing stances” (p. 237) that readers applied as
they attempted to understand texts and the recursive nature of these
“stances.” See also Flower, 1988, whose study suggests that a text’s
meaning changes as readers bring different purposes, goals, and
assumptions to their reading. ) Writing is a way to resee texts and so
lets us grapple with uncertainties, reflect on the complexities, deal
with the puzzlements, and offer approximative readings. By
providing us a means for working out a reading, writing allows
insights that may have been inaccessible or inchoate at the time that
the text was read. As Bartholomae and Petrosky (1986a) point out,
“The act of writing serves to justify, to repair or to speak into
coherence an experience that, while it occurred, felt more like loss
and confusion” (p. 17). By enabling us to formulate, judge, modify,
extend, and refine our responses, writing helps dispel the notion that
reading is a matter of getting something and getting it at the outset.


What specific classroom approaches can give readers experiences
with the dialogic and dynamic nature of reading? Students, for
example, can keep reading journals or logs in which reactions are
recorded and elaborated upon. Though journals or logs are by now
common practice in writing courses, assigned as a method for
fostering writing development, they need to be better understood
as a means for promoting reading as well. The following illustrates
one ESL student’s initial response to The Diary of Anne Frank
(Frank, 1967). Note how this entry begins with a straightforward
retelling, but then moves on to personal reflection, which helps
strengthen the student’s understanding of the author and her
She started the diary with by sketching in a brief the story of her life.
She started to introduce her family and the reasons why they had to
hidde. Also talk about the people who lived with her.


Anne feel strenge for the changes in her life who they left her house
and now they had to be hidding for their life also she feel sad because
they can’t go to anywhere and feel scare for her father and for her
When I read this pages I feel little upset because I don’t understand
why is this world people think they are perfect and the others people
had to be attack by words and action. In this case they paid to be Jewish
with their life.
In addition to responding to texts, these responses can serve as
evidence that reading is a constructive act; the various readings
students bring to a text can be shared so that students come to be
aware of the existence and legitimacy of multiple interpretations.
By becoming aware of these other interpretations, they have yet
new frameworks with which to go back and read the text.
Naturally, these reactions need not be recorded only after texts have
been read. On the contrary, students’ associations, questions, and
thoughts, written in progress, make immediate the connections
between reading and writing as well as inform and shape the text
being read. Again, sharing these “process” reflections may shed new
light on an understanding of the text.
Journals are a natural technique for dealing with literary texts,
texts which by their very nature invite subjective and alternative
interpretations (Spack, 1985). But this kind of written response
needs to be applied as well to reading assigned across the
disciplines. Even with reference to texts from which students are
expected to derive the same information, as is the case in much
school-based work, writing allows students to work through their
readings so that their understanding comes closer to approximating
what is assumed to be the intended meaning of these texts. Assumed
is used here in order to underscore the notion that even so-called
definitive interpretations are determined on the basis of a shared
body of knowledge, both about information in the text and about
how texts work, that we bring to a text. As Flower (1988) has
pointed out, even “autonomous texts” (p. 544) are subject to
readers’ individual interpretations. Given the inevitable influence
that our own perspectives, backgrounds, and knowledge have on
how we make meaning from texts, writing, sharing, and negotiating
these approximate readings provides a means not only for
understanding the content of texts but for learning about the
reading process, that is, what makes it possible to locate information
in a text, to determine a text’s meaning. Furthermore, giving
students the opportunity to respond to discipline-specific texts in
exploratory and tentative ways is more likely to make these texts
accessible. As Elbow (1991) argues, in order to promote genuine


understanding of the academic discourse of assigned texts, students
need to use informal, nonacademic writing to “render” rather than
explain experience, to translate the “discourse of the textbook and
the discipline into everyday, experiential, anecdotal terms”; to do
otherwise is to keep students from “experiencing or really internal-
izing the concepts they are allegedly learning” (p. 137). A graduate
student, writing about her growing awareness of the role writing
can play in her understanding of an assigned text, echoes Elbow:
While reading, I’m saying to myself, I understand exactly what he’s
saying, but if you ask me exactly what I agree with, I don’t know yet. I
think I have to write it out myself before I can internalize the points I
agree with. It’s still very fuzzy in my mind what I agree with. By writing,
I have to consciously think out what I’ve read, and by writing it down,
I internalize the meaning in my own words to make the ideas real to
me—so that I own the words.
Furthermore, giving students the opportunity to write about what
they find interesting/significant/moving/puzzling may help them
realize that their understanding of complex texts evolves as they
(re)read and that written reflection makes this understanding
possible. Note the following entry, an excerpt taken from an ESL
student’s journal responses to an assigned anthropology text:
The Chapter 6 “marriage” confuse me in some vocabulary, but I
understood the rule of marriage of the kungs women, and I found it
strange too. I think it is unfair for the parents to chose their daughters a
husband very young, if they travel with them, hunting and gathering
when the childrens are little, why don’t the parents keep their children
with them until they are able to understand the meaning of marriage, or
they are ready for it by their own, except give them away to be cared
and maintained by a strange man.
I also found it touching in some aspects, for example when Nisa
express her feelings about the times she was forced by her parents to live
with Tashay, her husband, and she ran away many times to sleep in the
bush. Also when she was living in his parents village, that she felt lonely
and sad without her mother. It’s was obvious that she still needed her
mother’s affection and care, but by that time the parents seem just to
worry about somebody or a man to maintain her, not about her feelings.
Numerous examples of entries of this sort appear in The Journal
Book (Fulwiler, 1987), a collection which convincingly demon-
strates how individual and personal responses to texts assigned
across academic content areas can be a more effective means for
understanding and learning from these texts than traditional
methods, such as note-taking. In one chapter, for example, a history
professor discusses the impact that short journal entries, written in


response to the assigned reading, has on students’ engagement with
the material. In yet another, an llth grade American Civilization
teacher describes the genuine involvement that is generated when
her students are given the opportunity to react, hypothesize,
interpret, and raise questions in their reading logs. It is clear that this
form of written response fosters the kind of intellectual and
meaningful activity that allows students to “enter” and engage these
challenging texts. Finally, I include the following excerpt, which
comes from the journal of a graduate student, not only because it
reflects her realization that writing plays a critical role in reading
assigned work, but because it illustrates how the very act of writing
this entry helped her establish connections with what she read:
For those courses where we have had to write journals, I am infinitely
more involved with the readings than those courses where no journal is
required and I do my obligatory underlining and marginal writing. One
course I am thinking about has an enormous amount of extremely
technical reading. No journal. As a result, it’s almost as if the readings are
one activity; class discussion is a separate activity and somehow,
although the course is based on the readings, they are never effectively
integrated into the course. In the article on reading, Freire says,
“Reading always involves critical perception, interpretation, and re-
writing what is read.” I have come to see that “re-writing” what one has
read is essential in interpreting and understanding that text.
Double-entry or dialectical notebooks (Berthoff, 1981) provide
another means for making us conscious of our reactions. In these
notebooks, students copy passages that have particular significance
for them in one column, and then respond to them in the other. The
following represents the ways in which two ESL students, studying
the history of racism and civil rights in the context of a course on the
American Dream, responded to an autobiographical excerpt by
Rosa Parks (reprinted in Gillespie & Singleton, 1991). On the left are
the passages each student copied, on the right, their reactions:
Copied text Reactions
My mother had a mind of her own. I like it because Rosa’s mother was
She always held to the belief that like a symbol of a life freedom!
none of us should be mistreated Her mind was very independent
because of our race. (Gillespie & and clear in front of society. I
Singleton, 1991, p. 372) think Rosa inherited her mother’s
courage. Rosa was as big as the
Statue of Liberty in front of her
black society that organized a
boycott on December 5, after she
was arrested because she opened
her “eyes to the prize.”


In reality we had to face the fact When Rosa Parks talks about
that we were not as free as the African-American, I had a horrible
books said. What they taught us in feeling. No other immigrants can
school didn’t apply to us as a race. feel about that. Think if you were
(Gillespie & Singleton, 1991, kidnapped to be a slavery from
p. 371) your country, how difficult the
situation would be? “This is not
the home of the blacks” is the
poem written by Langston Hughes
had expressed. Rosa had showed
her progressive action 12 years
before she arrested. But she was
taken off the bus. I was shocked
by the humiliating segregation law.
You have to stand up and give the
seat to somebody else because you
are black. What a racism! When I
was in China, even though there was
discrimination to the north people
who came down to the south, the
south people at most could call them
bad names and cheat them, but
could never show out.
These passages not only reveal the particular association each
student is making with the text, but demonstrate that the very act of
writing these reactions may have generated connections to previous
course content, in particular, the symbolic significance of the Statue
of Liberty, readings from Eyes on the Prize (Cohen, 1987), and a
poem by Langston Hughes (“Let America be America Again,”
There are a number of possible variations on this approach.
Students can summarize or react to certain passages in one column,
and then reflect on these reactions in the other. Or, students can be
asked to write about the associations a text calls forth after they
have summarized and reacted to it in separate and self-contained
pieces of writing (Bleich, 1986; Petrosky, 1982). Because so many
students have a limited notion of what it means to read, providing
them with opportunities to first summarize and retell what they
have read makes obvious that we want them to engage in a very
different process when they react to, associate with, and reflect
upon texts. After an extensive and detailed summary of one section
from Maxine Hong Kingston’s (1975) Woman Warrior, for example,
one student wrote the following moving account:
I identify with her experience and feelings about voices. More than
once in my life, I had to deal with that. Like her, before I picked up the


phone to talk to someone in English, I spent more than 15 minutes to
prepare what I was going to say but when my fearful voice came out,
word by word as a stutter and I forgot all. Moreover I’m now taking
English speaking, listening class and I know how important voicing is.
It’s very hard to practice. My throat once broken because the voices
come out fearfully now gets hurt again.
Also I have shared with Kingston her experience of going to school.
Right now, around me is silent too. I want to make friends but my
English doesn’t allow me to do that, I’m afraid of asking people and of
being asked question as well, so I keep quiet and do things by myself.
I’m always thinking of my old friends and my own happy, lovely world
I left behind. (I had studied chemistry for 4 years in college in Vietnam)
For 4 years, my class was home and my friends were sisters, brothers
and now the more I think of them, the more I feel lonely. Got into the
lab, looked around, I dreamed to see them, to hear them in my own
language, but only strangers and only English. More silence covered me
Yet another possibility is to ask students to mark certain passages
or insert marginal notations or some other form of reflective
comment as they read, just as experienced readers often do, and
then to go back to these passages and explore in writing why they
think these passages resonated for them. For example, after noting
a particular passage in Lost in Translation, in which the author, Eva
Hoffman (1989), describes her “spontaneous flow of inner
language,” (p. 107) one student wrote the following:
I read and reread it many times. It had given me a strange feeling. I also
have interior language, interior image sometime in my life but I could
never write it down as she did, even in my own language.
This form of response allows students to consider, weigh, and
interpret their reading and gives rise to reactions that they may not
have been aware of. It allows them to have a conversation with the
author and become actively involved in the reading. Students not
only come to realize what each viewed as important, what
represented key moments for them, but why. When the same
strategy is applied to passages that students found confusing or
difficult, not only do students come to see the legitimacy of raising
questions about texts (a powerful alternative to answering questions
about texts), but they may discover that going back to and writing
about these passages can help them work through their confusions
or difficulties. Furthermore, because each reader may have
responded to a different passage in the text, students can
understand the reader-dependent nature of reading and that what
speaks to each of us (or does not) may be different.


Students can be made aware of their associations with texts be-
fore they have read them if they are asked to write about experi-
ences that enable and contribute to their interpretations. Asking
them to write about an experience that figures in a text they are
about to read not only helps students explain the matter to them-
selves but sets up a connection, a readiness that may not have been
established otherwise. For example, Hudelson (1984) illustrates how
bilingual children learn by brainstorming and writing about the con-
tent they are about to study before they read their textbook, thus es-
tablishing a connection, a set of expectations, and background
knowledge that facilitates comprehension. ESL textbooks by
Raimes (1992), Spack (1990), and Withrow, Brookes, and Cum-
mings (1990) are notable for the various ways in which students are
invited to explore their understanding of and associations with
themes and topics that arise in the readings, not so much before they
read these texts, but in order to read them. Fulwiler’s (1987) book,
too, provides examples of the benefits of writing about academic
subjects before reading about them. In a recent ESL writing course,
prior to our consideration of civil rights in the United States, stu-
dents wrote about thoughts and associations related to this issue.
One Chinese student, for example, wrote about the 1989 demonstra-
tion that took place in Tiananmen Square, while a Vietnamese stu-
dent recounted a folktale that her grandmother had told her.
Inviting students to consider and weigh their own ideas first
demonstrates that what they bring to the reading not only is a valid
perspective but enables them to better understand, analyze, take
issue with the perspective that they read about. Asking students to
explore issues in writing before they read assigned texts makes it
possible for them to approach the reading from a position of
authority. Because they are allowed to bring their own experience,
knowledge, and cultural background into play, they can approach
the authors they subsequently read as other voices (rather than
definitive authorities) to which they add their own. The ways in
which students weave these voices together add depth, richness,
and particularity to their writing, and in so doing, point to the
problematic nature of setting up dichotomies between personal and
academic discourse (Elbow, 1991). The following, for example,
represents the conclusion of an essay in which the student does not
merely retell the Vietnamese folktale (which she wrote about prior
to our consideration of civil rights issues) but revises it in light of the
critical perspective she has developed
I remember a story which my grandmother told me in my childhood. “A
long time ago, people are the same color and live lovely together. Some


of them are lazy and some others are gluttonous. One day they gather
and bake a cake. The gluttonous ones eat their part of the cake when it
is still not well-done yet. So the color of their skin changes to white.
Meanwhile, the lazy ones eat their part of the cake when it is burned.
And their skin color changes to black.” Of course, it is just a popular
story of my childhood. But after studying the issue of Blacks and Whites
in class, everything changed in my mind. I felt bad for this unfair world.
If you ask me my dream, I would say that I dream one day the color of
skin changes back to that of a long time before, so all people may gather
and make a cake again, but eat the cake together at the same time.
By initially exploring her own values and associations in writing,
she is able to (reconsider her text with reference to those which she
subsequently reads, thus partaking in the kind of dialogic thinking
and active interpretation that is critical to the development of
reading/writing (Spellmeyer, 1989). Finally, writing about and
considering issues, themes, or concepts before reading about them
prepares students to view texts with “a writer’s eye” (McQuade &
Atwan, 1991, p. iii), to appreciate authors’ perspectives and
decisions, to read, in other words, like a writer.
Students can come to appreciate the ways in which readers use
predictions to make meaning by writing speculatively about what
will happen in a text and then comparing these predictions with
those of other students and with the original text. Written predictions
of this sort literally transform student writers into authors of the text.
This activity also gives students insight into how and why the texts
they themselves write produce expectations in readers. It thus further
underlines the fundamental principles that reading involves active
negotiation between reader and text and that who the reader is and
what he or she already knows affect that negotiation.
One final suggestion is that we sequence assignments around
readings so that students are guided to address these readings from
different perspectives. For example, before I assign excerpts from
Terkel’s (1972) Working, I ask students to write about their
recollections of and reactions to their own work experiences, thus
helping them uncover what they already know and think. In
addition to reading Terkel and relating his findings to the students’
own pieces, students conduct their own interviews in order to
explore the relationship between work and satisfaction. The essays
they finally write require them to reconsider both their initial
written papers and Terkel’s interviews in light of their own
interview data. Students are thus “pushed back” into the readings
and asked to view them in the context of a different framework and
to rethink their original responses. And because these recursive


assignments encourage the review of students’ own previous
writing, students’ texts become legitimate texts to consider and
examine alongside the published readings they are assigned. Other
suggestions for ways in which to build on and sequence assignments
can be found in Ways of Reading (Bartholomae & Petrosky, 1987),
an anthology of readings for students, and Facts, Artifacts and
Counteracts (Bartholomae & Petrosky, 1986b), a reference for
teachers. These exemplary works demonstrate how a dialectical
relationship is established through reading and writing, how reading
and writing can inform, be a source for understanding, and be
brought to bear upon one another.
These instructional strategies are suggestive of the ways in which
we can take advantage of writing and its inherent possibilities to
teach students not only about writing, but about reading, to teach
them that “any reading requires ‘writerly’ activities from the reader”
(Rosenblatt, 1978, p. 171). Given the interdependence of these two
language processes, the overriding implication is that the teaching
of reading and writing cannot be separated nor can they be
sequenced in linear fashion so that reading necessarily precedes
writing. As Scholes (1985) has argued, reading and writing are
“complementary acts” (p. 20); in the same way that writing a text
necessarily involves reading it, reading a text requires writing a
response to it. Thus, just as the teaching of writing should involve
the teaching of reading (as it has been conceptualized here), the
teaching of reading is necessarily the teaching of writing
(Bartholomae, 1986). Just as reading provides “comprehensible
input” for writing, writing can contribute comprehensible input for
reading. Just as students “need to become better readers” in order
“to become better writers,” as Spack (1988, p. 42) argues, they can
become better readers by becoming better writers. This realization
is critical, given the segregated ways reading and writing are
addressed in ESL research and pedagogy, and given the negligible
role writing continues to play in works on second language reading,
even when these works underline the active and reader-dependent
nature of reading.

The writing activities I have pointed to are ways to enter the text
in an organic and powerful way and help students to understand
that the reader acts upon and gives meaning to the text, that reading
is a process of composing. These activities demonstrate how
reading shares much in common with writing, that, for example,
both are affected by purpose and goal (Flower, 1988), that both


involve recursiveness, that both are malleable and open to
(re)interpretation, that both require experiencing a “felt sense”
(Perl, 1980, p. 365). Note how Rosenblatt’s (1978) characterization
of reading applies as well to writing
Nor does all this activity . . . go on in a simple linear progression. As the
text unrolls, there is not only the cumulative building up of effect
through the linking of remembered earlier elements to the new ones.
There is sometimes a backward flow, a revision of earlier understand-
ings, emphases, or attitudes; there may even be the emergence of a
completely altered framework or principle of organization. Sometimes
this very act of revision of the framework becomes an important aspect
of “the meaning’” of the work. (pp. 60-61)
Furthermore, because these activities allow students to actively
engage and grapple with texts, to explore how and why texts affect
them, students can make discoveries about what other readers do
with texts they compose. They come to realize that if reading
involves a reconstruction, they must help guide readers of their own
texts in that reconstruction, this despite the fact that even with
writers’ attempts to create “autonomous texts” (Flower, 1988,
p. 544), readers will nevertheless reconstruct a text on the basis of
their “alternative hypotheses.” They can begin to imagine their own
writing as potential reading. This gives them insight into the way
reading and writing work in tandem to promote and enhance one
another. In other words, writing like a reader becomes inextricably
bound up with reading like a writer.
Writing, because it gives students opportunities to discover that
reading is an active and generative process, teaches students a
critical lesson about reading. It allows students not just to learn
about something in a particular text, but to learn about how one
learns. This is what Bruner (cited in Raimes, 1983) meant when he
described as essential to learning the act of “climb[ing] on your own
shoulders to be able to look down at what you’ve just done and then
to represent it to yourself” (p. 537). Writing, because it allows us to
represent to ourselves our learning, our ways of making meaning,
teaches the most profound lesson about how we read, write, and use
language, about what it means to know. It teaches us, as Polanyi
(1958) has put it, that “into every act of knowing, there enters a
passionate contribution of the person knowing what is being
known”’ (p. viii). It is when students come to understand reading
and writing in this critical way, as acts of knowing, that they come
to see that reading lets us know writing, and writing lets us know


I would like to thank Elsa Auerbach and Ruth Spack for their “writerly” readings
of earlier drafts of this paper.

Vivian Zamel is Director of the ESL Program at the University of Massachusetts at
Boston, where she teaches writing courses for ESL students and graduate courses
on teaching ESL. She has published widely on writing theory, research, and

Atwell, N. (1987). In the middle. Portsmouth, NH: Heinemann.
Barnett, M. A. (1989). More than meets the eye: Foreign language reading.
Englewood Cliffs, NJ: CAL/Prentice Hall.
Bartholomae, D. (1986). Wanderings: Misreadings, miswritings, misunder-
standings. In T. Newkirk (Ed.), Only connect: United reading and
writing (pp. 89-118). Upper Montclair, NJ: Boynton/Cook.
Bartholomae, D., & Petrosky, A. (1986a). Facts, artifacts and counteracts:
A basic reading and writing course for the college curriculum. In
D. Bartholomae & A. Petrosky (Eds.), Facts, artifacts and counterfacts
(pp. 3-43). Upper Montclair, NJ: Boynton/Cook.
Bartholomae, D., & Petrosky, A. (Eds.). (1986b). Facts, artifacts and
counteracts. Upper Montclair, NJ: Boynton/Cook.
Bartholomae, D., & Petrosky, A. (Eds.). (1987). Ways of reading. N e w
York: St. Martin’s Press.
Belanoff, P. (1987). The role of journals in the interpretive community. In
T. Fulwiler (Ed.), The journal book (pp. 101-110). Portsmouth, NH:
Berthoff, A. E. (1981). The making of meaning. Upper Montclair, NJ:
Bleich, D. (1986). Cognitive stereoscope and the study of language. In
B. T. Peterson (Ed.), Convergences: Transactions in reading and
writing (pp. 99-114). Urbana, IL: National Council of Teachers of
Calkins, L. (1986). The art of teaching writing. Portsmouth, NH:
Carrell, P. L., Devine, J., & Eskey, D. E. (Eds.). (1988). lnteractive
approaches to second language reading. New York: Cambridge
University Press.
Carrell, P. L., & Eisterhold, J. C. (1988). Schema theory and ESL reading
pedagogy. In P. L. Carrell, J. Devine, & D. E. Eskey (Eds.), Interactive
approaches to second language reading (pp. 73-92). New York:
Cambridge University Press.


Cohen, S. (Ed.). (1987). Eyes on the prize: America’s civil rights years
1954-1965. Boston: Blackside.
Edelsky, C. (1982). Writing in a bilingual program: The relation of L1 and
L2 texts. TESOL Quarterly, 16 (2), 211-228.
Elbow, P. (1991). Reflections on academic discourse: How it relates to
freshmen and colleagues. College English, 53, 135-155.
Emig, J. (1977). Writing as a mode of learning. College Composition and
Communication, 28, 122-128.
Fish, S. (1987). How to recognize a poem when you see one. In
D. Bartholomae & A. Petrosky (Eds.), Ways of reading (pp. 222-234).
New York: St. Martin’s Press.
Flower, L. (1988). The construction of purpose in writing and reading.
College English, 50, 528-550.
Freire, P. (1983). The importance of the act of reading. Boston University
]ournal of Education, 165, 5-11.
Frank, A. (1987). Anne Frank: The diary of a young girl. New York:
Washington Square Press.
Fulwiler, T. (1987). The journal book. Portsmouth, NH: Boynton/Cook.
Gadamer, H. (1976). Philosophical hermeneutics. Berkeley: University of
California Press.
Gillespie, S., & Singleton, R. (Eds.). (1991). Across cultures: A reader for
writers. Boston: Allyn & Bacon.
Grabe, W. (1991). Current developments in second language reading
research. TESOL Quarterly, 25 (3), 375-406
Hansen, J. (1987). When writers read. Portsmouth, NH Heinemann.
Harste, J., Woodward, V. A., & Burke, C. L. (1984). Language stories and
literacy lessons. Portsmouth, NH: Heinemann.
Hirsch, E. D. (1987). Cultural literacy: What every American needs to
know. Boston: Houghton Mifflin.
Hoffman, E. (1989). Lost in transition: A life in a new language. N e w
York: Penguin.
Hudelson, S. (1984). Kan yu ret an rayt en ingles: Children become literate
in English as a second language. TESOL Quarterly, 18 (2), 221-238.
Hughes, L. (1992). Let America be America again. In G. Colombo,
R. Cullen, & B. Lisle (Eds.), Rereading America (2nd ed., pp. 776-778).
New York: St. Martin’s Press. (Original work published 1938)
Jones, P. (1991). The various benefits of dialogue journals. In J. K. Peyton
& J. Staton (Eds.), Writing our lives (pp. 102-128). Englewood Cliffs,
NJ: Prentice Hall Regents.
Kingston, M. H. (1975). The woman warrior. New York: Vintage.
Langer, J. A. (1990). The process of understanding: Reading for literary
and informative purposes. Research in the Teaching of English, 24,
McQuade, D., & Atwan, R. (1991). Resources for teaching The Win-
chester Reader. Boston: Bedford Books of St. Martin’s Press.
Mikulecky, B. S. (1990). A short course on teaching reading skills.
Reading, MA: Addison-Wesley.


Murray, D. M. (1991). All writing is autobiography. College Composition
and Communication, 42, 66-74.
Newkirk, T. (Ed.). (1986). Only connect; Uniting reading and writing.
Upper Montclair, NJ: Boynton/Cook.
Odell, L. (1980). The process of writing and the process of learning.
College Composition and Communication, 31, 42-50.
Perl, S. (1980). Understanding composing. College Composition and
Communication, 31, 363-369.
Peterson, B. T. (Ed.). (1986). Convergences: Transactions in reading and
writing. Urbana, IL: National Council of Teachers of English.
Petrosky, A. (1982). From story to essay: Reading and writing. College
Composition and Communication, 33, 19-36.
Polanyi, M. (1958). Personal knowledge: Toward a post-critical
philosophy. Chicago University of Chicago Press.
Raimes, A. (1983). Tradition and revolution in ESL teaching. TESOL
Quarterly, 17 (4), 535-552.
Raimes, A. (1992). Exploring through writing (2nd ed.). New York: St.
Martin’s Press.
Rigg, P. (1981). Beginning to read in English the LEA way. In C. W. Twy-
ford, W. Diehl, & K. Feathers (Eds.), Reading English as a second
language: Moving from theory (Monographs in Language and Reading
Studies No. 4, pp. 81-90). Bloomington, IN: Indiana University Press.
Rigg, P. (1991). Whole language in TESOL. TESOL Quarterly, 25 (3),
Rose, M. (1989). Lives on the boundary. New York: Penguin.
Rosenblatt, L. (1938). Literature as exploration. New York: Apple-
Rosenblatt, L. (1978). The reader, the text, the poem: The transactional
theory of the literary work. Carbondale, IL: Southern Illinois University
Scholes, R. (1982). Semiotics and interpretation. New Haven, CT: Yale
University Press.
Scholes, R. (1985). Textual power. New Haven, CT: Yale University Press.
Shannon, P. (1989). Broken promises: Reading instruction in twen-
tieth-century America. New York: Bergin & Garvey/Praeger.
Silberstein, S. (in press). Techniques in teaching reading. New York:
Oxford University Press.
Smith, F. (1971). Understanding reading. New York: Holt, Rinehart and
Smith, F. (1983). Essays into literacy. Portsmouth, NH: Heinemann.
Spack, R. (1985). Literature, reading, writing, and ESL: Bridging the gaps.
TESOL Quarterly, 19 (4), 703-725.
Spack, R. (1988). Initiating ESL students into the academic discourse
community: How far should we go? TESOL Quarterly, 22 (l), 29-51.
Spack, R. (1990). Guidelines: A cross-cultural reading/writing text. N e w
York: St. Martin’s Press.


Spellmeyer, K. (1989). A common ground: The essay in the academy.
College English, 51, 262-276.
Terkel, S. (1972). Working. New York: Avon Books.
Tierney, R. J., & Pearson, P. D. (1983). Toward a composing model of
reading. Language Arts, 60, 568-69.
Tobin, L. (1991). Reading students, reading ourselves: Revising the
teacher’s role in the writing class. College English, 53, 333-348.
Voices: New writers for new readers. Surrey, Canada: Lower Mainland
Society for Literacy and Employment.
Wallace, C. (1988). Learning to read in a multicultural society. New York:
Prentice Hall.
White, E. M. (1984). Post-structural literacy criticism and the response to
student writing. College Composition and Communication, 35, 186-208.
Withrow, J., Brookes, G., & Cummings, M. C. (1990). Changes. N e w
York: St. Martin’s Press.


TESOL QUARTERLY, Vol. 26, No. 3, Autumn 1992

Communicative Interaction and

Second Language Acquisition:
An Inuit Example
McGill University

This article reports on research findings that emerged during a

longitudinal ethnographic study on the role of cultural context in
the communicative interactions of young Inuit (Eskimo) children
and their caregivers. The study was conducted in two small
communities of arctic Quebec where Inuktitut, the native
language of the Inuit, is spoken on a routine, daily basis. The focus
of the research was on discourse features of primary language
socialization in Inuit families. The incongruity of these features
with the discourse in classrooms taught by non-Inuit second
language teachers surfaced repeatedly during the course of the
study. This incongruity raised several issues pertinent to the
learnability and teachability of second languages for Native
populations. Such issues are discussed here with reference to
related second language acquisition literature. In doing so, the
interface between the sociocultural aspects of communicative
interaction and second language acquisition is emphasized.

