Applied Linguistics, Volume 31

APPLIN-31(1)Cover.
qxd 1/27/10 3:06 PM Page 1
ISSN 0142-6001 (PRINT)
APPLI ED LI NGUISTICS
ISSN 1477-450X (ONLINE)
Applied
Linguistics
Volume 31 Number 1 February 2010
Published in cooperation with

AAAL American Association for Applied Linguistics
AILA International Association of Applied Linguistics
BAAL British Association for Applied Linguistics
OXFORD
www.applij.oxfordjournals.org
EDITORS
Ken Hyland, Director, Centre for Applied English Studies, KK Leung Building,
The University of Hong Kong, Pokfulam Road, Hong Kong
Jane Zuengler, Nancy C. Hoefs Professor of English, University of Wisconsin-Madison,
6103 Helen C. White 600 North Park Street Madison, WI, 53706 USA
Assistant to Jane Zuengler: Heather Carroll, University of Wisconsin-Madison
REVIEWS AND FORUM EDITOR

Stef Slembrouck, Professor of English Linguistics and Discourse Analysis, Universiteit
Gent, Vakgroep Engels, Rozier 44, B-9000 Gent, Belgium. <stef.slembrouck@ugent.be>
Assistant to Stef Slembrouck: Tine Defour, Universiteit Gent
ADVISORY BOARD
Guy Cook, British Association for Applied Linguistics
Aneta Pavlenko, American Association for Applied Linguistics
Martin Bygate, International Association for Applied Linguistics
Catherine Kneafsey, Oxford University Press
EDITORIAL PANEL
Karin Aronsson, Linköping University
David Block, London University Institute of Education
Jan Blommaert, University of Jyväskylä
Deborah Cameron, University of Oxford
Lynne Cameron, Open University (BAAL Representative)
Tracey Derwing, University of Alberta
Zoltán Dörnyei, University of Nottingham
Patricia Duff, University of British Columbia
Diana Eades, University of New England, Australia
ZhaoHong Han, Columbia University (AAAL representative)
Gabriele Kasper, University of Hawai’i at Manoa
Claire Kramsch, University of California at Berkeley
Angel Lin, City University Hong Kong
Janet Maybin, Open University, UK
Tim McNamara, University of Melbourne
Junko Mori, University of Wisconsin-Madison
Greg Myers, Lancaster University
Susanne Niemeier, University Koblenz-Landau (AILA Representative)
Lourdes Ortega, University of Hawai’i at Manoa
Alastair Pennycook, University of Technology, Sydney
Ben Rampton, King’s College, University of London
Steven Ross, Kwansei Gakuin University
Alison Sealey, University of Birmingham
Antonella Sorace, University of Edinburgh
Lionel Wee, National University of Singapore
Applied Linguistics is published five times a year in February, May, July, September and December
by Oxford University Press, Oxford, UK. Annual subscription price is £254/US$457/E381.
Applied Linguistics is distributed by Mercury International, 365 Blair Road, Avenel, NJ 07001, USA.
Periodicals postage paid at Rahway, NJ and at additional entry points.
US Postmaster: send address changes to Applied Linguistics (ISSN 0142-6001), c/o Mercury
International, 365 Blair Road, Avenel, NJ 07001, USA.
# Oxford University Press 2010
All rights reserved; no part of this publication may be reproduced, stored in a retrieval system, or
transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or
otherwise without prior written permission of the Publishers, or a licence permitting restricted
copying issued in the UK by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road,
London W1P 9HE, or in the USA by the Copyright Clearance Center, 222 Rosewood Drive,
Danvers, MA 01923.
Typeset by Glyph International, Bangalore, India
Printed by Bell and Bain Ltd, Glasgow
Applied Linguistics Journal online
The full text of Applied Linguistics is available online to journal subscribers. Online access has
a number of advantages:
. quality PDFs ensure articles look the same as the print original and are easy to print out
. access is easy—all you need is your subscription number or institutional IP address (see
below)
. online access is available ahead of print publication—so view while you await your print
version!
. access the text wherever you are (or from any part of your institution network if you have
a library subscription)
. perform searches by word or author across the full text of the articles of any part of the
journal
. download articles whenever you choose—you will be able to access past online issues as
long as you have a current subscription
. free sample copy available online
. fully searchable abstracts/titles going back to volume 1
. Table of Contents email alerting service.
The print version will continue to be available as previously. Institutions may choose to
subscribe to the print edition only, online only, or both. Individual subscribers automatically
receive both.
CONTRIBUTORS
There is no need for contributors to format their articles any differently; online files are
produced automatically from the final page proofs of the journal. However, if you know that
an item in your list of references is available online, please supply the URL. If you have your
own website, you are welcome to include the URL with your contact address in your
biodata.
ADVANCE ACCESS
Applied Linguistics now has Advance Access articles. These are papers that have been
copyedited and typeset but not yet paginated for inclusion in an issue of the journal.
More information, including how to cite Advance Access papers, can be found online at
http://www.applij.oxfordjournals.org.
Applied Linguistics
Subscription Information
A subscription to Applied Linguistics comprises Oxford University Press, Great Clarendon

5 issues. Street, Oxford OX2 6DP, UK.
Annual Subscription Rate Email: jnls.cust.serv@oxfordjournals.org.
(Volume 31, 5 issues, 2010) Tel (and answerphone outside normal
Institutional. working hours): +44 (0)1865 353907.
Fax: + 44 (0)1865 353485. In Japan, please
Print edition and site-wide online access:
contact: Journals Customer Services, Oxford
£254.00/US$457.00/E381.00;
Journals, Oxford University Press, Tokyo,
Print edition only: 4-5-10-8F Shiba, Minato-ku, Tokyo 108-8386,
£233.00/US$419.00/E350.00; Japan. Tel: þ81 3 5444 5858.
Site-wide online access only: Fax: þ81 3 3454 2929.
£212.00/US$382.00/E318.00.
Personal. subscribe to
Print edition and individual online access:
£82.00/US$164.00/E123.00. applied linguistics
Please note: US$ rate applies to US & Canada, For new subscriptions and recent single issues
Euros applies to Europe, UK£ applies to UK and Rest only. Current subscribers will automatically
of World. receive a renewal form.
Prices include postage by surface mail, or for
Please complete the form below and return it
subscribers in the USA and Canada by airfreight,
to: Journal Customer Service Department
or in India, Japan, Australia and New Zealand,
(please see above).
by Air Speeded Post. Airmail rates are available
on request. There are other subscription rates Please record my subscription to Applied
available for members of AAAL, BAAL, AILA, Linguistics, starting with Volume__________
and LSA, for a complete listing please visit (Subscriptions start with the March issue and
www.applij.oxfordjournals.org/subscriptions. can be accepted for complete volumes only.)
Full prepayment, in the correct currency, Please send me the following single issue(s)
is required for all orders. Orders are regarded Volume_________ Issue_________
as firm and payments are not refundable.
Name (BLOCK CAPITALS please)
Subscriptions are accepted and entered on
a complete volume basis. Claims cannot be _________________________________________
considered more than FOUR months after Adresss __________________________________
publication or date of order, whichever is later. _________________________________________
All subscriptions in Canada are subject to GST. _________________________________________
Subscriptions in the EU may be subject to City _____________________________________
European VAT. If registered, please supply Country _________________________________
details to avoid unnecessary charges. For
Postcode _________________________________
subscriptions that include online versions,
a proportion of the subscription price may be I enclose the correct payment of (see rates
subject to UK VAT. Personal rate subscriptions above):
are only available if payment is made by £/US/E __________________________________
personal cheque or credit card and delivery is Please debit my credit card:
to a private address. American Express / Mastercard / Visa
The current year and two previous years’ issues (delete as appropriate)
are available from Oxford Journals. Previous
Card number:
volumes can be obtained from the Periodicals
__|__|__|__|__|__|__|__|__|__|__|__|__|__|__|__|__
Service Company at http://www.periodicals.
Expiry date: |__|__|__|
com/oxford.html or Periodicals Service
Company, 11 Main Street, Germantown, NY Signature ________________________________
12526, USA. Email: psc@periodicals.com. Tel: œ Please tick this box if you do NOT wish to
+1 (518) 537 4700. Fax: +1 (518) 537 5899. receive details of related products and services
For further information, please contact: of OUP and other companies that we think
Journals Customer Service Department, may be of interest.
Aims
Applied Linguistics publishes research into language with relevance to real-world problems.
The journal is keen to help make connections between fields, theories, research methods,
and scholarly discourses, and welcomes contributions which critically reflect on current
practices in applied linguistic research. It promotes scholarly and scientific discussion of
issues that unite or divide scholars in applied linguistics. It is less interested in the ad hoc
solution of particular problems and more interested in the handling of problems in a
principled way by reference to theoretical studies.
Applied linguistics is viewed not only as the relation between theory and practice, but also
as the study of language and language-related problems in specific situations in which
people use and learn languages. Within this framework the journal welcomes contributions
in such areas of current enquiry as: bilingualism and multilingualism; computer-mediated
communication; conversation analysis; corpus linguistics; critical discourse analysis;
deaf linguistics; discourse analysis and pragmatics; first and additional language learning,
teaching, and use; forensic linguistics; language assessment; language planning and policies;
language for special purposes; lexicography; literacies; multimodal communication; rhetoric
and stylistics; and translation. The journal welcomes both reports of original research and
conceptual articles.
The Journal’s Forum section is intended to enhance debate between authors and the
wider community of applied linguists (see Editorial in 22/1) and affords a quicker
turnaround time for short pieces. Forum pieces are typically responses to a published article,
a shorter research note or report, or a commentary on research issues or professional
practices. The Journal also contains a Reviews section.
Applied Linguistics is covered by the following abstracting/indexing services: Bibliographie
Linguistique/Linguistic Bibliography, BLonline, British Education Index, Current Index to
Journals in Education, ERIC (Education Resources Information Centre), International
Bibliography of the Social Sciences, ISI: Social Sciences Citation Index, Research Alert,
Current Contents/Social and Behavioral Sciences, Social Scisearch, Sociological Abstracts:
Language and Linguistics Behaviour Abstracts, Language Teaching, MLA Directory of
Periodicals, MLA International Bibliography, PsycINFO, Sociological Abstracts, Zeitschrift für
Germanistische Linguistik.
ADVERTISING
Inquiries about advertising should be sent to Linda Hann, Oxford Journals Advertising,
60 Upper Broadmoor Road, Crowthorne, RG45 7DE, UK. Email: lhann@talktalk.net.
Tel/Fax: +44 (0)1344 779945.
PERMISSIONS
For information on how to request permissions to reproduce articles/information from this
journal, please visit www.oxfordjournals.org/permissions.
DISCLAIMER
Statements of fact and opinion in the articles in Applied Linguistics are those of the respective
authors and contributors and not of Applied Linguistics or Oxford University Press. Neither
Oxford University Press nor Applied Linguistics make any representation, express or implied,
in respect of the accuracy of the material in this journal and cannot accept any legal
responsibility or liability for any errors or omissions that may be made. The reader should
make his/her own evaluation as to the appropriateness or otherwise of any experimental
technique described.
APPLIN-31(1)Cover.qxd 1/27/10 3:06 PM Page 2
APPLIN-31(1)Cover.qxd 1/27/10 3:06 PM Page 1
ISSN 0142-6001 (PRINT)
APPLI ED LI NGUISTICS
ISSN 1477-450X (ONLINE)
Applied
Linguistics
Published in cooperation with

AAAL American Association for Applied Linguistics
AILA International Association of Applied Linguistics
BAAL British Association for Applied Linguistics
OXFORD
www.applij.oxfordjournals.org
APPLIED LINGUISTICS
CONTENTS
Articles
Textual Appropriation and Citing Behaviors of University
Undergraduates
LING SHI 1
Practices of Other-Initiated Repair in the Classrooms of Children with
Specific Speech and Language Difficulties
JULIE RADFORD 25
Style Shifts among Japanese Learners before and after Study Abroad
in Japan: Becoming Active Social Agents in Japanese
NORIKO IWASAKI 45
The Relationship between Applied Linguistic Research and Language
Policy for Bilingual Education
DAVID CASSELS JOHNSON 72
Seizure, Fit or Attack? The Use of Diagnostic Labels by Patients with
Epileptic or Non-epileptic Seizures
LEENDERT PLUG, BASIL SHARRACK, and MARKUS REUBER 94
English in Advertising: Generic Intertextuality in a Globalizing
Media Environment
AN H. KUPPENS 115
A Subject–Object Asymmetry in the Comprehension of wh-Questions
by Korean Learners of English
JIN-HWA LEE 136
REVIEWS
F. Christie and J. Martin (eds): Language, Knowledge and Pedagogy:
Functional Linguistic and Sociological Perspectives
RICHARD BARWELL 156
S. Makoni and A. Pennycook (eds): Disinventing and Reconstituting
Languages
STEPHEN MAY 159
Alison Wray: Formulaic Language: Pushing the Boundaries
PHILIP DURRANT 163
Ruth Wodak, Rudolf de Cillia, Martin Reisigl and Karin Liebhart: The
Discursive Construction of National Identity
ALEKSANDRA GALASIŃSKA 166
NOTES ON CONTRIBUTORS 169

Applied Linguistics 31/1: 1–24 ß Oxford University Press 2008
doi:10.1093/applin/amn045 Advance Access published on 4 December 2008
Textual Appropriation and Citing Behaviors

of University Undergraduates
Downloaded from applij.oxfordjournals.org at Anhui University of Finance and Economics on December 31, 2010
LING SHI
University of British Columbia
This article explores the citing behaviors of 16 undergraduates in a North

American university. After completing a research paper for their disciplinary
courses, each participating student was interviewed to identify in his/her writing
words and ideas borrowed from source texts and to explain why and how the
relevant texts were appropriated with or without citations. Analysis of students’
writing and comments illustrates how they relied on source texts for various
aspects of their essays, some of which they believed required citations while
some of which did not. Results showed that they tried to strike a balance
between the need to cite published authors to gain credit for the scholarly
quality of their writing and the desire to establish their own voice by limiting
the extent to which they cited other texts. Some students also reported how
they chose between quoting and paraphrasing (though the latter sometimes
contained direct copying) on the basis of their ability to rephrase other’s
words and their understanding of the different roles played by the two.
The study indicates the degree to which citational acts are discursive markings
of learning and knowledge construction.
INTERTEXTUALITY
One theoretical perspective for research on students’ citing behaviors is the
concept of intertextuality which suggests that each academic text is populated
with other texts organized to generate new knowledge claims (Fairclough
1992). Researchers have found the concept useful in interpreting how stu-
dents acquire academic literacy. For example, Starfield states that university
students need to learn to interact with various texts which ‘circulate as
currency’ (2002: 125) and develop citation skills to accumulate their own
‘textual capital’ (2002: 126). Young and Gaea (1998) suggest that citation
skills develop along with one’s disciplinary knowledge and students achieve
knowledge transformation by integrating content as interpreted evidence.
While working with multiple source texts, many students, however, have
produced ‘patchwriting’ by copying from a source text and then deleting or
changing a few words and altering the sentence structures, a textual strategy
that leads to accusations of plagiarism (Howard 1992, 1995). Such accusations,
as Pennycook (1996: 226) states, are ‘inadequate and arrogant’ especially to
L2 students from a culture that views borrowing and memorizing others’
words as a legitimate learning strategy. In order to understand issues
embedded in students’ citation practices, Chandrasoma et al. (2004: 171)
2 STUDENT’S TEXTUAL APPROPRIATION AND CITING BEHAVIORS
propose that we should ‘do away with the notion of plagiarism in favour of
an understanding of transgressive or nontransgressive intertextuality’.
PREVIOUS RESEARCH
Reliance on source texts
A number of researchers compared student-generated texts with their source
texts and found that students relied on source texts in their writing (Campbell
1990; Moore 1997; Shi 2004). Among them, Campbell (1990) noted that copy-
ing was a major strategy used by both L1 and L2 writers. In two other studies
(Moore 1997; Shi 2004), L2 undergraduates were found to rely more than
their L1 counterparts on source texts in their summary writing. Close compar-
isons of students’ texts also revealed that L2 students tended to use implicit
attributions, such as ‘it is said’ or ‘the first idea is . . .’, to summarize proposi-
tions compared with their L1 peers who tended to mention the individual
authors explicitly (Moore 1997; Shi 2004). The varied performances of the
participants lead to the question of how students conceptualize textual bor-
rowing or plagiarism.
Inability to identify plagiarized texts

To explore students’ perceptions of plagiarism, a second strand of research
explored students’ abilities to identify plagiarized texts by using question-
naires that contained examples of edited rewritten versions with their
sources (McCormick 1989; Deckert 1993; Roig 1997; Chandrasegaran 2000).
The edited rewritten texts contained what researchers believed to be either
models of correct use of sources or texts that were copied or paraphrased with
or without acknowledgement. For example, McCormick (1989) included three
rewritten samples that had direct copying of phrases of 7, 13, and 15 words,
respectively, without quotation marks. Together, these questionnaire surveys
revealed that many students, either with L1 (McCormick 1989; Roig 1997)
or L2 or a non-western background (Deckert 1993; Chandrasegaran 2000),
had problems distinguishing the ‘plagiarized’ texts. Gray areas surrounding
plagiarism concern the degree of reformulation or paraphrasing. Students,
especially those from an L2 culture, are sensitive to the complexities of plagia-
rism (Pennycook 1994).
Nontransgressive plagiarism
A third strand of research used students’ comments to clarify their inappro-
priate textual borrowing or plagiarized texts identified by researchers. These
studies could be further distinguished based on whether the analysis focused
more on students’ texts (Pecorari 2003, 2006) or interview data (Hull and Rose
1989; Dong 1996; Spack 1997; Currie 1998; Angélil-Carter 2000; Starfield
2002; Chandrasoma et al. 2004; Petrić 2004; Thompson 2005). Those that
L. SHI 3
focused more on students’ voices from interviews were case studies. The rich
interview data in these case studies suggest evidence of unintentional pla-
giarism. Students were reported to defend their unacknowledged use of
source information in terms of how they understood common knowledge
(Chandrasoma et al. 2004) or relied on memorized and internalized knowledge
(Petrić 2004). There was also evidence that inappropriate use of citations was
tied to students’ confusion of how to cite (Thompson 2005), unsuccessful
attempts to develop authorial selves (Starfield 2002), underdeveloped skills
of reading comprehension as well as critical thinking in relation to the authors’
stance (Spack 1997; Angélil-Carter 2000), and limited content knowledge that
hindered them from selecting relevant and important references (Dong 1996;
Chandrasoma et al. 2004). Being primarily concerned with how their teachers
would judge their work, some students actually found patchwriting an effi-
cient strategy for survival (e.g. Hull and Rose 1989; Currie 1998; Chandrasoma
et al. 2004; Petrić 2004). These findings suggest that plagiarism is a problem of
academic literacy rather than that of dishonesty (Angélil-Carter 2000).
The aforementioned case studies, however, provided limited or a less com-
prehensive description of students’ texts. Recognizing this limitation, Pecorari
(2003, 2006) analyzed excerpts of 17 graduate students’ thesis writing in
different content areas. Based on comparisons of passages from student texts
and the cited sources, Pecorari (2003) found that students’ citations were not
transparent as most of them had passages in which 50 per cent or more of
the words were from the sources without acknowledgment or quotations.
However, like those in the case studies, these students indicated no intention
to deceive at the interview. The interview data revealed individual student
writers’ understanding of the nature of source use. Citation is, as Pecorari
(2003: 324) put it, ‘an occluded feature of academic writing’ because the
‘real nature of source use is only known to the writer’. Further exploring
these students’ occluded use of citations, Pecorari (2006) identified that
about 18 per cent of the total source mentions in students’ texts were actually
from secondary sources without acknowledging the original author. Uncertain
about such practice, one student said she relied on her advisor for guidance.
However, since such occluded features were a blind spot for readers, Pecorari
(2006) expressed a concern about how these students might graduate with the
wrong impression that their citation practices were in compliance with the
standard practice.
The present study

The above review suggests that researchers that used both textual and inter-
view data have provided more insightful information since citational acts are
subjective and private. However, these studies have either provided a limited
description of students’ text in terms of quantities and weightings of various
types of use of citations or textual appropriation, or solicited students’ com-
ments based on seemingly plagiarized texts identified by the researcher who
examined passages of students’ texts that contained citations. In the latter

case, portions of texts were excluded from the analyses when they ‘included
no citations’ (Pecorari 2003: 322). What is missing here is a great amount
of textual borrowing not indicated by citations. Since textual borrowing is
a subjective act guided by, what Pecorari (2006) has noted, students’ own
inferred rules, research needs to explore how students identify their own
textual borrowing and explain why they make the borrowing and whether
each borrowing is cited or not cited. Such research could also symmetrically
compare students’ views in terms of how they apply textual borrowing or
citations to quotations, paraphrases or summaries. To tap these issues, the
present study uses undergraduates’ self-reflections to explore three research
questions:
(i) Why do students appropriate source texts and cite them in their
writing?
(ii) Why do students appropriate source texts but not cite them in their
writing?
(iii) How do students apply principles of textual borrowing to quotations,
paraphrases or summaries?
METHOD
Participants
Sixteen undergraduates in a North American university responded to an
advertisement and volunteered to participate in the study. Of the 16 partici-
pants, 4 were science majors and the rest were in Arts and Social Science
(Table 1). The initial letters of the pseudonyms indicate the first language of
the students: ‘E’ for English (Elmer, Edward, and Eddy), ‘C’ for Cantonese
(Carmen, Carl, Candy, Carol, Cathy, and Cary), ‘M’ for Mandarin (Martin,
Mark, and May), ‘J’ for Japanese (Jane), ‘K’ for Korean (Kate), ‘R’ for
Romanian (Rose), and ‘P’ for Polish (Polly). Participants who regarded
English as their second language, including Carmen and Carl who were
born in North America, all spoke their first language at home. Among them,
Jane, Kate, and Rose had just arrived as international students and were
literate in their home language. The rest were not literate in their first
language as they had all or most of their schooling in local North American
schools. These students, unlike those who were literate in their first language
and might therefore bring different ideas about citation practices from their
home cultures, were more like their North American born monolingual peers
in terms of learning to cite, a difficult task for all university students.
Writing assignments
Participants were asked to bring to the interview a research paper they had
just written for a course as well as the source texts they used (see Appendix A
L. SHI 5
Table 1: Participants’ profiles

ID Major Year of study at L1 Age Age arriving
the participating
university
Elmer Commerce 1 English 18 Born in North
America
Edward Commerce 1 English 18 Born in North
America
Eddy History 3 English 29 Born in North
America
Carmen Science 1 Cantonese 19 Born in North
America
Carl Arts 2 Cantonese 19 Born in North
America
Candy Science 1 Cantonese 18 7
Carol Arts 1 Cantonese 17 3
Cathy Arts 1 Cantonese 18 3
Cary Science 1 Cantonese 18 10
Martin Commerce 3 Mandarin 20 10
Mark Science 1 Mandarin 18 12
May Arts 2 Mandarin 19 9
Polly Economics 2 Polish 25 7
Jane Law Exchange Japanese 23 23
Kate English Literature Exchange Korean 21 21
Rose Political Science 2 Romanian 19 18
in the supplementary data for the online version of the article). Depending
on the courses they were taking at the time, most students brought their
research papers for 100 level (first year) courses in English (8), history (1),
film studies (1), and biology (2). The rest brought their research papers for 200
or 300 level (second or third year) courses in political science (3) or women’s
studies (1). The average length of these papers was 1,805 words. The course
handouts that explained the assignments all had a warning against plagia-
rism and directed students to the university or faculty websites on plagiarism.
In other words, the participants were aware of the importance of attributing
appropriated materials in their research paper assignments.
Interviews and data analysis