This article reports on the role of cultural context in the commu-

nicative interaction of young Inuit1 children, their caregivers, and
their non-Inuit teachers. The study on which the article is based was
designed to answer questions that arose during Inuit teacher educa-
tion courses on the development of language. Ethnographic
methodology was used to gather information on Inuit children’s
communicative interactions in the situational context of their
everyday lives.
The research concerned language socialization. As such, it went
beyond but encompassed Inuktitut language acquisition. Although
the study was centered on Inuit family socialization of language and
communicative interaction, information about the differences in the

communicative interaction of Inuit families and non-Inuit second
language teachers surfaced repeatedly during the course of the
study. Statements were made in the context of the research which
reflected the frustration felt by both parents and teachers and shed
some light on the difficulties that Inuit children have in learning a
second language in the ways that their non-Inuit teachers expect
them to. These emergent findings point to a significant link between
the process of language socialization at home and second language
teaching in school. The central questions that this article addresses
are: How do Inuit families socialize their children to become
cultural members whose communicative interactions differ
markedly from those of their non-Inuit teachers? What can this tell
us about the teachability of a second language for Inuit learners?
The acquisition of a second language implies the acquisition of a
second culture for many learners. For these children, school
becomes a form of secondary socialization where the pragmatics of
the first language interfere with the learning of the second language.
The idea that school is a secondary form of socialization for
nondominant-culture children is not new. Several authors (Au &
Jordan, 1981; Duranti & Ochs, 1988; Heath, 1983, 1989; Ochs &
Schieffelin, 1984; Philips, 1983) have documented how certain
universal notions of childhood, with their corresponding educa-
tional and language acquisition corollaries, are false assumptions.
Yet, such language socialization literature, with its description of the
discontinuities of home and school, has not always been recognized
in second language acquisition literature. This has meant that the
interface of the sociocultural dimensions of language learning and
second language acquisition have not been fully explored. Cultural
variation in second language learning is the main concern of this
article. Looking closely at the Inuit culture’s particular ways of using
and learning language serves as a reminder that there is more than
one way to learn a language, be it an L1 or an L2.


Approximately 23,000 Inuit inhabit the Canadian North. Yupik,
Inupiaq, Aleuts, and other closely related peoples, are also found in
northern Russia, Greenland, and Alaska. Canadian Inuit live in the
Northwest Territories, arctic Quebec, and Labrador. The research
discussed in this article was carried out in Kangirsuk and Quaqtaq,
two small communities on the Ungava Bay of arctic Quebec.
Kangirsuk lies approximately 1,000 miles north of Montreal, and
Quaqtaq, approximately 100 miles north of Kangirsuk. There are
325 Inuit living in Kangirsuk and 200 in Quaqtaq. Half the


population is under the age of 15. Both of these communities have
fewer than 20 non-Inuit people living in them, except during peak
periods of construction.
Education in northern Quebec is under the jurisdiction of the
Inuit-controlled Kativik School Board. The school in Kangirsuk has
10 Inuit educators out of a staff of 18. The Quaqtaq school is headed
by an Inuk principal with a staff of six Inuit teachers and six
Qallunaat (non-Inuit; singular, Qallunak) teachers.
Inuktitut, the language of the Inuit of arctic Quebec, has been
described as one of the few languages of Native North Americans
with a possibility of long-term survival (Foster, 1982; Priest, 1985).
The Ungava Bay dialect of Inuktitut is spoken routinely and
constantly in Kangirsuk and Quaqtaq. It is used for all formal and
informal transactions that take place between Inuit living in these
two communities.
In schools, kindergarten, Grade 1, and Grade 2 are taught
exclusively in Inuktitut. After second grade, children receive most
of their instruction in either French or English by teachers who are,
in the large majority, Qallunaat. The shift to a second language
comes, for the most part, with no systematic preparation of the
children or the teachers. Usually the only preparation that children
have for the shift is through informal exposure to second language
on television and, occasionally, through their siblings’ talk. From
Grade 3 until the end of secondary school, however, Inuit children
continue to be instructed in their first language during religion, Inuit
culture, Inuktitut language, and sometimes physical education
classes. Religion, Inuktitut, and Inuit culture courses are viewed as
the courses whose content is most meaningfully taught by Inuit
teachers in the native language (J. Howden, personal communica-
tion, June 1991). This means that although the children’s contact
with fluent adult speakers of Inuktitut is concentrated in the lower
elementary grades, it is, nevertheless, maintained to a certain degree
throughout the entire upper elementary and secondary cycle. In
Grades 3 through 7, instruction in Inuktitut comprises 6 out of a total
of 23.5 hours of instruction per week, or roughly one quarter of a
child’s instructional time. In high school, first language instruction
diminishes to one fifth of a child's total instructional time, or 5 out
of 25 hours per week. More extensive first language instruction is
constrained by the limited number of trained Inuit teachers and the
limited amount of instructional materials available in Inuktitut
(Taylor, 1990),
Personnel from the Kativik School Board report that the choice of
a second language (English or French) is made by the children’s
parents, some of whom decide randomly and some of whom base


their choice on an attempt to balance language competency in the
family, choosing French for certain of their children and English for
others. Parents are also reported to base their decisions on the
reputed abilities of certain teachers in their communities. Still others
maintain that their children already learn English from television
and from other Inuit in their communities who were instructed in
English under the original federal school system. For these parents,
selecting French language instruction is done to maximize their
children’s possibility of a trilingual future with optimum
employment possibilities. The number of children enrolled in the
French language stream has grown in the last 10 years until it
represents approximately half of all school children above Grade 3.
As a result, all schools in northern Quebec, no matter what the
enrollment, provide instruction in three languages: Inuktitut,
English, and French.

Materials and Procedures
The present research consisted of a longitudinal study of 4 Inuit
children and their families. The study of these 4 children is based on
the tradition of the developmental psycholinguistic studies (Bloom,
1970; Brown, 1973; Halliday, 1975) that focused on a few children
over a length of time and in which recordings (in this case,
videotapes) were made from which detailed verbal and nonverbal
information was transcribed. The research, however, departed
from its psycholinguistic roots. The primary focus was not strictly
on the linguistic aspects of the child's acquisition but, instead, on the
communicative interactions that existed between the children and
their caregivers. Furthermore, in this study, these communicative
interactions were contextualized ethnographically.
The ethnographic and communicative interfactional information
of the study was gathered from three sources: 80 hours of
transcribed and translated videotapes of 4 children and their
families, 20 ethnographic interviews of Inuit women conducted in
Inuktitut by a native speaker, and several hundred pages of
observation and informal interview notes (some of these made for
other research and consulting purposes). This body of data was
collected between 1985 and 1987 on six different trips that averaged
3 weeks in length. These data included school information that was
not collected in a systematic fashion but rather was a by-product
of the numerous conversations that I had during the course


of my home socialization research. An additional body of data in
the form of observation notes was collected in an informal manner
during my work as a consultant for the Kativik School Board and as
the coordinator for the Project for Hearing Impaired Inuit of
northern Quebec.

Videotapes were made every 3 months of 2 Inuit boys, 2 Inuit
girls, and their families. At the outset of the study, all 4 children
were between 1 and 2 years of age. Two of the children had
mothers in their early 20s and 2 had mothers who were considerably
older (see Table 1). Only one child lived in a small nuclear, one-
generational family. Such family structure is a nontraditional and
only recently emerging phenomenon in northern Quebec. The other
3 children lived in larger families. Two of these larger families
spanned three generations. All 4 children came from unicultural,
unilingual homes and none of them showed apparent evidence of
mental, physical, or emotional disability.
In addition, 20 women out of the possible 45 women living in
Quaqtaq were interviewed in Inuktitut by the northern Inuk re-
search coordinator, who is a native speaker of Inuktitut. The
coordinator decidal to interview first all 10 of the women in the
settlement over 45 years of age. In her estimation, these women
were the most knowledgeable about child rearing. Later, after
viewing the videotapes and observing the differences between
young mothers and older ones, the coordinator selected two other
age groups to interview (ages 18–25 and 25-45). The age groupings
reflected her sense of the ages and corresponding stages of modern
Inuit motherhood. Five additional women were then selected in
each of these two groups. The two mothers and one of the
grandmothers of the two children studied in Kangirsuk were also
interviewed. Four Inuit educators happened to be among the
mothers who were interviewed.
Observation notes were also made of the same four families
featured on the videotapes. In addition, a variety of episodes and
encounters with other people in several communities of northern
Quebec formed a part of the observation notes. These included
Inuit’s actions and reactions at home, around the community, in
Inuit campsites, and in school. Informal conversations and
interviews with Inuit and non-Inuit educators were included in
these notes. The non-Inuit educators were both French- and


Family Constellations

English-speaking and interactions with them took place in the

language they preferred.

Results and Discussion

It is beyond the scope of this article to explain all the results of this
research. Instead, those findings that have the greatest relevance to
the sociocultural dimensions of second language acquisition will be


Early Evidence of Language Learning
In the interviews, the women of Quaqtaq were asked how they
knew whether a child had learned language. The responses of
younger women in the two younger groups differed from those of
the women over the age of 45. They stated their ideas as follows:
1. I think it is possible to tell what the baby is trying to say when she tries
to talk.
2. When she is able to say something meaningful.
Older women, on the other hand, believed that comprehension
was an indicator of the attainment of language learning. The
majority of them made statements like the following
3. Try and see if they can recognize something and use clothing or
things and ask them what they are and try to make them understand
what they are for.
4. When she is able to understand, then if she was told to get a mitten
and if she went and got it, and when she was told to bring it over to
the person who sent her to get it and if she understood that, then we
can know that she has learned language in that way today.
These older women’s comments show that they value a child’s
ability to understand and to do what is asked. As a part of their
socialization, Inuit children are encouraged to follow directives. On
the videotapes, the imperative form of verbs was used more
frequently with Inuit children than were comments that followed a
child’s lead. The older women indicated that it was not as important
that children talk as it was that they show comprehension, obedi-
ence, and performance of socially useful tasks.

Limited Responses to Young Children’s Early Vocalizations

Suusi and Jini, the two girls studied, were the youngest of the 4
children. In several of the tapes I made of them, they frequently
made unintelligible vocalizations. The large majority of these
vocalizations went unheeded. Many times their parents did not
respond, not even looking up at the children.
When their vocalizations were responded to, both Suusi’s and
Jini’s mothers occasionally imitated their child’s vocalizations. In
only one instance, on any of the tapes, was a child’s unintelligible
vocalization interpreted as a word. Jini’s older sister, Miaji, was
playing with Jini when the younger child vocalized, tatata. Miaji
responded to this with ataata (father). No father lived in their home.
Clarification of unintelligible vocalizations did not take place on
any of the videotapes. Intentions, then, were not imputed to these


early unintelligible utterances nor did they elicit a communicative
response from the caregivers in most instances.

Talk Not Meant to Be Understood

Because Inuit children are present in many multi-age situations,
they are exposed to a great deal of talk by older people. Yet, it
became apparent in this study that they were neither expected to
participate nor to ask questions of adults who were speaking
together. If they did ask questions, the adults ignored them, leaving
their questions unanswered. One older woman explained this kind
of exclusion to us:
5. Now it is very different. In those days we were not supposed to talk
even a little bit, and when the Bible was being read no one was to say
a word and no one was to move around. Today it is not like that any
more. When people are talking together telling a story children just
join in nowadays. I say something sometimes when that happens,
when a child participates in adult conversation. It never was like that
as I remember it. . . . Children were never allowed to join in adult
I: Why not?
They were told they were children and that was why they were not
to participate in adult conversation. They did not want to have the
ways of adults because they were small.
My videotapes showed numerous examples of children being
ignored while adults conversed. Insuring that children do not
actively participate in conversations they are not ready for is a
communicative modification that some older Inuit women said
served to protect the child from adult concepts and adult ways.
Younger women have quite different conceptions of keeping
children from interfering in adult conversations. We asked them the
same question that we asked the older women, “Are there situations
in which children are not supposed to talk?” They replied with
statements like
6. I do not really believe in having children not talk because one cannot
keep from talking when people talk.
7. Maybe when people are not well.
8. Sometimes in church.
Surprised by their answers, the Inuk interviewer rephrased her
question. This time she asked, “Were there times when you were not
supposed to talk as a child?” They responded as follows:


9. I don’t seem to have.
10. I don’t think so.
11. I don’t remember having to not talk.
The reactions of the younger women were in direct contrast to those
of the older women. All the women over 45 years of age spoke with
conviction about the complete unacceptability of children
interfering in adult conversation.

Changing Forms of Discourse

On the videotapes, the two older mothers only very occasionally
asked their children a question. Furthermore, they only asked one
single interrogative form, “What’s the matter?” This was asked
when their children evidenced distress. The two younger mothers,
however, requested that their children display their knowledge with
interrogative forms such as Who’s this? or How old are you
Questions to which adults knew the answer such as Where’s the so
and so? and What’s this? were interpreted by one young mother as
a necessary part of preparing the young child for school. Conse-
quently, this woman had begun drilling her child by asking him
such display questions. She told us that
12. the younger women talk more like that to their children.
One older woman whom we interviewed described the
difference between Qallunaat and Inuit ways of talking to their
children in these terms:
13. She [the Qallunaat] talks in a teaching way. Sometimes I do that
with my children when they are learning to talk. I teach them things
like the names of the colors and make them say them to see if they
know them well.
This woman did not seem totally convinced about the utility of this
“teaching way” of talking. She concluded by saying,
14. Sometimes they learn to remember them and sometimes they forget.
Inuit parents, then, do not normally ask their children to display
knowledge that they as parents already know. Consequently, the
classroom discourse patterns that are customarily used by white
middle-class North American teachers (Cazden, 1988) involve a
form of communicative interaction (question asking and answering)
that differs sharply from the pattern of early language socialization
in Inuit families. Although it has been documented (MacLure &
French, 1981) that there is a certain amount of disjuncture between


home and school discourse even within the same culture, it has also
been documented that early literacy practices in dominant-culture
homes involving, for example, question asking and answering, are
not carried out in certain families from nondominant cultures and
that this difference represents a particularly noteworthy disjuncture
between home and school discourse patterns for dominant-culture
teachers and their students from nondominant cultures (Heath,

Various Perspectives on Appropriate Language Behavior

The educational world of northern Quebec is filled with the
various perspectives of parents and teachers, Inuit and non-Inuit,
some of which concern children as language learners. The
comments quoted below give an indication of how different
participants in the educational process view second language
classroom communication. Some of the non-Inuit teachers who are
quoted are French speaking and some are English speaking. To
start, a comment from a sixth-grade Qallunak teacher who was in
her first year of teaching in northern Quebec:
15. What I found hardest in teaching up here was trying to do this L2
thing when all they do is sock you with their interminable silence.
This next quote comes from a Qallunak pedagogical counselor
who has worked for 15 years in northern Quebec:
16. The kids are coming right along in most schools. They are really
starting to speak up now. They are less shy. It’s great.
Contrast this with a parent’s comment at a report card interview
concerning his fifth-grade son:
17. Non-Inuit teacher: Your son is talking well in class. He is speak-
ing up a lot.
Inuk parent: I am sorry.
Or with this older Inuk parent’s description of the difficulty she
has in handling her younger children:
18. Today children will just start asking questions to the two people who
are trying to tell each other something and they are made to be this
way nowadays, and also, since there are teachers [here nowadays] it
is said that this child has to know real language and real ways of
doing things and so the child seems to be just like the adult these
This comment is taken from an interview conducted in Inuktitut
with an older Inuk woman who is the mother of seven children, the


youngest of whom was in the seventh grade. In it, she described her
opinions about extracurricular activities:
19. Inuk woman: We don’t want them in school after three
o’clock. We have to get them home and wipe it
all out.
Inuk interviewer: Wipe out what?
Inuk woman: The language and the behaviour . . . those
ways of talking and being with people that we
Inuit don’t have and don’t want our children to
A Qallunak teacher taught in one of the bigger settlements where
his kindergarten class was one of the only mixed (Inuit and non-
Inuit) kindergartens and one of the only ones where a second
language was the language of instruction. In his French kinder-
garten, he had some children whose first language was English. In
reference to this situation, he openly admitted:
20. These little white kids, even if they are learning in a second language,
they talk circles around the Inuit kids. The Inuit kids just sit there
looking at me with big eyes, saying nothing. I hate to admit it, but one
day I just couldn’t stand their silence any longer, I ended up shaking
this Inuk boy and screaming, “Talk to me for heaven’s sake.”
Finally, an Inuk educator in her thirties said:
21. School was tough when we were kids. Our parents had never come
at us like that to teach us.

Misunderstanding of Classroom Discourse

When non-Inuit teachers were asked to comment on the speech
and language abilities of their students, they indicated that as many
as 30% of the children in their classes might have language
disabilities, citing as evidence that the children did not talk in class.
An Inuk teacher who was called upon to evaluate the children’s first
language abilities looked at the list of children’s names that had
been referred by the teachers and said:
22. These Qallunaat teachers never seem to learn that well-raised Inuit
children should not talk in class. They should be learning by looking
and listening.

Communicative Interaction in Inuit Teaching Situations

How, then, does learning take place in Inuit home-teaching
situations? I asked Inuit men (aged 20 to 30) how their fathers had
taught them to hunt. They responded like this:


23. We watch. We watch those that are able.
24. It is mostly watching, observing. They don’t say anything to the boys.
25. You look, that’s all, just look. Like an igloo, they never tell you
anything and you don’t ask.
26. You just watch and one day you do it by yourself.
On the other hand, the men I questioned about learning to hunt said
that they had learned a lot of language with their fathers and other
older men. One put it this way:
27. When they talked among themselves we weren’t supposed to listen.
We did in the tents. You couldn’t help it. They tell lots of stories in
the tents. You learn hard words out on the land. That is where I still
learn words.
An Inuk woman described how in times past in the igloos and
now in tents most talking takes place at night after the children are
in bed. She remembered the grown-ups crouched near the stove
drinking tea. She remembered lying there listening to their
conversations and stories. Not all learning, then, was done by
observation. There was a great deal of information that was learned
verbally through listening. However, information was not directed
specifically at children nor were children to obtain information by
questioning. Instead, learning occurred by overhearing the
conversations and stories of adults or, as the Scollons (1981) wrote,
by having “a grandparent in the ear” (p. 101).
Important verbal explanations were formulated in two genres of
narrative, in the stories and in the recounts that form part of adult-
to-adult conversations. In certain situations, it was tolerated that
children listen to this information as long as they remained
inconspicuous and unobtrusive. Their participation and their
questions, however, were neither fostered nor tolerated, nor was the
production of these genres of narrative shaped or elicited by Inuit
parents from their young children.
With the evolution of the Inuit culture, there are indications of
changes occurring in the communicative interactions of Inuit
caregivers with their children. Some women attributed this change
to the advent of schools and the infiltration of Qallunaat ways into
their communities. Indeed, the stated concepts and the behavior of
young mothers imply that they are analyzing and, to some degree,
adapting their communicative interactions to suit the schools’ styles.
Unfortunately, until now, the schools for their part have not
reciprocated. They have neither studied nor analyzed Inuit styles of
communication, nor have they been as likely to adapt classroom
patterns to suit the Inuit ways.


The findings from this research suggest a number of speculative
implications for second language teaching.2 In brief, the study adds
to the growing body of work demonstrating that all cultures do not
have the same patterns of communicative interaction. The nature of
caregiver accommodations to the child found in North American
dominant society are not the same in Inuit families. Inuit adults do not
use questions to draw their children’s attention to such things as
names of objects and events. Both French- and English-speaking non-
Inuit second language teachers in northern schools, however, often
still use a “transmission” style of teaching in which a child is singled
out to respond to a question whose answer the teacher already
knows. Moreover, the cultural role and status of the Inuk child are
different from the North American dominant-culture child. So, too,
are the structures of caregiving. Inuit children are raised, like children
in other hunting and gathering societies (Heath, 1989; Scollon &
Scollon, 1981), to take the role of spectators, observing and listening
to the performance of knowledgeable adults. Heath (1989) has
pointed out that spoken language in these societies is not considered
the key evidence to prove that a child is learning language. Further-
more, the early physical competence of children is more highly
2 This research and the conclusions, although informed by community members, are
nevertheless the work of a non-Inuk. It is instructive to see how this research has been
interpreted and acted upon by Inuit educators and policy makers. In keeping with Kativik
School Board policy, this research was reported to the Council of Education Committees
and to the executive of the school board as well as to the educational commissioners. The
council, which represents education committees from all 14 communities in northern
Quebec, found the research to be a very accurate representation of educational issues and
cultural realities as they understood them. This group, then, voted that the research should
be presented to all Inuit and non-Inuit teachers in all settlements of northern Quebec. A
course on the educational ramifications of Inuit child rearing and cultural values was
initiated and made available through McGill University to all non-Inuit teachers in northern
Quebec. I was asked to team teach the course with an Inuk educator. Courses on
cooperative learning have slso been instituted for non-Inuit teachers and content from the
research has been integrated into courses for Inuit teachers. The Inuit directors of the school
board decided that there was a need to assure as much cultural continuity as possible to
students in the elementary grades. However, they felt that students in secondary grades
should be exposed to styles of classroom discourse derived from the dominant culture. It
was their opinion that this would help the students to become prepared for post-secondary
education in southern Canada and for other forms of interethnic dialogue. In addition, the
research council of the school board decided to fund research on classroom discourse in
Inuit-taught classrooms. These studies have as their goal the documentation of Inuit styles
of discourse that can be used to inform the board, the community, and non-Inuit educators
about the classroom styles that have been adopted and formulated by Irsuit educators for
use in the early elementary grades (Crago & Eriks-Brophy, in press). It is the hope of the
school board that when Inuit and non-Inuit educators become mutually informed about
their ways of teaching that they will be able to make the cultural adjustments that students
experience more educationally productive. The school board, then, has had a number of
reactions to this research.


valued than the display of spoken language. The talk of children has
a place but primarily with their peers and sibling caregivers.
The difference in Inuit patterns of communicative interaction
must affect the teaching of a second language since L2 teachers are,
by necessity, the communicative partners of the children whom
they teach. In the instruction of a second language, teachers need to
consider the cultural attitudes conveyed by the communicative
interactions that they engage in and propose to others. Perceived
problems in communication between Native students and their non-
Native teachers have been documented for almost a decade in both
North America and Australia (Christie & Harris, 1985; Erickson &
Mohatt, 1982; Philips, 1983, 1985; Saville-Troike, 1985). Teaching
strategies for second language acquisition in Native children, then,
need to be grounded in the knowledge of what comprises cultural
membership for these children. Patterns of communicative
interaction are central to cultural identity. Hence language teaching
and learning are intertwined in a fundamental manner with
culturally integral ways of communicating.
Other issues about the teachability of a second language are also
raised by the study of this particular cultural group. Many Inuit
children resemble a group of successful second language learners
that Wong Fillmore (1983) described in her study on individual
variation in second language learning. They are, by white North
American middle-class standards, shy, uncommunicative children
who have been socialized to be attentive listeners in the presence of
adults. Wong Fillmore has suggested that in educational circumstan-
ces such as those in northern Quebec, where the teacher is the only
fluent second language speaker, such children need ample input
that is well structured by the teacher.
This suggestion is reminiscent of Krashen’s (1985) theory of
second language learning that emphasizes the importance of
exposure to comprehensible input that contains structures on the
learning edge of a person’s ability. Comprehension-based second
language programs of the type documented by Lightbown (1989)
would appear to be a culturally congruent format for teaching
second language to children from cultures such as the Inuit that
emphasize learning by listening.
On the other hand, Inuit children are raised in a society where
sibling caregiving is still a norm and peer-oriented talk, in the
absence of adults, is recognized as an appropriate and very
necessary part of language learning. In keeping with this, a
culturally congruent form of classroom interaction for L2 teaching
with this population would consist of cooperative, peer-interactive


activities (Kagan, 1986). However, it is important to bear in mind
that efforts to adapt the group learning model developed with
Native Hawaiians for use in the Navajo context required
considerable modification of the original cooperative learning
arrangements (Tharp & Gallimore, 1988; Vogt, Jordan, & Tharp,
1987). Thus, it is to be expected that successful implementation of
cooperative learning strategies will require accommodation to the
precise cultural repertoire of the students involved.
One needs to ascertain whether the value of these two approaches
to teaching Inuit children a second language represents a
contradiction or whether two such alternative strategies can be used
in combination in much the same way as they are in the children’s
home culture. Perhaps each approach supplies the children with a
different set of competencies and each one signals to the children
the appropriate interactional hierarchy and the interactional roles of
adults and children in the communicative events of their lives. In the
“unpackaging,” to use Weisner, Gallimore, and Jordan’s (1988)
term, of the critical variables that will maximize Inuit students’
learning and minimize their cultural loss, both of these teaching
strategies for L2 need to be assessed.
Furthermore, for schools like those of northern Quebec that have
both first language and second language instruction, there is inter-
esting potential for using Inuit children’s L1 competencies to en-
hance their learning of their L2 (Cummins, 1981). For instance,
early readers that children master first in Inuktitut might be trans-
lated and used to introduce children to an L2. This past year, the
Kativik School Board has inaugurated the simultaneous trilingual
development of instructional materials for use in an ecology course
at the eighth-grade level. In this course, modern Qallunak knowl-
edge about such things as beluga whales will be taught to the stu-
dents after they have first learned traditional knowledge and vocab-
ulary about the same subject taught to them by Inuit teachers in
Inuktitut. At the elementary school level, the number of trained
Inuit teachers with the Kativik School Board has become large
enough to allow a small number of Inuit to teach in English at the
third- and fourth-grade levels. Because these teachers are bilingual,
they are able to explain difficult concepts to their students in L1 be-
fore these concepts have to be mastered in the L2. The presence of
such teachers and the innovation of the kinds of bilingual instruc-
tional materials and curriculum described above should help to
make the transition from L1 to L2 less abrupt and educationally
hazardous because it can be presumed that the conceptual under-
standing derived from the Inuktitut material will transfer to the
learning of the same subject matter in the second language.


However, when addressing the second language learning of
minority culture children like the Inuit of northern Quebec, we need
to recognize, as Cummins (1989) has pointed out, that the “lack of
second language fluency may be only a secondary contributor to
children’s academic difficulty but the fundamental causal factors of
both success and failure lie in what is communicated to children in
their interactions with educators” (p. 33). More research is needed
on the kinds of communicative interactions in school that cause
minority students and their parents to “mentally withdraw from
academic effort” (Cummins, 1989, p. 34). The teaching and learn-
ing of a second language may lead to more successful outcomes for
the children involved if research and educational efforts continue to
explore the cultural and communicative dimensions of second
language learning.
In conclusion, non-Native educators of Native children from
around the world need to become aware of a form of prejudice
endemic to much of the education of children from nondominant
cultures. This prejudice has been called “ethnicism” by Skutnabb-
Kangas (1988). Ethnicism is the conscious or unconscious denial of
a person’s cultural ways. Schools for Native North Americans have
exercised ethnicism of a particularly virulent variety in the past.
Still, today, it behooves educators who are dedicated to a
multicultural society and to the empowerment of Native people to
look at and learn from the indigenous patterns of instruction and
upbringing that in many societies long preceded the institution we
know as school. The importance of this premise is summarized in
the following statement by Okpik (1989):
There are only very few Inuit, but millions of Qallunaat, just like
mosquitoes. It is something very special and wonderful to be an Inuk—
they are like the snow geese. If an Inuk forgets his language and his Inuit
ways, he will be nothing but just another mosquito. (p. 10)

This research was supported by funds from the Toronto Sick Children’s Hospital
Foundation (Grant No. 8614), the McGill University Faculty of Medicine, and the
Kativik School Board of northern Quebec. Special thanks go to Betsy Annahatak
and Lizzie Ningiuruvik for their contributions to the research. My thanks also go to
Elisabeth Handsley for her help in the manuscript preparation, to James Howden
for information on the second language programs of the Kativik School Board, and
to a careful and informative anonymous reviewer. I am especially indebted to the
Inuit families who opened their homes to me and my children so that we could
learn from them.


Martha B. Crago is Assistant Professor in the School of Human Communication
Disorders and the Northern and Native Education Program of McGill University.
Her research includes ethnographic studies of language socialization and classroom
discourse in the homes and schools of Inuit, Mohawk, and Algonquin communities.
She has published many articles on the cultural dimensions of language and learning.

Au, K. H., & Jordan, C. (1981). Teaching reading to Hawaiian children:
Finding a culturally appropriate solution. In H. T. Trueba,
G. P. Guthrie, & K. H. Au (Eds.), Culture and the bilingual classroom
(pp. 139-152). Rowley, MA: Newbury House.
Bloom, L. (1970). Language development: Form and function in emerging
grammars. Cambridge, MA: MIT Press.
Brown, R. (1973). A first language. Cambridge, MA: Harvard University
Cazden, C. (1988). Classroom discourse. Portsmouth, NH: Heinemann.
Christie, M., & Harris, S. (1985). Communication breakdown in the
Aboriginal classroom. In J. B. Pride (Ed.), Cross-cultural encounters:
communication and miscommunication (pp. 81-90). Melbourne,
Australia: River Seine.
Crago, M. B., & Eriks-Brophy, A. (in press). Culture, conversation, and
children’s co-construction of interaction. In J. Duchan (Ed,.), Pragmat-
ics: From theory to therapy. Englewood Cliffs, NJ: Prentice Hall.
Cummins, J. (1981). The role of primary language development in
promoting educational success for language minority students. In
California State Department of Education (Ed.), Schooling and
language minority students: A theoretical framework. Los Angeles:
California State University, Evaluation, Dissemination, and Assessment
Cummins, J. (1989). Empowering minority students. Sacramento, CA:
California Association for Bilingual Education.
Duranti, A., & Ochs, E. (1988). Literacy instruction in a Samoan village. In
E. Ochs, Culture and language development: Language acquisition and
language socialization in a Samoan village (pp. 189-209). Cambridge:
Cambridge University Press.
Erickson, F., & Mohatt, G. (1982). Cultural organization of participation
structures in two classrooms of Indian students. In G. Spindler (Ed.),
Doing the ethnography of schooling (pp. 132-174). New York: Holt,
Rinehart and Winston.
Foster, M. (1982). Indigenous languages: Present and future. Language
and Society, 7, 7-14.
Halliday, M. (1975). Learning how to mean: Explorations in the
development of language. London: Edward Arnold.
Heath, S. B. (1983). Ways with words: Language, life and work in
communities and classrooms. Cambridge: Cambridge University Press.


Heath, S. B. (1989). The learner as culture member. In M. L. Rice &
R. L. Schiefelbusch (Eds.), The teachability of language (pp. 333-350).
Toronto: Paul H. Brookes.
Kagan, S. (1986). Cooperative learning and sociocultural factors in
schooling. In California State Department of Education (Ed.), Beyond
language: Social and cultural factors in schooling minority students. Los
Angeles: California State University, Evaluation, Dissemination, and
Assessment Center.
Krashen, S. (1985). The input hypothesis: Issues and implications. London:
Lightbown, P. (1989). Can they do it themselves? A comprehension-based
ESL course for young children. Paper presented at the Conference on
Comprehension-Based Second Language Teaching: Current Trends,
University of Ottawa, Canada.
MacLure, M., & French, P. (1981). A comparison of talk at home and at
school. In G. Wells (Ed.), Learning through interaction (pp. 205-239).
Cambridge: Cambridge University Press.
Ochs, E., & Schieffelin, B. B. (1984). Language acquisition and socializa-
tion: Three developmental stories and their implications. In
R. A. Schweder & R. A. Levine (Eds.), Culture theory: Essays on mind,
self, and emotion (pp. 276-322). New York: Cambridge University
Okpik, A. (1989). What it means to be an Inuk. Inuktitut, 70, 10.
Phillips, S. U. (1983). The invisible culture. New York: Longman.
Phillips, S. U. (1985). Interaction structured through talk and interaction
structured through “silence.” In D. Tannen & M. Saville-Troike (Eds.),
Perspectives on silence (pp. 205-214). Norwood, NJ: Ablex.
Priest, G. E. (1985). Aboriginal languages in Canada. Language and
Society, 15, 13-19.
Saville-Troike, M. (1985). The place of silence in an integrated theory of
communication. In D. Tannen & M. Saville-Troike (Eds.), Perspectives
on silence (pp. 3-20). Norwood, NJ: Ablex.
Scollon, R., & Scollon, S. (1981). Narrative, literacy, and face in
interethnic communication. Norwood, NJ: Ablex.
Skutnabb-Kangas, T. (1988). Multilingualism and the education of
minority children. In T. Skutnabb-Kangas & J. Cummins (Eds. ), From
shame to struggle (pp. 9-41). Clevedon, England: Multilingual Matters.
Taylor, D. M. (1990). Carving a new Inuit identity: The role of language in
the education of Inuit children in Arctic Quebec. Montreal, Canada:
Kativik School Board.
Taylor, D. M., Crago, M., & McAlpine, L. (1991). Education in Native
communities: Dilemmas around empowerment. Unpublished manu-
script, McGill University, Montreal, Canada.
Tharp, R., & Gallimore, R. (1988). Rousing minds to life. Cambridge:
Cambridge University Press.
Vogt, L., Jordan, C., & Tharp, R. (1987). Explaining school failure,
producing school success. Two cases. Anthropology and Education
Quarterly, 8, 276-286.


Weisner, T. S., Gallimore, R., & Jordan, C. (1988). Unpackaging cultural
effects on classroom learning: Native Hawaiian peer assistance and
child-generated activity. Anthropology and Education Quarterly, 19,
Wong Fillmore, L. (1983). The language learner as an individual:
Implications of research on individual differences for the ESL teacher.
In M. A. Clarke & J. Handscombe (Eds.), On TESOL ’82: Pacific
perspectives on language learning and teaching ( p p . 1 5 7 - 2 7 3 ) .
Washington, DC: TESOL.


TESOL QUARTERLY, Vol. 26, No. 3, Autumn 1992

Learning to Teach: Instructional

Actions and Decisions of Preservice
ESL Teachers
The Pennsylvania State University

This study examines the instructional actions and decisions of

preservice English as a second language teachers during their
initial teaching experiences. Six preservice ESL teachers viewed
videotapes of their own teaching and provided recall comments
that detailed their instructional decisions while teaching.
Transcriptions of videotaped lessons along with corresponding
recall comments were examined to determine the ways in which
these teachers perceived and responded to student input during
second language instruction, the instructional decisions they made,
and the prior knowledge they considered while making those
decisions. The results suggest that preservice ESL teachers’
instructional actions were directed by unexpected student
responses and the desire to maintain the flow of instructional
activities. Their instructional decisions were overwhelmingly
influenced by the need to ensure student understanding, to
increase student motivation and involvement, and to maintain
control over instructional management. The findings of this study
highlight the cognitive demands placed on preservice ESL
teachers and support the need for second language teacher
preparation programs to provide opportunities for preservice ESL
teachers to understand the dynamics of how they think and act as
they learn to teach.

Recent concern over the status of teacher education in the field of

second language teaching has focused on the lack of a theoretical
framework to serve as a basis for second language teacher
preparation programs (Bernhardt & Hammadou, 1987; Freeman,
1989; Richards, 1987; Richards & Nunan, 1990). While most teacher
educators agree that such programs need to provide opportunities
for preservice teachers to learn the skills and competencies of
effective second language teaching, the field has yet to establish a

clear understanding of what effective second language teaching is,
or how second language teachers actually learn to teach.
To date, much of the classroom-based research on second
language teaching has sought to describe effective teaching
behaviors, positive learner outcomes, and teacher-student
interactions that are believed to lead to successful second language
learning (Chaudron, 1988). Only recently have researchers begun to
recognize the importance of exploring the cognitive dimensions of
how second language teachers’ thoughts, judgments, and decisions
influence the nature of second language instruction. Freeman (1990)
suggests that to understand second language teachers’ instructional
behaviors, we must first understand the thoughts and judgments
that shape those behaviors. Moreover, Richards and Nunan (1990)
argue that to build a theory of second language teaching, research is
needed to discover the higher level concepts and thinking processes
that guide teachers during second language instruction.


Insight from first language educational research on teachers’ de-
cision making offers important promise for understanding the cog-
nitive dimensions of second language teaching. Clark and Peterson’s
(1986) comprehensive review of the research on teachers’ thought
processes purports that teachers are decision makers who process
information and act upon those decisions within complex
environments. Much of this research has focused on teachers’
interactive thoughts, judgments, and decisions as representations of
the cognitive information processing that occurs during classroom
instruction. Recent attempts to devise a conceptual model of
teachers’ cognitive information processing have led to the
characterization of teachers’ interactive decisions as reflecting well-
established instructional routines (Clark & Peterson, 1986; Joyce,
1978; Shavelson, 1983; Shavelson & Stern, 1981). This model
suggests that during instruction, teachers monitor the classroom,
seeking cues—usually in the form of student participation—to
determine whether a routine should proceed as planned. If student
behavior becomes unacceptable, teachers call up another routine
developed from previous experience or, if no routine is available,
teachers react spontaneously. Such a characterization suggests that
to reflect on an alternative decision drastically increases the
cognitive demands placed on teachers. Therefore, instructional
routines help minimize the sheer number of conscious decisions that
need to be made during instruction (Shavelson & Stern, 1981).


While the notion of instructional routines is generally accepted as
an accurate characterization of teachers’ interactive decisions, some
have questioned whether unexpected student behavior is the sole
antecedent of teachers’ instructional decisions (Clark & Peterson,
1986). To establish a more accurate model of teachers’ interactive
decisions, calls have been made for “research which is able to
describe the process whereby a given antecedent condition results
in an interactive decision by the teacher” (Clark & Peterson, 1986,
p. 277).
Attempts to gain a fuller understanding of the antecedent
conditions of teachers’ interactive decisions have focused on both
experienced and preservice teachers. Much of this research suggests
that experienced teachers possess a well-organized knowledge base
regarding the students and the classroom environment that enables
them to simplify, differentiate, and transform information during
instruction and make alternative choices without disrupting the
flow of instruction (Calderhead, 1981, 1983; Clark& Peterson, 1978;
Doyle, 1977; Morine & Vallance, 1975). Unlike experienced
teachers, preservice teachers have not developed a schema for
interpreting and coping with what goes on during instruction, nor
do they possess a repertoire of instructional routines upon which
they can rely (Calderhead, 1981; Doyle, 1977; Morine & Vallance,
1975). Instead, preservice teachers must consciously make an
overwhelming number of instructional decisions while teaching
and, consequently, tend to focus on unexpected student behavior
and maintaining the flow of activities during instruction (Fogarty,
Wang, & Creek, 1983; Housner & Griffey, 1983).
To enable preservice teachers to cope with the cognitive
demands of teaching, some researchers have proposed that
preservice teachers be trained in effective instructional decision-
making skills (Bishop & Whitfield, 1972; Sutcliffe & Whitfield,
1979). However, others suggest that teacher preparation programs
create opportunities for preservice teachers “to perceive, analyze,
and transform their perceptions of classroom events in ways similar
to those used by effective teachers” (Clark & Peterson, 1986,
p. 281).
Given the concern over the current status of teacher education in
second language teaching, research on teachers’ decision making
from L1 educational literature becomes an important starting point
for explorations into the cognitive dimensions of second language
teaching. Such research is essential for the preparation and training
of second language teachers. At present, the practice teaching
experience, or practicum, of most second language teacher
preparation programs functions without a clearly articulated


rationale of how to teach second language teachers to teach
(Freeman, 1989; Richards & Crookes, 1988). More often, the
practicum reflects the notion that once preservice teachers have
mastered the required subject-matter content (i.e., theories of
second language acquisition, linguistics, applied linguistics,
methodology) they will be able to use that knowledge to lead their
students to successful second language learning. Calls continue to
be made for explorations into what takes place during the
practicum in order to better understand the ways in which
preservice teachers conceptualize their initial teaching experiences,
to recognize the unique instructional considerations of second
language teaching and, finally, to determine the contribution that
such field experiences have on the professional development of
second language teachers (Richards & Crookes, 1988). To date, no
research of this type has been carried out.

The purpose of this study is thus twofold. First, by coming to un-
derstand the ways in which preservice teachers perceive, process,
and act upon information during their initial teaching experiences,
the field of second language teacher education stands to gain a great-
er understanding of the instructional processes that preparatory pro-
grams and preservice teaching experiences need to address. Second,
by recognizing the antecedent conditions that lead preservice
teachers to make instructional decisions, the field also gains a great-
er understanding of the cognitive dimensions of second language
This study examines the instructional actions and decisions of pre-
service ESL teachers during their initial teaching experiences. It
attempts to identify student behaviors and/or responses that lead
preservice ESL teachers to implement specific instructional actions.
In so doing, it examines the instructional decisions and information
teachers considered during interactive teaching. While it is
recognized that knowledge about preservice ESL teachers’
instructional actions and decisions can in no way provide a
comprehensive characterization of the complex conceptual
processes of second language teaching, it is, however, an important
starting point for understanding how second language teachers
conceptualize their initial teaching experiences and, in essence,
learn to teach. Therefore, this study explored the following research
1. What student performance cues do preservice ESL teachers
utilize during second language instruction?


2. What instructional actions do preservice ESL teachers imple-
ment in response to certain types of student performance cues?
3. What instructional decisions and prior knowledge do preservice
ESL teachers consider during second language instruction?

Over the course of two semesters, 6 preservice ESL teachers—5
female and 1 male—enrolled in a practicum as part of a 2-year
master’s degree program in TESOL voluntarily partiticipated in this
study. Four of the teachers were native speakers of English (3 of
U.S. English and 1 of British English), and 2 possessed nativelike
fluency in English but spoke either Japanese or Spanish as a native
language. Three of the teachers had no previous teaching
experience, while 3 others had limited teaching experience (no
more than 2 years) in elementary or adult-level English as a foreign
language instruction in their native countries. During their
preservice teaching experience they were concurrently enrolled in 6
credit hours of course work in theory and practice in TESOL. They
were teaching in a one-semester practicum in which they were
assigned to a cooperating teacher in either a precollege intensive
English language program or an adult-level ESL community
education program.

The sources of data for this study included (a) videotaped
observations of actual classroom instruction, (b) teachers’ stimulus
recall reports of the instructional decisions they recalled making and
the prior knowledge they recalled considering as they viewed their
own videotaped instruction, and (c) retrospectives written by the
teachers on what they perceived to have had the greatest influence
on their decisions during classroom instruction.
Collection of teaching samples. At midsemester, each preservice
ESL teacher was videotaped teaching one regularly scheduled 55-
minute lesson. Four of the lessons were of intermediate-level ESL
reading and writing classes and 2 were of intermediate-level ESL
speaking and listening classes. Prestudy observations had been
conducted, and teachers and students were accustomed to the
equipment and to an observer being in the room.
Collection of stimulus recall reports. Following the procedures
established by Shavelson and Stern (1981), teachers participated in


stimulus recall procedures. To familiarize the teachers with these
procedures, a demonstration videotaped lesson from a pilot study
was shown and sample audiotaped stimulus recall reports were
played at predetermined decision points identified on the
videotape. Four to six segments of the demonstration videotape
were played for each teacher. The teachers were given ample
opportunity to ask questions about the stimulus recall procedures.
Teachers were then asked to view their own videotaped lessons
and to stop the tape at any points where they recollected their
thoughts and decisions while teaching. They were asked to focus
their attention on their instructional decisions and to explain why
they chose to make these decisions. All stimulus recall reports were
audiotaped, and videotape footage was noted. At a later date, the
stimulus recall reports were transcribed, and tape footage was used
to match stimulus recall reports to corresponding points in the
transcribed videotaped lessons.
Collection of written retrospectives. Additional descriptive data were
collected to ascertain the teachers’ perceptions of factors
influencing their instructional decisions. Immediately after teaching
their lessons, teachers wrote a one-page written retrospective in
which they responded to two questions: What was your greatest
concern during this lesson? and What influenced your instructional
decisions most? A second written retrospective, based on the same
questions, was written after the teachers viewed their own
videotaped lessons and completed the stimulus recall procedures.

Data Analysis
The data collected in this study was interpreted using a modified
version of the classification scheme created by Fogarty et al. (1983).
The intent of their classification scheme was to establish “a
methodology for describing competent classroom performance and
analyzing components of competent classroom decision making”
(p. 22) covering four dimensions of interactive teaching. These
included descriptive categories under Student Performance Cues
(behaviors and/or responses exhibited by students during
instruction), Instructional Actions (behaviors and/or responses
exhibited by teachers during instruction) that teachers implemented
in response to Student Performance Cues, Instructional Goals, and
Prior Knowledge that teachers recalled considering during
An initial analysis of the data revealed that it was necessary to
make several modifications in order to tailor the original class-
ification scheme to second language instructional contexts. The


modified classification scheme used to analyze the data collected in
this study is presented in the Appendix.
Analysis of videotaped data. Transcribed videotaped data were
analyzed by identifying and coding (a) the type of student
performance cue(s) and (b) the type of instructional action(s)
implemented by the teacher. The researcher and one rater, trained
to use the modifid classification scheme on data from a pilot study,
coded four 10-minute videotaped teaching segments. Interrater
agreement of 83% on student performance cues and 76% on
instructional actions were obtained with an overall average
agreement of 80%. The remainder of the data was coded by the
researcher. As the data were analyzed, the sequence of instructional
actions implemented in response to specific types of student
performance cues was noted. That is, the analysis revealed not only
the frequency with which different types of student performance
cues and instructional actions occurred but also the sequence of
instructional actions that followed specific student performance
Analysis of standards recall data. The researcher and the same
trained rater classified all 6 stimulus recall reports according to the
instructional decisions and prior knowledge categories, with an
average agreement of 90%. The frequency with which instructional
decisions and prior knowledge comments occurred was tabulated,
and the types of instructional decisions and prior knowledge
comments that followed different types of instructional actions
were then identified.
Analysis of written retrospectives. The written retrospectives were
examined to ascertain the teachers’ stated focus of concern during
instruction and their perceptions of what influenced their
instructional decisions. Their perceptions were categorized by
means of inductive analysis procedures (Bogdan & Biklen, 1982) by
which descriptive categories were identified in the teachers’ written
responses regarding their greatest concern during instruction and
what they perceived as having the greatest influence on their
instructional decisions.

The results of these analyses are presented in terms of the type
and frequency of (a) teachers’ response to student performance
cues, (b) teachers’ instructional actions in response to student
performance cues, and (c) teachers’ instructional decisions and


consideration of prior knowledge about those instructional
decisions. Excerpts from the transcripts of videotaped instruction
along with corresponding stimulus recall reports are included to
highlight the instructional actions and decisions of preservice ESL
teachers during second language instruction.

Frequencies of Student Performance Cues

Table 1 presents the frequencies of student performance cues that
were utilized by teachers during second language instruction. As
one might expect, the results indicate some differences in the
frequency with which individual teachers utilized different
categories of student performance cues. Teachers 2, 4, and 6
utilized fewer student performance cues than Teachers 1, 3, and 5.
In addition, teacher-elicited responses accounted for almost half of
each teacher’s overall utilization of student performance cues,
except for Teachers 4 and 6. For these teachers, teacher-elicited
responses accounted for only one third of their overall utilization of
student performance cues.

Frequencies of Student Performance Cues Responded to by Preservice ESL Teachers

A significant difference was found in the frequencies of student

performance cues across all teachers, x 2(3, N = 223) = 81.87,
p <.01. Overall, teachers tended to utilize more student perfor-
mance cues that occurred after being elicited by the teacher
(Elicited Response, 48%) than those in which no response was made
after a direct elicitation (Deficient Response, 28%). Likewise,
teachers tended to utilize elicited and deficient responses more than
spontaneous student-initiated questions, comments, or behaviors


(Initiation, 15%). Except for Teacher 6, teachers utilized student
responses that were incorrect, insufficient, or superfluous (Error,
9%) only on occasion.
The categories of student performance cues may be argued to
represent approximately independent events: The occurrence of
one type of student performance cue did not substantially affect the
probability of the occurrence of any other. This became especially
important for the error category, since errors represented responses,
either initiated or elicited, that were incorrect, insufficient, or
superfluous. Errors were judged according to the accuracy of the
student’s response, the teacher’s signaling that the response was
incorrect, and/or reference to the inaccuracy of the student’s
response during the stimulus recall report; errors therefore did not
overlap with any other category.
However, teachers’ responses to student performance cues may
not have been independent, since individual teachers may have
been predisposed to utilize particular categories and/or sequences
of student performance cues. Yet, despite this concern, the
frequencies reported in Table 1 indicate no striking differences
between teachers for each type of student performance cue. In fact,
only Teachers 4 and 6 were found to have deviated from the general
pattern mentioned above: Teacher 4 utilized slightly more deficient
responses, and Teacher 6 utilized slightly more errors.

Frequencies of Teacher Instructional Actions

Table 2 presents the frequencies of instructional actions that
teachers implemented during second language instruction. Again,
the results indicate some differences in the frequency with which
individual teachers implemented different categories of instruc-
tional actions. Teacher 6 contributed fewer instructional actions
overall; Teacher 3 contributed the most.
When examining general patterns, a significant difference was
found in the frequencies of instructional actions across all teachers,
x 2(5, N = 356) = 131.98, p <.01. Overall, teachers’ instructional
actions were fairly equally distributed across (a) providing an
explanation of a concept or procedure (Explains Concept/
Procedure, 28%), (b) questioning student knowledge of a concept,
topic, or procedure (Checks Knowledge, 24%), and (c) encouraging
and incorporating student initiations (Elicits & Incorporates Input,
24%). Teachers gave less instructional attention to providing
students with information regarding their performance (Gives
Feedback, 18%). They rarely, if at all, implemented instructional
actions involving directing student attention or encouraging


persistence (Focuses Attention/Effort, 3%), or applying new
concepts, extending instruction, or planning future instruction
(Applies, Extends, Plans, 3%).
Once again, while individual teachers may be predisposed to
implement particular categories and/or sequences of instructional
actions, the frequencies reported in Table 2 indicate no striking
differences between teachers for each type of instructional action.
Teachers 1, 2, 4, and 5 were found to support the general pattern
mentioned above, while Teachers 3 and 6 deviated slightly—they
were more likely to encourage and incorporate student initiations.
An additional set of chi-square tests was used to examine
differences between teachers with varying amounts of teaching
experience. Teachers 2, 3, and 5, who had no previous teaching
experience, were compared as a group to Teachers 1, 4, and 6, who
had fewer than 2 years of teaching experience. No significant
differences were found between the two groups of teachers for the
utilization of student performance cues, x 2(3, N = 356) = 0.31, ns,
or for the implementation of instructional actions, x 2 ( 5 ,
N = 356) = 0.75, ns. This suggests that no systematic differences
were found between teachers who had a limited amount of teaching
experience and those who had none.

Instructional Actions Implemented in Response to Student

Performance Cues
The instructional actions implemented in response to student
performance cues are illustrated through the following excerpts of


transcribed videotaped instruction along with corresponding
stimulus recall reports. The following excerpts show the ways in
which these preservice ESL teachers carried out instructional
actions and made instructional decisions in response to deficient,
initiated, elicited, or error student performance cues.
Deficient responses. When no behavior or response was made after
a direct elicitation, teachers generally checked student knowledge
and explained a concept or procedure. In Excerpt 1, the teacher re-
sponded to a student’s deficient response by providing an
additional explanation of the assignment while continually
questioning the student’s knowledge. In addition, this teacher
recalled considering the student’s language skill, academic ability,
and probable knowledge when deciding to repeat the directions in
order to ensure student understanding.
Initiations. When a spontaneous question, comment, or behavior
was initiated, teachers tended either to elicit and incorporate
student initiations or to explain a concept or procedure. An in-depth
analysis of the types of instructional actions implemented in
response to student initiations indicated that teachers’ instructional
actions differed if the initiation was a question or a comment. If the
student initiation was a question, teachers tended to respond with
an explanation. For example, in Excerpt 2, the student asks a
question about the procedures for unscrambling a comic strip, and
the teacher provides an explanation of the procedures for
completing the task.
However, if the student initiation was a comment or contained
additional information, teachers tended to incorporate the student’s
remarks and then elicit more information. In Excerpt 3, this teacher
recalled deciding to encourage elicited responses as a means of
maintaining student motivation and involvement, but perceived one
student’s continued initiations as interrupting the flow of the
instructional activity. She recalled then questioning the appropriate-
ness of her own teaching strategy and therefore deciding not to
utilize the student’s initiations but to continue with the instructional
activity as planned.
Elicited responses. Student responses to an elicitation by the teacher
were generally followed by instructional actions in which teachers
tended to either elicit and incorporate student input or give feed-
back on student performance. In Excerpt 4, the teacher recalled
perceiving the student’s elicited responses as an indication that she
was on task and engaged in the instructional activity. After
considering the student’s affective needs and language skills, the
teacher decided not to correct the response, but paraphrase it to
ensure that the other students understood.


Errors. When students produced incorrect, insufficient, or superflu-
ous responses, teachers tended to check student knowledge,
provide an explanation, and give feedback. In Excerpt 5, the teach-
er recalled perceiving the student’s continued errors as an indication
of misunderstanding and, therefore, continued to provide feed-
back, explain the concept, and check the student’s knowledge.
However, she also recalled her perceived ineffectiveness with these
instructional actions, and commented on her mounting concern
with not only the appropriateness of her teaching strategy, but also
her inability to maintain control over the instructional activity.
These descriptive data are supported by the data presented in
Table 3 that report the frequencies of instructional actions imple-
mented by teachers in response to specific categories of student
performance cues. A two-way chi-square test for association re-
vealed a significant association between instructional actions and
student performance cues, x 2(12, N = 576) = 134.74, p <.01.
Moreover, the ratios reported in Table 3 support the general pat-
terns mentioned above. Teachers tended to implement approxi-
mately two instructional actions (2.05) in response to a deficient re-
sponse, generally by questioning student knowledge (Checks
Knowledge, 40%) and providing an explanation (Explains Concept/
Procedure, 38%).

Frequencies of Instructional Actions Implemented by Preservice ESL Teachers
in Response to Specific Categories of Student Performance Cues


In addition, they tended to implement, on average, approximately
one instructional action (1.15) in response to an initiation, either
utilizing student-initiated questions by providing an explanation
(Explains Concept/Procedure, 29%) or utilizing student-initiated
comments by encouraging and incorporating student input (Elicits
& Incorporates Input, 34%). Moreover, they tended to implement
only one instructional action (1.09) in response to an elicited
response, either by encouraging and incorporating student input
(Elicits & Incorporates Input, 42%) or by providing feedback (Gives
Feedback, 36%).
On the other hand, teachers tended to implement, on average,
more than three instructional actions (3.65) in response to an error.
Generally by implementing a sequence of questioning student
knowledge (Checks Knowledge, 34%), providing an explanation
(Explains Concept/Procedure, 33%), and giving feedback (Gives
Feedback, 20%). Overall, these ratios provide evidence that teachers
tended to implement twice and sometimes three times as many in-
structional actions in response to incorrect student performance

Instructional Decisions and Prior Knowledge

Table 4 reports the frequencies of Instructional Decisions and
consideration of Prior Knowledge recalled by preservice ESL
teachers. As the table shows, teachers most frequently recalled
making instructional decisions to increase student understanding
(Student Understanding, 37%) or to increase student motivation and
involvement (Student Motivation & Involvement, 17%. To a lesser
extent, teachers recalled making instructional decisions with
consideration for the flow of the instructional activity (Instructional
Management, 15%). Only on occasion did teachers recall making
instructional decisions while considering knowledge related to (a)
the level of student language skill and ability (Student Language
Skill & Ability, 8%), (b) the student’s social, affective, and
developmental needs (Student Social Needs, 6%), (c) the content of
the subject matter (Subject Matter Content, 8%), or (d) the sequence
of the lesson content and instructional materials (Curriculum
Integration, 9%).
Teachers frequently recalled that when making instructional
decisions, they considered the appropriateness of a particular
teaching strategy (Appropriateness of Teaching Strategy, 65%), and
the student’s language skill, academic ability, and probable
knowledge about the content (Student Language Skill, Academic
Ability, & Probable Knowledge, 19%). To a much lesser extent, they


considered knowledge related to the lesson content (Important
Content, 18%), or instructional principles and/or specific advice or
suggestions from the cooperating teacher (Pedagogical Principles,

Frequencies of Instructional Decisions and Prior Knowledge
Reported by Preservice ESL Teachers

Since stimulus recall reports were not collected at predetermined

points in the transcribed videotaped lessons but at instances
selected spontaneously by the teachers, each instructional action
could not be accounted for in the stimulus recall reports. However,
the matching of instructional actions and stimulus recall reports
indicates that teachers recalled making instructional decisions
primarily to encourage student understanding, motivation and
involvement, and instructional management in response to
deficient, initiated, elicited, or error student performance cues.
Other information considered by these teachers while making
instructional decisions focused on the appropriateness of their
teaching strategies and the language skills and academic abilities of
their students.


Teachers’ Written Retrospectives
The analysis of initial and final written retrospectives provides
confirmatory evidence of these teachers’ concerns with student
understanding, motivation and involvement, and the appropriate-
ness of their teaching strategies. These data revealed three
categories that represented teachers’ perceptions of their greatest
concern during the lesson and what influenced their instructional
decisions most. The first category, Ensuring Student Understand-
ing, included written comments by all 6 teachers that their greatest
concern during classroom instruction was to ensure that their
students understood their instructional activities. One teacher
1. My greatest concern was getting the students to understand the
vocabulary as well as the connections between family relationships.
These concerns were closely tied to the second category,
Maintaining Student Motivation and Involvement, and included
written comments reflecting instructional decisions to maintain
student motivation and involvement in the instructional activity.
This same teacher wrote:
2. I made the decision to let them practice the words together because
they needed to get involved in the lesson.
The third category, Appropriateness of Teaching Strategy, included
self-evaluation comments that all 6 teachers made about the
appropriateness of their instructional behavior. One teacher wrote:
3. I was worried that I was doing all the talking so I decided to put them
in small groups.
Overall, all 6 teachers recalled that their instructional decisions
were most influenced by whether or not the students understood
the instructional activities and whether they were actively involved
in the lesson. In addition, they questioned the appropriateness of
their teaching strategies and recalled deciding to shift activities on
the basis of student understanding and involvement. As another
teacher wrote:
4. I tried to key into the students’ indirect ways of showing lack of
understanding, as well as their direct ways, i.e., asking questions. This
appeared to influence my instructional decisions most, because
whenever they appeared confused, I had to come up with a new
strategy. In some cases, I decided to rephrase what I said, but in
others I had to come up with a different way of presenting the


These data suggest that issues related to student understanding,
motivation and involvement, as well as concerns over the appropri-
ateness of teaching strategies were paramount in the minds of these
teachers as they reflected on their classroom instruction.