I met the 16 students separately and conducted discourse-based interviews
(Odell et al. 1983). Each interview lasted for an hour during which the student
was asked to compare his/her own text with the source texts to identify and
comment on texts appropriated, whether cited or not cited (see Appendix B
in the supplementary data for the online version of the article for inter-
view questions). Since citations without quotation marks signal the borrowed
content whereas those with quotation marks signal the borrowed language
in which content is presented, I also solicited students’ comments on their
decisions of using quotes, paraphrases, or summaries for texts appropriated.
The term ‘plagiarism’, though mentioned by some students, was carefully not
used by me so as not to bias students’ views. Based on the interview data,
a total of 187 units of textual appropriation were identified together with
students’ comments. There were occasions when students identified units of
textual borrowing but did not make any comments. Such units were excluded
as the lack of explanation from students in these cases might be a significant
problem worthy of another study.
There is no consensus on how much language must be copied to be deemed
plagiarism. For example, students have been advised to cite when copying a
string with a minimum of three words from a source text (e.g. Hodges 1962).
For the present study, a unit of textual appropriation was defined as a sentence
or several sentences that contained words or ideas borrowed from source texts.
The longest unit in the present data contained seven sentences with a total of
164 words. The boundaries of the units were set by students themselves who
connected each unit to one specific borrowing and citation decision. After
repeated readings of students’ explanations, a coding scheme was developed
based on some key words used by students (e.g. support; new information).
The purpose of using students’ own language was to be more accurate about
students’ perspectives. This approach would also help produce ‘consensual
readings’ of the narrative data (Denzin 1997: 232) and, therefore, catch
‘both the variation and central tendency or typicality’ (Watson-Gegeo 1988:
585). To check intercoder reliability, a research assistant was trained to use
the scheme and then coded 10 per cent of the data on her own. The agreement
between the researcher and the research assistant reached 84 per cent. The
discrepancies were then solved by revising some of the categories.
Table 2 presents the revised coding scheme. Fourteen reasons were identi-
fied under three major factors that influenced students’ use of source texts and
citation decisions. First, the participants were concerned about the functional
or rhetorical role of the borrowed texts in terms of whether it could provide
support, help develop or form one’s own idea or form the basis of a key point.
Such concerns demonstrate that the present students, unlike those in previous
research (e.g. Chandrasegaran 2000), did have an understanding of using tex-
tual borrowing rhetorically to legitimize their own claims.
In addition, the participants were making interpretations of the source
information to determine whether it was new information, fact, research finding,
background information, common knowledge, or information from a credible source.
These reasons indicate that, apart from background information that was identi-
fied in previous research (Pecorari 2003), there were various other inferences
Table 2: Coding of students’ explanations and examples
Explanations Definitions Examplesa Formatb
Functions
1 Support To provide support for one’s point These are all quotes from their websites. Quote
I used them to support what I wrote
here. (Polly)
2 Form one’s own idea To build up one’s own idea It’s part of my own argument . . . I Summary
mean, I took it from their argument
but incorporated it into my own . . .
overarching description of the
situation. (Eddy)
3 Key point To form the basis of one’s key point I used that [citation] because it forms
the basis of my argument. (Elmer)
Interpretations of source text
4 New information New information for the writer I just took all the information and Summary
put it in my own words. . . . I just
cite the author because I didn’t
know any of the stuff really before.
(Edward)
5 Fact Facts or events This is just like facts . . . so I put the Summary
reference here. . . . I am summarizing
her points. (Edward)
6 Research finding Previous research findings I cite that, saying that who did the
research and what they found. (Cary)
7 Background information Background information or I got that from this article. . . . It’s just
established theory in the field like the background information.
(Edward)
I rephrased these from the article. . . . Paraphrase
I mean it’s . . . fairly well established
L. SHI
theory . . . That’s why I kind of

cited the textbook. (Polly)
7
(continued)
Table 2: Continued
Explanations Definitions Examplesa Formatb
8 Credible source Credible or reliable source This is a citation from a book written by Paraphrase
a concert artist. . . . I think he’s very
credible. . . . It’s paraphrased here.
(Cathy)
9 Common knowledge or term Knowledge or words commonly I didn’t cite it because that’s common
known sense. (Mark)
I didn’t cite it because it’s an actual
term that people use. (Martin)
Reasons related to learning
10 Other’s words/ideas Words and ideas directly taken These are the actual words [from Quote
from the source the source]. You have to quote them,
right? (Edward)
11 Result of learning Knowledge accumulated by And this is kind of I just knew from my
learning mum . . . and from previous school . . . I
don’t think I really have to cite it. (Cary)
12 Reference cited earlier A source cited earlier or in the I used my own words and it’s a summary. Summary
in the text or in the reference list I didn’t cite because I already
reference list mentioned it in the first paragraph.
(Carol)
I don’t think I was supposed to cite this.
We just indicate the book that we
used at the end. (Cary)
13 No need to cite everything Not everything needs to be cited Or else I would have to cite every single
sentence. I think it’s kind of distracting.
(Candy)
14 Teachers’ preference Teacher prefers citations They (teachers) prefer you rephrase Paraphrase
and cite it. (Cary)
a
Keywords that result in relevant coding are highlighted.
b
Coding of the example comments in terms of quote, paraphrase, and summary when relevant information is available.
L. SHI 9
that the present students made about which type of source information needed
to be borrowed with or without citations.
Finally, the participants described how their learning experiences influ-
enced their use of source texts and citations. Such experiences included how
they identified words of others, wrote from knowledge accumulated as result of
learning, followed teachers’ preference, and understood that there was no need to
cite everything including references that were cited earlier in the text or in the reference
list. Table 2 shows that the relevant mentions were also coded in terms of a
quote, paraphrase or summary1 when information was available. The fact that
the relevant information was not available for some units will be discussed in
the findings section.
FINDINGS AND DISCUSSION

Since students’ decisions were based on more than one reason for some units,
the 187 units of textual appropriation generated a total of 236 explanations
of why and how source texts were used (Table 3). Over one-third of the
explanations given (85) were related to the functions served by using the
source texts, about one-third (77) were related to the interpretations of pre-
vious works, and slightly under one-third (74) were related to reasons of
learning to cite. Of the 236 explanations provided, 159 (67 per cent) illustrate
why students ended up citing the work. Students also explained how they
applied the principle of textual borrowing by quoting, paraphrasing, or sum-
marizing the source texts a total of 122 (out of 236) times. I will analyze how
participants used source texts, focusing on their understanding of whether
citing was required or not.
Reasons mentioned more frequently for citing texts borrowed

The present students, as a group, seemed to mention some reasons for taking
something from the source text more frequently for citing and others more
frequently for not citing. The only reason that did not favor either citing or
not citing was background information that was mentioned three times for citing
and three times for not citing. Figure 1 presents the eight reasons mentioned
more frequently by students to explain why a unit of textual borrowing was
acknowledged.
Securing support for one’s writing

Support was the most frequently mentioned reason (48) by students for using
source texts with citations (Figure 1). The two occasions when it was men-
tioned for not citing were double coded with reasons of no need to cite everything
and result of learning that typically led to no citations. Among those referring to
support as the main reason for citing, many commented on how they searched
for materials to cite from various published authors to support their arguments.
Table 3: Classification of reasons students gave for using source text,

by whether source is cited or not
Explanation Cited Not cited Total No. of
mentions
No. % No. % No. % of format
of textual
borrowinga
Functions
1 Support 48 96 2 4 50 100 33
2 Form one’s own point 7 30 16 70 23 100 6
3 Key point 12 100 0 0 12 100 10
Subtotal and percentage of 236 67 79 18 21 85 100 49
Interpretations of source text

4 New information 18 82 4 18 22 100 12
5 Fact 9 64 5 36 14 100 7
6 Research finding 5 100 0 0 5 100 1
7 Background information 3 50 3 50 6 100 4
8 Credible source 15 100 0 0 15 100 7
9 Common knowledge or term 0 0 15 100 15 100 2
Reasons related to learning

10 Other’s words/ideas 37 100 0 0 37 100 32
11 Result of learning 0 0 18 100 18 100 1
12 Reference cited earlier or in the 0 0 3 100 3 100 1
reference list
13 No need to cite everything 0 0 11 100 11 100 3
14 Teachers’ preference 5 100 0 0 5 100 3
Total 159 67 77 33 236 100 122
a
Students mentioned in 236 instances how they relied on source texts. Among them, 122 were
accompanied with explanations of whether they used source texts by quoting, paraphrasing,
or summarizing.
The following is a typical example of how Jane took pains to find something
to cite.2
Example 1
A unit of textual appropriation in Jane’s paper titled ‘Domination of the Liberal
Democratic Party in Japanese Politics’:
Keeping the dominance in the diet [parliament] made LDP [Liberal
Democratic Party] able to influence the policies dramatically.
L. SHI 11
50
Frequency of mentions
40
30 With citing
20 Without citing
10
0
rt in
t
ion ct ng e as e
o
po at Fa i ur
c
de nc
pp fi nd so s /i
e re
Su Ke
y
fo
r m
ch le rd re
f
in ear dib wo s 'p
w s e 's r
Ne Re Cr er he
th ac
O Te
Figure 1: Leading reasons given for use of source text, with citing
50
Frequency of mentions
40
30
With citing
20 Without citing
10
0
e
t
g
g
t
lis
in
dg
in
n
hi
po
rn
ce
le
yt
ea
ow
en
n
er
ow
fl
ev
kn
r
fe
to
s
re
te
on
e'
ul
ci
on
e
m
es
th
to
om
R
m
in
ed
r
C
Fo
or
ne
er
o
rli
N
ea
d
te
ci
ce
en
er
ef
R
Figure 2: Leading reasons given for use of source text, when not citing
The dominant party is generally able to get its own way in policy-
making issues in the parliament. (Stockwin 1998: 121)
Jane’s reason for citing, which has been classified as an instance of support:
This is something else that I got from the textbook . . . I have to find
what was said previously to back up my writing. So that’s why
I keep putting in references. . . . I had this idea that if you have
dominance in the Diet, you must have big influence in the par-
liament. I just assume that it should be the case but I was just
worried because it might not be the case. So I just went through
all the books trying to find if somebody was actually saying the
same thing. . . . So those are the little things that I went through
and made me so tired.
The fatigue Jane expressed at the end shows the enormous time and energy
she had spent searching for citations that would support her basic intuition
about parliamentary influence. Jane’s comment suggests that she had learned
to both trust and check her insights, even as she realized that further evidence
beyond her own hunches needed to be found and included in her work. This
contrasts markedly with the participant in Thompson (2005) who felt that it
was necessary to always find a support because, as a student, he had nothing
original to contribute, Jane seemed to have a scholarly sense of following one’s
hunches, with enough confidence that one did not give up until either finding
the evidence or realizing it was not there.
The students’ interest in securing support from source texts was also evident
when in 12 instances participants commented on how they presented a key
point through the voice of a cited author. For example, in explaining how she
argued for the importance of parents’ encouragement for kids learning to play
the piano, Cathy said that she decided to use the citation ‘to show it was an
important factor’. The reliance on source texts to frame important points was
also shown in five instances of students citing research findings and 15 instances
of drawing information from a credible source. Martin believed that citations
‘gave some authenticity or authority over [his] essay’. Like Martin, Carol
explained that she cited from Milroy and Milroy (1999) about how English
was an essential tool for working-class children because ‘quotes or cita-
tions . . . [had] been proven and recognized’. By using such a citation, Carol
said her message was that ‘[i]t is not just what I am saying, it is from a
published author’. These comments reveal how students draw on others to
show that they are not alone in thinking and thus to gain credit for their own
writing.
Learning to cite
Students were also motivated to cite materials for two additional reasons
related to what I am terming learning (Figure 1). As learners, the participants
chose to cite when they recalled teachers’ preference for extensive citation (5) or
L. SHI 13
when they saw source texts as others’ words and ideas that were worth quoting
directly (37). In the following example, Edward explained how he attributed
the ownership of the word ‘localization’ to an author (the co-authors of the
original text were missing in both his writing and comment):
Example 2
A unit of textual appropriation in Edward’s paper titled ‘Nowhere to Turn:
Drug Abuse as a Cause of Homelessness and Poverty in Vancouver’:
In his essay ‘‘Drug Dealers,’’ Caulkins describes the ‘‘localization’’
of certain areas of the city for drug use as a major problem lead-
ing to the augmented organization of drug dealers in Vancouver
(Caulkins 326).
Edward’s reason for quoting, which has been classified as an instance of
other’s words:
I used quotations for ‘‘localization’’ because it’s a word Caulkins
used. It is his word. . . . He describes the ‘‘localization’’ of certain
areas of the city for drug use . . . I usually cite if it’s something that
most people wouldn’t use. . . . Just like if I was to hear ‘‘localization’’
before, I probably wouldn’t know it means that much before
reading.
It is interesting to note that Edward felt that the original author owned the
word because any average person like himself would not have used the
word in such a context. Also regarding themselves as average people, Martin
and Mark said that citing words from published authors would save them
the trouble of providing supporting details. As a student, Martin said that
instructors always cast doubt on ideas he claimed as his own. For example,
in explaining why he cited Hirsch (1990) when writing about Woody Allen’s
family, Martin said ‘If I didn’t cite that, the instructor would say, ‘‘Where did
you get this? Who said this? How did you know who said this?’’’ In compar-
ison, Mark acknowledged the sources when ‘the language is powerful’
because, as he explained, ‘If you just present something really stunning
people won’t believe you’. Mark’s lack of confidence in using his own words
echoed that of L2 students in previous studies (e.g. Angélil-Carter 2000;
Chandrasoma et al. 2004).
Learning ‘facts’ and ‘new information’ in various academic disciplines

Compared with the above reasons, students were divided on whether fact
and new information required citing, although they favored citing (Figure 1).
The participants distinguished facts (things exist or performed) from ideas
and opinions. For example, Edward interpreted chemical characteristics of
drugs as facts that had to be cited, whereas Eddy regarded current or historical
events as facts that needed no acknowledgement. The different decisions made
by students imply that the debt owed to authors for facts differs across
disciplines. If facts about drugs are derived from scientific research, facts of
current or historical events are public knowledge. Students’ citing behaviors
are thus connected to their variable learning and practices in different disci-
plinary communities.
Like the notion of fact, the concept of new information reveals the role
of individual learning experiences. For example, the present students cited
such new information as:
‘Influences from early in our lives such as childhood trauma . . . or alcohol
abuse can contribute to initial experimentation with a drug’ [a citation
given by Edward and the material to which the citation applied came
from Jones (1998)];
‘The most significant side effects of antibiotics, however, is [sic] the
depression of the immune system’ [a citation given by Mark and
the material to which the citation applied came from Hauser and
Reminngton (1982)]; and
that the reformed Indian Constitution ‘articulated the principle of equality
of all citizens irrespective of caste, community, race or sex’ [a citation
given by Rose and the material to which the citation applied came
from Desai (1973)].
One can easily imagine that, on the one hand, some of these students’ peers
might not consider such information as new knowledge and, on the other
hand, the same students might choose not to cite later when they become
acquainted with such information. By comparing students’ citing behaviors,
the present study suggests a dynamic process of learning and claiming owner-
ship of knowledge. Judgements on textual appropriation, whether appreciated
or illegitimate, are grounded in the context of epistemologically and socially
constructed academic literacies.
Reasons mentioned more frequently for not citing

texts borrowed
Since all of the reasons were offered to explain chunks of texts which the
students identified as appropriated, a citation would be expected in every
case. It is, therefore, highly significant to examine cases when students
did not give a citation. Of the 14 reasons for using source texts, five were
mentioned more often for not citing than citing (Figure 2).
Interpreting source information as common knowledge/term

In 15 instances, the present students interpreted source texts as common
knowledge/term and, therefore, did not cite them in their writing (Figure 2).
For example, Cathy said that she did not cite the statement that ‘musical suc-
cess is an innate gift’ which she read from a book because it was ‘something
that [she had] grown up with’. Other appropriated texts that students
L. SHI 15
considered common knowledge included statements that ‘the industrial policy

[of the government] is the driving force for economic growth [of Japan]’
(Kate); that ‘[t]he baby boom that took place from the mid 1940s throughout
France had created, by the middle of the 1960s, many thousands of new
undergraduates’ (Carl); and that ‘the human use of antibiotics helped the
natural selection of bacteria’ (Mark). The following example illustrates how
May generalized Starbucks’ way of preparing coffee to a common sense of
how to keep food fresh:
Example 3
A unit of textual appropriation in May’s paper titled ‘Starbucks Coffee’:
Moreover, Starbucks prepare the coffee beans in a sanitary and safe
environment that keeps the coffee fresh.
May’s reason for not citing, which has been classified as an instance of
common knowledge:
I got this from the book. . . . But actually it’s kind of common sense
because you have to put food in a bag or something so it doesn’t
go stale. I think it’s just kind of related to all the food.
Compared with May and those students who distinguished common knowl-
edge based on their own understandings and interpretations, other students
made similar decisions based on whether the source information was intro-
duced in a textbook (Eddy) or a lecture (Mark), or frequently mentioned in
reading materials (Jane). Jane, for example, did not cite the appropriated text
about the dominance of the Liberal Democratic Party in Japan because, in her
words, ‘I read them in so many places and textbooks so I thought that was
just like a general understanding of this whole issue.’ Such comments illus-
trate that when certain information is recognized by students as public or
owned by many, it becomes base knowledge of which students would claim
ownership. Although previous research has also reported how one student
did not cite common knowledge acquired through personal experiences
(Chandrasoma et al. 2004), the present study provides insights of how a
group of students across disciplines arrive at such an interpretation in each
specific case.
Learning not to cite

Students also attributed some of their decisions not to cite to various learning
experiences (Figure 2). Specifically, they chose not to cite when the appro-
priated texts matched their knowledge accumulated as a result of learning (18)
or was something they believed not worth citing otherwise they had to cite every-
thing (11). To avoid using too many citations, three students (Carol, Cary, and
Martin) said that they did not cite a reference cited earlier in the paper or included
in the reference list. As an example, Cary referred to her learning experiences

when commenting on the following unit of textual appropriation:
Example 4
A unit of textual appropriation in Cary’s paper titled ‘One Type of Spontaneous
Mutations is Point Mutation’:
However, spontaneous mutation occurs more frequently than
induced because spontaneous mutations can occur simply due to
natural radiation, and during replication in DNA.
Cary’s reason for not citing, which has been classified as instances of the result
of learning and no need to cite everything:
Well, I didn’t cite these because I know that there is natural
radiation. . . . I got it from lecture and the books I used. . . . I learned
since I was in high school. . . . I don’t think we really have to
cite everything, right?
Although the above comment might be interpreted as common knowledge, it
is coded as result of learning based on Cary’s own account of how she learned
the information in high school. Like Cary, Eddy explained that he did not cite
the term ‘unitary actors’ because it was ‘a part of lingo of political science
[he had] been studying for a number of years’. Also commenting on the
development of one’s own knowledge based on learning, Carol said, ‘When
you read an article, you absorb the ideas then it becomes your own’. The
participants believed that they were entitled to claim ownership of words
and ideas learnt previously or internalized. As two other students explained
explicitly when commenting on certain texts appropriated, ‘I’ve learned this
in class so it is considered personal knowledge’ (Carmen); ‘I know it before
then I don’t think I really have to cite it’ (Cary). These explanations for not
citing illustrate how citation practices go hand in hand with students’ learning
and accumulation of knowledge.
Claiming authorship of ideas and learning to construct knowledge

Compared with the above reasons, the reason to use course texts to form one’s
own point was bidirectional (for both citing and not citing) though it was men-
tioned more often for not citing than citing (Figure 2). Among those who
mentioned how they appropriated texts to form their own points, Carmen,
in his Biology research paper, explained how he came up with the statement
about ‘a direct relationship between light intensity and accumulation of
newrosecretions’ based on the information from Cymborosky (1983) that
‘accumulation of neurosecretions is the highest during the night when crickets
are supported to move at the highest levels’. In Carmen’s words, ‘I came up
with this based on what this person found. . . . So it was what I inferred from
the study. This was what I came up with myself’. Although it was arguable
L. SHI 17
how different the two statements were, Carmen believed that, as a legitimate
process of learning and constructing knowledge, he was entitled to take the
ownership of the inference he made. Carmen’s choice also reveals, as one
anonymous reviewer of the paper noted by citing from Sinclair (1986) and
Tadros (1993), an intention to aver or to put forward a claim on the basis of
one’s own authority.
Unlike Carmen, other students mentioned how they appropriated texts to
form their own points together with other reasons for either citing or not
citing. The relevant cases further illustrate how students support or claim
their ownership of ideas in the context of learning. For example, Mark cited
a reliable or credible source to support his understanding that antibiotics can be
harmful. In comparison, Cary, in the following example, claimed ownership of
the knowledge of point mutation in DNA as a result of learning and, therefore,
did not cite the source:
Example 5
A unit of textual appropriation in Cary’s paper titled ‘One Type of Spontaneous
Mutations is Point Mutation’:
Point mutation also divides into different categories that include
base-pair substitution in the DNA, and insertion of deletion in a
base pair. Base pair mutation happens when there is a replacement
of one nucleotide in a chain of amino acids. . . .
Cary’s reasons for not citing, which have been classified as instances of forming
one’s own point and result of learning:
I know it from a course last term. I just vaguely remember it.
I think it’s mine, the idea and wording. I think it’s just my idea.
The above examples illustrate how students might make different citation
decisions when appropriating source texts to form their own points. If students’
citation decisions are based on various combinations of reasons related to
their complex process of learning and constructing knowledge, the present
findings highlight the important role of self-reflections in exploring the
subjective act of citing.
Quotations, paraphrases and summaries

A total of 122 (out of 236) mentions of reasons were accompanied with students’
explanations of whether the relevant units of textual appropriations were pre-
sented as quotes, paraphrases or summaries. Most of these identifications
referred to texts cited (Table 4). Of the 122 mentions, only 14 (3 for paraphrases,
11 for summaries) were for texts not cited, which was about 18 per cent of total
mentions (77, Table 3) for not citing. In addition, probably because quotations
were visible, about 86 (out of 122) mentions were for textual borrowing quoted,
leaving only 36 for texts either paraphrased (17) or summarized (19).
Table 4: Citation types, format of textual borrowing, and students’

explanations
Citation types No. of units with explanations of format No. of units Total
with no
explanations
of formata
Paraphrase Summary Quote Subtotal
Cited Not Cited Not Cited

cited cited
Functions
1 Support 3 1 29 33 17 50
2 Form one’s 1 3 2 6 17 23
own point
3 Key point 2 1 7 10 2 12
Subtotal 6 2 3 38 49 36 85
Interpretations of
source text
4 New information 3 3 6 12 10 22
5 Fact 1 2 4 7 7 14
6 Research finding 1 1 4 5
7 Background 1 1 1 1 4 2 6
information
8 Credible source 1 1 5 7 8 15
9 Common 1 1 2 13 15
knowledge
or term
Subtotal 5 2 6 4 16 33 44 77
Reasons related
to learning
10 Other’s 32 32 5 37
words/ideas
11 Result of learning 1 1 17 18
12 Reference cited 1 1 2 3
earlier or in
the reference
list
13 No need to cite 3 3 8 11
everything
14 Teachers’ 3 3 2 5
preference
Subtotal 3 1 4 32 40 34 74
Total 14 3 8 11 86 122 114 236
a
Of the 236 units or instances in which students explained how they relied on source texts,
122 were accompanied with explanations of whether they used source texts by quoting,
paraphrasing, or summarizing, whereas 114 were left unexplained.
L. SHI 19
Limited identifications for formats of textual borrowing

The limited identifications for the formats of textual borrowing not cited or
as paraphrases or summaries are worth noting. During the interviews, when
prompted to identify how they appropriated the source texts, some students
failed to do so with typical responses, such as ‘I remember reading about that
but I don’t remember whether I rephrased it or not’ (Polly), or ‘The idea is
from the source but I don’t remember I used the words or not’ (Carol). Martin
said he could not specify whether an appropriated text about Martin Scorsese
and Woody Allen’s childhood was paraphrased or summarized because he
‘got the idea from different books’. Such responses or excuses could suggest
students’ hesitations in talking to the researcher about how they missed
the citations at the interview, an indication of the epistemological status of
the data.
One might also think that these students need to upgrade their note-taking
skills so as to record source information accurately (e.g. Pecorari 2003).
However, a close reading of students’ comments revealed that their inability
to remember the source was frequently mentioned together with reasons for
using source texts to present result of learning and common knowledge or term
which typically led to no citations. For example, arguing that it was impossible
and unnecessary to remember and cite all source information, Eddy said,
‘I guess my vocabulary was applied to IR (International Relationship). It’s
part of me now. Obviously it was something that I picked up through a
lecturer, through experts, but I couldn’t tell you where I got that because
it’s just part of my vocabulary.’ Eddy’s comment implies that while some
borrowed texts could be accurately cited by improving the note taking process,
others are appropriated without acknowledgement deliberately by students
who choose to establish their own voice by claiming ownership of the rele-
vant knowledge or achievement of academic literacies.
Understanding the roles of summaries, quotes and paraphrases