The results of this study illustrate the ways in which these
preservice ESL teachers utilized student performance cues, the
instructional actions they implemented in response to student
performance cues, and the instructional decisions they made and
other information they considered during second language
instruction. Overall, these teachers utilized more student responses
that were teacher elicited than student initiated. They paid more
instructional attention to errors and deficient responses by
implementing a cycle of instructional actions that included checking
knowledge, providing an explanation, and giving feedback. They
paid less instructional attention to elicited and initiated responses by
either encouraging and incorporating student initiations or giving
feedback. When faced with unexpected student responses such as
deficient responses, continued student initiations, or errors, they
relied on a limited number of instructional actions, and often
perceived such responses as obstacles to maintaining control over
the instructional activities. Moreover, their instructional decisions
were overwhelmingly concerned with ensuring student understand-
ing, motivation and involvement, and instructional management. At
the same time, they continually questioned the appropriateness of
their teaching strategies, appeared to judge their own effectiveness
in relation to student responses and, therefore, sought confirmation
throughout the instructional process that students understood and
were actively engaged.
These findings support previously held characterizations in L1
educational literature that unexpected student behavior is the most
prominent antecedent condition of preservice teachers’ instruction-
al behavior. In addition, these findings support the notion that pre-
service teachers tend to rely on a limited number of instructional
routines, and are primarily concerned with maintaining the flow of
instructional activities (Clark & Peterson, 1986). These similarities
imply, at least to some extent, that the cognitive demands of
learning to teach may cut across instructional contexts and that the
cognitive strategies utilized by preservice teachers during their
initial teaching experiences may be quite similar. Such similarities
suggest that the field of second language teacher education can, and


should, look to L1 educational literature as it continues to explore
the cognitive dimensions of second language teaching.
Yet despite these similarities, the results of this study also suggest
patterns of instructional actions and decisions that may be unique to
second language teaching. Most noticeable was the way in which
these teachers interpreted and responded to deficient responses
during second language instruction. They recalled perceiving
deficient responses as indicating a lack of student understanding
due to limited English proficiency and, therefore, being more likely
to respond by providing an explanation and/or checking student
knowledge. While no direct comparisons can be made to the
Fogarty et al. (1983) study, it is interesting to note that preservice
elementary school teachers were more likely to respond to deficient
responses by providing feedback on a student’s performance or
focusing a student’s attention and effort. This apparent differential
behavior by preservice elementary school and ESL teachers may be
the result of differing interpretations of deficient responses. That is,
in the Fogarty et al. (1983) study, preservice elementary school
teachers were more likely to perceive deficient responses as
reflecting a lack of student attention or effort, whereas in the
present study, preservice ESL teachers were more likely to perceive
deficient responses as a lack of student understanding due to limited
English proficiency.
A similar pattern was found in the ways in which preservice ESL
teachers interpreted and responded to student errors. These
teachers recalled perceiving student errors as reflecting a lack of
student understanding due to limited English proficiency and,
therefore, tended to repeat a cycle of checking knowledge,
providing an explanation, and giving feedback. In contrast, pre-
service elementary school teachers in the Fogarty et al. (1983) study
were more likely to perceive errors as indicating a lack of student
attention and, therefore, responded by providing feedback on the
student’s performance or by focusing the student’s attention.
A comparison of elementary school and ESL instructional
contexts remains speculative at best; further research is warranted
in order to determine the extent to which differing interpretations
of student performance cues result in differential patterns of
instructional actions by preservice teachers in different instructional
contexts. To date, no research of this type has been conducted. If,
in fact, future research can establish that differing interpretations of
student responses result in differential patterns of instructional
actions, then the type of student response, or what students actually
say or do, may not be the overriding consideration that shapes
teachers’ instructional behaviors. Instead, teachers’ interpretations


of student responses may be the determining factor in understand-
ing teachers’ instructional behaviors. Therefore, while unexpected
student responses are likely to occur in most instructional contexts,
how teachers interpret and cope with such responses may depend
upon instructional considerations that are unique to that instruc-
tional context.
In this study, the importance of ensuring student understanding in
the face of limited English proficiency appeared to be an overriding
consideration that shaped the instructional actions and decisions of
these preservice ESL teachers. Not only did this appear to shape
their instructional decisions, but it also shaped their perceptions of
the effectiveness of their own instructional behavior.
Moreover, the extent to which ESL teachers are forced to
interpret and respond to deficient responses may be a unique
instructional consideration in second language teaching; that is, due
to limited English proficiency, ESL students may be more likely to
remain silent than risk making an error during second language
instruction. Therefore, deficient responses may pose difficulties for
preservice ESL teachers because they are forced to interpret and
respond to silence frequently during second language instruction.
Overall, the patterns of instructional actions and decisions found in
this study suggest that these preservice ESL teachers did not possess
a clearly defined cognitive schema for interpreting unexpected
student responses. Moreover, they lacked a repertoire of instruc-
tional routines for coping with such responses. Therefore, they
tended to repeat an often ineffectual cycle of instructional actions in
response to unexpected student responses until concerns over in-
structional management took precedence.
The results of this study also suggest that preservice ESL teachers’
utilization of student initiations may pose unique concerns for
second language instructional contexts. Given the importance of
“authenticity” (van Lier, 1991; Widdowson, 1979) and self-selected
initiations (Ellis, 1990; van Lier, 1988) in instructed second language
learning, it becomes essential for second language learners to be
given the opportunity to use the target language for reasonably
authentic communicative purposes. In this study, preservice ESL
teachers tended to perceive continued student initiations as off-task
behavior and a threat to instructional management. In such
instances, they were more likely to ignore student initiations in favor
of maintaining control over the instructional activity. This sort of
instructional behavior may in fact limit second language learners’
opportunities for authentic and self-selected use of the target
language during classroom instruction. One of the many challenges
for future teacher preparation programs is to enable preservice ESL


teachers to encourage and incorporate student initiations into their
instructional activities without perceiving such initiations as a threat
to instructional management.
While this study has begun to provide insights into the
instructional actions and decisions of preservice ESL teachers, it
also points to the need for further research on the cognitive
dimensions of second language teaching. In order to understand the
extent to which differing interpretations of student performance
cues result in differential instructional actions by preservice
teachers, comparison studies must be carried out between ESL and
non-ESL instructional contexts. Such studies should provide more
definitive evidence of the unique instructional considerations of
second language teaching that may shape the instructional actions
and decisions of second language teachers. Future research must
also explore differences in the instructional actions and decisions of
both experienced and preservice second language teachers. Such
research should provide insight into differences in the cognitive
schema that experienced and preservice second language teachers
rely on to interpret information during second language instruction.
If differential patterns of instructional actions and decisions can be
established, such findings may become the basis upon which
teacher preparation programs can provide opportunities for pre-
service teachers to develop a cognitive schema for processing
information during second language instruction.
Future research must also recognize that the characterization of
preservice ESL teachers’ instructional actions and decisions
presented in this study is in no way a comprehensive description of
the complex conceptual processes of second language teaching. As
in the first language educational literature, future research on
second language teaching must include explorations into the sub-
areas of teacher cognition, including (a) teacher thinking prior to
instruction (teachers’ instructional planning), (b) the cognitive
information processing that occurs during instruction (teachers’
interactive thoughts and decisions), and (c) the cognitive filter
through which teachers’ instructional plans, judgments, and
decisions are made (teachers’ theoretical beliefs). Such work must
begin descriptively, across a broad spectrum of second language
instructional contexts, by researchers and teachers for whom the
study of teacher cognition is an essential goal for the field of second
language teacher education.

Implications for Second Language Teacher Education Programs

Teacher preparation programs concerned with enabling pre-
service teachers to cope with the cognitive demands of second


language teaching can create opportunities for preservice teachers
to become conscious of the ways in which they perceive and act
upon information during second language instruction. The goal of
such opportunities is for preservice teachers to understand not only
what they do during second language instruction but also why they
do it, that is, how their thoughts and judgments about what occurs
during second language instruction shape their instructional actions.
One way to accomplish this goal is through a series of videotaped
observations about which preservice teachers recall their
instructional thoughts and judgments as they watch themselves
teach. This sort of self-analysis and reflection can enable preservice
teachers to recognize how they interpret and respond to the variety
of student performance cues that occur during second language
instruction. In consultation with a supervising teacher or their peers,
preservice teachers can begin to consider alternative instructional
actions for dealing with unexpected student responses and issues
related to instructional management.
A second important way for preservice teachers to understand
the cognitive demands of second language teaching is for them to
be given opportunities to study experienced second language
teachers as they teach. This entails more than simple classroom
observations. It involves training preservice teachers to recognize
the instructional routines that experienced teachers rely on to lessen
the number of conscious decisions necessary during second
language instruction. It also involves providing opportunities for
preservice teachers to recognize the ways in which experienced
teachers interpret and respond to unexpected student responses.
This can be accomplished by requesting that experienced teachers
videotape their own teaching and provide a commentary of their
instructional decisions while teaching. Debriefing sessions between
preservice and experienced teachers can highlight the unique
considerations that influence teachers’ instructional actions and
decisions during second language teaching.

Limitations of the Study

These findings must be interpreted in light of several limitations
of the study. The small number of subjects provides rich descrip-
tions of individual cases, but the specific findings may not be gen-
eralizable to the broader population of preservice ESL teachers. In
addition, all of the instructional observations took place in precol-
lege or adult-level ESL classroom hence, similar studies conducted
at different instructional levels are necessary. Finally, the stimulus
recall technique utilized in this study may have generated


teachers’ thinking that occurred during the time when the videotape
was viewed; the thoughts expressed may not represent a pure mea-
sure of actual interactive thinking (Ericsson & Simon, 1980; Nisbett
& Wilson, 1977). Thus, threats to the validity of retrospective data
must be considered when interpreting the findings of this study.

Despite these limitations, the results of this study point in
promising directions. Among these are the need for continued
exploration into the cognitive dimensions of second language
teaching. Further research along the lines suggested by the present
study may eventually provide a broader understanding of not only
the processes of second language teaching but how second language
teachers learn to teach. This study represents a necessary first step
toward understanding the initial teaching experiences of preservice
second language teachers.

I gratefully acknowledge Patricia A. Dunkel, David Nunan, and Dennis S. Gouran
for their incisive and substantive suggestions on earlier drafts of this paper.

Karen E. Johnson is Assistant Professor of Speech Communication at The
Pennsylvania State University, where she teaches courses in the MA TESOL
program. Her research focuses on teacher cognition in second language
instructional contexts and communication in multicultural classrooms.

Bernhardt, E., & Hammadou, J. (1987). A decade of research in foreign
language teacher education. The Modern Language Journal, 71, 291-299.
Bishop, A. J., & Whitfield, R. C. (1972). Situations in teaching. London:
Bogdan, R., & Biklen, S. K. (1982). Qualitative research for education: An
introduction to theory and methods. Boston, MA: Allyn & Bacon.
Calderhead, J. (1981). A psychological approach to research on teachers’
classroom decision making. British Educational Research Journal, 7,


Calderhead, J. (1983, April). Research into teachers’ and student teachers’
cognition: Exploring the nature of classroom practices. Paper presented
at the annual meeting of the American Educational Research
Association, Montreal, Canada.
Chaudron, C. (1988). Second language classrooms: Research on teaching
and learning. NY: Cambridge University Press.
Clark, C. M., & Peterson, P. L. (1978). Teachers’ reports of the cognitive
processes during teaching. American Educational Research Journal, 15,
Clark, C. M., & Peterson, P. L. (1986). Teachers’ thought processes. In
M. C. Wittrock (Ed.), Handbook of research on teaching (pp. 255-296).
NY: Macmillan.
Doyle, W. (1977). Learning the classroom environment An ecological
analysis. ]ournal of Teacher Education, 28, 51-55.
Ellis, R. (1990). Instructed second language acquisition: Learning in the
classroom. Cambridge, MA: Basil Blackwell.
Ericsson, K. A., & Simon, H. A. (1980). Verbal reports as data. Psycholog-
ical Review, 87, 215-251.
Fogarty, J. L., Wang, M. C., & Creek, R. (1983). A descriptive study of
experienced and novice teachers’ interactive thoughts and actions.
Journal of Educational Research, 77, 22-32.
Freeman, D. (1989). Teacher training, development and decision making:
A model of teaching and related strategies for language teacher educa-
tion. TESOL Quarterly, 23 (l), 27-45.
Freeman, D. (1990). ‘Thoughtful work’: Reconceptualizing the research
literature on teacher thinking. Unpublished monograph. School for
International Training, Brattleboro, VT.
Housner, L. D., & Griffey, D. C. (1983, April). Teacher cognition:
Differences in planning and interactive decision making between
experienced and inexperienced teachers. Paper presented at the annual
meeting of the American Educational Research Association, Montreal,
Joyce, B. (1978). Toward a theory of information processing in teaching.
Educational Research Quarterly, 3, 66-67.
Morine, G., & Vallance, E. (1975). Special study B: A study of teacher and
pupil perceptions of classroom interaction (Tech. Rep. No. 75-11-6) San
Francisco: Far West Laboratory.
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know:
Verbal reports on mental processes. Psychological Review, 84, 231-259.
Richards, J. C. (1987). The dilemma of teacher preparation in TESOL.
TESOL Quarterly, 21 (1), 9-27.
Richards, J. C., & Crookes, G. (1988). The practicum in TESOL. TESOL
Quarterly, 22 (l), 9-27.
Richards, J. C., & Nunan, D. (1990). Second language teacher education,
Cambridge: Cambridge University Press.
Shavelson, R. J. (1983). Review of research on teachers’ pedagogical
judgments, plans and decisions. Elementary School journal, 83, 392-413.


Shavelson, R. J., & Stern, P. (1981). Research on teachers’ pedagogical
thoughts, judgments, decisions, and behavior. Review of Educational
Research, 51, 455-498.
Sutcliffe, J., & Whitfield, R. (1979). Classroom-based teaching decisions.
In J. Eggleston (Ed.), Teacher decision making in the classroom: A
collection of papers. London: Routledge & Kegan Paul.
van Lier, L. (1988). The classroom and the language learner: Ethnography
and second-language classroom research. London: Longman.
van Lier, L. (1991). Inside the classroom: Learning processes and teaching
procedures. Applied Language Learning, 2, 29-68.
Widdowson, H. G. (1979). Teaching language as communication. Oxford:
Oxford University Press.


Modified Classification Scheme for Four Dimensions of Interactive Teaching
in Second Language Instructional Contexts


TESOL QUARTERLY, Vol. 26, No. 3, Autumn 1992

Avoiding Obstacles to Student

Comprehension of Test Questions
Tel Aviv University

We present preliminary criteria to consider in the writing of

English for academic purposes (EAP) reading comprehension
tests. Such criteria are necessary as a frame of reference for a large
and diverse teaching staff. The criteria pertain to the test as a
whole as well as to the form of individual questions. The choice of
items derives from our classroom and test-writing experience and
reflects the comprehension problems that students encounter with
test questions. These problems may indeed be stumbling blocks
for the student because they constitute processing constraints.
Pragmatic theory can provide a theoretical framework to explain
these processing constraints.

In a university setting where the medium of instruction and

communication is not English and where undergraduate students
are required to read content-course bibliographies in English, the
goal of the foreign language reading comprehension course is to
provide students with the skills and strategies needed to meet this
requirement. In such a setting, the English reading course is often
compulsory. This is indeed the case at Tel Aviv University, where
courses are provided for thousands of students in various content
areas and at various proficiency levels. Such conditions require
extensive and varied testing in order to maintain student motivation
and cooperation and to ensure some degree of standardization
among courses with a large teaching staff (see Wexler & Kirschner,
As coordinators of two English for academic purposes (EAP)
courses that service hundreds of students, we found that many of
the tests submitted by the teachers did not meet our expectations,
either on the basis of classroom experience or theoretical consider-
ations. We suspected that this situation might well be symptomatic
of more serious underlying problems. Perhaps it indicated that as
course coordinators, we had not defined the course objectives clear-
ly enough or that the staff had not effectively applied these

objectives in the classroom. We therefore attempted to formulate a
unified set of criteria for writing and evaluating tests. We hoped
that the adoption of such criteria by the teaching staff would have
a positive backwash effect on classroom activities (Carroll, 1980;
Hughes, 1989; Krashen, 1982). Preliminary criteria are presented
We found a lack of specific guidelines in the literature for
formulating or assessing items for reading comprehension tests at a
high-proficiency level, where authentic academic texts are used.
Furthermore, the recommendations that we did find (e.g., Hughes,
1989; Nuttall, 1982; Perkins & Jones, 1985; Shohamy, 1985) were
not, on the whole, accompanied by a comprehensive theoretical
framework which could be relayed to teachers, thus eliciting their
We anchored our work by responding to the following questions:
What cognitive and linguistic constraints on communication apply
to the testing situation? What constitutes a good reading
comprehension test for academic texts? What are the possible
explanations as to why a particular item is good or, conversely, what
makes a question problematic? How can better test writing be
facilitated among teachers in order to ultimately improve the
quality of classroom teaching?
The purpose of this paper is to present criteria, along with a the-
oretical rationale, to be considered by the test writer during the test-
writing process. These criteria pertain to the global form of the test
as well as to the form of specific questions. At the practical level, we
hope that such a list of considerations or criteria will facilitate the
test-writing process such that teachers and course coordinators
become more aware of potential difficulties for the student. Such
awareness will ultimately improve the final product, which should
have a positive effect on the quality of classroom teaching.
Reading comprehension tests used in our courses comprise full-
length, unedited academic texts. The accompanying questions
attempt to assess student ability to process a text on macro- and
microlevels. Both the choice of text and the tasks implicit in the test
questions should reflect tasks students are expected to perform in
their academic studies. Thus, students can use dictionaries during
the test. Spolsky (1985) claims that language tests by definition are
not authentic. Although we acknowledge this claim, practical
constraints dictate that we must administer a reading comprehen-
sion test in the form of a text plus questions. We thereby attempt to
approximate the pedagogical goal of authenticity widely discussed
in the literature (Bachman, 1990; Carroll, 1985; Jones, 1985; Oller,
1979, 1980).


Strict adherence to the goal of authenticity should dictate the use
of L1 in both the test items and the students’ responses (Hughes,
1989; Nuttall, 1982; Shohamy, 1984, 1985) because in academic
settings such as ours, content-course bibliographies are read in L2
and discussed and tested in L1. However, due to practical
constraints (e.g., staff members who are not completely proficient
in the students’ Ll), the test questions and the students’ written
responses are in English. Student responses are evaluated on the
basis of the students’ apparent comprehension of the ideas in the
text and not for syntax, spelling, and so forth.


In order to facilitate more authentic language testing, Bachman
(1990) calls for “a theoretical framework that provides a coherent
rationale for the identification and definition of critical features of
language use, whether this be in test or nontest contexts” (p. 315).
(Seliger, 1985, is a step in that direction.) Accordingly, we posit the
1. Test questions should constitute a communicative interchange
between the test writer and the test taker.
2. Test questions should be as easy for test takers to process as
3. It is the test writer’s task to define, identify, and subsequently
remove any potential difficulties inherent in the test questions.
If, following our first assumption, the test questions constitute a
communicative interchange between the test writer and the test
taker, then all the cognitive and pragmatic constraints on
interpersonal communication should apply to the testing situation.
In line with Bachman’s call, we utilize insights from pragmatic
theory to form the basis for our theoretical framework. In order to
substantiate our claims, we will refer to theoreticians such as the
Prague School linguists, Chafe (1976), and Prince (1981), rather than
to the more sociolinguistic approach to pragmatics as expressed in
Oller (1979).
In his discussion of test items, Widdowson (1978) seems to reject
the notion of test questions as communicative interpersonal
interchanges. He favors question types “directed toward internal,
mental response” rather than “an external social response” (p. 95).
However, the very introduction of test questions by an external
source (i. e., the test writer) in the form of a physical entity (i. e., the
questions) do indeed render the test an instance of interpersonal


communication. Anecdotal evidence supporting this claim is the
fact that during tests, students often approach the proctor for
clarification by asking “What do you mean?” rather than “What
does this mean?” (emphases added) regardless of whether the
proctor personally wrote the test.
The reading comprehension testing situation is similar to written
correspondence because it is addressed to and therefore tailored to
a specific type of respondent. Moreover, unlike in oral communica-
tion during which a breakdown in communication can be repaired
immediately, the addressee’s response is delayed. However, testing
situations differ from regular written correspondence because (a)
the interlocutors are not equal (Seliger, 1985) in proficiency in the
language of the test (although in a nontest interchange it is also
possible to have nonequals; (b) the testing situation is structured, if
not completely closed, by virtue of the fact that the interchange
constitutes metacommunication (i.e., the questions are about a text,
and so there is an extremely limited range of questions and
responses, whereas a regular interchange could develop freely in
any direction as its dynamics are not restricted by another text); (c)
the question/response channels are unidirectional; and (d) although
the test questions constitute a text, albeit a metatext about a reading
passage, they seem to be less cohesive in relation to one another
than regular continuous discourse.
In ordinary communication, speakers and writers tailor their
utterances to fit their assessment of the addressee’s knowledge of
the topic (Prince, 1981) as well as of how accessible the topic is.
Accessibility is a cognitive notion that requires further qualification.
Whereas a detailed definition of this concept and the factors
affecting it are not within the scope of this paper (see Ariel, 1985,
1990; Kirschner, 1987), we shall use the term accessibility to denote
an attribute of the mental entities referred to by linguistic elements.
The level of accessibility is the amount of processing addressees
require in order to retrieve the entity in question from their short- or
long-term memories. Thus, a highly accessible entity does not
require extensive processing and can immediately be retrieved,
whereas low accessibility entails more processing. An inaccessible
entity does not exist in the addressees’ short- or long-term memories
because they have never learned it. Although Ariel (1985, 1990) does
not include the notion of inaccessibility in her continuum of
accessibility, some mental entities are completely inaccessible
because the addressee has never been exposed to them (Kirschner,
The notions of pragmatic theory as presented above can be
applied to communication between teachers and learners of L2 and


must be extended to include assumptions about the addressee’s
knowledge of the target language per se. The difficulty or incom-
prehensibility of test questions could be explained in terms of prag-
matic variables such as those proposed in this theoretical
From the students’ perspective, they enlist all their linguistic and
extralinguistic resources in approaching test questions. These
resources include knowledge of the world (e.g., various frames,
Minsky, 1977), knowledge of linguistic and communicative conven-
tions carried over from L1 (e.g., Grice’s (1975) cooperative princi-
ple), constraints of relevance (e.g., Sperber & Wilson, 1986), as well
as students’ knowledge of L2 and previous experience with tests.
Our second working assumption, that the test questions should be
as easy to process as possible, is consistent with the notion of the test
as a communicative interchange because in ordinary discourse
speakers and writers do indeed take the addressee’s knowledge into
account in formulating their utterance (Ariel, 1985, 1990; Prince,
1981). Furthermore, any difficulty with processing should be
inherent in the text because students are being tested on their ability
to read and comprehend an academic text and not on their ability to
decipher test questions in English. Although we found agreement
on this perspective in the literature (e.g., Hughes, 1989; Shohamy,
1984, 1985), it is not universally shared by teachers in our programs.
In order to fulfill our goal of improving tests, however, we shall
posit the above assumption as a given and proceed from there.
If one accepts the above premise, then our third assumption fol-
lows—it is the test writer’s task to define, identify, and subsequent-
ly remove any potential difficulties inherent in the test questions.
We suggest that such difficulties are processing constraints which
hinder the comprehension processes. Instead of having direct access
to the task they are asked to perform, the students must first work
their way through a series of hurdles. (The same point is made by
Pollitt, Entwhistle, Hutchinson, & De Luca, 1985).
Not every processing constraint or hurdle can always be re-
moved. Removal of one particular difficulty may introduce
material more problematic than the original difficulty. Thus, the
ultimate decision to remove or keep a potential stumbling block for
the student is often a matter of compromise.


Elucidating the Text
Our experience has been that many testers in educational
organizations, whether consciously or unconsciously, write tests in


which they seek to expose the student’s weaknesses—that is, to
reveal gaps in the student’s knowledge. In contrast, we propose a
more positive approach, one in which the purpose of the test is to
show how well the student can cope with a reading task which
approximates the real-life task of the university student (Heaton,
1975; Hughes, 1989; and Munby, 1979, make the same point).
In other words, the test should elucidate the text, helping the
student work through its hierarchy of ideas. By elucidate we do not
mean give away. The mere presence of test questions necessarily
entails some guidance. For example, a test question asking for the
similarities between X and Y indirectly establishes for the student
that X and Y are similar. However, the extent of the guidance which
the test questions offer should be the result of a conscious decision
on the part of the test writer. This approach may have a positive
backwash effect in the classroom because students will ultimately
be more cooperative when they perceive classroom tests as con-
structive learning experiences rather than occasions to expose their
Furthermore, the types of questions should reflect the overall
purpose of the test. Thus, an integrative reading comprehension test
should not include explicit questions on specific vocabulary items or
grammatical points for two reasons: (a) In dealing with the ideas
expressed in the text, the student must necessarily cope with
vocabulary and grammar. Direct vocabulary or grammar questions
would mean that the student is being tested twice on the same point.
(b) Students’ abilities to answer explicit grammar and vocabulary
questions may not necessarily be related to their abilities to
understand the ideas expressed in the text.
A variety of question types is necessary in order to mitigate the
problems inherent in any one question type (Pollitt et al., 1985;
Shohamy, 1984). In this way, the focus remains on comprehension
of the text because the student is not given the chance to become
test wise to any given question type. The reverse problem, where
the student is not given sufficient opportunity to become
acquainted with a given question type, may also constitute a
processing difficulty. This problem is overcome by prior exposure
to various test or question types in the classroom (Hughes, 1989). In
addition, having a variety of question types reduces the amount of
student writing which would be required on a test with exclusively
open-ended questions (Pollitt et al., 1985).

Hierarchical Versus Linear Progression

One dilemma the writer of a reading comprehension test must
often resolve is whether to compose questions according to the


linear sequence of paragraphs or according to the hierarchical
organization of ideas. Research has suggested that good readers
process the text hierarchically (Armbruster, Anderson, & Ostertag,
1987; Meyer, 1984; Richgels, McGee, Lomax, & Sheard, 1987; and,
indirectly, Minsky, 1977). Accordingly, the test questions should
follow the hierarchical progression of the text in order to help the
student work through the ideas of the text, which is the task the
student is expected to perform in real life. Another argument is that
the hierarchical approach results in a clear parallelism between the
hierarchy of the text and the relative importance of the test items.
This parallelism is crucial to the face validity of the test as a whole.
The importance of face validity, attested to in the literature
(Bachman, 1990; Carroll, 1980; Hughes, 1989; Morrow, 1979), en-
hances student motivation and facilitates a positive backwash effect
(Krashen, 1982).
Thus, the first step in the test-writing process is to establish the
hierarchy of ideas within the text in question (i.e., the main ideas,
the subordinate ideas and their interrelationships). The most
important questions on the test will be those about the main ideas,
that is, the top level of the hierarchy. The test should not omit
questions on any of the main ideas of the text as established by the
hierarchy of ideas. For example, in an experimental research article,
not including a question about the hypothesis (whether the
hypothesis is explicitly stated in the text or merely implied) would
be a serious omission. The importance of these questions should be
reflected in the point distribution for scoring purposes. The relative
weight of the questions should be clearly stated so that it is apparent
to the student. The parallelism between the hierarchy of ideas
presented in the text and the test questions should also be retained
with regard to supporting ideas, examples, details, and so forth.
We have found that texts in which the hierarchy of ideas is not
immediately apparent are more difficult to process. This processing
difficulty arises (a) when the organization of the text is not explicitly
delineated by the authors in their discussion of the purpose of the
text; (b) when explicit markers of rhetorical organization (Olin,
1986) or graphic devices, such as subtitles are lacking; (c) when the
hierarchical structure or schema of ideas does not seem to
correspond to the linear progression of the text; or (d) when the
particular text seems to deviate from a conventional schema or
hierarchy with which the reader is familiar.
The above suggestion that the test questions should correspond to
the hierarchical organization of the ideas as presented in the text
may serve as a guiding principle in test writing (Hughes, 1989,


makes the same point). However, in texts in which the linear order
does not correspond to the expected schema, asking questions
linearly rather than hierarchically may help the student (especially
the weak student) impose order on the text.


In this section, we present a series of considerations for the
writing of specific test questions as opposed to the global
considerations outlined earlier. The extension of pragmatic theory
to nonnative speakers’ processing of L2 can serve as a theoretical
foundation for these considerations.
The practical problem faced by the test writer is how to ask a test
question such that the students can most easily process and thereby
have immediate access to the ideas of the text and to the task they
are expected to perform. However, rather than immediately
proposing the best forms, our strategy in solving this problem has
been first to attempt to identify and analyze those forms that inhibit
the students’ processing of the questions and access to the ideas of
the text. We hope thereby to approximate a characterization of the
optimal form of a test item.
Some of the difficulties in processing discussed below stem from
two different sources which may operate either separately or
simultaneously. One source of difficulty is with “structures which
differ most from equivalent structures in a learner’s native
language” (Pica, 1984, p. 689). This is discussed in the section below
on native language interference. The other is the linguistic
complexity of a given structure in L2 (Pica, 1984), discussed in the
sections below on syntactic forms and on anaphoric reference.

Native Language Interference

Because our student body is relatively homogeneous (i.e., the
majority share the same native language), it is often possible to
foresee or at least identify specific areas of interference from the
native language. As noted previously, L1 interference is a potential
source of confusion and should be carefully considered in the
process of constructing test items. Interference from the native
language may manifest itself in different aspects of language
proficiency—namely, syntax, lexicon, morphology, and phonology.
Because such interference plays a part in many of the difficulties
discussed, it warrants special mention as a factor in the test-writing


Difficult Vocabulary
Avoiding the use of “difficult” vocabulary items may seem to be
a rather obvious suggestion (Shohamy, 1984). However, we have
found that this principle is not universally shared by teachers or test
writers. Some teachers claim that understanding the test questions is
part of the test. If the text includes difficult vocabulary, then why
shouldn’t the questions? Our working assumption, however, is that
any processing constraint should be inherent in the text and not in
the questions.
For the purpose of this discussion, difficult vocabulary is any
word or phrase which is not immediately accessible to the student
in the sense defined in our discussion of our theoretical framework.
If one accepts the premise that the test is a communicative
interchange, then it follows that, as in oral conversation, for
example, one should not talk above the heads of the students.
According to Grice (1975), any violation of the cooperative
principle will result in either a communication breakdown or the
addressee’s attribution of intent to convey some underlying mean-
ing (e.g., name dropping to convey superiority or irony). In the test-
ing situation, miscalculation of the student’s level of vocabulary
usually results in a breakdown in communication; that is, the student
does not understand the question and therefore cannot effectively
perform the task required. In addition, the mere fact that a particu-
lar lexical item was taught during the course does not necessarily
mean that the student can readily recall it during the test.
Although use of a dictionary may alleviate the problem of
unfamiliar vocabulary, it also entails additional processing, takes up
valuable test time, and exacerbates an already stressful situation.
The use of the dictionary ideally should be reserved for difficult
vocabulary in the text.
The avoidance of difficult vocabulary poses a practical problem
because any given test is not addressed to a single student but rather
to a large student body (ranging from one class to hundreds of stu-
dents), and it is virtually impossible to consistently use vocabulary
which is immediately accessible to every student. Through
classroom contact with the target audience of the test, the teacher/
test writer ultimately gains insight into which lexical items are not
within most students’ immediate domain. Even so, a difficult vo-
cabulary item may present itself as such the first time the test is ad-
ministered and can subsequently be changed.
In addition to individual lexical items that may be unfamiliar to
students, two other kinds of vocabulary may also constitute
obstacles for the students: idioms and polysemous words. An idiom


must first be recognized as such by the student. Students often
interpret idioms literally. Polysemous words are also often
potentially confusing because students may initially associate the
word with one of its alternative meanings. Interference from L1
may play a role because a given lexical item in L1 may diverge into
two or more counterparts in L2. The reverse situation (namely, one
word in L2 with two or more counterparts in Ll) is also possible and
may mislead the student.