In instances when students did identify the format of their textual borrowing,
some described how they arrived at such decisions. Students’ accounts suggest
that while summaries were used to report a large amount of information
concisely, quotes and paraphrases seemed to play different roles to individual
students. For example, some students said that quoting was chosen over para-
phrasing when they ‘did not know how to paraphrase it’ (Carl) or could not
‘think of another way to say it’ (Cathy). In comparison, when explaining their
choices of paraphrasing over quoting, Rose and Carmen said that paraphrasing
was the skill preferred at the university compared with quoting which they
learned to use at high school. Two other students, Elmer and Carol, said that
they chose to use paraphrases when citing secondary sources because they
did not know if they could use quotations for such information. In the
following account, Polly said she decided to paraphrase because she believed
that quoting was not appropriate at the beginning of the paper:
Example 6
Polly’s reason for paraphrasing a unit of textual borrowing
I rephrased these from the article. . . . I think I maybe copied that
word for word and then rephrased it. I changed the wording around
a bit and then cited it. Because it’s really early in the paper, and
I figure it would have looked really silly to start quoting . . . It doesn’t
look as nice, I think. It’s really a general introduction and
background.
Concern for plagiarism

While explaining their choices between paraphrasing and quoting, some stu-
dents expressed a concern for plagiarism or an uncertainty about whether they
had paraphrased some source texts enough. Although they believed that quot-
ing was irrelevant when they changed the source text slightly, these students
were uncomfortable when they noted a close resemblance between the source
texts and their writing. As an example, the following is a comparison of the
source text and the unit of textual borrowing that Polly commented on in the
previous quote (words identical in Polly’s text and source text are highlighted):
Example 6 (continued)
A unit of textual appropriation in Polly’s paper titled ‘GM Crops or Franken
Foods: An analysis of the Rhetoric of Activis’:
The first successful cross-species gene transfer to plants took place
in the early 1980’s, and by 1998 more than 26 percent of the cotton
and 40 percent of the soybean acreage in the US was planted with
GM crops. (Shields 2000)
The corresponding source text from Shields (2000: 18):
The first successful gene transfers in plants took place in the
early 1980’s, and by 1998 more than 26 percent of the cotton
and 40 percent of the soybean acreage in the US was planted
with GM crops containing a gene for herbicide resistance.
Noticing the direct copying in their writing such as the above, Polly and other
students (e.g. Carmen, Eddy, Mark), wondered whether they should have
quoted the source texts. In the following comment, Eddy also expressed
uncertainty about whether he should have acknowledged the source when
summarizing the information about how an Iraqi citizen would be eliminated
if he/she openly criticized Saddam Hussein:
I got my idea from The Economist. . . . I mean, I took it from their
argument but incorporated it into my own . . . overarching
L. SHI 21
description of the situation. . . . And I guess that’s sort of a border-

line. If you really want to be strict, then I probably should have
quoted or cited that. . . . Use the judgment call. ‘Am I plagiarizing or
not?’ Maybe I should have cited that.
The concern about plagiarism expressed by Eddy and others further explains
the limited identifications of the format of textual borrowing among the pres-
ent students. Participants might be reluctant to identify how they borrowed
source texts because they were not sure how much to paraphrase, so the
relevant text did not need to be quoted, nor were they sure if it was OK not
to cite when certain information was summarized using entirely one’s own
words. In addition, they might not want to talk about their plagiarized texts
with the researcher who might disagree with them on this sensitive issue. The
epistemological status of the data complicates the findings of a dynamic learn-
ing process of choosing between not only citing and not citing but also quoting
and paraphrasing. As the process is a blind spot for teachers, students have to
rely on their own judgement call in making these choices.
CONCLUSION
The present study illustrates the sophistication and range of the participating
students’ understanding of the role of textual borrowing and citation by ana-
lyzing students’ self-reflections on not only units of textual appropriation cited
but also those not cited. Results show that the citing behaviors of novice
scholarly writers (in the case of undergraduates) are guided by a complex set
of factors including functional uses of cited works, citers’ interpretations of
source texts, a learning process to accumulate one’s own knowledge and tex-
tual capital, as well as a choice between quoting and paraphrasing. The study
indicates the extent to which students attempt to maintain a balance between
a reliance on source texts for support and an attempt to establish their own
voice by choosing not to cite. On the one hand, the participants cited source
texts that contained similar ideas so that their opinions were bolstered and
secured. On the other hand, the participants understood that not everything
needed to be cited and would simply draw on, rather than cite directly, source
texts that matched either common knowledge or what they had learnt pre-
viously. Students’ citational acts are situated in a learning context where cer-
tain information seems supportive or irrelevant, new or learnt, owned by a
particular author or shared by many others. As there are no fixed rules, some
participants wondered when a citation would be appropriate and whether they
should quote or paraphrase texts borrowed.
Citing a source text is more than providing a name and a date; it is a
subjective process of deciding how to make meaning out of the available
resources. Students are still learning the general principles and guidelines.
As they learn the parameters of appropriate use, students can easily go
across the unmarked borders of appropriate borrowing and lapse into
unintentional plagiarism. For example, some people might disagree with Cary
who considered spontaneous mutation that occurs during replication in DNA
as knowledge learnt previously, and therefore did not cite the source when
presenting the information; others would see Polly’s text (Example 6), though
cited, as almost a direct copy of the original text and, therefore, a clear example
of plagiarism. The present study suggests that appropriate or inappropriate
textual borrowing cannot be determined by general context-free criteria.
The judgement is negotiated, localized, and contingent. Given the students’
understanding of appropriate citation found among this sample, students’
textual appropriation, appropriate or inappropriate, could be viewed as
evidence that students are learning this important academic writing skill of
intertextuality and learning about it in ways that relate to the expression
of their own ideas and acquisition of academic literacies.
Like other studies that focus on participants’ post hoc self-reports, the pres-
ent data analyses might have not represented students’ citation practices with
accuracy. Holstein and Gubrium (2004), among many others, have pointed out
that responses from the interviewees represent a dynamic meaning-making
process done in collaboration with the interviewer or in the direction designed
by the interviewer. For example, participants might have also used ‘familiar
narrative constructs’ to comment on their textual appropriation rather than
providing the lived experiences or ‘meaningful insights into their subjective
view’ (Miller and Glassner 2004: 127). As I pointed out earlier, some students
might have hesitated to comment on their textual borrowing for fear that
the interviewer would disagree with their citation practices. Readers are thus
advised to be aware of the epistemological status of the interview data.
Despite the limitations, the study sheds some light on how and why
some students borrow texts and, therefore, provides teaching implications.
Instructors are advised to use specific examples or cases, such as those identi-
fied in the present study, to help students learn how to make value judgements
around the use of prior texts based on the degree to which those texts belong
to others, or represent either new or common knowledge. Since there is no
hard and fast set of rules on citation practices in scholarly texts, students need
assistance or direct instruction. I tried out this teaching strategy in a university
writing workshop where the discussion on whether to cite or not to cite in
each case aroused heated discussions. Many students said it was their first
experience to discuss, share, and clarify their individual and subjective acts
of textual borrowing.
The present study also generates implications for follow-up research. One
research focus could be on how citing behaviors mark high-level students’
(juniors, seniors, or grad students) development of disciplinary knowledge.
Case studies comparing students’ self-reflections on their use of citations in a
longer period of time would identify how students learn to use citations as
standard symbols of their discipline and shape their own positions among
recognized networks of references. Another research focus could be the
L. SHI 23
teacher’s perceptions of the students’ citing behaviors in terms of, for example,
how successful the student writers are in learning the conventional citing
forms and whether they are right to omit citations for information they
think is common knowledge. Relevant findings would reveal discipline-
related differences that may well speak to distinctions that instructors in
these fields may wish to address directly with students. Together, these stud-
ies should build up a theory of citing as a process of learning for novice
academic writers.
SUPPLEMENTARY DATA
Supplementary data is available at Applied Linguistics online.
ACKNOWLEDGEMENTS
This study is part of a larger project on students’ textual appropriation funded by a Standard
Research Grant of Social Sciences and Humanities Research Council of Canada. I thank the
participating students, Sin Heng Celine Sze for her help in transcribing and analyzing the data,
and John Willinsky, Lynne Earls, and the anonymous reviewers for their valuable feedback on
an earlier draft of the article.
NOTES
1 One student (Kate) referred to four 2 In all examples cited in this article,
of her units as translations from a students’ writing and comments are
source in her first language. Another presented verbatim. For easy read-
student (Martin) said he created a ing, keywords in students’ comments
quote himself but attributed it to that result in relevant coding are
a credible author. Since these cases highlighted and alternatives for
raise different issues and also repre- special terms in student writing are
sent a small proportion of the data, provided in brackets. See Appendix C
I decided not to deal with them in in the online version of this article for
the present analyses. cited works in student writing.
REFERENCES
Angélil-Carter, S. 2000. Stolen Language? of plagiarism,’ Asian Journal of English
Plagiarism in Writing. Essex: Pearson Language Teaching 10: 91–113.
Education Limited. Chandrasoma, R., C. Thompson, and
Campbell, C. 1990. ‘Writing with others’ words: A. Pennycook. 2004. ‘Beyond plagiarism:
Using background reading text in academic transgressive and nontransgressive intertex-
compositions’ in B. Kroll (ed.): Second Language tuality,’ Journal of Language, Identity, and
Writing: Research Insights for the Classroom. Education 3: 171–93.
Cambridge: Cambridge University Press. Currie, P. 1998. ‘Staying out of trouble: apparent
Chandrasegaran, A. 2000. ‘Cultures in contact plagiarism and academic survival,’ Journal of
in academic writing: students’ perceptions Second Language Writing 7: 1–18.
Deckert, G. D. 1993. ‘Perspectives on plagiarism writing,’ Journal of Second Language Writing 12:
from ESL students in Hong Kong,’ Journal of 317–45.
Second Language Writing 2: 131–48. Pecorari, D. 2006. ‘Visible and occluded cita-
Denzin, N. K. 1997. Interpretive Ethnography: tion features in postgraduate second-language
Ethnographic Practices for the 21st Century. writing,’ English for Specific Purposes 25: 4–29.
London: Sage Publications. Pennycook, A. 1994. ‘The complex contexts of
Dong, Y. R. 1996. ‘Learning how to use cita- plagiarism: a reply to Deckert,’ Journal of Second
tions for knowledge transformation: non- Language Writing 3: 277–84.
native doctoral students’ dissertation writing Pennycook, A. 1996. ‘Borrowing other’s words:
in science,’ Research in the Teaching of English text, ownership, memory, and plagiarism,’
30: 428–57. TESOL Quarterly 30: 201–30.
Fairclough, N. 1992. Discourse and Social Change. Petrić, B. 2004. ‘A pedagogical perspective on
Cambridge: Polity Press. plagiarism,’ NovELTy 11: 4–18.
Hodges, J. C. 1962. Harbrace College Handbook. Roig, M. 1997. ‘Can undergraduate students
5th edn. New York: Harcourt, Brace & World determine whether text has been plagiarized?’
Inc. Psychological Record 47: 113–23.
Holstein, J. A. and J. F. Gubrium. 2004. Shi, L. 2004. ‘Textual borrowing in second
‘The active interview’ in D. Silverman (ed.): language writing,’ Written Communication 21:
Qualitative Research: Theory, Method and Practice. 171–200.
London: Sage Publications. Sinclair, J. M. 1986. ‘Fictional worlds’
Howard, R. M. 1992. ‘A plagiarism pentimento,’ in M. Coulthard (ed.): Talking about Text.
Journal of Teaching Writing 11: 233–45. Birmingham: University of Birmingham ELR.
Howard, R. M. 1995. ‘Plagiarism, authorships, Spack, R. 1997. ‘The acquisition of academic
and the academic penalty,’ College English 57: literacy in a second language: a longitudinal
788–806. case study,’ Written Communication 14: 3–62.
Hull, G. and M. Rose. 1989. ‘Rethinking reme- Starfield, S. 2002. ‘ ‘‘I’m a second-language
diation,’ Written Communication 6: 139–54. English speaker’’: negotiating writer identity
McCormick, F. 1989. ‘The Plagiario and the and authority in Sociology One,’ Journal of
professor in our peculiar institution,’ Journal Language, Identity, and Education 1: 121–40.
of Teaching Writing 8: 133–45. Tadros, A. 1993. ‘The pragmatics of text averral
Miller, J. and B. Glassner. 2004. ‘The ‘‘inside’’ and attribution in academic texts’ in M. Hoey
and the ‘‘outside’’: finding realities in inter- (ed.): Data, Description, Discourse. London:
views’ in D. Silverman (ed.): Qualitative HarperCollins.
Research: Theory, Method and Practice. London: Thompson, C. 2005. ‘‘Authority is everything’:
Sage Publications. a study of the politics of textual ownership and
Moore, T. 1997. ‘From test to note: cultural knowledge in the formation of student writer
variation in summarization practices,’ Prospect identities,’ International Journal for Educational
12: 54–63. Integrity 1. Available at http://www.ojs.unisa.
Odell, L., D. Goswami, and A. Herrington. edu.au/journals/index.php/IJEI/article/view/18.
1983. ‘The discourse-based interview: a Accessed 15 December 2007.
procedure for exploring the tacit knowl- Watson-Gegeo, K. A. 1988. ‘Ethnography in
edge of writers in non-academic settings’ ESL: defining the essentials,’ TESOL Quarterly
in P. Mosenthal, L. Tamor, and S. Walmsley 22: 575–92.
(eds): Research on Writing. New York: Longman. Young, K. M. and L. Gaea. 1998. ‘Writing from
Pecorari, D. 2003. ‘Good and original: plagiarism primary documents: a way of knowing in
and patchwriting in academic second-language history,’ Written Communication 15: 25–68.
Practices of Other-Initiated Repair in the