Presupposing Outside Knowledge

Because most academic texts are addressed to an audience which
shares a common academic and/or cultural background, they often
make reference to concepts which might not be completely familiar
to the EAP student. (Examples of such concepts are the Protestant
Ethic, affirmative action, and the empirical approach to medicine.)
Even when such concepts are explicitly defined in the text, they are
often presented in a manner presupposing familiarity and
Thus, test writers should avoid using a term or concept in a test
question that presupposes any outside knowledge (i.e., whose
definition is not available in the test), even if the term is presented
or used in the text. Such a term, however, can be used in a test
question if it is possible to first elicit its definition in the text by
means of an additional test item. This strategy renders the
previously unfamiliar concept highly accessible in the sense
discussed earlier and in Ariel (1990). For example, one could first
ask the student to extract the author’s definition of the term
affirmative action and only then ask for the author’s opinion
regarding the concept.

Syntactic Form of Test Questions

Students can most easily comprehend test questions when the
syntactic form of the question is relatively simple. We have identi-
fied three main problem areas which constitute a preliminary list of
problematic syntactic forms: clauses, passive constructions, and
subjunctive or hypothetical constructions. We have observed that
these particular structures appear with high frequency on teacher-
made tests.
A large number of clauses in an item contribute to processing
difficulty. A possible reason for this is that each clause represents an
assertion or logical proposition which the student must process.
Therefore, a practical recommendation would be to keep the


number of clauses to a minimum. Specifically, multiple embedded
relative clauses and/or prepositional phrases are particularly
difficult. This problem is exacerbated when there are no explicit
markers of subordination (i.e., reduced relative clauses). The
difficulty in parsing and processing such complex constructions
may be compounded by interference from L1. We have found this
to be the case with our Hebrew-speaking population, possibly
because Hebrew does not allow relative clauses without explicit
markers of subordination.
We have found that active constructions are usually easier for the
student to process than passive constructions. In their L1 research,
Silver, Phelps, and Dunlap (1989) found that “a positive statement
with passive voice (B is now followed by A) will be processed more
slowly than a positive statement with active voice (A in this case
follows B)” (p. 70). Similarly, DeVito (1969) found that it takes
longer to process a passive sentence.
Furthermore, the absence of an explicit agent may constitute an
impediment to comprehension; in other words, the student may not
immediately understand who does what to whom. Moreover, there
may be the additional problem of L1 interference. For instance,
with our Hebrew-speaking population, the pragmatic function and
the distribution of the passive and active constructions in Hebrew
do not coincide with those in English.
Subjunctive or hypothetical constructions also seem to be
problematic for students in the test-taking situation. DeCarrico
(1986) points out that such forms are indeed a source of confusion
for L2 learners. This problem may be particularly pronounced
among Hebrew speakers due to first language interference because
English has two hypothetical or conditional forms, whereas Hebrew
has only one. On the other hand, this might be a processing
difficulty which also pertains to L1 because as Dulay, Burt, and
Krashen (1982) note, this construction is among those acquired at a
later stage of the language acquisition process in both L1 and L2.

The Use of Pronominal Forms for Anaphoric Reference within and

across Test Items
In an attempt to elucidate the text and reflect its hierarchical
organization, teachers often ask questions that are thematically
linked. In so doing, they may use pronominal forms (pronouns or
definite noun phrases) which refer anaphorically to a previous test
item. Recovering the referent in the case of anaphoric reference
requires processing and often considerable cognitive effort on the
part of the student, even if the anaphoric reference is to an element


within the same sentence. If the student must execute the mental
operation of retrieving the referent from the previous item, this may
unnecessarily confound the task of answering the question.
Because the test questionnaire constitutes a metatext intrinsically
linked to another text (i.e., the reading passage), several potential
ambiguities for the student may pose further constraints on process-
ing: (a) The anaphoric pronominal forms in the test may seem to
have several referents; that is, the referent of the pronoun or definite
noun phrase used in the test item may not be uniquely recoverable
(see Quirk, Greenbaum, Leech, & Svartvik, 1972). The accessibility
of a discourse entity is also inhibited by the size of the set of poten-
tial choices for referents (Kirschner, 1987). (b) A reference which to
the students appears anaphoric to previous items in the test
questionnaire may, in fact, be exophoric (Halliday & Hasan, 1976),
having its referent in the reading passage. (c) The reading passage
may in turn make reference to other texts, which the student may
also find difficult to retrieve.
Thus, when feasible, we recommend avoiding pronominal forms
in test items, especially in cross-item reference, because the
explicitly specified entity facilitates processing.

Types of Test Items

Shohamy (1984) demonstrated that different testing methods
significantly affect student achievement on reading comprehension
tests. The discussion below is restricted to those formats relevant to
our teaching/testing context.
Interrogative form versus sentence completion. As noted earlier, the
test questions constitute a communicative interchange between the
student and the examiner which, in many ways, is similar to any
other speech event, whether oral or written. When a speaker wishes
to elicit specific information, the most natural and/or effective way
to do so is to ask a question. Many if not all languages have a
syntactic, phonological, or lexical interrogative form. Accordingly,
we suggest that the easiest test-question format for students to
process is an interrogative sentence—despite the prevalence of the
sentence-completion format on reading comprehension tests. It has
been our experience that the sentence-completion format presents
processing constraints which are not inherent in the interrogative
Statman (1988) found that students performed more successfully
if the stem of a multiple-choice question was an interrogative rather
than an incomplete declarative sentence. A possible explanation for


her results may be that the students reading a sentence-completion
item must reformulate it in their minds as an interrogative in order
to complete it. A cognitive process involving two steps rather than
one might constitute an unnecessary burden on the student.
A related explanation is that, unlike interrogatives and
imperatives, the sentence-completion construction is not a natural
part of students’ syntactic apparatus either in L1 or L2. It is a
construction that students encounter only on tests or in “teacherese”
(i.e., the particular register employed by many teachers in the
classroom and supplemented by intonation and extralinguistic cues
such as gesturing). Students may not have internalized this
unnatural and register-specific construction, making it less
accessible and more difficult to process. Moreover, the blank space
in the sentence-completion task signals students that they must
generate a wh- question, to which the deleted part of the sentence is
the answer. This is an additional cognitive task.
Our experience administering tests confirms the difficulty of
sentence-completion items. The sentence-completion items often
evoke a puzzled reaction on the part of the students who then
approach the proctor for further clarification. It is interesting to
note that the proctor’s spontaneous response is usually to
reformulate the particular item as an interrogative construction,
which almost always results in immediate comprehension on the
part of the student. This further attests to the possibly universal
function of interrogatives to extract information.
Another popular question type is the cloze. When used in a
metatext to test reading comprehension of another text, the cloze is,
in effect, a kind of sentence-completion task. We therefore
recommend cautious use of the cloze to test reading comprehen-
Although we have found the interrogative construction to be
generally more accessible to the student, it is not always possible in
actual practice to avoid using sentence completions or cloze. The
choice of the interrogative construction may sometimes involve
complex structures (e.g., multiple clauses), which are also difficult
for the student to process. Thus, the ultimate decision regarding the
format to be used is often a practical compromise.
If interrogative forms are the most effective and authentic ways
to elicit information in natural language, then the open or
unstructured response should also be the most desirable. However,
in our situation, this requires the student to formulate an answer in
L2, which often poses an additional burden. The ultimate decision
regarding the proportion of questions with unstructured responses
depends on various factors such as the language of response (Ll or


L2), the length of the text and test, the interrelationship between the
test and text, and so forth.
True/false questions. One kind of test-question format we have
often encountered on reading comprehension tests is the true/false
question, in which the student is asked to classify a series of
statements as either true or false according to the text. We have two
objections to this question format. First, it does not seem to simulate
authentic communication and is therefore an unnatural and
inauthentic task (cf. Widdowson, 1978, who rejects the notion of the
test as a social exchange and advocates the use of true/false items as
reflecting reader’s thought processes.)
Second, in many cases true/false test items group together a list of
statements from different and often unrelated parts of the text.
They do not reflect the hierarchical and rhetorical organization of
the text and hence do not elucidate it.
A solution would be to devise true/false items which do adhere to
the hierarchical structure of the text. However, the dichotomous
nature of the true/false item makes it difficult to compose items that
are not too difficult (because they require too much processing) or
too easy (in any true/false item, the student has a 50% chance of
simply guessing the correct answer; see Shohamy, 1984, and
Widdowson, 1978).
A viable alternative to the true/false format is the multiple-choice
question with one or more correct answers. Although this test-item
format has its own disadvantages (see Heaton, 1975; Munby, 1979;
Nuttall, 1982; Oller, 1980), its advantage over the true/false item is
that the multiple-choice stem may be interrogative and therefore
more authentic and more easily processed.
Tables as a test-question format. An additional type of test question
is a table with empty boxes or slots which the students are required
to fill in with information extracted from the text. The table is
appealing as a test format for several reasons: (a) it has a certain
conciseness and visual elegance; (b) it serves as a means of
conceptually organizing the text and thus constitutes a simulation of
a real-life reading task in which one transfers the information
conveyed in a text to a more concise and therefore manageable
form (Widdowson, 1978); (c) the table works very well as a
classroom exercise, and therefore teachers assume that it will also be
effective on a test.
However, our experience in testing situations has revealed
problems with this test format. Students very often approach the
examiner for further clarification. In explaining the table, the
proctor/teacher usually responds by explicitly formulating a


question for each space in the table. The fact that this response
usually alleviates students’ confusion has led us to suspect that filling
in a table is a cognitive task which is more complex than processing
separate and unlinked questions. In addition to understanding the
question, they must reformulate as a question the material
presented in the column headings in order to fill in each space in the
table; they must also transfer the information from the text to the
appropriate parts of the table. Therefore, it may be unfair to expect
nonnative speakers to perform such complex tasks, especially in a
stressful test situation.
Short of avoiding tables altogether on tests, there are several pos-
sible ways to mitigate the processing and performance difficulties
they engender. Instead of using conventional column headings as
they usually appear in tables (e.g., noun phrase postmodified by a
reduced relative clause— evidence found in excavations), w e
recommend using complete interrogative sentences as column
headings or at the beginning of rows (What evidence was found in
excavations?) The conventional column heading form may turn the
table into a series of problematic sentence-completion tasks.
Another way to alleviate the processing constraints is to fill in
several spaces in the table with correct answers to guide the student
in completing the table. Finally, we would advocate using tables
only with texts for which the correlation between the hierarchical
structure of the text and table is explicit.

Prefacing the Question with Explanatory Declarative Sentences

In an attempt to help the student elucidate the text, some teachers
preface the test question with declarative sentences. In a regular
communicative interchange, this is an intuitively sound strategy:
Only after the information has been introduced into the discourse
and is therefore no longer new information can one ask questions
about it (e.g., “There’s a party on Friday night at my brother’s. Do
you want to come?” [Kirschner, 1987, p. 145]). Ariel (1985, 1990)
notes that markers of linguistic givenness (i. e., “given” to the
addressee by virtue of its explicit introduction into the current
discourse) serve to signal the highest level of accessibility. In
prefacing a question with declarative sentences, the test writer is in
effect raising the information obtained from a more distant source
to the current discourse, that is, from the reading comprehension
passage to the test questionnaire (which is the metatext). This
communicative device may particularly characterize discourse in
which the interlocutors are unequal partners.


Paradoxically, we have found that in the case of language learn-
ers, such prefacing remarks may actually result in more confusion
for the students because they have difficulty extracting the focus of
the question and understanding the task they are expected to per-
form. Two factors may account for this confusion: (a) there is addi-
tional material to process before the student reaches the question;
and (b) the student must process two separate speech acts (the term
speech acts as used by Austin, 1962, and Searle, 1971), namely, an
assertion (the introductory statements) and a question. The test
writer should be aware of the potential problems that can arise in
the use of this device and weigh each case individually.

The Interrelationship Between the Text and the Test Questions

Another major source of processing difficulty exists in the
interrelationship or interface between the text and the test
questions. An analysis of the exact nature of this interrelationship
and the parameters which affect it is beyond the scope of this
paper. However, the following questions address some potential
sources of processing constraints which involve this interrelation-
ship in some way:
1. Is the test question ambiguous? The question may be ambiguous
for two reasons: First, the wording of the question may be vague
or not informative enough, so that it does not allow the student
to pinpoint the intended meaning of the question. Second, the
ambiguity may lie in the fact that within the text there is more
than one possible answer to the question as worded.
2. Are there overlapping questions which elicit similar answers? We
have found that student: are confused by multiple questions
tapping into the same ideas. Ordinary discourse allows for a
greater range of functions and rules regarding the distribution of
questions. For instance, although rhetorical questions are
acceptable in texts or speech, they are intolerable on a test.
Moreover, the student expects an exclusive answer for each
question, so an overlap in responses to different questions is
3. Have explicit directions been omitted even though they may be
essential? Are the directions vague? The need for directions and
the degree of their specificity are a function of the interrelation-
ship of the text and test. If the desired response is clearly
articulated and signaled in the text, the need for highly specific
directions may decrease.


We have presented preliminary criteria for both writing and eval-
uating reading comprehension tests. An extension of pragmatic the-
ory served as a theoretical framework to clarify the nature of the
communicative interchange between the test writer and the exam-
inee in an L2 testing context. This theoretical rationale also facili-
tated identification of forms which constitute possible constraints
on cognitive processing. The identification of such forms has prac-
tical implications for teacher training and staff management.
We have summarized our approach in a checklist, which can be
used to guide teachers in the test-writing process and to bring
course objectives into greater focus (see the Appendix). This in turn
may have a positive effect on the quality of classroom teaching.
However, specific items on such a checklist must undergo periodic
review and revision.
Although our study stems from practical experience in a specific
educational context, we hope that our recommendations can be
applied to other contexts and stimulate additional discussion on
both theoretical and practical levels. Future studies should
empirically test the specific criteria included in this paper.

We would like to thank Dulcie M. Engel for her helpful comments on this paper.

Michal Kirschner, Carol Wexler, and Elana Spector-Cohen are EFL course
coordinators in the Division of Foreign Languages at Tel Aviv University, Israel.
They specialize in curriculum design, course administration, and teaching reading
comprehension of academic texts.

Ariel, M. (1985). Givenness marking. Unpublished doctoral dissertation,
Tel Aviv University, Israel.
Ariel, M. (1990). Accessing noun-phrase antecedents. London: Routledge.
Armbruster, B. B., Anderson, T. H., & Ostertag, J. (1987). Does text
structure/summarization instruction facilitate learning from expository
texts? Reading Research Quarterly, 22, 331-346.


Austin, J. L. (1962). How to do things with words. Oxford: Clarenden
Bachman, L. (1990). Fundamental considerations in language testing.
Oxford: Oxford University Press.
Carroll, B. J. (1985). Second language performance testing for university
and professional contexts. In P. Hauptman, R. LeBlanc, & M. B.
Wesche (Eds.), Second language performance testing (pp. 73-88).
Ottawa, Canada: University of Ottawa Press.
Carroll, J. B. (1980). Foreign language testing: Persistent problems. In
K. Croft (Ed.), Readings on English as a second language (pp. 518-530).
Boston: Little, Brown.
Chafe, W. (1976). Givenness, contrastiveness, definitiveness, subjects, top-
ics and points of view. In C. N. Li (Ed.), Subject and topic (pp. 25-56).
New York: Academic Press.
DeCarrico, J. S. (1986). Tense, aspect, and time in the English modality
system. TESOL Quarterly, 20 (4), 665-685.
DeVito, J. A. (1969). Some psycholinguistic aspects of active and passive
sentences. Quarterly Journal of Speech, 55, 401-406.
Dulay, H., Burt, M., & Krashen, S. (1982). Language two. New York:
Oxford University Press.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan
(Eds.), Syntax and semantics 3: Speech acts (pp. 41-58). New York:
Academic Press.
Halliday, M. A. K., & Hasan, R. (1976). Cohesion in English. London:
Heaton, J. B. (1975). Writing English language tests. London: Longman.
Hughes, A. (1989). Testing for language teachers. Cambridge: Cambridge
University Press.
Jones, R. (1985). Second language performance testing An overview. In
P. Hauptman, R. LeBlanc, & M. B. Wesche (Eds.), Second language
performance testing (pp. 15-24). Ottawa, Canada: University of Ottawa
Kirschner, M. (1987). Form and function: An analysis of the linguistic
means of introducing entities into the discourse for the first time.
Unpublished master’s thesis, Hebrew University, Jerusalem, Israel.
Krashen, S. D. (1982). Principles and practice in second language
acquisition. New York: Pergamon Press.
Meyer, B. (1984). Reading research and the composition teacher: The
importance of plans. In S. McKay (Ed,), Composing in a second
language (pp. 63-78). Rowley, MA: Newbury House.
Minsky, M. (1977). Frame-system theory. In P. N. Johnson-Laird &
P. C. Wason (Eds.), Thinking: Readings in cognitive science (pp. 355-
376). Cambridge: Cambridge University Press.
Morrow, K. (1979). Communicative language testing: Revolution or
evolution? In C. L. Brumfit & K. Johnson (Eds,), The communicative
approach to language teaching (pp. 143-157). Oxford: Oxford Universi-
ty Press.


Munby, J. (1979). Teaching intensive reading skills. In R. Mackay,
B. Barkman, & R. R. Jordan (Eds.), Reading in a second language
(pp. 142-158). Rowley, MA: Newbury House.
Nuttall, C. (1982). Teaching reading skills in a foreign language. London:
Olin, R. (1986). The effects of connectors on reading comprehension.
Unpublished master’s thesis, Tel Aviv University, Israel.
Oller, J. (1979). Language tests at school: A pragmatic approach. London:
Oller, J. (1980). An interview with John Oller: Language testing today. In
K. Croft (Ed.), Readings on English as a second language (pp. 481-491).
Boston: Little, Brown.
Perkins, K., & Jones, B. (1985). Measuring passage contribution in ESL
reading comprehension. TESOL Quarterly, 19 (1), 137-153.
Pica, T. (1984). L1 transfer and L2 complexity as factors in syllabus design.
TESOL Quarterly, 18 (4), 689-704.
Pollitt, A., Entwhistle, N., Hutchinson, C., & De Luca, C. (1985). What
makes exam questions difficult: An analysis of “O” grade questions and
answers. Edinburgh: Scottish Academic Press.
Prince, E. (1981). Toward a taxonomy of given-new information. In
P. Cole (Ed.), Radical pragmatics (pp. 223-256). New York: Academic
Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1972). A grammar of
contemporary English. London: Longman.
Richgels, D. J., McGee, L. M., Lomax, R. G., & Sheard, C. (1987).
Awareness of four text structures: Effects on recall of expository texts.
Reading Research Quarterly, 22, 177-196.
Searle, J. R. (1971). What is a speech act? In J. R. Searle (Ed.), T h e
philosophy of language (pp. 39-53). Oxford: Oxford University Press.
Seliger, H. W. (1985). Testing authentic language: The problem of
meaning. Language Testing, 2, 1-15.
Shohamy, E. (1984). Does the testing method make a difference? The case
of reading comprehension. Language Testing, 1, 147-170.
Shohamy, E. (1985). A practical handbook in language testing for t h e
second language teacher. Tel Aviv, Israel: Shaked.
Silver, N. C., Phelps, G. R., & Dunlap, W. P. (1989). Baddeley’s gram-
matical reasoning test: Active versus passive processing differences re-
examined. Language Testing, 6, 164-171.
Sperber, D., & Wilson, D. (1986). Relevance: Communication and
cognition. Oxford: Basil Blackwell.
Spolsky, B. (1985). The limits of authenticity in language testing. Language
Testing, 2, 31-40.
Statman, S. (1988). Ask a clear question and get a clear answer: An inquiry
into the question/answer and sentence completion formats of multiple
choice items. System, 16, 367-376.


Wexler, C., & Kirschner, M. (1989, August). The supervisor as course
designer: Resolving the conflict between academic needs and pragmatic
constraints in a mass EAP reading course at an Israeli university. Paper
presented at the Conference of English for Specific Purposes, Budapest,
Widdowson, H. G. (1978). Teaching language as communication. Oxford:
Oxford University Press.

Global Considerations for Writing Tests


TESOL QUARTERLY, Vol. 26, No. 3, Autumn 1992

L2 Tense and Time Reference

The Ohio State University

The meaning and forms of tenses are complex and often difficult
for nonnative speakers to acquire. The concepts associated with
time which differ among language communities can present an
additional level of complexitv for learners. In a survey, 130 ESL
students were asked to describe the meanings of English tenses in
terms of time concepts used in ESL grammar texts. The results
suggest that speakers of Chinese, Japanese, Korean, Vietnamese,
and Arabic associate different temporal relationships with the
terms right now, present, and past than do native speakers. An
implication of this finding is that grammar teaching that utilizes
descriptions of time accepted in English-speaking communities to
explain usages and meanings of English tenses can produce a low
rate of learner comprehension.

Few ESL researchers doubt that learners’ L1 conceptualization of

time and lexical and/or grammatical time markers have an impact
on their acquisition of English tense. In all languages, time is
referred to in some fashion. However, time attributes (i.e.,
perceptual, conceptual and cultural divisions of time) differ among
societies. One obvious example of this is the boundary of a day. In
nonsecular Muslim and Jewish cultures, days begin at sunset and not
at midnight as in Western civil convention. On the other hand, the
Japanese consider sunrise the beginning of a new day.
Time attributes are bound to reflect on the systems through which
languages represent these divisions (Levinson, 1983). Linguistic
references to time attributes can take many forms: Some languages,
such as Chinese and Japanese refer to time lexically by employing
nouns and adverbs; others, like English, also utilize grammatical
references (i.e., verb tense). If both L1 time attributes and their lin-
guistic references differ from those in L2, learners may find them-
selves in an environment where they cannot pick out the temporal at-
tribute to which tense is a grammatical reference (Donnellan, 1991).
English aspect can be morphologically marked as well. For
example, the verbs in both sentences He runs and He is running are

in the present tense. However, the present simple runs carries
iterative (or habitual) implicature, whereas -ing in the second
sentence imparts progressive implicature to the verb’s present tense
meaning. Aspects, which Comrie (1976), Lyons (1977), and
Richards (1987) view as additional features of time deixis (or means
of locating events in time), can present the same potential
dichotomy between the time attributable and its reference.
In order to gain insight into how ESL learners acquire morpho-
logical tense, numerous studies have examined the order of
morpheme acquisition (e.g., Andersen, 1977; Bailey, Madden, &
Krashen, 1974; Dulay & Burt, 1974; Larsen-Freeman, 1976; Makino,
1979; Pienemann, 1985). In addition, a great deal of research has
been devoted to ESL learner acquisition of tense and morpheme
meaning (Andersen, 1983; Bailey, 1989a, 1989b; Hatch, 1978).
Whereas some specialists on language and tense acquisition believe
that learners acquire tense meanings before their morphological
forms, others hold the opposite view. This paper will address the
issue of whether nonnative speakers (NNSs) who have received
extensive L2 training and have achieved a relatively high L2
proficiency intuitively perceive English conceptualization of time
and its grammatical references to deictic (or indexical) time, that is,
morphological tense, in ways similar to native speakers (NSs).
Another focus of this study is NNSs’ perceptions of English
aspectual implicature.
ESL teachers and L2 researchers recognize that English tenses are
difficult to acquire (DeCarrico, 1986; Richards, 1981; Riddle, 1986).
Guiora (1983) notes that speakers of Hebrew encounter difficulty
mastering the meanings and usages of several of the English past
tenses which, to them, seem redundant and without an easily
discernible function. He also notes that speakers of Chinese may be
faced with establishing an entirely new hypothesis of how time is
used and referred to. Sharwood Smith (1988) indicates that his
Polish students had difficulty relating to the past progressive and its
form. Richards (1981) discusses the complexity of introducing
English progressive tenses and their explicit and implied meanings.
Dialect variations even within English-speaking societies make for
significant differences in tense usage and meanings (Leech, 1971).
Coppetiers (1987), who conducted a study of highly educated
NNSs with near-native proficiency in French, found that whereas
they had obviously acquired tense forms, their perceptions of tense
meanings were not NS-like. Coppetiers contends that the NNSs’
perceptions of tense meanings were strongly affected by tense
meanings in the L1 so that the speakers of Romance languages
interpreted the meanings of French tenses differently from speakers
of Germanic and tenseless languages (pp. 560-561).


To date, whether speakers of the many languages without mor-
phological tenses can fully master the English verbal system of
tenses has not been determined. Richards (1973) observes that ESL
learners’ omissions of tense markers represent a damaging and
confusing type of error. Chappel and Rodby (1983) note that ESL
students’ tense-related errors often detract from the overall
comprehensibility of their text. They further mention that despite
the fact that verb tenses occupy a prominent role in the teaching of
ESL, students seem to choose verb tenses arbitrarily. In their view,
tense errors may result from the learners’ lack of understanding of
the impact of tense on text.


The issue of the relationship between grammatical tense and time
and the acquisition of tense systems is complex. Whether a
connection exists between the detailed marking of time in English
and its morphological tense as a grammatical category has not been
established with certainty. Comrie (1985) mentions that various
cultural groups “have radically different conceptualizations of
time” (p. 3) and only some measure time and occurring events with
exactitude. Fillmore (1975) notes that, in most languages, lexical
markers, such as today, tomorrow, and yesterday, can refer to a
variety of time lengths within a relevant span. These relevant spans,
however, differ from one language to another. Levinson (1983)
claims that in “languages without true tenses, for example Chinese
or Yoruba” (p. 78), the concept of time is realized through adverbs
and implicit and contextual assumptions. Southeast Asian languages
require a strict discourse frame which delineates time and,
therefore, the time reference.
The numerous studies of the meaning relationships in English
between attribute and reference—the thing and its name—have
demonstrated that they are vague (Bach, 1981) and language
specific. Kripke (1991) views notions of meanings as “determined
by the conventions of the language” which can be treated only in
conjunction with the related linguistic phenomena of the language
(p. 84). Bach (1981) advances this argument stating that, in order to
be understood, the speaker and his audience must have mutual
contextual beliefs. Linguistic meanings of tense also include the
mutual beliefs and shared perceptions of the members of a speech
community. The expression of such beliefs and perceptions may not
be shared by members of other speech communities (Searle, 1979).
As Donnellan (1991) notes, if descriptions of time are used
referentially, the subjects to whom these descriptions are addressed


are thus enabled to “pick out” (p. 60) the references and their
attributes. However, if the reference does not fit the subjects’
perceptions of the attribute, they may be unable to establish a
correspondence between them.
For example, if the NS instructor states that the morpheme -ed
marks the verb for past simple tense but the student’s conceptual-
ization of past differs from the instructor’s, the student may not use
this morpheme in the contexts where the instructor would. Learners’
abilities to establish the referential relationships between L2 time
deixis, tense, and morphological markers necessarily affects their
perceptions of the meanings and functions of tense morphemes.
Recanti’s (1991) availability principle assumes that linguistic
meanings must be available or accessible to our “ordinary,
conscious intuitions” (p. 106). Because time-span conceptualizations
and their lexical references differ for NSs and NNSs, English
grammatical references to time may not be readily available for
pragmatic interpretation by speakers of tenseless languages. If this
is the case, morphological time reference (i.e., linguistic meaning of
tense) may not be accessible to these speakers’ conscious intuitions.
Another complication is that even developed morphological tense
structures in two languages may differ greatly (Fillmore, 1975).
Levinson (1983) sees English time reference as calendrical
reckoning and observes that most Amerindian languages, Japanese,
and Hindi differ from it and one another in names and lengths of
days and time spans. In his brief examination of how the time
attribute corresponds to tense, Levinson mentions that in languages
with tense, sentences are anchored to a context by morphological
tense, whereas other languages utilize other linguistic and social
means of contextual anchoring. If mutual contextual beliefs (Bach,
1981 ) and calendrical time deixis (Levinson, 1983) are necessary for
picking out a time attribute and its morphological reference in
English, NNSs lacking intuitions and access to knowledge
associated with the English time deixis and linguistic tense may face
problems in using and interpreting English time references.
Usually, instructors teach tenses by presenting rules, explaining
the meanings of tenses, and by identifying the time deixis and
lexical contexts in which certain tenses are called for (Eisenstein,
1987). Such presentations are usually accompanied by exercises in
which the students are expected to apply the instructor’s explana-
tions. In order to do so successfully, the students have to perform a
series of tasks. They need to be aware of the lexical and syntactic
markers of time and their environments in the sentence, understand
their meanings and implications, analyze them for time and tense


reference and aspect implicature, pick out one or more correspond-
ing auxiliaries or morphemes, put them in the relevant form, and
produce a correct verbal structure. In this study, ESL students were
asked in a questionnaire to reverse this process and describe the
meanings and implications which tenses and aspects have for them
through the reference terms associated with English time deixis.
(The descriptions of English time deixis and the framework of
temporality were adopted following Leech, 1971, and Comrie,

Questionnaire Design
In the questionnaire, the students were asked to describe four
sentences for each of the 8 English tenses excluding future, a total of
32 sentences: 4 present (present simple, present progressive, pres-
ent perfect, and present perfect progressive) and 4 past (past sim-
ple, past progressive, past perfect, and past perfect progres-
sive). If responses for 2 sentences with the same tense and aspect
differed, they were averaged independently for tense and aspect. In
order to circumvent the issue of the respondents’ possible confu-
sion when performing the required task, responses to the first 2 sen-
tences per tense were considered invalid and excluded from data
In the questionnaire, time attributes and references were listed
with the immediate present first, moving back to the past perfect,
which is the most deictically distant from the present moment. To
assure that the tense descriptors were accessible to the NNSs, the
selection of terms describing the meanings of tenses and aspectual
implicatures were chosen from intermediate/advanced ESL and
grammar texts: right now (Azar, 1989) and at the moment of
speaking (Leech, 1971); in the present and in the past (Leech &
Svartvik, 1975); in the past and before another past event (Azar,
1989; Leech, 1971; Leech & Svartvik, 1975); progressive (Azar, 1989;
Leech, 1971; Leech & Svartvik, 1975); and repetitive/habitual
(Azar, 1989; Leech& Svartvik, 1975).
The semantics of the contexts were made uniform for grammat-
ical gender, animacy, and number. The choice of sentences in the
questionnaire reflected several considerations:
1. The verbs did not carry momentary or durational meanings
(Leech, 1971) (as in, respectively, blink or love) and only three
verbs were used: walk, talk, and visit.


2. Explicit time markers were excluded, with the exception of
before, to motivate the past perfect tenses (Azar, 1989; Leech,
3. Vocabulary was restricted to fewer than 100 high-frequency
(See the Appendix for a listing of the questionnaire sentences. These
are presented in an order different from that in the actual
The NNS and the NS controls were instructed to choose however
many of the multiple-choice items they wished and thus describe
their own perceptions of temporal references and the progressive
and iterative/habitual aspects (Comrie, 1985; Leech, 1971;
Richards, 1981). However, true to the multiple-choice testing
tradition, almost all participants selected only one answer per
multiple-choice selection. The first multiple-choice selection had a
general heading, The time of the action is, and required the subjects
to identify the English verb time reference regressively from the
present to the past. The second selection had the heading The action
is and dealt with the respondents’ perceptions of aspect. The
aspects addressed in the questionnaire included the progressive
aspect and the iterative/habitual aspect. The perfective aspect and
Ø aspect were not included and, for the purposes of this study, are
termed nonprogressive / nonhabitual. (The selection in the question-
naire corresponding to these aspects was none of the above.) The
multiple-choice options remained uniform for all 32 sentences.
For example, the students read the sentence Bob is talking to his
brother. Then they saw two multiple-choice selections for tense and
aspect descriptors, respectively
1. The time of the action is:
a. right now/at the moment of speaking
b. in the present and in the past
c. in the past
d. before another past event
e. cannot decide
2. The action is:
a. progressive
b. repetitive/habitual
c. none of the above
d. cannot decide
The survey was administered at the conclusion of the Autumn
Quarter, immediately following 9 weeks of instruction in daily or
thrice-weekly ESL classes. There was no time limit for the subjects
to respond to the questions.


Of the 130 ESL students who participated in this study, 70
students were speakers of Chinese (CH); 17, Korean (KR); 13,
Japanese (JP); 11, Vietnamese (VT); 12, Spanish (SP); and 7, Arabic
(AR). Of the 21 NS included as controls, 19 were graduate students
enrolled in various departments at The Ohio State University
(OSU), most of whom had minimal training in linguistics. The
remaining 2 were ESL instructors. The total number of participants
was 151.
All NNS participants had been admitted to OSU and were taking
classes at the university. Their TOEFL scores ranged from 500 to
617, with a mean of 563. Unlike the majority of NNSs, the
Vietnamese and some speakers of Spanish were U.S. resident aliens
or citizens and thus were not required to take the TOEFL.
The NNS subjects’ ESL training ranged from 4 to 18 years with a
mean of 9.6 years. All NNS students included in the study, with the
exception of the Vietnamese, had been residing in the U.S. for a
period of time ranging from 2.5 to 30 months, with a mean of 6.3
months. The Vietnamese students’ residence in the U.S. ranged
from 4 to 11 years, with an average of 5.7 years, and the duration of
their formal ESL training ranged from 9 to 33 months, with a mean
of 10.3 months.


The sizes of the NNS groups were not equalized. After the data
were compiled for each sentence, they were converted to percent-
ages. The NS values were compared to those for other groups. The
temporal reference for each tense chosen by the highest number of
NSs was accepted as the tense temporal reference against which all
those of the NNSs were compared. (See Table 1.)
Only in the present progressive were the NNSs’ perceptions of
tense meanings close to those of NSs. Otherwise, NSs generally
chose descriptions of temporal references substantially differently
from members of all groups of trained NNSs. In fact, the
differences between NSs and NNSs were statistically significant
(p < .01) for each row of Table 1 except the present progressive,
which is not significant.1 The NNSs’ temporal reference for the
present progressive right now/at the moment of speaking indicates
1 Thisis based on Fisher’s exact test for each row, grouping all NNSs together. A chi-square
test for independence would not have been appropriate due to small cell sizes associated
with percentages near 0 or 100%. Since results for 2 sentences were used and averaged in
Table 1, care was taken to perform the test separately for each sentence.


Temporary Reference (%)
(N= 151)

that, for them, it is the most intuitively accessible deictic point. This
finding is consistent with that of Olshtain (1979) whose case study
showed that even a speaker of a language without aspect acquired
the present progressive earlier than other tenses. The past simple
attribute provides the second most easily available point of
reference because present, past, and future are the basic tense
meanings within the conceptualizations of linear temporality
(Comrie, 1985). The unanimity of the Japanese, none of whom
perceived it to mark the past, may be explained by the Japanese
system of naming a certain number of days back from today which
can be included in both the present and the past (Fillmore, 1975).
The Chinese perceived the deictic time of the present progressive
and past simple most nearly approximating NS perceptions. In
terms of distance from the NS values, these two tenses were
followed by the past perfect and the past progressive, then the
present simple, present perfect progressive, past perfect progres-
sive, and present perfect, respectively. Interestingly, the values for
Koreans followed approximately the same pattern. The Vietnamese
values for the present progressive and past simple are also the
highest for this group and are similarly followed by the past pro-
gressive. As has been mentioned, the Japanese are somewhat differ-


With the exception of the speakers of Spanish and Arabic, the
values for the other present tenses reflect the considerable difficulty
most NNSs had when choosing the temporal descriptions listed
within the selections. The past perfect tenses presented less
difficulty, which can be partially explained by the lexical (as
opposed to grammatical) reference of before. Levinson (1983)
indicates that most tenseless languages provide for lexical and
discourse sentence anchors. In this case, the adverb before is an
explicit lexical marker congruent with the concepts of time
reference intuitively available to the speakers of such languages.
Linear conceptualizations of time may not be common to all
societies (von Stutterheim & Klein, 1987). Among the 6 groups of
NNSs, only the speakers of Spanish and Arabic were speakers of
languages with developed morphological tenses. The very fact that
Spanish and Arabic have deictic time reference provides an
established conceptual structure and morphological temporal
reference which the speakers of these languages can draw on when
exposed to L2 conceptualizations of time and morphological tense.
To some degree, they share more mutual conceptualizations of time
with NSs and were more successful in picking out appropriate L2
time attributes than speakers of Chinese, Korean, Japanese, and
The NSs’ behavior in the analysis of tense-marked temporality
demonstrates that they appear to know that auxiliaries and
morphemes represent deictic time reference and were, therefore,
able to pick out the more appropriate time attribute (Donnellan,
1991). They appear to have access to the linguistic meanings which
auxiliaries and morphemes encode in English. The NNSs, however,
do not seem to have the NS-like intuitive knowledge of the linear
conceptualization of time and its linguistic references.
Morphological references to deictic time are inextricably linked
to tense reference. If a grammatical reference to temporality
implies a deictic time, we assume that the NNS knows and intends
that meaning (Recanti, 1991); that is, we assume that NNSs’ choice
of morphemes implies their knowledge of morphological meanings.
Even if the NNSs’ intuitive knowledge of deictic time attribute is
NS-like but their choice of morphemes is not, their NS-like intuitive
knowledge of deictic time would still appear seriously flawed.
In the second task, the study participants were requested to assign
aspectual implicature (Comrie, 1976, 1985) to each temporal
reference of tense. The implicature of linear temporal aspects tends
to increase the distance between the NS and NNS perceptions of
temporality. (See Table 2.)


Aspectual Implications (%)
(N= 151)

The NNSs’ perceptions of aspectual implicature indicated by

their choices were also analyzed against the choices made by the
majority of NSs. NSs chose descriptions of aspectual implicature
significantly differently (p< .01) from NNSs in every case,
including present progressive (based on Fisher’s exact test). A cell-
by-cell comparison of same-tense values associated with the NNSs’
perceptions of L2 aspectual implicature (see Table 2) shows an
average decline of 7.8% compared to values associated with NNSs’
perceptions of temporality (see Table 1). This finding is consistent
with Bailey’s (1989a, 1989b) account of NNS acquisition of past
simple and past progressive, which notes that the progressive aspect
combined with the meaning of the past presents an additional level
of complexity for L2 learners. The NNSs’ perceptions of the
progressive aspect were generally closer to those of NSs than were
their perceptions of the habitual and nonprogressive/nonhabitual.
Durative and continuative, and iterative and repetitive aspects, in
some form, can be found in all Lls represented in the data with the
exception of Vietnamese. NNSs whose Lls have aspect as
referential implicature thus have access to the associated linguistic
conceptualization. Chinese (Li & Thompson, 1981) and Spanish
(Comrie, 1976, 1965) have the durative and continuative; Korean
(Joo Hwang, 1987), Japanese (Inoue, 1984), and Arabic (Kaye,
1987) have both types of implicature—durative and continuative,
and iterative and repetitive. However, the aspectual implicature in


these languages is different from that of English—so much so that,
as the cited authors indicate, it is rather difficult to describe in terms
of English. Vietnamese, however, is unique in that it does not have
tenses or aspects, and its word order is the sole means of indicating
grammatical relations (Nguyen, 1987).
The NNS perceptions of aspect and temporality in the present
perfect are the most distant from those of NSs. Among the speakers
of tenseless languages, the Koreans and the Japanese more closely
approximated NS values over the range than did the Chinese, who
have only durative L1 aspect. In turn, the values for the Chinese
were nearer NS values than the Vietnamese. The fact that the Span-
ish continuative and the situational repetitive are not similar to the
English progressive and iterative (Comrie, 1976, 1985) is presum-
ably reflected in the values for the Spanish speakers.
Leech (1971) and Comrie (1976, 1985) strongly distinguish be-
tween the basic meanings of tenses and the secondary meanings of
aspects. The NNSs’ interpretations of L2 time deixis that are, in
Donnellan’s (1991) framework, restricted by their L1 conceptualiza-
tion are made additionally difficult by the need to infer aspectual
implicature. The fact that the distance between NNS and NS
perceptions was greater in regard to aspectual implicature than with
temporal reference supports the earlier observation that NNSs’
intuitions regarding morphological references to deictic temporality
may not be fully developed by years of L2 training.

Independent of the NNSs’ perceived meanings of time spans,
morphological references to time impose obvious constraints on L2
learner performance. The fact that NNSs with extensive language
training and TOEFL scores above 500 consistently made temporal
reference analyses and choices of time attributes significantly
different from those of NSs in nearly all cases can be accounted for
by four interrelated hypotheses which require further investigation.
1. NNSs’ intuitive conceptualizations of time are not linear and/or
deictic and, therefore, removed from those of NSs. Extensive L2
instruction may diminish this conceptual distance only to a
limited extent.
2. Because English, unlike some other languages, requires morpho-
logical reference to time deixis, NNSs’ intuitions associated with
deictic tense may not be based on linear temporality and mor-
phological tense as fully as those of NSs are.


3. Despite their years of language training, compared to NSs, NNSs
have limited access to the means of interpreting morphological
deictic time.
4. As is apparent from the data for the Vietnamese speakers, many
years of exposure to L2, combined with instruction, may “have a
limited impact on NNSs’ perceptions of L2 deictic tense.
The marked differences which have been noted between the NS
and NNS perceptions of time and its associated morphology as
described in the terms accepted in L1 research and L2 methodology
can also imply that tense-related instruction does not always strike
a familiar chord or provide for a point of reference in NNSs’ con-
ceptualizations of time and its grammatical encoding.


The data presented in this study are preliminary and require
further investigation. For this reason, only some general suggestions
and implications for teaching can be offered. The substantial
differences between NS and NNS perceptions of tense meanings
seem to indicate that NSs and NNSs view time spans and their
divisions and measurements differently. If this is the case, the
teacher cannot assume that the terminology and the conceptualiza-
tions associated with English time deixis are understood by NNS
students in the same way as they are understood by NSs. Specifical-
ly and thoroughly explaining English time attributes and notions,
the reference terms used to describe them, and their impact on the
meanings of tenses can possibly help L2 learners associate the word
labels and morphemes which refer to time divisions.
The data further show that for these L2 learners, the present pro-
gressive, past simple, and past progressive, respectively, repre-
sented the most accessible deictic time spans. It is reasonable that
the teaching of English tenses should begin with these three tenses.
As has been noted, Japanese speakers may have particular difficulty
with the meanings and morphology associated with the past simple.
Because NNSs tend to rely on lexical time markers such as before
and after when interpreting the meanings of tenses and their mor-
phological references, these may be included in the initial
explanations of the English tense system to facilitate the learners’
understanding of time-span relationships and tense meanings.
Because morphological tense markers impose constraints on
learner performance, they may be specially addressed in conjunc-
tion with tense meanings. The speakers of Spanish seem to have dif-
ficulty distinguishing between English tense-related morphemes


and false cognates associated with the Spanish tense system and its
morphology (Andersen, 1983; Comrie, 1985). The intuitions of
Vietnamese speakers regarding tense morphological markers seem
to be notably different from those of other groups of NNSs, pre-
sumably due to the absence of morphological deixis in their L1. For
speakers of Arabic, as opposed to Chinese and Korean, English no-
tions of temporality seem to impose somewhat reduced constraints
associated with notions of temporality. However, their acquisition
of the meanings and forms for the perfect tenses, such as the past
perfect, past perfect progressive, and present perfect, appears to
present substantial difficulty. In very general terms, the teaching of
English conceptual notions of time, its divisions, and the relation-
ships between these divisions can underlie or even precede the
teaching of the tense system and its morphological references.

Eli Hinkel received her PhD in linguistics from The University of Michigan in 1984
and has taught in intensive and ITA-training programs for the past 10 years. Her
research interests include concept-based transfer and L2 teaching methodologies.
She is employed as Coordinator of the ESL Composition Program at The Ohio
State University.

Andersen, R. (1977). The impoverished. state of cross-sectional morpheme
acquisition/accuracy morphology. In C. Henning (Ed.), Proceedings of
the Los Angeles Second Language Research Forum (pp. 308-320). Los
Angeles: University of California.
Andersen, R. (1983). Transfer to somewhere. In S. Gass & L. Selinker
(Eds.), Language transfer in language learning (pp. 177-202). Rowley,
MA: Newbury House.
Azar, B. (1989). Understanding and using English grammar (2nd ed.).
Englewood Cliffs, NJ: Prentice Hall.
Bach, K. (1981). Referential/attributive. Synthese, 49, 219-244.
Bailey, N., Madden, C., & Krashen, S. (1974). Is there a “natural sequence” in
adult second language learning? Language Learning, 27, 235-244.
Bailey, N. (1989a). Theoretical implications of the acquisition of the
English simple past and past progressive. In S. Gass, C. Madden,
D. Preston, & L. Selinker (Eds.), Variation in second language
acquisition: Vol. 2. Psycholinguistic issues (pp. 109-124). Clevedon,
England: Multilingual Matters.


Bailey, N. (1989b). Discourse conditioned tense variation: Teacher
implications. In M. Eisenstein (Ed.), The dynamic interlanguage:
Empirical studies in second language variation (pp. 279-295). New York:
Plenum Press.
Chappel, V., & Rodby, J. (1983). Verb tense and ESL composition: A
discourse-level approach. In M. A. Clarke & J. Handscombe (Eds.), On
TESOL ’82. Pacific perspectives on language learning and teaching
(pp. 309-320). Washington, DC: TESOL.
Comrie, B. (1976). Aspect: An introduction to the study of verbal aspect
and related problems. Cambridge: Cambridge University Press.
Comrie, B. (1985). Tense. Cambridge Cambridge University Press.
Coppetiers, R. (1987). Competence differences between native and non-
native speakers. Language, 63, 544-4573.
DeCarrico, J. (1986). Tense, aspect and time in the English modality.
TESOL Quarterly, 20 (4), 665-682.
Donnellan, K. (1991). Reference and definite descriptions. In S. Davis
(Ed.), Pragmatics (pp. 52-64). Oxford: Oxford University Press.
Dulay, H., & Burt, M. (1974). Natural sequences in child second language
acquisition. Language Learning, 24, 37-53.
Eisenstein, M. (1987). Grammatical explanations in ESL: Teach the
student, not the method. In M. Long & J. Richards (Eds.), Methodology
in TESOL. New York: Newbury House.
Fillmore, C. (1975). Santa Cruz lectures on deixis. Bloomington: Indiana
University Linguistics Club.
Guiora, A. (1983). The dialectic of language acquisition. Language Learn-
ing, 33, 3-12.
Hatch, E. (1978). Discourse analysis and second language acquisition. In
E. Hatch (Ed.), Second language acquisition: A book of readings.
Rowley, MA: Newbury House.
Inoue, K. (1984). Some discourse principles and lengthy sentences in
Japanese. In S. Miyagawa & C. Kitagawa (Eds.), Studies in Japanese
language use. Edmonton, Canada: Linguistic Research.
Joo Hwang, S. J. (1987). Discourse feature of Korean narration. Dallas, TX:
The Summer Institute of Linguistics.
Kaye, A. (1987). Arabic. In B. Comrie (Ed.), The world’s major languages.
Oxford: Oxford University Press.
Kripke, S. (1991). Speaker’s reference and semantic reference. In S. Davis
(Ed.), Pragmatics (pp. 77-96). Oxford: Oxford University Press.
Larsen-Freeman, D. (1976). An explanation of the morpheme acquisition
order of second language learners. Language Learning, 26, 125-134.
Leech, G. (1971). Meaning and the English verb. London: Longman.
Leech, G., & Svartvik, J. (1975). A communicative grammar of English.
London: Longman.
Levinson, S. (1983). Pragmatics. Cambridge: Cambridge University Press.
Li, C., & Thompson, S. (1981). Mandarin Chinese. A functional reference
grammar. Berkeley: University of California Press.
Lyons, J. (1977). Semantics (Vols. 1 & 2). Cambridge: Cambridge
University Press.


Makino, T. (1979). English morpheme acquisition order of Japanese
secondary school students. TESOL Quarterly, 13 (3), 428-449.
Nguyen, D-H. (1987). Vietnamese. In B. Comrie (Ed.), The world’s major
languages. Oxford: Oxford University Press.
Olshtain, E. (1979). The acquisition of the English progressive: A case
study of a seven-year-old Hebrew speaker. Working Papers in
Bilingualism, 18, 81-102.
Pienemann, M. (1985). Learnability and syllabus construction. In
K. Hyltenstam & M. Pienemann (Eds.), Modeling and assessing second
language acquisition. San Diego, CA: College-Hill Press.
Recanati, F. (1991). The pragmatics of what is said. In S. Davis (Ed.),
Pragmatics (pp. 97-120). Oxford: Oxford University Press.
Richards, J. C. (1973). Error analysis and second language strategies. In
J. W. Oller & J. C. Richards (Eds.), Focus on the learner: Pragmatic
perspectives for the language teacher. Rowley, MA: Newbury House.
Richards, J. C. (1981). Introducing the progressive. TESOL Quarterly,
15 (4), 391-402.
Richards, J. C. (1987). Introducing the perfect: An exercise in pedagogic
grammar. In M. H. Long & J. C. Richards (Eds.), Methodology in
TESOL. New York: Newbury House.
Riddle, E. (1986). Meaning and discourse function of the past tense in
English. TESOL Quarterly, 20 (2), 267-286.
Searle, J. (1979). Referential and attributive. Monist, 62, 190-208.
Sharwood Smith, M. (1988). Imperative versus progressive: An exercise in
contrastive pedagogical linguistics. In W. Rutherford & M. Sharwood
Smith (Eds.), Grammar and second language teaching. New York:
Newbury House.
von Stutterheim, C., & Klein, W. (1987). A concept-oriented approach to
second language studies. In C. Pfaff (Ed.), First and second language
acquisition process. Cambridge, MA: Newbury House.





The TESOL Quarterly welcomes evaluative reviews of publications relevant to
TESOL professionals. In addition to textbooks and reference materials, these
include computer and video software, testing instruments, and other forms of
nonprint materials.


University of Washington

Process and Experience in the Language Classroom (Applied Lin-

guistics and Language Study Series).
Michael Legutke and Howard Thomas. London: Longman, 1991.
Pp. xv + 332.

■ Every now and then a book appears which is immediately

recognizable as a distinctive contribution to the field. This is such a
book. It is a book with an important message. It is also a book with
The central message of the book is that language learning is an
educational endeavor and that educational principles and values lie
at the heart of what makes language classrooms work. It therefore
sets itself apart from those works in which the (usually implicit)
message is that learning a language is so different from learning
anything else that the language educators have little to learn from
the educational mainstream.
The task which the authors set themselves is to show us what we
as language teachers have to gain from adopting a broad view of the
language learning process. Their philosophical orientation is
towards an experiential view of learning in which the learner “learns
by doing.” In making a case for experiential learning, they draw on
traditions of experiential learning going back to Dewey (1916).
They also introduce the reader to a rich vein of work in Europe
which has been largely ignored in the United States and Britain (see,
for example, Kohonen, 1992). In addition, they have reactivated and
revitalised the debate over the nature and direction of communica-
tive language teaching (Breen & Candlin, 1980).
Supporting themes within the book are as follows:
1. The classroom is an authentic psychosocial environment in its
own right. The authors point out that classrooms have their own
sociocultural reality and that this reality can and should be

utilized for the purposes of learning and teaching. This theme
reinforces their attempt to transcend what they see as the rather
narrowly conceived, psycholinguistically oriented approaches to
pedagogy and research which have dominated task-based
learning to the present time.
2. Learners can make important contributions to the content and
the process of learning. Consistent with this view of the class-
room as a community, the authors demonstrate ways in which
learners themselves can contribute to the ongoing construction of
the curriculum by making choices and contributions not only
about classroom content but also about pedagogical procedures.
3. Language classrooms should provide a dual focus on language
content and learning process. If learners are to make choices
about what to learn and how to learn, then they need to have
opportunities for focusing not only on language content but also
on learning processes. The necessity for this dual focus is dealt
with consistently throughout the book and receives detailed
treatment in the penultimate chapter of the book.
4. Project-based learning provides coherence to task-based lan-
guage programs. The pedagogic “task” has become an important
organizing principle in language curriculum development. At the
present time, the theoretical and empirical view of task pre-
sented in the literature is very much based within a psycholin-
guistic, cognitivist view of language learning and use. Legutke
and Thomas seek to provide a counterpoint to this view by dem-
onstrating the sociocultural dimensions of task-based learning.
One of the dangers of task-based learning is that the curriculum
can become a collection of disconnected classroom activities. In
this book, the authors show how the potential for lack of coher-
ence can be avoided by integrating tasks through projects.
5. Experiential teaching should be mirrored by experiential teacher
education. In the final chapter, the focus shifts from the learners
to the teacher, and the authors make the point that the
philosophy and approach which is advocated for learners should
also find expression in the education of teachers.
The theoretical heart of the book is to be found in chapter 3,
where the authors take a long, hard look at the contribution of
humanism to language education. They are extremely critical of the
“touchy feely” philosophy of the 1970s, which was, in most
instances, highly manipulative while pretending to be emancipa-
tor. While agreeing with the thrust of much of what the authors
have to say in this chapter, I must say that I found the level of
personal debate rather vigorous at times.


One possible criticism of the approach advocated in this book
might be that it reflects a particular Western intellectual and educa-
tional tradition, and that these ideas are of questionable relevance in
other contexts, cultures, and traditions. This may or may not be so,
and is a matter for empirical investigation rather than theoretical de-
bate. All too often, however, the claim that “it wouldn’t work here”
is an excuse for maintaining the status quo. In fact, it is instructive
to look at what some of the research has to say about what learners
think is legitimate pedagogic activity. In a recent survey in Japan,
for instance, Widdows and Voller (1991) found that
students do not like classes in which they sit passively, reading or
translating. They do not like classes where the teacher controls
everything. They do not like reading English literature much, even when
they are literature majors. Thus it is clear that the great majority of
university English classes are failing to satisfy learner needs in any way.
Radical changes in the content of courses, and especially in the types of
courses that are offered, and the systematic retraining of university EFL
teachers in learner-centred classroom procedures are steps that must be
taken, if teachers and administrators are seriously interested in
addressing their students’ needs. (p. 135)
Perhaps the greatest strength of this book is the way in which the
authors have managed to provide a balance between theory,
research, and practice. For this reason, it has something for teach-
ers, teacher educators, curriculum designers, materials developers,
and researchers themselves. By providing a theoretical and empiri-
cal rationale for the classroom procedures and tasks which make up
the latter part of the book, the authors guard against the danger of
the work ending up as little more than a series of “tricks of the
trade.” At the same time, the fact that the book is lavishly illustrated
with practical ideas and techniques makes it an intensely practical
book for all those involved in language education.

Breen, M., & Candlin, C. (1980). The essentials of a communicative
curriculum in language teaching. Applied Linguistics, 1, 89-112.
Dewey, J. (1916). Democracy and education. Macmillan New York.
Kohonen, V. (1992). Experiential language learning Towards second language
learning as learner education. In D. Nunan (Ed.), Collaborative language
learning and teaching (pp. 14-39). Cambridge: Cambridge University Press.
Widdows, S., & Voller, P. (1991). PANSI: A survey of ELT needs of
Japanese university students. Cross Currents, 18 (2), 127-141.

Macquarie University, Australia

Developing Communicative Competence in a Second Language.
Robin C. Scarcella, Elaine S. Anderson, and Stephen D. Krashen
(Eds.). New York: Newbury House, 1990. Pp. xviii+ 356.

This book is the first compilation of studies I have seen specifical-

ly oriented to the exploration of second language communicative
competence and to its elaboration in terms of the framework devel-
oped by Canale and Swain (1980) and elaborated by Canale (1983).
The Canale and Swain framework includes four competencies:
grammatical competence, the understanding of vocabulary, pro-
nunciation, and syntax; sociolinguistic competence, the knowledge
of speech acts and the appropriate use of language; discourse comp-
etence, the ability to produce coherent and cohesive texts; and
strategic competence, the ability to enhance communication or deal
with breakdowns in it. The editors have chosen not to include a sec-
tion on grammatical competence, an area which, they note, has
already received a great deal of attention. They add sections on
communicative competence in first language acquisition and in the
workplace and on developing communicative competence in the
language classroom.
This book is useful for ESL teachers and researchers, and for
courses in L2 acquisition and research design. The questions at the
end of each section are mostly useful for testing the reader’s
comprehension of chapter contents although some address research
design. One drawback is that the chapters are of unequal quality
and only Scarcella’s refers to other chapters. This leads to a curious
lack of discussion about the usefulness of the Canale and Swain
framework or about whether the new competencies included in this
book have features which distinguish them from the original ones.
In addition, the decision to illustrate Canale and Swain’s much
quoted framework with papers not written specifically for this
volume poses a problem discussed only briefly in the introduction—
one chapter seems to describe both discourse and strategic
competencies. This raises the question of whether there are in fact
two separate competencies.
In the section on first language acquisition, Anderson’s “Acquiring
Conversation Competence: Knowledge of Register Variation”
addresses the pressing question of whether this competency is
developmental in nature. She finds children first using prosody,
then topic and the lexicon, and finally syntactic adjustments to
indicate register. Although Anderson does not connect her work to
nonnative speakers, we could conjecture that since these features
are clearly salient in child language acquisition, ESL teachers may


want to work on register recognition in approximately the same
In “Pragmatic Transfer in ESL Refusals,” part of the section on
sociolinguistic competence, Beebe, Takahashi, and Uliss-Wetz find
that in Japanese, excuses are generally more vague than in U.S. Eng-
lish and this formula can be inappropriately transferred when Jap-
anese speakers make excuses in English. The authors cite question-
naire responses to a situation in which a salesman tries to bribe the
boss of a company by inviting him to an expensive restaurant. One
Japanese respondent wrote the excuse: “I have to go to Europe
soon.” When teaching Japanese speakers, ESL teachers might need
to explain that U.S. English speakers tend to name the place, time,
or parties involved in a conflicting engagement.
The section on discourse competence addresses assessment. In “A
Quantitative and Qualitative Analysis of Speech Markers in the
Native and Second Language Speech of Bilingual,” Olynak,
D’Anglejan, and Sankoff underscore the usefulness of examining
hesitation phenomena for their effect on fluency. They find that
speakers use the same speech markers in their second language that
they do in their first language. It appears student performances in
oral proficiency interviews should be considered in light of their
performances in their L1.
The section on strategic competence offers some interpretations
of this elusive and broadly defined category in two task-oriented
studies and an ethnographic one. Yule and Tarone’s chapter,
“Eliciting the Performance of Strategic Competence,” emphasizes
the importance of comparing native speaker (NS) and nonnative
speaker (NNS) responses to various research designs to avoid
interpreting every omission or mistake by an NNS as a lack of one
competence or another.
The section on developing communicative competence in the
language classroom will be especially welcome to ESL teachers
because it applies theory to practice. Scarcella’s “Communication
Difficulties in Second Language Production, Development, and
Instruction,” written for this volume, reviews literature on cultural
conventions that could pose problems in cross-cultural conversa-
tions and then suggests useful guidelines for teachers.
One area absent from this volume, but important to many
teachers who use communicative methodologies, is research on
conversational adjustments. Schegloff, Jefferson, and Sacks (1977)
show conversational adjustment to be a common occurrence in NS-
NS interaction. It is also common to NS-NNS interaction (e.g.,
Fiksdal, 1989) and NNS-NNS interaction (Varonis and Gass, 1985).
This volume attempts to illustrate the work done in the field in a
representative way, and it succeeds for the most part. The authors

will, I believe, achieve a stated purpose: The book should encour-
age active exploration of communicative competence by both
researchers and teachers.