Classrooms of Children with Specific
Speech and Language Difficulties
JULIE RADFORD
Institute of Education, University of London
Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Repair practices used by teachers who work with children with specific speech
and language difficulties (SSLDs) have hitherto remained largely unexplored.
Such classrooms therefore offer a new context for researching repairs and con-
sidering how they compare with non-SSLD interactions. Repair trajectories are
of interest because they are dialogic sites where the child’s meaning is being
negotiated and, therefore, where adults might create opportunities for language
learning. The interactions take place during activities, such as story writing,
where teachers elicit children’s ideas and orient to their lack of clarity. From a
data set of 78 cases, four significant patterns of teacher repair initiation emerged.
First, non-specific repair initiators (RIs), such as ‘say that again’, target any
aspect of the prior turn and reveal the adult’s lack of grasp of its content.
Next, specific RIs (‘she has’) that are constructed with minimal components of
the child’s turn, pinpoint the location of the trouble but provide no new lexical
information. In contrast, specific RIs that are constructed as ‘wh’ questions
(‘down where’), target the nature of the trouble and elicit further information.
Finally, offers of candidates (‘do you mean X’) do provide new models of lexis
but do not elicit repetition from the child.
INTRODUCTION
Many children in schools experience specific speech and language difficulties
(SSLDs) (Dockrell et al. 2006), also widely referred to as children with specific
language impairment. These children are characterized, in particular, by lexical
and grammatical language difficulties (Leonard 2000) that may impact on
interpersonal communication. Education is typically delivered in schools
where they receive additional resources from trained language teachers and
speech and language therapists/pathologists (Law et al. 2000). The priority of
specialised professionals is to develop the children’s understanding and use of
grammar and vocabulary, to develop interpersonal communication skills and
to facilitate access to the curriculum (Dockrell and Messer 1999).
The fact that children with SSLDs spend a large proportion of their time in
regular classrooms, where the vehicle of instruction is primarily oral (Edwards
and Westgate 1994; Cazden 2001), places them at a significant disadvantage.
Their dilemma for accessing the curriculum is exacerbated when teacher–pupil
discourse is primarily targeted at whole class groups, such as during
26 REPAIR PRACTICES IN SSLD
mathematics and literacy lessons (Mroz et al. 2000; Smith et al. 2004). Arguably
the most important factor concerns the quality of the interaction itself.
International research continues to report the prevalence of interrogative
and evaluative styles of talk in classrooms (Alexander 2001), despite many
efforts to foster the increased participation of children during interactions. A
well-documented system, called the ‘IRE/F’, constrains the participation of
teachers and students (Sinclair and Coulthard 1975; Mehan 1979). It is char-
acterized by the teacher initiating a ‘test’ question (I) that is followed by a
known answer/response (R) (Macbeth 2004). The third turn position, follow-
ing the child’s response, is where evaluation (E) typically occurs, although
there is potential for other devices that build upon the child’s turn, such as

follow-up (F) or extension moves (Wells 1993) and uptake questions
(Nystrand 1997). In IRE/F discourse contexts, however, children with
SSLDs run the risk of becoming passive and lacking engagement in their learn-
ing (Radford and Ireson 2006). Minimal participation may lead to reduced
opportunities for language learning, because topic-related exchanges are
shortened and trouble sources may remain unresolved (Sadler and Mogford-
Bevan 1997).
An alternative to interrogative and evaluative talk is dialogic discourse that
is evident during some types of classroom activities, for example in France,
England and India (Skidmore 2000; Alexander 2001). Dialogic interactions
provide higher quality learning experiences for children than talk that is
teacher-led (Mercer 1995, 2000; Mercer and Littleton 2007). Dialogic talk is
characterized by (at least some) questions that provoke thoughtful answers,
the sharing of ideas between teachers and students, children’s free articulation
of ideas and the chaining of ideas for a common purpose (Mercer 2000;
Alexander 2006). Children with SSLDs could benefit from dialogic discourse
provided that the language is accessible and allowance is made for their expres-
sive difficulties. There is preliminary evidence that specialist language teachers
engage in dialogic teaching with children with specific language difficulties
during small group activities, such as story writing and news telling (Radford
et al. 2006). Drawing on studies into the co-construction of topic using con-
versation analysis (CA) (Button and Casey 1984, 1985; Radford and Tarplee
2000; Ridley et al. 2002), the authors demonstrate how the children and their
teachers co-construct topic on a turn-by-turn basis.
Two examples are now provided that are relevant to the data analysed in the
current study because children’s own ideas are generated in these contexts.
Owing to the children’s language difficulties novel ideas, as we shall see in the
later analyses, frequently necessitate the initiation of repair. The resulting
trajectory of talk that deals with the repair is, arguably, an example of dialogic
discourse. The first example, an ‘invitation’ sequence, has been found when
teachers construct a story with a group of children [Extract (A), taken from
Radford et al. 2006]. Invitation sequences differ from question-with-known-
answer sequences because, although they are suited to constraining the child’s
response in terms of adherence to a story-line, they accomplish the generation
J. RADFORD 27
of the child’s novel story ideas. As such, the teacher is not predicting precisely
what ideas the child will offer, and therefore what s/he will subsequently
receive as a relevant contribution.

Extract (A) begins with the teacher’s (T) story starter line (‘there was a’), that is
designed to be incomplete and, therefore, completed by a nominated child. The
child (line 2) treats T’s turn as a request to contribute an idea concerning the
story’s character. Other examples of next turns include children’s ideas rele-
vant to the story’s title, the setting or the plot. The teacher’s response to the
child’s ideas is not an evaluation, but an ‘acceptance’ which can take a variety
of forms. In line 3 it is a repeat of the child’s idea and, in some cases, the verbal
idea is drawn onto a flipchart to make a visual representation of the story
constructed so far (lines 5 and 6).
The second example is found during a one-to-one speaking-book
activity where the teacher and child are discussing pictures and photographs
that the child has chosen. These pictures have been brought from the
child’s home and the teacher has not seen them before. Extract (B) shows
the teacher using a topic initial elicitor (TIE) (Button and Casey 1984)
which is heard by the child as a request to offer ideas. The TIE sequence differs
from a question-with-known-answer sequence because the child treats it
as an opportunity to choose his/her own topical material, albeit within the
constraints of relevancy to the picture. In the case of photographs, the
child may offer personal news ideas whereas, in the case of pictures,
the ideas may concern a description of any detail of the picture (Radford
et al. 2006).
In line 2, the teacher’s TIE at first appears to resemble an instruction, although it

could take the form of a request for a telling ‘can you tell me. . .’. However, the
child treats it as an open inquiry and offers her own ideas to describe the picture
(line 2). The teacher responds to the ideas with topicalization of the material
generated so far (Button and Casey 1984; Radford and Tarplee 2000). Here we
find ‘mhm’, a continuer that does the job of topicalizing in this context. The
child orients to the topicalizer and the silence with continuation of her descrip-
tion (line 6). If, instead of topicalization, the teacher had elected to evaluate the
child’s response and asked a test-like question, the child would have been
denied the opportunity to pursue the topic that she had nominated herself.
As invitation and TIE sequences are suited to the generation of ideas by the

children, they provide potentially richer resources for language learning than
known-answer questions that generate short, unexpanded responses.
Sequences that have further potential, but have not yet been investigated,
are where acceptance or topicalization of the children’s ideas is withheld.
Instead, the participants suspend the usual sequence in order to deal with a
trouble (see Schegloff 1972, on insertion sequences). Children with SSLDs,
owing to the very nature of their linguistic problems, produce turns that con-
tain grammatical or lexical errors or risk being unclear to the listener. When
teachers initiate repair (usually) in the next turn, the insertion sequence deals
contingently with the child’s trouble. During the insertion, the teacher might
provide correction (Jefferson 1987), which in language teaching would take
the form of a more competent model of grammar or lexis, as compared with
the child’s version. Another possibility is that the teacher might initiate repair
such that the child is placed to revise her/his original move.
This article aims to explicate, in detail, a series of practices where
teachers and children are dealing with repair, when acceptance or topicaliza-
tion of the child’s idea is withheld. More specifically, the article will pay parti-
cular attention to the design of the teachers’ initiator techniques and the
actions accomplished by them. Owing to the pedagogical interests of the
researcher, the analysis will highlight the relative strength of the repair initia-
tor (RI) to locate the problem with speaking/hearing/understanding (Schegloff
et al. 1977; Schegloff 1997). Techniques that provide the least help in locating
the trouble, such as open class RIs (e.g. ‘pardon’) (Drew 1997) will be com-
pared with those that locate the repairable more specifically and/or supply a
candidate offer to the child. Some tentative suggestions will be made concern-
ing the ordering of these devices in oral language lessons. It is of further
empirical interest to compare the findings to studies conducted by conversa-
tion analysts in first and second language discourse contexts since there is no
parallel work in the classrooms of children with SSLD.
PARTICIPANTS, DATA SET AND METHOD

The adult participants were three specialized teachers who had all worked
in language resource provision for at least 10 years. Whilst they were not
J. RADFORD 29
qualified as speech and language therapists/pathologists, they did have

post-training qualifications in the area of language development and SSLDs.
There were six children, aged 4 years 4 months to 8 years 7 months at the
time of the first visit, who all had ‘statements’ of special educational need that
indicated the need for language resource provision. Information about
the nature of their language difficulties was collected by questionnaire that
was completed by the teacher and speech and language therapist/pathologist
together. Table 1 shows that the children had mainly receptive and expressive
language difficulties,2 characterized also by semantic problems. Children
with significant speech difficulties, such as verbal dyspraxia, phonological dif-
ficulties and pragmatic difficulties3 were excluded.

Story writing, book-sharing and circle-time activities were selected by the
teachers as best examples of oral language teaching. Story writing and circle-
time activities were conducted in small groups of five to six children, although
two target children in each activity were the focus for the analysis. Book-
sharing was conducted on a one-to-one basis with the teacher. Video-record-
ings were made with a digital camera on four separate occasions, yielding a
data set of 24 lessons (8–45 min long). Radio-microphones were positioned on
the teacher and the quietest of the two target children in the group activities
to help increase the accuracy of the recordings. Repeated viewings were made
followed by detailed transcriptions of selected segments.
Given that, apart from aforementioned work, there are no detailed studies
of interaction in the classrooms of children with specific language
difficulties, one possible analytical approach would be to borrow or adapt a
coding scheme used in second language classroom research. This was rejected
for several reasons. In terms of children’s moves, coding single turns, for
instance as phonological, lexical or grammatical ‘errors’ (Lyster 1998), would
be inappropriate because (i) many of the turns contain multiple sources of
trouble and (ii) unclear content is a common source of trouble in specific
language difficulties data, but it only becomes apparent from the teacher’s
display of difficulty in the subsequent turn or turns. Similarly, coding single
teacher moves as, for example, repetitions, recasts or requests for clarification
(Loewen 2004) would be problematic as shown in the analysis. This study thus
adopts CA perspective in order to represent ‘how participants themselves pro-
duce and interpret each other’s actions’ (Pomerantz 1988: 361). First, the
author made a collection of 245 instances of repair of which 78 cases appeared,
superficially at least, to be dealing with repair initiated in response to the
child’s idea. The analysis next proceeded by repeated viewings of the video
sequences, which uncovered how apparently similar practices, such as repeti-
tions, accomplish completely different actions (Schegloff 1997). It also showed
how actions can co-occur, notably checking/correction/clarification.
Furthermore, a rich and detailed analysis of participants’ moves was made,
including linguistic detail, non-verbal action and prosody, which is essential to
fully document the competences of children with language difficulties and
their teachers.
30
Table 1: Children’s SSLDs

Child Age Gender Receptive Expressive Phonology Verbal Naming Word Pragmatics
syntax dyspraxia meaning
A 8 years 7 months F 3 2 0 0 1 2 1
REPAIR PRACTICES IN SSLD
B 8 years 3 months F 2 2 0 0 2 2 1
C 4 years 4 months M 2 2 1 0 2 2 1
D 4 years 8 months F 3 3 2 0 3 3 0
E 5 years 10 months F 3 3 1 0 2 2 1
F 5 years 5 months F 3 3 1 0 3 3 3
Note: 0 = none, 1 = mild, 2 = moderate, 3 = severe.