Canale, M. (1983). From communicative competency to communicative
language pedagogy. In J. C. Richards & R. W. Schmidt (Eds.),
Language and communication (pp. 2-28). New York: Longman.
Canale, M., & Swain, M. (1980). Theoretical bases of communicative
approaches to second language teaching and testing. Applied
Linguistics, 1, 1-47.
Fiksdal, S. (1989). Framing uncomfortable moments in cross-cultural
gatekeeping interviews. In S. Gass, C. Madden, D. Preston, &
L. Selinker (Eds.), Variation in second language acquisition (190-207).
Philadelphia, PA: Multilingual Matters.
Schegloff, E., Jefferson, G., & Sacks, H. (1977). The preference for self-
correction in the organization of repair in conversation. Language, 53,
Varonis, E. M., & Gass, S. M. (1985). Task variation and nonnative/
nonnative negotiation of meaning. In S. M. Gass & C. G. Madden
(Eds.), Input in second language acquisition (pp. 149-161). Rowley, MA:
Newbury House.

The Evergreen State College


The TESOL Quarterly welcomes short evaluative reviews of print and nonprint
publications relevant to TESOL professional. Book notices may not exceed 500
words and must contain some discussion of the significance of the work in the
context of current theory and practice in TESOL.

Discourse Analysis for Language Teachers (Cambridge Language Teach-

ing Library). Michael McCarthy. Cambridge: Cambridge University
Press, 1991. Pp. x + 213.

This volume, whose author is highly experienced in the field, is a

practical and informative introduction to what discourse analysis is and
how it is relevant to language teaching.
There are six chapters in all. A first overview chapter (covering topics
ranging from speech acts, to conversational analysis, to textual cohesion) is
followed by ones on grammar and vocabulary (both of these are
predominantly Hallidayan); phonology (a balanced assessment of work on
intonation by Bolinger, G. Brown, Brazil, and others); spoken language
(from adjacency pairs to anecdotes, with an interesting section on the
important issue of “transactions and topics”); and written language.
Throughout these chapters, clear explanations of native-speaker discourse
patterns and phenomena pave the way for helpful commentaries on how
those same patterns may cause difficulties for L2 learners. Appended to
each chapter is a briefly annotated list of suggestions for further reading on
specific issues, which will be a boon to master’s-level students and their
teachers; these lists and the referencing throughout are admirably up-to-
All the key issues (e.g., goals and authenticity in transcription; top-down
model-driven theory versus bottom-up attention to specific discoursal
“troubles”; the role of signaling devices in discourse structuring; genre-
based characteristics and constraints) seem to be addressed—again, in an
appropriately clear and straightforward way. Consequently, the theoreti-
cal controversies do not displace the very reasonable main concern, which
is to show how a discourse-analytic perspective can greatly enrich lan-
guage classes for both teachers and learners.
The book is teeming with excellent Reader Activities: samples of
discourse (from a range of sources, oral or written), helpfully prefaced by
questions or pointers which alert the reader to those aspects of textual
content or structure the author wishes to foreground. Together with
changes of typeface, formatting, and so on, the Reader Activities help the
book to practice what it preaches—a stimulating back-and-forth in which
author and reader take turns doing something and keeping the
conversation going. And, as a “third turn” or feedback move, there are

useful suggestions at the end of the book as to how one might complete the
Reader Activities, with appropriate reminders that these are only possible
responses and that there is “no single right answer.” For me, those Reader
Activities are the book’s winning feature: Although the author does not set
out to demonstrate specific ways of using discourse analysis in teaching,
anyone who works through these activities will have acquired a range of
texts and procedures (and the ability to identify related discourse samples)
which can be so used.
One of the virtues of this book is its acknowledgment of a range of
useful language-focused studies contributing to the understanding of that
multifaceted process, discourse. Samples and examples are overwhelm-
ingly from or keyed to British English or Englishes (i.e., many varieties
besides RP and “good” usage), but this finally seems a strength rather than
a weakness, even for non-Britons: the reader is always examining directly
related voices and styles, rather than Hawaiian chalk and Ulster cheese.
For the student approaching the area without a linguistics background,
it is an ideal book with which to begin, before proceeding to more detailed
or technical works; while for those soon heading back to the language
classroom, the book offers just about the right amount of coverage on its

University of Washington

Sound Advantage: A Pronunciation Book. Stacy A. Hagen and Patricia E.

Grogan. Englewood Cliffs, NJ: Prentice Hall, 1992. Pp. xviii + 200.

Stacy Hagen and Patricia Grogan have succeeded in creating a

comprehensive pronunciation textbook that is accessible to teachers who
are not pronunciation specialists. Designed for high-intermediate and
advanced learners, the book may be particularly useful to university
programs for international teaching assistants, which often have an
emphatic need to boost speech intelligibility. The text also offers focused
preparation for the Educational Testing Service’s SPEAK Test.
Central to the authors’ approach is the belief that bottom-up listening
skills exercises must be integrated into a useful pronunciation text.
Throughout the book we see the two skills, pronunciation and listening,
building on one another. Another distinctive and admirable feature of the
text is the way it introduces the articulation of new sounds. Here the
authors usefully incorporate aspects of Catford’s (1987) “intensive silent
introspection” technique.
The complicated nature of pronunciation and discrete-sound listening
skill development, and the comprehensiveness of the text, come through in
the sheer number of language points considered. The following are some
of the less commonly found aspects of pronunciation covered: unstressed
syllables with nonreduced vowels, nonreleased final consonants,


consonant combinations in phrases and sentences, stress timing, reductions
in consonant clusters, and, in one of the appendices, general rules for
predicting stress. In fact, as a result of handling so many important
concepts, the text lacks somewhat in practice and review. But teachers will
be pleased to note that the publisher offers, free of charge to instructors,
supplementary instructional materials.
Concerning methodology, this already excellent text could be improved
by occasionally incorporating into pronunciation instruction the
opportunity for imitation. The learner might listen to a (meaningful/
authentic) model, record an imitation of it, listen and compare the two,
record another imitation, and finally respond to a teacher evaluation. Of
course, the materials offered (the text is supplemented by a cassette tape
series) can be fashioned into exercises following this procedure.
Sound Advantage makes a solid base upon which to build. The book is
designed as a companion to Hagen’s earlier Sound Advice: A Basis for
Listening (1988), but the incorporation of listening skills into the new text
allows it to stand on its own. An Instructor’s Manual and the
aforementioned supplementary materials are available free of charge.

Catford, J. C. (1987). Phonetics and the teaching of pronunciation: A systematic
description of English phonology. In J. Morley (Ed.), Current perspectives on
pronunciation (pp. 83-100). Washington, DC: TESOL.
Hagen, S. A. (1988). (Ed.), Sound advice: A basis for listening. Englewood Cliffs,
NJ: Prentice Hall.

The Pennsylvania State University

Language Transfer: Cross-linguistic Influence in Language Learning

(Cambridge Applied Linguistics Series). Terence Odlin. Cambridge:
Cambridge University Press, 1989. Pp. xii + 210.

As ESL teachers, we are often aware of the different ways of talking and
thinking of our students from various language backgrounds. One wonders
if these differences result from students’ native languages, and if so, how
serious such influence is in the study of a second language. Further, how
well can we, as teachers, predict and control the influences? Odlin’s
Language Transfer: Cross-linguistic Influence in Language Learning
provides some insightful answers to these questions. Like the other books
in the Cambridge Applied Linguistic Series edited by Michael Long and
Jack Richards, the subject of discussion is closely related to language
teaching and learning and is of direct concern and interest to both
researchers and language teachers.


Adopting a comprehensive approach to language transfer, the author
starts with a historical overview which leads the reader through the
dramatic change from the 1950s, when transfer was considered the most
important factor in learning a second language, to the 1960s, when
empirical investigations suggested that little or no evidence of transfer
from the learner’s native language can be observed. This introduction sets
the stage for a more enlightened and balanced discussion of the subject.
With a broad audience in mind, the author focuses on the fundamental
issues of transfer and presents them in a quite straightforward way.
Clearly, the author has made efforts to avoid extensive theoretical
discussions and keep out highly abstract and controversial concepts and
jargon such as markedness, parameter setting, schema, and so on. Readers
with little background in linguistics will find the book readable and
helpful. It is a good place to begin work on the general subject before
embarking on more in-depth and more specific studies.
One of the strong features of this book is its clear layout. The discussion
is conducted at various levels—discourse, semantics, syntax, phonetics,
phonology, and writing systems—which enables the reader to go straight
to any of the subdomains of interest. In each of these subdomains, the
author incorporates presentation and analysis with data obtained by
empirical studies and demonstrates in a convincing way the existence of
transfer and its subtlety.
Although it is an excellent reference and a useful guide for the
understanding of the issues surrounding language transfer, pedagogical use
of this book is limited because, as the author points out, “relatively little is
known about the best way to make use of transfer research in the
classroom” (p. xi). Maybe this is the reason why language teachers should
contribute to the field in areas where researchers often fail.

Northern Arizona University

Understanding Computers: A Text for Developing Critical Reading,

Thinking, and Reasoning Skills in English. Ellen McGill and Donald
DiCristoforo. New York: Collier Macmillan, 1989. Pp. xxii + 272.

The principal goal of Understanding Computers is not only to help

students develop critical reading, thinking, and reasoning skills in English
but also to give learners the opportunity to acquire basic computer
knowledge. Students can get essential and detailed information on the
following topics regarding computers in the text’s five units: “What We
Know About Computers,“ “Analyzing Problems and Creating Solutions,”
“Changing Solutions into Programs,” “How Computers Perform
Programs,” and “Computers in our Lives.”
According to its authors, this book may be used as a primary or
supplemental text in programs of English for general purposes, reading


courses, or in classes preparing students for computer-related course work.
Nevertheless, the preface suggests that it may be more appropriate for
advanced secondary students preparing for postsecondary programs
(native or nonnative English speakers), students in universities, colleges,
and other postsecondary institutions (native or nonnative English
speakers), and computer professionals wanting to develop English
language skills (nonnative English speakers).
Since the book contains clear information and instructions on its use (in
the Introduction, Glossary, and Answer Key), no previous background is
needed. In addition, a Student’s Self Study Guide and a Teacher’s Guide
provide detailed explanations about the entire course material.
Most of the 15 chapters are divided into five sections: Thinking and
Discussing, Reading, Checking Comprehension, Understanding Lan-
guage, and Discussing and Writing. The Thinking and Discussing section
presents prereading questions which prepare learners for active discussion
about what they will be reading. In the Reading section, paragraphs are
numbered to promote the development of skimming and scanning skills.
Style and vocabulary tend to be simple in order to encourage reading
fluency. The Checking Comprehension section presents different kinds of
exercises related to
extracting main ideas and specific information, recognizing inferences and
drawing conclusions, making generalizations, and recognizing patterns of
organization common to academic texts, such as comparison/contrast, cause/
effect, purpose, classification, and time and spatial relationships. (p. xv)
The Understanding Language section deals with “technical and nontech-
nical vocabulary development, word families, reference and signal (tran-
sition) words, restatement of ideas, affixes, antonyms, synonyms, specific
determiners, and connotation versus denotation” (p. xv). The Discussing
and Writing section is designed to promote discussion and role plays. Writ-
ing activities vary from descriptions of important memories to writing let-
ters and short papers.
If we consider this book as a primary text for general purpose English,
it will represent a useful guide only from the point of view of the language
learning process and product. Regarding the content dimension it will
prove too limited. Nevertheless, the most interesting features of
Understanding Computers is that reading is viewed as an interactive
process in which all language skills are integrated.

Purdue University


The TESOL Quarterly invites readers to submit short reports and updates on their
work. These summaries may address any areas of interest to Quarterly readers.
Authors’ addresses are printed with these reports to enable interested readers to
contact the authors for more details.


San Francisco State University

Bias-free Teaching as a Topic in a Course

for International Teaching Assistants
Vanderbilt University

■ ESL professionals working with international teaching assistants (ITAs)

can help shape the future of science and technology in the United States.
Three interrelated trends lead to this conclusion. The first trend is the
increasing shortage of scientists and engineers in the United States
(Malcom, 1990; “Shortage,” 1990). This trend is accompanied by the
second, an increasing shortage of faculty in the scientific and technological
fields (Mooney, 1989). The third trend is the increasing recognition, based
on a growing body of research, that many potential scientists and
engineers are discouraged from pursuing careers in these areas by the
behaviors—probably mostly unconscious—of science and engineering
faculty (Malcom, 1990). The unconscious discriminatory behavior begins
early and lasts beyond graduate school: As Shirley M. Malcom (1990) of
the American Association for the Advancement of Science writes, “there
exists a vast and strangely invisible talent pool that remains virtually
untapped.” This pool consists of minorities, girls and women, and the
handicapped, who, according to her, have received education in science
that has been “most dismal” (p. 112).
ESL professionals working with ITAs may have a significant impact on
the training of future scientists and technicians in the United States because
the shortage of science, mathematics, and engineering faculty is predicted
to lead to a greater proportion of foreign-born faculty being hired
(Mooney, 1989). Many of these faculty will have been teaching assistants in
U.S. universities, as is clear from the sheer numbers of foreign students
earning PhDs in science and engineering—in 1988 about 50% of the
recipients of PhDs in engineering were non-U.S. citizens (Watkins, 1990).
How can ESL teachers make a difference? We can help by introducing
into our ITA training programs lessons and activities designed to

encourage a commitment to providing equal opportunity to all categories
of students in their classes. A 1982 publication of the Association of
American Colleges, “The Classroom Climate: A Chilly One for Women?”
(Hall, 1982) provides convincing evidence that we need to address this
question. The report documents that both obvious and subtle discrimina-
tory actions, as well as unrecognized patterns of behavior, lead to unequal
opportunity for well-qualified women, as compared to nonminority males,
in university math, science, computer science, engineering, and other
technical programs. Most of the discriminatory faculty behaviors
described in this report are, in all likelihood, unconscious. As the authors of
Removing Bias: Guidelines for Student-Faculty Communication (Jenkins,
Gappa, & Pearce, 1983) write, instructor bias “does occur in the classroom,
not deliberately, but because of misinformation or a lack of awareness”
(p. 4). Some discriminatory behaviors are behaviors of commission; for
example, responding more favorably to questions from male students than
to those from female students. Some are behaviors of omission, such as not
asking female students to participate in research projects or not crediting
female students for their ideas as often as crediting male students for theirs
with such “authoring;’ references as, “As John pointed out” (Hall, 1982).
The “chilly climate” found for women exists as well for other groups not
traditionally found in the sciences, such as African Americans and
Hispanics (Jenkins, Gappa, & Pearce, 1983).
Those of us who teach ITAs can help them to learn how to create
classroom climates that are equally beneficial to all categories of students.
We can do this by encouraging attention to bias when doing activities that
focus on classroom interaction patterns, both verbal and nonverbal. By
doing so, we can create an awareness of how interaction patterns reflect
unconscious biases and stereotypes.
Such a solution was attempted in a semester-long course for ITAs which
I have taught at two institutions. The course, although called Advanced
Diction, in fact includes training in cross-cultural issues, oral presentation
techniques, and teaching techniques, as well as English language
instruction. The course gives a good deal of attention to how teachers ask
questions, how they respond to answers (correct or incorrect), and how
they respond to questions from students. In addition, the course includes
observation activities in which ITAs note in detail the behaviors of
university teachers by observing them both in the classroom and on video.
The purpose of the observation activities has always been to make the TAs
more aware of the following: (a) the variety of teacher styles, (b) the effect
of various teacher styles and behaviors on their students, (c) the
importance of nonverbal behavior in communication, (d) the variety of
ways in which teachers and students interact, and (e) the unconscious
nature of much of a teacher’s behavior. Recently, I decided to follow the
observation segment of the course with a segment that would consider
classroom equity.
At the time this new segment was introduced, the class consisted of 13
ITAs, 12 male and 1 female. All were TAs in math, science, or


technological fields. Five of them were from the People’s Republic of
China or Taiwan, 5 from India, and 1 each from Morocco, Greece, and
Korea. Most had been in the United States for only a few months.
In introducing the topic of classroom equity, I made a great effort not to
convey to the ITAs that I assumed they were all more racist, sexist, or
ethnocentric, than their U.S. counterparts. In fact, I began by claiming that
all people, including all teachers, have biases, and explained the meaning
of the word bias. I admitted to having biases of my own. Nevertheless, I
continued, it was the job of teachers in U.S. universities and colleges, in
spite of their biases, to give equal opportunity to all the students in their
classrooms. I explained that it was the U.S. ideal (though not an actuality)
that all groups have equal educational opportunity: The groups I named
were women, racial and ethnic minorities, the handicapped, older
students, and college-age white males. Terms other than bias which relate
to equal opportunity were defined and commented upon: equal
opportunity, equity, discrimination, nondifferential treatment, sexual
harassment, and stereotyping. Following this, I distributed a handout that
described some of the many ways in which teachers discriminate unfairly,
usually unknowingly, against certain categories of students. (Much of the
information on this handout came from the Hall, 1982, report. ) We
reviewed the handout and together related many of the behaviors
described in it (such as giving minimal responses to the comments and
questions of certain categories of students) to the work we had done
previously on student-teacher interactions, both verbal and nonverbal. I
also reminded the students of one of the major insights they had gained
from their teacher-observation activities, that is, that so much of a teacher’s
behavior is unconscious. I asked them in the future to use their observing
skills to notice whether or not their teachers interacted similarly with all
categories of students. The activity concluded with an appeal to them to
be as aware as possible of their own biases and, in spite of them, to attempt
to provide equal encouragement and support to all categories of students
in their classes and labs.
The student response to the segment of the course devoted to classroom
equity was gratifying. First, there did not appear to be any defensiveness
on the part of the ITAs, all but one of whom were male. One reason for the
lack of defensiveness may have been that I admitted to having my own
biases. This seemed to achieve my aim of linking the ITAs and myself in
the common enterprise of attempting to become better teachers. In
addition, throughout the discussion, I used the inclusive we: “We all have
biases, but we all need to be aware of our behaviors so that we do not
discriminate unfairly as teachers.” Another reason for the lack of
defensiveness may have been that the ITAs identified with the groups of
students who have traditionally been discriminated against: When I
mentioned that they might have felt discriminated against as international
students, I heard a murmur of assent. (In the future, if the same response
occurs, I may encourage further discussion of the discrimination they have
encountered and how it made them feel.) The ITAs also appeared to


accept enthusiastically the ideal of equal opportunity for all. Earlier
discussions in the class about who goes beyond high school in their
countries and in the U.S. had introduced them to the fact that in the U.S.
the accepted attitude appears to be that everyone should be able to go as
far educationally as he or she is able (Althen, 1966). On the whole, the ITAs
appeared to approve of this stance. Another reason the ITAs were
receptive may have been that they felt that they were acquiring valuable
information when the terms relating to equity and discrimination were
explained; they may have believed that this information would help them
to negotiate successfully in U.S. culture.
In the course evaluation completed by the ITAs, the portion of the
course described as “Talking about promoting equity in the classroom”
received a high score, 5.33 on a Likert-like scale, with 1 being low and 6
being high. One student commented, “I appreciate that someone let me
know something about this and I will try to avoid differential treatment for
different groups in my teaching.” In sum, the ITAs appeared to be
receptive to the notion that providing equal educational opportunity for all
their students was part of their obligation as teaching assistants.
It would be naive to claim that a few class sessions will turn all ITAs into
models of equitable teachers. But at least they might become more aware
of how they behave toward their students, both in the classroom and out,
and it is hoped, attempt to treat all groups of students equally. Ideally, all
TAs and all faculty, not only international TAs, would participate in
activities designed to increase sensitivity to equity issues, but clearly such
a circumstance is beyond the authority of ESL professionals to bring
about. Yet ESL professionals can increase the sensitivity of students in their
ITA classes, who will become a significant percentage of the university
faculty of the future.
The topic of avoiding discrimination in the classroom will become a
regular part of the ITA course on the campuses where I have taught this
course, but attempts will be made to improve how it is taught. Most useful
would be videotapes or films illustrating discriminatory behaviors in a
university setting. I was unable to locate such tapes. Certainly, videotapes
sensitively designed for a university setting would be worth many hours of
lecture and discussion. Another improvement, one easier to implement,
would be the development of a questionnaire by which the ITAs could
analyze and evaluate their own teaching behaviors and the climate of their
own classrooms and labs.
The addition of a component dealing with bias-free teaching seems to
me an important addition to the standard kind of curriculum for ITA
courses which has been developing over the past few years. ITAs are
currently in a position to seriously affect the numbers of women and
minorities who go into the sciences, mathematics, engineering, and other
technical fields; moreover, many of our ITAs will become the United
States’ future professors and will have even more effect. ESL professionals
can thus play an important role in helping present and future students of
ITAs receive equal opportunities to learn and to succeed.


Althen, G. (1988). Manual for foreign teaching assistants. Iowa City, IA: University
of Iowa.
Jenkins, M. M., Gappa, J. M., & Pearce, J. (1983). Removing bias: Guidelines for
student-faculty communication. Annandale, VA: Speech Communication
Hall, R. M. (1982). The classroom climate: A chilly one for women? Washington,
DC: Association of American Colleges.
Malcom, S. M. (1990, February). Who will do science in the next century?
Scientific American, p. 112.
Mooney, C. J. (1989, January 25). Uncertainty is rampant as colleges begin to
brace for faculty shortage expected to begin in 1990’s. Chronicle of Higher
Education, pp. A14-A17.
Shortage of scientists seen if action isn’t taken. (1990, January 3). Chronicle of
Higher Education, p. A2.
Watkins, B. T. (1990, January 3). Half of 1987-88 Ph.D. graduates owed no money
for education, study finds.Chronicle of Higher Education, pp. A13-A14.

Author’s Address: Department of Teaching and Learning, George Peabody

College for Teachers, Vanderbilt University, Nashville, TN

Discourse Style and Patterns of Participation

on ESL Interactive Tasks
University of Panama in Chiriqui

Second language acquisition theory, with its emphasis on input and

interaction, has brought about changes in the roles of ESL teachers and
students, and the types of activities they engage in. Increasingly common
are communicative tasks performed by small groups and pairs. The
advantage of this type of activity is that students produce and receive
more language than in traditional, teacher-directed classes and, more
importantly, they negotiate meaning—to avoid or repair communication
breakdowns, they rephrase the message using synonyms, paraphrases,
explanations, and examples. Researchers such as Long (1981), Long and
Sato (1983), Hatch (1983), and Doughty and Pica (1986) believe that these
modifications facilitate second language acquisition. Accordingly, ESL
classroom research has recently focused on the discourse of participants
during small-group tasks to determine under which conditions negotiation
is most likely to occur.
In a series of studies, Pica and Doughty (1985a, 1985b), Doughty and
Pica (1986), and Pica (1987) found that the two most important variables
were task type and teacher participation. More negotiation occurred in
two-way tasks in which participants had to exchange information with


each other than in discussion tasks in which participants shared all the
information from the outset. In addition, more negotiation occurred when
the teacher was not a participant.
In discussing the results, Pica (1987, p. 16) strongly implied that the need
to exchange information and the equal social role of the participants led to
balanced conversations. Doughty and Pica (1986) also found, however,
that when the teacher was absent from discussion tasks, one or two more
assertive students took over the teacher’s role and dominated interactions
so that less assertive students barely participated.
Many questions remain. For instance, could more assertive students also
dominate two-way tasks? That is, while all speakers have to participate to
solve the task, more assertive students could still monopolize conversations
by managing the direction and content of the interaction, including the
negotiation. The research cited above was not designed to address these
concerns. Yet if small-group communicative tasks continue to gain
acceptance in the ESL classroom, we must know more about the dynamics
of interaction: how students participate and which factors influence them.
More and varied studies are needed.
The following pilot study was designed to complement previous
research on students’ patterns of participation in small-group tasks. It does
so in two ways. First, it is a qualitative rather than a quantitative analysis,
and second, it explores different questions, the principal of which is
whether the discourse style of the students is another significant factor in
determining the manner in which they interact.

Ten ESL students were studied: 6 from the University of Florida’s
English Language Institute and 4 from the University of Panama’s Centro
Regional Universitario de Chiriquí.
The subjects were classified by two of their instructors as among the 5
most active (assertive) or nonactive (nonassertive) members of their class.
The subjects were then organized into sets in which a “nonactive” speaker
was paired alternatively with an “active” and a “nonactive” partner, for
Set 1
Pair 1: Student 1 (Nonactive) + Student 2 (Nonactive)
Pair 2: Student 1 (Nonactive) + Student 3 (Active)
Students were given 5 commonly used communicative tasks:
Task 1: Two-way, speakers found differences between their pictures
Task 2: One-way, one speaker described a picture for the other to draw
Task 3: Nondirectional, speakers ordered a list of candidates


Task 4: Nondirectional, speakers narrated a story about a picture
Task 5: Nondirectional, speakers solved a current events problem

All sets were audiotaped for two versions of each task, one with each
pair of the set. The order of the pairs alternated after each task so that the
first pair to do Task 1 would be the last to do Task 2, and so forth. After
finishing all 5 tasks in English, students from the University of Panama
repeated the same procedure in Spanish, their native language.
Transcripts of the first 15 minutes of the conversations were prepared
for analysis.

The analysis consisted of two parts. In the first, the speech production of
each speaker per task was measured by counting the T units, fragments,
phrases, and words, The second part, by far the most important, consisted
of a qualitative analysis of three aspects of discourse
1. Conversation management: Directing and controlling the conversation
2. Information presentation: Contributing ideas to the conversation
3. Interaction: Working with the other speaker to solve the problem jointly
Conversation management refers to overt moves to control the conver-
sation by regulating the other speaker’s participation or establishing proce-
dures for carrying out the task. Information presentation refers to how
much content each participant contributed. It was broken down into three
categories: paraphrases of immediately preceding ideas; information obvi-
ous from the context; and new information such as judgments, examples,
and opinions. Finally, interaction refers to the degree both speakers par-
ticipated in making decisions towards accomplishing the task.


The results were similar in all sections. Students judged to be verbally
active dominated by managing conversations, contributing information,
and making decisions more than their nonactive partners. Moreover,
speakers maintained similar patterns of participation regardless of the
discourse styles of their partners, the task type, or the language.
First, in active-nonactive pairs, passive speakers followed a respond and
react mode. The little they did contribute was often cut off or overruled by
their active partners:
Task 1 (Two-way)
E (Active): umm what more you have what color is (continues)
Y (Nonactive): we have— (Alvarado, 1990, p. 51)


However, nonactive speakers did no better with nonactive partners.
Nonactive speakers spoke no more and sometimes less with nonactive
than with actives, and nonactive-nonactive conversations tended to be
short or rambling, with little direction or on-task discussion. Thus when
paired, neither nonactive speaker took an assertive role.
In regard to task type, actives effectively dominated their nonactive
partners on the 3 discussion tasks and on the two-way task. In addition, 2
of the active speakers dominated on the one-way task, even though the
nonactive had all the information (the picture). The actives, nevertheless,
dictated what to elaborate, when to move on, and how to proceed,
maintaining their passive partners in response roles:
Task 2 (One-way)
J (Nonactive): draw the sweater
L (Active): bu— but what kind of sweater, the sweater
have flower, has line . . . or is white or or black (Alvarado,
1990, p. 101)
Finally, students who performed the tasks in the first and second
languages maintained similar patterns of participation in both. Indeed, the
degree of domination was often more marked in the native language.

The finding that speakers maintained similar patterns of participation
with different partners, task types, and languages is evidence that
discourse style is another significant factor in students’ performance on
interactive tasks. If true, important implications follow. For research, the
discourse style of the students should be considered in future studies. For
methodology, (a) the task type alone (e.g., two-way tasks) may not ensure
balanced conversations, (b) pairing nonassertive students may not
improve the quantity nor the quality of their participation, and (c) practice
alone may not be enough to improve students’ performance on small-
group interactive tasks.
The purpose of this pilot study was not to present definitive answers but
to explore new questions about students’ interaction in small-group ESL
tasks. Given the nature of the results and the importance of the
implications, I suggest that this line of research be pursued.

This pilot study is based on my doctoral dissertation, defended in December 1990
at the University of Florida. I appreciate the guidance of my committee, especially
the chair, Robert de Beaugrande.


Alvarado, C. (1990). Discourse styles and interactive tasks in the classroom
acquisition of English as a second language. Unpublished doctoral dissertation,
University of Florida, Gainesville.
Doughty, C., & Pica, T. (1986). “Information gap” tasks: Do they facilitate second
language acquisition? TESOL Quarterly, 20 (2), 305-325.
Hatch, E. (1983). Psycholinguistics: A second language perspective. Rowley, MA:
Newbury House.
Long, M. H. (1981). Questions in foreigner talk discourse. Language Learning, 31,
Long, M. H., & Sato, C. J. (1983). Classroom foreigner talk discourse: Forms and
functions of teachers’ questions. In H. W. Seliger & M. Long (Eds.), Classroom
oriented research in second language acquisition (pp. 268-265). Rowley, MA:
Newbury House.
Pica, T. (1987). Second language acquisition, social interaction, and the classroom.
Applied Linguistics, 8, 3-21.
Pica, T., & Doughty, C. (1985a). Input and interaction in the communicative
language classroom: A comparison of teacher-fronted vs. group activities. In
S. Gass & C. Madden (Eds.), Input and second language acquisition
(pp. 115-132). Rowley, MA: Newbury House.
Pica, T., & Doughty, C. ( 1985b). The role of group work in classroom second
language acquisition. Studies in Second Language Acquisition, 7, 233-248.