J. RADFORD 31
FINDINGS
A series of other-initiated repair (OIR) practices are described, of
interest for language learning because they are dialogic sites where the
child’s meaning is potentially being negotiated. First, there is an account of
OIRs that are ‘non-specific’ in the sense that each RI has an open design
that orients to the teacher’s problems hearing/understanding the child’s
turn(s). Non-specific OIRs are, for the most part, treated as prompts
to repeat. In contrast, the next examples are classed as ‘specific’ OIRs since
the teacher pinpoints, in various ways, the location of the trouble, either
because she leads the child into its location with a non-completed utterance,
or because she uses a ‘wh’ question to specify its nature. Finally, examples

where candidate offers are made will be explored, including requests
for definition.
Non-specific (‘general’) RIs

What is distinctive about non-specific RIs is that when teachers
signal trouble with the child’s response turn, they do not, by any specific
design indicator, locate the item to be repaired. As such they target
potentially any aspect of the child’s prior utterance. Children have different
options in response: they can either repeat or reformulate their own
previous material in the following turn or turns. Extract (C) illustrates
three examples of ‘general’ RIs: a repetition request at line 3 and two state-
ments of uncertainty (lines 5 and 7). The context of the extract is story
writing with a small group of six children. The teacher (T) has already
written and drawn several of the children’s ideas onto a flipchart
regarding the characters and story-line and now asks the children a question.
T’s question (line 1) would typically be heard by the child as a request

for an idea concerning the story’s title. As the child (Ch) offers a verbal
contribution at line 2, T’s relevant next would be to accept the idea (often
in the form of a repeat). Instead, an insertion sequence suspends production
of T’s acceptance to deal with a trouble. T’s use of ‘say that again’ gives
the child an unambiguous message that a repeat is needed. In fact, line 3

has the appearance of an instruction, given the absence of lexical or
syntactic markers of politeness. Although the child offers a response in line
4, T indicates continued trouble with a further non-specific marker of
uncertainty (line 5). Ch next produces a non lexical item (‘ee::’), so T
responds with a third RI, this time clearly displaying her difficulty with
understanding. This also fails to solve the trouble so it is left to T to
suggest a story name idea. A notable feature is that each of T’s RIs target
the whole of Ch’s prior as opposed to marking out specific elements for Ch’s
attention.
Extract (C) showed three examples of non-specific RIs: a statement that a

repeat is needed, and two statements that display general trouble with hearing
or understanding. As the child’s responses were unintelligible, further data are
needed to clarify the sequential consequences of such actions. Extract (D) has a
clearer outcome because the teacher indicates trouble with an open class
RI (‘pardon’, line 3) and this is taken up by the child as an opportunity to
reformulate her grammar. The sequence takes place where the children are
invited to suggest a plot line for the group’s story. The child’s subsequent ideas
include both a novel character and an event. That the name ‘Dina’ that
she supplies is searched-for is indicated by the stretched sound at the end of
‘com:::’ and the brief silence. The relevant next, a teacher acceptance, is once
again withheld and an insertion sequence begins.
‘Pardon’ could indicate that the trouble-source entails an issue related to T’s
hearing and thus be heard as a request for a repeat. Yet, the child’s subsequent
move is designed differently: while she does repeat some lexical components of
her idea, she also reformulates her original word order and verb form. An
explanation for the reformulation might be that open class RIs are claimed
to treat ‘the whole of the prior turn as in some way problematic’ (Drew 1997).
The child thus orients to T’s signalling of trouble with potentially any element
of her turn at line 2. As in Extract (C), the open class RI is not targeting a
specific component of the child’s message for repair. Features of all non-spe-
cific designs are that T’s options for repair initiation are constrained: if the
child’s turn is not heard or understood clearly, T is unable to incorporate
elements of the turn in a subsequent move. In this sense, T’s response is a
‘general’ repair initiation, as opposed to the more specific formats that will be
presented next.
J. RADFORD 33
Specific RIs (designedly incomplete utterances) that pinpoint

the location of the trouble source
When teacher’s acceptance of a child’s idea is withheld, a common design
for initiation of an insertion sequence, is a designedly incomplete utterance
(DIU). Whilst acknowledging that there is a small data set and more research
is needed, the DIUs found in the classrooms of children with SSLDs share
the following characteristics to those reported in second language writing
conferences (Koshik 2002). First, they are constructed with minimal
components, either a phrase or an individual word, that borrow fragments
of the child’s prior turn. Next, they are used to perform different actions:
on the one hand, through partial repeat of the child’s turn, a DIU may

elicit repetition of prior talk [Extract (E)]; on the other hand, it may be
used to target an error and prompt the child to self-correct [Extract (F)].
Extract (E) takes place during the early phase of the group story
writing session where characters are being decided. The teacher targets Tom,
using an oral cloze technique (incomplete utterance and vowel stretch), to
select a character. In his turn, Tom begins with a repeat of T’s modelled
story starter ‘there was a:::’. Prolongation of the vowel sound, constructed
with continuing intonation, and the brief silence indicate that he is searching
for a novel idea to offer the teacher.
Although Tom offers the name of an animal, T withholds acceptance of this

idea. Instead, she initiates an insertion sequence in line 3 with a DIU. It is a
single word that is formed in a similar fashion to T’s original version and the
child’s production: it repeats the element that preceded the lexical item, pre-
serving the lengthening of the vowel and continuing intonation. Her trouble
could have entailed hearing and is confirmed by her body language: a shake of
the head and leaning posture. The child’s response shows her interpretation of
the DIU, thus she supplies a repeat of the noun phrase (line 4), rather than a
reformulation. The repeat supplies T with information and she offers her can-
didate hearing, proffered for confirmation. Although the hearing is confirmed
by the child, T pursues a definition, as if to find out if the child knows what a
cheetah is. One notable feature of the DIU is that it targets the trouble source
very precisely by leading the child, at least syntactically, up to the location of

the trouble. Furthermore, the DIU has pedagogical value in so far as it gen-
erates both confirmation and definition sequences that afford further oppor-
tunities for language learning.
A similar design of DIU is found in Extract (F) but the work that it does is
different from Extract (E) because the child hears it as a request for revision
rather than for repetition. The teacher and child have just embarked on the
speaking-book activity so, as they turn the page to reveal a new picture to talk
about, they embark on a TIE sequence.

The child responds to T’s topic elicitor with a description of a girl (line 2),
although the turn is partially unclear to both T and the researcher. That Ch’s
gaze remains on the picture suggests that she is engaged in solitary searching
during the silence. At line 3 T withholds topicalization and, instead, produces a
DIU. It is formed by using two words of Ch’s turn up to the point at which
there was potentially a missing adjective, as seen in Ch’s subsequent turn (line
4). Continuing intonation and gaze directed at the child, rather than the pic-
ture, suggests that the turn may be actioning a prompt. Ch hears it as a prompt,
perhaps owing to the syntactic incompleteness of ‘she has’, not to produce the
noun but a relevant next adjective ‘long’. T hears Ch’s turn and displays
acceptance of the child’s idea with a repeat of both the child’s adjective and
noun from line 2. In terms of pedagogical value, by generating further infor-
mation from the child, the DIU sets up an opportunity for T to expand the
material and supply further contingent models of language.
Like Koshik’s (2002) DIUs, these two examples are not complete turn con-
structional units, but are recognizably complete actions that are designed, with
the child’s own words, to be incomplete and thus completed by the child. In
response, the child is not ‘interrupting’, even though T’s turn construction unit
is not completed. The child orients to T’s design (where it stops) and continues
with a syntactically fitted contribution. The first example (‘a::’) led the child to
re-produce her noun and therefore complete the noun phrase. The second, a
phrase, was not only an oral invitation to complete the sentence, but also
taken as an opportunity by the child to revise the next item of her original
phrase. DIUs are prosodically marked here, by sound stretches and by continu-
ing intonation which position the child to continue the phrase or sentence.
DIUs, thus, target very specifically the location of a trouble. Yet, even when
they are heard as invitations to self-correct, they do not give the child semantic
J. RADFORD 35
information that could be usable in the subsequent correction. Both examples

have pedagogical value in so far as T indicates the trouble source very specifi-
cally and generate further opportunities for targeted language practice. Finally,
it is worth mentioning how the current examples are different from the writing
conference examples reported by Koshik (2002). In most of the writing con-
ference data, a written text was an additional available resource that was
visible (at least on occasion) to the participants. In contrast, the language
difficulty data are taken from oral language lessons, so the teachers are dealing
with online discourse problems. That the children have language difficulties
places an additional burden on the listener to process the children’s turns
accurately, especially when novel ideas are being generated in an activity

like story writing.
Specific RIs (‘wh’ questions) that pinpoint the nature of the

trouble source
The next sequences show three types of RIs that are formatted as ‘wh’ ques-
tions: a request to provide specific information and two requests for definition.
Instead of being heard as requests for repetition, the ‘wh’ questions are treated
as requests to supply further information. The analysis will reveal how their
designs serve to pinpoint precisely, not so much the location of the trouble
source, as in DIUs, but the nature of the repairable. The analysis will also
show how the child orients to the RI by making a revision that provides addi-
tional information in terms of novel aspects to the story-line. Extract (G) takes
place in the context of genuine news enquiry and reporting. As the child’s
reports are potentially ‘news’ to the teacher, any new ideas are susceptible to
lack of clarity. The turn of interest is at line 6, and is part of the emerging news-
telling sequence. The child and teacher are looking at a photograph in the
‘speaking-book’ and Ch is reporting a personal news account about taking
her dog, Penny, for a walk.
The teacher responds to Ch by asking a genuine question since she has

not experienced the walk-with-the-dog. The enquiry is itemized in so far as
she targets the location of the walk as a topic for further talk. The child’s
pause and searching behaviour (‘uh::m’) show that a response is being
searched-for. To assist the child’s search T offers a candidate answer of a
potential location in line 4. However, as the reported event is the child’s experi-
ence, Ch holds the expert knowledge about what happened and rejects T’s
version (‘no’). Following the rejection, Ch does supply some information that
answers the itemized enquiry. At line 6, given T’s puzzled look, she seems to
treat this turn as providing partial information regarding the location of the
walk. She thus appears to be making an itemized enquiry to elicit specific infor-
mation relevant to the news report. Although the request has a minimal format,
recycling just one lexical item of the child’s turn (‘down’), it is informative to
the child. There are two important dimensions: lexically, ‘whe:re’ indicates that
a location is the source of trouble; prosodically, loudness on ‘whe:re’ marks out
the item that lacks clarity. In response, (line 7), Ch answers the question by

supplying potentially clarifying information. Even though the analyst found it
impossible to tell, because the ambiguity of the homophone ‘see/sea’ in English,
whether or not the child was using a verb (see) or was self-repairing use of the
determiner ‘the’ before the noun ‘sea’, T displays an understanding which is
confirmed by the child. T’s repair initiation at line 6, then, is heard by the child
as an invitation to offer further information that has the additional effect of
clarifying her news report. Furthermore, at line 8, when T pursues the informa-
tion generated by the RI, the child is afforded another language model that
is contingent.
Extracts (H) and (J) illustrate typical definition-type requests that were
widely used by the specialist teachers. Whilst the author acknowledges that
more examples are needed to be confident about claims regarding their
systematicity, both examples adopt a similar grammatical format (‘what’s X’)
and initiate repair on a lexical item just used by the child which displays the T’s
lack of understanding of Ch’s usage in that context. In the first example, Ch
supplies different information that revises the earlier version of her lexical
item. In the second example, there is no revision but, instead, it is heard as
a request to provide additional information concerning the trouble source.
In Extract (H), the exchange takes place in the context of picture description
where, earlier in the sequence, the child had contributed a range of ideas.
At line 20, Ch is continuing her description and is currently talking about
the activities of a frog.
J. RADFORD 37
In UK English, ‘neighbourhood’ (line 20) is a rare item of vocabulary and

its use with respect to frogs is particularly unusual. The teacher in line 21,
drawing presumably on this world knowledge, offers a candidate hearing
by repeating the final two words of Ch’s turn. The candidate hearing is con-
firmed by Ch in line 22 (‘yeah’). Despite Ch’s confirmation, T continues to
remain confused. She displays this confusion with a specific request in line 23
to define the problematic phrase (‘his neighbourhood’). The grammatical
format ‘what’s X?’ supplies Ch with information that a word definition is
needed. In line 24, the child does not, however, supply a definition but
responds by reformulating the trouble source. The child’s self-repair is pref-
aced by ‘I mean’ that appears to indicate a degree of metalinguistic awareness

on her part about self-correction. That Ch produces the lexical item
(‘neighbour’) at line 24 gives T opportunity for a repeat and, therefore, a
further module of the correct lexis.
Extract (J) illustrates an apparently comparable request at line 5. It also
emerges from prior talk that is dealing with checking mutual understanding
and has the same grammatical form (‘what’s X’). What is different here is that,
instead of being heard as a request for revision, the child is able to provide a
partial definition of item ‘X’. At the beginning of Extract (J) we find a TIE
sequence that generates a plot idea from the child. Although Ch offers a novel
idea (given that ‘sheds’ have not been talked about so far), T withholds accep-
tance to initiate an insertion sequence. She repeats the last two words that the
child said with additional loudness on ‘SHED’ and final rising intonation, as if
to check for confirmation. The child offers non-verbal confirmation rather
than a revision of her turn (line 4).
The teacher’s question at line 5 signals her on-going trouble with Ch’s
idea, also confirmed by T’s simultaneous non-verbal behaviour (shaking her
head). The question has the appearance of a request for definition. How Ch
heard T’s request is unclear. Perhaps, owing to her language difficulty, she was
unable to supply a standard definition (e.g. a small wooden building).
The action accomplished here, interestingly, is that she supplies a novel
idea that is relevant to the story-line. T’s subsequent reformulation (line 7)
incorporates the ideas from both line 2 as well as the new material produced
in 6. The work done by the repair initiation is therefore to achieve a definition

that has the additional effect of contributing to the ongoing story-line (T later
displays acceptance by drawing both ideas onto the flipchart). The RI also
creates a sequential opportunity for the teacher (line 7) to supply
an expanded model of the child’s ideas, which has value in terms of language
learning.
A further observation is that a candidate hearing/understanding (proffered
for confirmation) precedes clarification in each of these sequences and
appeared with some regularity elsewhere in the data set. As an illustration,
Extracts (H) and (J) involve two teachers working with very different children
of different ages and yet the pattern is similar. First of all, there is the first part of

an adjacency pair, formed with a repeat of the child’s noun phrase [Extract (H),
line 21, ‘his neighbourhood’; Extract (J), line 3, ‘a SHED?’]. This provides
a candidate hearing or understanding that makes relevant the child’s confirma-
tion or correction. What appears to be happening is that, in the context of
content troubles, a first move is to check that the trouble does not reside
with the teacher herself: could it have entailed her own problem with hearing
or understanding? In both cases confirmation is given, so a hearing/under-
standing problem is ruled out. Yet, T’s pursuit of elaboration of the item persists,
which has the effect of initiating a pedagogical side sequence that targets
the child’s current understanding of the lexical item. A design like ‘what’s X’
does two jobs. First, it invites the child to provide further information and, as
we see here, the information is forthcoming, either as a revision or as material
relevant to the topic. Next, since X is a component of the child’s prior turn,
it marks out very specifically the item to be talked about, which renders
it pedagogically useful for displaying the child’s understanding. This distin-
guishes the practice from the single item ‘what?’ technique that targets
the whole of the prior (Schegloff 1997). A final, somewhat tentative observa-
tion, is that there appears to a ‘natural ordering’ at work, based on the relative
strength or power of the teachers’ techniques to locate the repairable (Schegloff
et al. 1977). The teachers are employing their moves in order of increasing
strength.
Offers of candidates that supply a linguistic model

The examples shown in Extract (K) are distinctive because there is material in
the teacher’s turns that has not been mutually shared in prior talk, according
to the available data. There are two teacher moves of particular interest: at line
5 there is a candidate understanding and, at line 10, an offer of a candidate that
does the job of correction. Offers of candidates can be useful when a speaker
anticipates that a hearer will experience difficulty supplying the relevant infor-
mation and the speaker wants to offer help (Pomerantz 1988). The example at
line 10 is formed as a modulated other-correction ‘Do you mean X?’ (Schegloff
et al. 1977), although at line 5 there is a variation of form. T and Ch are talking
about a picture of the seaside that Ch has drawn to represent her story setting.
J. RADFORD 39
At the start of the sequence the setting is being described and an additional
character suggestion is made at line 3.

Ch’s idea, ‘a nice snake’ is treated as problematic by T who withholds
acceptance. An insertion sequence is initiated in line 4 with a repeat of the
lexical item which is formed with high pitch rise, as if to signal surprise.
T makes two specific queries concerning the snake’s whereabouts in lines
4 and 5. The first, beginning with ‘wha d’y’mea:n’, at first appears as a chal-
lenge to the credibility of Ch’s account (Schegloff 1997). A more likely inter-
pretation is that T is targeting ‘in the water’ for clarification of the child’s
meaning. Instead of waiting for a response, T directly offers a candidate under-
standing ‘in the sea?’ (line 5) of the Ch’s turn at line 3. Ch does not respond
pragmatically to T’s queries or take up the offer of the candidate. Instead,
she treats the turn as a hearing or understanding check, with a repeat of
her phrase (line 6). No new topical material is produced by Ch, so mutual
understanding has yet to be reached which is indicated by T re-visiting the
problem at line 9. This turn signals T’s confusion (‘I’ve never see:n a snake in
the sea.’) now making it explicit that her own world experience does not fit
Ch’s claim. In line 10, T offers a further candidate with the modulated other
correction format ‘do you mean X.’ In contrast with line 5, it has falling
intonation which confirms its status as a correction. Use of the verb ‘mean’
indicates clearly to Ch that the issue concerns T’s understanding rather than
hearing. In order to correct ‘snake’ to ‘eel’ T needed to guess by drawing on a
range of contextual information.
How are candidate offers significant for a discussion of language learning
opportunities? In lengthy repair sequences, given the new material that they
include, they occasion further opportunities for language practice. When offer-
ing a candidate item, the teacher is providing a model to the child of what is
relevant and appropriate to produce as a response (Pomerantz 1988). In the
case of ‘d’y’mean an eel?’ the model is a single lexical item, although gram-
matical units could equally be modelled in this way. The teacher leads the child
to focus on the item, and further exposes it, by placing it in the final position of
the turn and marking it with additional prosodic emphasis. Yet, in the example
presented here and others in the data set, it is not repeated, even following
teacher repetition at line 12. An explanation is that a relevant response to

this type of yes/no question is either for the child to confirm T’s guess as the
correct target, or to give a different version if the guess is incorrect.
SUMMARY AND CONCLUSIONS

This analysis has detailed, for the first time, a variety of techniques adopted
by specialist teachers to initiate repair in response to troubles hearing or
understanding children’s novel ideas. Given that children with SSLD typically
display problems with their production of phonology, syntax and semantics,
their original ideas commonly occasion insertion sequences that deal with

the business of correction, other-initiation and self-repair. Such sequences
afford contingent opportunities for dialogic discourse and, therefore, provide
potential opportunities for language learning. Four distinct types of RI
practices emerged as prevalent patterns in a substantial database of 78 cases.
It is now worth summarizing the key characteristics of these techniques.
(i) Non-specific (‘general’) RIs:
constructed as an open class RI (‘pardon’) or as a message that
the trouble source entails hearing (‘say that again’) or understand-
ing (‘I don’t know what that means’);
target the whole, or potentially any aspect, of the child’s prior turn;
reveal a ‘substantial lack of grasp of what was said’ (Schegloff 1997;
524);
used to elicit a repeat or reformulation of any (or all) of the prior.
(ii) Specific RIs: DIUs:
constructed with minimal components of the child’s prior turn
(‘a:::’; ‘she has’);
used as a clue that pinpoints the location of the trouble source;
provide no new lexical information;
used to elicit a repeat of the child’s prior or to target an error and
prompt self-correction.
(iii) Specific RIs: ‘Wh’ questions:
constructed as ‘wh’ questions (‘down whe:re.’; ‘what’s his
neighbourhood?’);
used as a clue that pinpoints the nature of the trouble source;
used to elicit further information about the trouble.
(iv) Offers of candidates:
constructed as modulated other correction ‘do you mean X’;
provide a new model of the type of information that is needed
(Pomerantz 1988), such as a lexical item, but could be grammatical,
for example morphology:
not suited to eliciting a repeat (uptake) of the model since formed
as a yes/no question.
J. RADFORD 41
Non-specific RIs are distinguished from specific RIs principally by the degree
of information afforded to the child. The former target the trouble source more
generally, so the child has a wider domain of prior utterance(s) from which to
select in order to formulate a repeat or revision. Specific formats are more
focused: they either target the location of the trouble (DIUs) or the nature of
the trouble (‘wh’ questions). DIUs are used to both elicit repeats and to prompt
self-correction. Furthermore, ‘wh’ questions include requests for word defini-
tion, and these are suited to eliciting the child’s display of understanding of
lexical items. Whereas offers of candidates provide a linguistic model, they are
not taken as opportunities for repeat (or ‘uptake’, as in Loewen 2004).
General’ RIs are preceded by less clear child contributions, where the only

option is to target the whole/any of the prior. In contrast, specific RIs are
possible in contexts where the child has been, at least partially, heard or
understood. This affords the opportunity to repeat part of the prior, as in
DIUs, or to target a specific component such as in offers of candidates (‘Do
you mean X?’) or ‘wh’ questions.
Detailed descriptions of repair practices have important implications for
practitioners. Studies of repair trajectories in classrooms where children are
learning a second language have identified distinctive patterns that are asso-
ciated with different communicative tasks (Seedhouse 2004). As far as the field
of speech and language therapy is concerned, comparable work has provided
preliminary insights into interactions between adults and children with specific
language impairment (Gardner 2005) and adults with acquired speech disor-
ders (Bloch 2005). In terms of the current study, both teachers and speech and
language therapists will gain a richer understanding of adults’ responses to
children’s ideas during language interventions, especially when dealing with
sources of unclear meaning.
From a methodological perspective, using CA has afforded fresh insights
into the participants’ behaviours. Whereas Radford et al. (2006) broadened
the notions of teacher initiation (I) and child response (R), this study provides
fresh insight into feedback (F) practices that initiate repair and the actions
accomplished by a range of designs. Although not a primary aim, the
study also demonstrates the rich interactional competencies of the children
to engage in repair sequences, despite their evident language difficulties.
Furthermore, it is evident that in analysing the data of children with speech
and language difficulties, external judgements of ‘errors’ would be misleading.
The broader construct of trouble source may be better suited to ‘messier’ data,
whilst the participants’ perspectives are key to interpreting the actions accom-
plished. Yet, as CA has rarely been adopted in studies of such children, there is
a considerable amount of work still to do. Future research aims to explore a
wider range of lesson and participant contexts and examine in more
detail other design features, such as the non-verbal and prosodic features of
the interaction.
APPENDIX
Transcription key1
(0.5) The number in brackets indicates silence by tenths of seconds.

(.) A dot enclosed in a bracket indicates a gap in the talk of less than two-
tenths of a second.
= The equals sign indicates ‘latching’ between utterances.
[] Square brackets between adjacent lines of concurrent speech indicate
the onset and end of a spate of overlapping talk.
.hh A dot before an ‘h’ indicates speaker in-breath. The more h’s the

longer the breath.
hh An ‘h’ indicates an out-breath. The more h’s the longer the breath.
((