Author’s Address: A.P. 595, David, Chiriqui, Republic of Panama


The TESOL Quarterly invites commentary on current trends or practices in the
TESOL profession. It also welcomes responses or rebuttals to any articles or
remarks published here in The Forum or elsewhere in the Quarterly.

Comments on Antony John Kunnan's

“DIF in Native Language and Gender
Groups in an ESL Placement Test”
The Limits of Biased Item Analysis

University of Chicago

Differential item functioning (DIF) occurs when some special

ability comes into play in answering an item on an examination
other than the one that the item was” intended to test. A typical
example of this is when the examinees’ native languages enable
them to identify the correct answer to a question on a vocabulary
test because there is a cognate in L1 with meaning similar to the
word being tested. In this situation, it is common to say that the item
is “biased” towards speakers of a particular language. A more
neutral preferred term these days is DIF, but both terms will be
used interchangeably here. Being able to detect DIF and to
eliminate test items that exhibit it is important, especially when test
results contribute to important decisions such as admission to
school, licensing, job qualification, and promotion. We would like
to know that the decisions we make are based on sound data that
represent actual measurements of the quality that we are intending
to measure. Antony John Kunnan’s brief report in the TESOL
Quarterly (Vol. 24, No. 4, Winter 1990) is admirable in what is
intended, but the fact that statistical and subjective procedures for
identifying DIF cannot be relied upon dilutes the significance of his
conclusions and of the conclusions of nearly all studies which
investigate the causes of DIF.
There is nothing wrong with Kunnan’s method of identifying
biased items, which relies on selecting items which lie outside the
95% confidence interval. In fact, most statistical tests of significance

and hypothesis tests involve comparing data to known distributions
and deducing the likelihood of those data occurring by chance
alone. In the social sciences, if such results could have occurred by
chance no more than 1 time out of 20, the results are considered
“significant” and may be published in a scholarly journal. However,
even significant results may actually be the result of chance and not
the result of what is being investigated; at the 5% level of
significance, we expect this to happen 1 time out of 20. In the same
way, if we use the 5% significance level to identify items that are
characterized by DIF, we expect 5% of the items, even if there is no
bias, to be selected by this method.
Luppescu (1991) has verified this principle in a simulation study
with data representing the performance of 1000 people on a 75-item
test. In the simulation, 900 of the people were imagined to belong to
the population majority group, and 100 to a minority group. The
data were simulated so that there was no bias favoring either of the
groups. The Rasch analysis difficulty shift method of detecting DIF
was used. In this method, the item difficulties are obtained for each
of the groups separately. (See Wright & Stone, 1990, for details on
this procedure.) Items for which the difference in item difficulties
for the two groups is greater than 2 times the combined standard
error are considered to be candidates for DIF. 1 In this simulation of
1000 people and 75 items with zero bias, 6 items were identified as
containing DIF. That is, despite the fact that there was no DIF at
work in the data, because of the ordinary, expected stochastic
variation, 8% of the items were selected by the procedure used to
identify bias. It is this writer’s position that of the 13 items Kunnan
identified as exhibiting native language group bias, and the 23 items
exhibiting sex bias on the 150-item test, at least 7 or 8 items do not
display DIF but were selected by the procedure because of
ordinary variation. This writer does not claim that there is no DIF
operating at all in Kunnan’s test, but that some items identified by
statistical procedures as biased are actually not, and that it is
impossible to tell by nonstatistical procedures which are and which
are not.
Plake (1980) has looked into the problem of identification of
biased test items by visual inspection. Investigating the Iowa Test of
Basic Skills math concepts subtest, she found that there was very
little agreement between subjective and statistical procedures in
identifying biased items. Moreover, L. Hedges (personal commu-
nication, spring 1991), the chairman of the commission charged


with overseeing the implementation of the so-called Golden Rule
procedures (Linn & Drasgow, 1987; Weiss, 1987) in Illinois insur-
ance licensure examinations, reports that the commission members
were generally unable to identify the sources of DIF in test items
that had been so identified by statistical methods.
No one could oppose eliminating test items which display DIF.
And for this purpose, the statistical procedures used by Kunnan, by
this writer, and by many others, are valuable. When this is done,
however, researchers or test developers are surely aware that they
are necessarily eliminating some items that are not biased along
with those that are. This is an inevitable consequence of the process
of test construction. Ideally, such items will amount to not more
than 5% of the total item pool. Because of this, and because some
items will invariably be deemed unsuitable for other reasons and
will have to be eliminated, the test developer will always prepare
more than the final number of test items. Although it is not possible
to be certain of retaining all the unbiased items, one can usually be
certain on statistical grounds of having eliminated all the items that
display substantial DIF.
However, since some of the items selected by the statistical
process are selected as a result of statistical accident and not
because of the presence of actual DIF, and because it is impossible
by inspection alone to determine which items are biased, Kunnan
may not be justified in drawing the conclusions which he does about
the sources of DIF in the University of California, Los Angeles’ ESL
placement examination.

Linn, R. L., & Drasgow, F. (1987). Implications of the Golden Rule
settlement for test construction. Educational Measurement: Issues and
Practice 6 (2), 13-17.
Luppescu, S. (1991). Eliminating test bias. Unpublished manuscript.
Plake, B. S. (1980). A comparison of a statistical and subjective procedure
to ascertain item validity: One step in the test validation process.
Educational and Psychological Measurement 40, 397-404.
Weiss, J. (1987). The Golden Rule bias reduction principle: A practical
reform. Educational Measurement: Issues and Practice 6 (2), 23-25.
Wright, B. D., & Stone, M. (1990). Identification of item bias (Rasch
Measurement Practice Research Primer No. 5). Chicago: MESA
Psychometric Laboratory.


The Author Responds


The University of Michigan

Though Stuart Luppescu seems to share the common concern of

developing and using tests and test items without differential item
functioning (DIF), there are serious problems with his evidence,
arguments, and interpretation regarding my study, in particular,
and DIF research in general.
At the very outset in his commentary, in the first paragraph, he
concludes with a sweeping statement: “the fact that statistical and
subjective procedures for identifying DIF cannot be relied upon
dilutes the significance of his conclusions and of the conclusions of
nearly all [italics added] studies which investigate the causes of
DIF.” Before responding substantially to this statement, I would
first like to know what research evidence he has to support the
assertion that there is a “fact” about statistical and subjective
procedures for identifying DIF which compels him not to rely upon
them. Second, I would like to know which studies are excluded
from his assertion since he leaves the door open with “nearly all”
Luppescu pointedly criticizes the method I used in my study. My
study was a posteriori analysis of an ESL placement test with 150
multiple-choice items. I used a method similar to the Delta-plot
method within an overall one-parameter item response or Rasch
model (see Angoff, 1982; Angoff & Ford, 1973; Chen & Henning,
1985, for procedural details). In addition to the justification I
provided in my article for this approach, I would add that the Rasch
model is quite suitable for multiple-choice items (see Henning,
1989). Luppescu specifically argues that because I used the 95%
confidence interval for the regression plot to identify DIF based on
item-difficulty indices, “we expect 5% of the items, even if there is
no bias, to be selected by this method.” This is misinterpreting and
confounding the level of significance chosen for statistical tests of
hypotheses with the percentage of items identified as having DIF.
The two are not related and comparable and, therefore, cannot be
used in the manner Luppescu does. Wainer’s (1991) “isthmus of
acceptance” proposal, however, is an interesting one in this regard.
Luppescu further argues that his simulation study of 1000 subjects
and 75 items with’ no bias identified 6 items as containing DIF. He
then concludes that since there was no generated bias in his data,
“ordinary, expected, stochastic variation” identified DIF for items


that did not have such a characteristic. Based on this finding, he
argues that in my study, of the 36 items that exhibited DIF, “at least,
7 or 8 items do not display DIF but were selected by the procedure
because of ordinary variation.” Once again, there are several
problems with this conclusion. First, Burrill’s (1982) excellent
review of simulation studies shows that identification of items that
have induced DIF is quite accurate (see, for example, McCauley&
Mendonza, 1985; Rudner, Getson & Knight, 1980a, 1980b;
Subkoviak, Mack, Ironson, & Craig, 1984). These studies show that
studies based on simulated data do not have an inherent problem in
the way Luppescu argues. Besides, as Burrill (1982) correctly points
out, simulation studies have their limitations.
Second, Luppescu incorrectly assumes that if a statistical
procedure identifies items that display DIF, then those items are
biased. He argues, referring to my study, that “some items
identified by statistical procedures as biased are actually not, and
that it is impossible to tell by nonstatistical procedures which are
and which are not.” The argument he makes here implies that he
sees statistical procedures and nonstatistical procedures as two
separate ways of identifying DIF. Again, this is a misreading of the
ways in which the two procedures work: Statistical procedures are
empirical, internal methods that strictly examine items for DIF in
“context” (that is, item sets must be homogeneous, belonging to the
same content or construct), whereas judgments (by experts or test
reviewers) are external and often made at the item level, ignoring
context or construct (see Coffman, 1982, for directions to review
panels judging the Iowa Tests of Basic Skills). Thus, these two
approaches could yield different though valuable results, but
reliance on any one of the approaches would be wrongheaded.
In addition, the measurement literature is full of caution
regarding total dependence on statistical as well as nonstatistical
approaches in identifying DIF. As Shepard (1982) states, “there is
no foolproof statistical bias detection method. Item bias techniques
themselves require validation” (p. 22). This point of view has been
articulated through validation and reliability studies (for example,
Hoover & Kolen, 1984; and Shepard, Camilli, & Williams, 1985) and
through recent attempts to find the most appropriate method (for
example, Ryan, 1991; Swaminathan & Rogers, 1990; and Wainer,
Sireci & Thissen, 1991). The use of judgments, too, has been
questioned (see Reynolds, 1982, and Sandoval & Miille, 1980)
though test publishers use item-review forms and test-sensitivity
reviews (see Berk, 1982), sometimes to the exclusion of statistical
procedures. Therefore, for the best results, both statistical and
judgmental approaches should be used in combination.


This combinatory approach was used in my study: First, items
that were aberrant for the different native language and gender
groups were identified through statistical procedures using the
Rasch model. These items were then examined for construct or
content differences from other items in the set so that potential
sources of DIF could be hypothesized through nonstatistical
procedures. Three potential sources of DIF (instructional
background, major field, and native language) for 22 (61%) out of
the 36 items were identified, leaving 14 (39%) of the items with no
hypotheses and explanation. Thus, my study, in general, and my
conclusions, in particular, were exploratory and speculative (similar
in approach to Scheuneman, 1987) rather than confirmatory.
To conclude, a DIF study would be seen as critical to language
testing research when it is conceptualized as a special case of
construct validation because if a test or test items exhibit DIF, not
only do the test or the test items have the potential for bias but test
invalidity could occur. And, this would mean that test scores would
be distorted for all groups, and decisions made on the basis of such
results could be invalid. This conceptualization could help
transform the role of DIF research from the narrow focus of being
fair to all groups to the broad view of validating tests and test-score
use. Thus, DIF research could contribute to construct validation as
much as other validation studies which model test performance, test
methods, and test-taker characteristics (for example, Kunnan, 1992).
Then, researchers like Luppescu would consider DIF research not
merely as a way of developing “culture-fair” tests without “biased”
items but as a way of developing acceptable construct validity for
test-score use.

Angoff, W. H. (1982). Use of difficulty and discrimination indices for
detecting item bias. In R. A. Berk (Ed.), Handbook of methods for
detecting test bias (pp. 96-116). Baltimore: The Johns Hopkins
University Press.
Angoff, W. H., & Ford, S. F. (1973). Item-race interaction on a test of
scholastic aptitude. Journal of Educational Measurement, 10, 95-106.
Berk, R. A. (Ed.) (1982). Handbook of methods for detecting test bias.
Baltimore: The Johns Hopkins University Press.


Burrill, L. E. (1982). Comparative studies of item biased items. In
R. A. Berk (Ed.), Handbook of methods for detecting test bias (pp. 161-
179). Baltimore: The Johns Hopkins University Press.
Chen, Z., & Henning, G. (1985). Linguistic and cultural bias in language
proficiency tests. Language Testing, 2, 155-163.
Coffman, W. E. (1982). Methods used by test publishers to “debias”
standardized tests: Riverside Publishing Company/Houghton Mifflin. In
R. A. Berk (Ed.), Handbook of methods for detecting test bias (pp. 240-
255). Baltimore: The Johns Hopkins University Press.
Henning, G. (1989). Does the Rasch model really work for multiple-choice
items? Take another look: A response to Divgi. Journal of Educational
Measurement, 26, 91-97.
Hoover, H. D., & Kolen, M. J. (1984). The reliability of six item bias
indices. Applied Psychological Measurement, 8, 173-181.
Kunnan, A. J. (1992, April). A case for construct validation of tests through
structural modeling. Paper presented at the Department of Linguistics
Colloquium Series, University of Michigan, Ann Arbor.
McCauley, C. D., & Mendonza, J. (1985). A simulation study of item bias
using a two-parameter item response model. Applied Psychological
Measurement, 9, 389-400.
Reynolds, C. R. (1982). Methods for detecting construct and predictive
bias. In R. A. Berk (Ed.), Handbook of methods for detecting test bias
(pp. 199-227). Baltimore: The Johns Hopkins University press.
Rudner, L. M., Getson, P. R., & Knight, D. L. (1980a). Biased item
detection techniques. Journal of Educational Statistics, 5, 213-233.
Rudner, L. M., Getson, P. R., & Knight, D. L. (1980b). A Monte Carlo
comparison of seven biased item detection techniques. Journal of
Educational Measurement, 17, 1-10.
Ryan, K. E. (1991). The performance of the Mantel-Haenszel procedure
across samples and matching criteria. Journal of Educational Measure-
ment, 28, 325-337.
Sandoval, J., & Miille, M. P. W. (1980). Accuracy of judgments of WISC-R
item difficulty for minority groups. Journal of Consulting and Clinical
Psychology, 48, 249-253.
Scheuneman, J. D. (1987). An experimental, exploratory study of causes of
bias in test items. Journal of Educational Measurement, 24, 97-118.
Shepard, L. A. (1982). Definitions of bias. In R. A. Berk (Ed.), Handbook
of methods for detecting test bias (pp. 9-30). Baltimore: The Johns
Hopkins University Press.
Shepard, L. A., Camilli, G., & Williams, D. M. (1985). Validity of
approximating techniques for detecting item bias. Journal of Education-
al Measurement, 22, 77-105.
Subkoviak, M. J., Mack, J. S., Ironson, G. H., & Craig, R. D. (1984).
Empirical comparison of selected item bias detection procedures with
bias manipulation. Journal of Educational Measurement, 21, 49-58.
Swaminathan, H., & Rogers, H. J. (1990). Detecting differential item
functioning using logistic regression procedures. Journal of Educational
Measurement, 27, 361-370.


Wainer, H. (1991). The isthmus of acceptance: A graphical tool for func-
tion-based item analysis and test construction. Journal of Educational
Statistics, 16, 109-124.
Wainer, H., Sireci, S. S., & Thissen, D. (1991). Differential testlet func-
tioning: Definitions and detection. Journal of Educational Measurement,
28, 197-219.

Research Issues
The TESOL Quarterly publishes brief comments on aspects of qualitative and
quantitative research. For this issue, we asked three researchers to address the fol-
lowing question: How should qualitative researchers in our field understand relia-
bility and validity?


University of Hawaii at Manoa

Validity and Reliability in Qualitative Research

on Second Language Acquisition and Teaching

Two Researchers Comment. . .


University of Arizona

Because research on second language acquisition and teaching (SLAT)

draws on and contributes to a variety of disciplines, it is important to study
differing views on the nature of inquiry. Researchers and teacher-
researchers in this field should be able to read, assess, conduct, and benefit
from research with an understanding of different views about what
constitutes high-quality inquiry. Eisner and Peshkin (1990) suggest that
being bimethodological or multimethodological is a mark of scholarly
sophistication. This idea is worth considering for SLAT students who need
to know about methods and standards of inquiry in linguistics, education,
the humanities, anthropology, psychology, sociology, and so on. We focus
here on the notions of validity and reliability as standards in research.
Although our own research perspective is essentially qualitative in nature,
we will argue for the potential utility of auxiliary quantitative procedures
in achieving these standards. We will also argue that qualitative procedures
are important for establishing the validity of research conducted from an
essentially quantitative perspective. In other words, the two approaches
should be seen as complementary rather than mutually exclusive (see, for
example, Jaeger, 1988).
Notions of validity differ substantially in different research traditions,
but the generally accepted view derives from a positivist-realist


philosophical position which identifies validity with truth. In quantitative
research, a test or measure is considered valid if it measures what it is
intended to measure, if one can infer that a score on an indicator directly
represents the construct that the researcher intends to assess. Quantitative
researchers strive for high levels of validity of various types, including
content, face, concurrent, and predictive validity; construct validity, of
central importance to qualitative researchers, is often assumed rather than
Qualitative research does not have a precise equivalent to the notions of
validity in quantitative measurement and experimental research. Rather,
many who write about qualitative approaches hold the view that notions
of truth are problematic (e.g., Guba & Lincoln, 1989). In using the term
qualitative research within the field of second language acquisition and
teaching, we are referring to a range of approaches, such as naturalistic
inquiry, longitudinal case studies, educational ethnography, the
ethnography of communication, discourse analysis, and other approaches
that employ qualitative methods.
In qualitative research, judgments of validity focus primarily on the
interpretation of findings: on the extent to which such interpretations
adequately account for observations in relation to relevant contextual
factors, minimize potential researcher bias, and provide explanatory
coherence within a larger theoretical frame. It is important to emphasize
that validity in either quantitative or qualitative research is not an absolute
notion nor can validity be “proven.” Rather, a high level of validity is a
goal to strive for.
Over the years, some of our colleagues in related fields who operate
within a quantitative paradigm have indicated their belief that validity can
be established internally by using powerful statistical procedures, but we
believe this to be false. From our perspective, internal threats to validity in
quantitative research include underlying assumptions about random
selection, means of data collection, and lack of knowledge about what
kind of questions to ask. Validity problems can come, for example, from
self-report census data which does not consider the distinction between
“ideal” and “real” culture, or the effects of interviewer ethnicity on
responses. Another obvious negative example involves misinterpretation
of the Flanders (FIAC) scale (Flanders, 1970) in multicultural classroom
contexts, where an Anglocentric interpretation of “use of student names”
has been interpreted as positive rapport between student and teacher,
instead of as a negative control measure (as it is in some Hispanic cultures).
Intensive participant observation that gets beyond overt behavior and
methods which otherwise tap the “meaning” of events for participants/
experiences provide important ways to strive for validity in research on
On the other hand, we are inclined to agree with those who point out
that the reliability factor in qualitative research is often neglected (e.g.,
Kirk & Miller, 1986). The reliability of qualitative data is always suspect
because of nonrandom or unexplained sampling, and the validity of


interpretation is suspect because of possible observer bias and observer
effects. Researchers minimally have the responsibility to provide
information concerning themselves, the circumstances of data collection,
and the representativeness of examples presented. Further, sound
qualitative research requires depth of involvement (as opposed to what
Rist, 1980, calls “blitzkrieg ethnography”) and extensive training in
appropriate field procedures.
We believe that quantitative methods may be useful (even essential) in
some aspects of data collection and analysis, especially when variable
features of language are being explored. Our position is, however, that if
quantitative measures are, to be used, they must first be developed and
validated by qualitative procedures. In other words, while we accept the
importance of assessing reliability in research, we believe that the concept
is meaningless if the researcher does not know what s/he is measuring.
We urge particular caution in the use of quantitative measures when
studying people of different cultural backgrounds. Research on language
attitudes, for instance, may call for subjects to make judgments about
speakers’ intelligence, personality, and suitability for particular occupa-
tions (as in “matched guise” studies). Interpreting such judgments pre-
sumes a hierarchy in terms of prestige that must be determined anew for
each culture being investigated. Similarly, determining the social meaning
of code markers requires qualitative study before statistical patterns in
sociolinguistic data can be interpreted, since “quantitative techniques can
only sensibly be applied after a prior examination of the dependencies that
a linguistic variable’s significance has on other aspects of interaction struc-
ture and process” (Brown & Levinson, 1979, p. 333).
Finally, we agree with Eisner and Peshkin (1990) that creative
scholarship requires an openness and willingness to explore many forms of
inquiry. They suggest that “appraising the products of our scholarly effort
should not be in the spirit of ‘This is it,’ but rather, in that of ‘What have we
here?’ and ‘if this, then what possibilities are next?’” (p. 367). We conclude
by endorsing Eisner and Peshkin’s statement that our commitment should
be not simply to qualitative research but to research of quality.

Donna M. Johnson (PhD, Stanford University) teaches in the MA/ESL and SLAT
PhD programs at the University of Arizona. Her current research interests include
interpersonal strategies in written and cross-cultural communication. Her
publications include Approaches to Research in Second Language Learning
(Longman, 1992) and Richness in Writing (edited with D. Roen; Longman, 1989).

Muriel Saville-Troike (PhD, University of Texas) directs the Interdisciplinary PhD

Program in Second Language Acquisition and Teaching at the University of
Arizona and conducts research on first and second language acquisition and
attrition. Her publications include The Ethnography of Communication (Basil
Blackwell, 1989) and Perspectives on Silence (edited with D. Tannen, Ablex, 1985).


Brown, P., & Levinson, S. (1979). Social structure, groups and interaction. In
K. R. Scherer & H. Giles (Eds.), Social markers in speech (pp. 33-62). London:
Cambridge University Press.
Eisner, E., & Peshkin, A. (Eds,). (1990). Qualitative inquiry in education: The con-
tinuing debate. New York: Teachers College Press.
Flanders, N. A. (1970). Analyzing teaching behavior. New York: Addison-Wesley.
Guba, E. G., & Lincoln, Y. S. (1989). Fourth generation evaluation. Newbury
Park, CA: Sage,
Jaeger, R. M. (Ed.). (1988). Complementary methods for research in education.
Washington, DC: American Educational Research Association.
Kirk, J., & Miller, M. L. (1986). Reliability and validity in qualitative research.
Beverly Hills, CA: Sage.
Rist, R. C. (1980). Blitzkrieg ethnography: On the transformation of method into a
movement. Educational Researcher, 9 (2), 8-10.

Another Researcher Comments. . .

University of Hawaii at Manoa

Quantitative and qualitative linguistic inquirers continue to miscommu-

nicate because of differences in philosophical assumptions and, thus, the
definitions they use in talking about research. I hope to clarify here some
assumptions underlying validity and reliability in both paradigms and,
briefly consider their implications for second language acquisition
Probably the most fundamental philosophical difference between quan-
titative and qualitative research concerns the nature of reality (ontology).
According to Lincoln and Guba (1985), the quantitative researcher
assumes that “there is a single tangible reality ‘out there’” (p. 37). This
contrasts with qualitative researchers who assume that “there are multiple
constructed realities” (Lincoln & Guba, 1985, p. 37). These assumptions
about the nature of reality are directly related to the ways in which
quantitative and qualitative researchers view concepts of validity and
Internal validity is defined by quantitative researchers as the best
available approximation of the truth or falsity of a statement (Cook &
Campbell, 1979). The purpose of a research design is to control or
randomize factors which may affect the outcome such as changes due to
historical effects, respondent maturation, and calibration of a research
instrument. Thus, a relationship is postulated, then tested against reality.
In qualitative research, since reality is a multiple set of mental
constructions, to demonstrate “truth value” researchers must show that
their reconstructions in the form of findings and interpretations are


credible to those being researched—credibility becomes the salient test of
reality. To enhance credibility, qualitative researchers use procedures such
as prolonged engagement, persistent observation, and triangulation.
Prolonged engagement and persistent observation involve the investment
of a sufficient amount of time to build trust with respondents, learn the
culture, and test for misinformation introduced by both the researcher and
researched. This permits documenting the multiple influences that may
impinge upon the phenomenon being studied and identifying those
characteristics that are most relevant to the research problem or issue. In
investigating the multiple influences affecting a study, qualitative re-
searchers focus on just those factors which quantitative researchers at-
tempt to control, namely, context and time variables. To ensure credibil-
ity, qualitative researchers also triangulate, utilizing multiple sources,
methods, and investigators. Triangulation may involve the use of multiple
copies of one type of source (e.g., interview respondents), different
sources for the same information (e.g., interview and documents), differ-
ent methods for triangulation such as interviewing, questionnaires, obser-
vations, and testing, or even different investigators who compare data in
efforts to formulate and test emerging hypotheses about the phenomenon
being studied.
External validity for quantitative researchers involves the degree to
which the findings can be generalized to other contexts and/or subjects.
Quantitative researchers try to ensure that the study has high internal
validity and subjects chosen for the study are representative of the
population to which the generalization is to apply. Thus, the belief in one
reality and efforts to ensure that the study is time and context free underlie
the notion that a generalization will apply to all contexts within that same
population. The researcher is required to make a relatively precise
specification of external validity.
For qualitative studies, rather than attempting to specify external
validity, the researcher focuses on transferability. Assuming multiple
realities and the centrality of context, qualitative researchers feel that only
working hypotheses about a particular situation may be abstracted from
the study. The degree to which working hypotheses can transfer to other
times and contexts is an empirical matter, depending on the degree of
similarity between contexts. The researcher is expected to provide a “thick
description” of the study with sufficient details to allow the reader to
determine whether transfer can be considered a possibility. The reader is
expected to accumulate empirical evidence about the contextual similarity
between the described situation and the situation in which transfer is to be
applied. In this way, the responsibility of the original investigator ends in
providing sufficient descriptive data to make such similarity judgments
The degree of reliability for the quantitative researcher inevitably
depends on the extent of internal and external validity—an unreliable
measure cannot be valid. The concepts underlying tests for reliability
include stability, consistency, and predictability, and are typically


demonstrated by replication; if two conditions yield similar findings then
reliability is established.
In qualitative research, rather than assuring reliability, the inquirer
attempts to ensure that findings are dependable, by way of the cyclical
process of formulating hypotheses through multiple methods (triangula-
tion) and then testing those hypotheses in continuing data collection
through prolonged and persistent observation. Qualitative theorists have
suggested additional techniques such as peer debriefing, member checks,
and inquiry audit (Lincoln & Guba, 1985). Peer debriefing involves a
critical analysis of research methodologies, actual data, and hypotheses by
a peer of the researcher during a “debriefing” session. Member checking is
based on the assumption of constructed realities and is performed by an
ongoing process of testing out with informants the research analysis,
interpretation, and conclusions. An inquiry audit is a detailed accounting
of the degree of consistency among methodologies, data, hypotheses, and
interpretations by an external “auditor,” that is, a qualitative researcher
engaged specifically for this purpose.
Second language acquisition studies have tended to focus either on
cognitive or sociocultural factors and, clearly, quantitative and qualitative
research methodologies are needed to understand them both as they affect
language choice and language proficiency for an individual and within a
community or nation. However, use of quantitative and qualitative
research methodologies suggests the need for considering a number of
issues. First, researchers, teachers, journal editors, and others working in
the field need a basic understanding of the assumptions and definitions
used in both quantitative and qualitative paradigms in order to apply
appropriate criteria in judging the quality of research. Second, to
effectively utilize methods, researchers must have a clear understanding of
the philosophical and theoretical assumptions underlying these methods.
Finally, use of multiple methods in a single study assumes knowledge of
the paradigms being drawn on as well as appropriate choice of method
according to the question being asked.
A model of multiple methods applied to sociocultural inquiry is the
research by Fishman, Gertner, Lowy, & Milan (1985) on ethnic/linguistic
revival. In examining the sources for language maintenance across four
U.S. ethnic groups, Gertner, Fishman, Lowy and Milan (1985) utilized
quantitative methods to determine the extent of publications according to
type and language used. To understand the ethnocultural dimensions of
biliteracy acquisition, Fishman, Riedler-Berger, Koling and Steele (1985)
conducted ethnographies of four New York City schools. Studies of both
cognitive and sociocultural factors affecting L2 learning clearly would be
enhanced by efforts on the part of quantitative and qualitative researchers
to acknowledge the criteria appropriate for each paradigm and coopera-
tively engage in inquiries into the complex nature of language acquisition.


Kathryn A. Davis is Assistant Professor in the Department of ESL at the University
of Hawaii at Manoa. Her research and teaching interests include qualitative
methods, language policy and planning, literacy, and bilingual education.

Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design and
analysis issues for field settings. Chicago Rand McNally.
Fishman, J. A., Gertner, M. H., Lowy, E. G., & Milan, W. G. (Eds.). (1985). The
rise and fall of the ethnic revival. New York: Mouton.
Fishman, J. A., Riedler-Berger, C., Koling, P., & Steele, J. M. (1985). Ethnocultu-
ral dimensions in the acquisition and retention of biliteracy: A comparative
ethnography of four New York City schools. In J. A. Fishman, M. H. Gertner,
E. G. Lowy, & W. G. Milan (Eds.), The rise and fall of the ethnic revival
(PP. 377-441). New York: Mouton.
Gertner, M. H., Fishman, J. A., Lowy, E. G., & Milan, W. G. (1985). Language
and ethnicity in the periodical publications of four American ethnic groups. In
J. A. Fishman, M. H. Gertner, E. G. Lowy, & W. G. Milan (Eds.), The rise and
fall of the ethnic revival (pp. 305-342). New York: Mouton.
Lincoln, Y. S., & Guba E. G. (1985). Naturalistic inquiry. Newbury Park, CA: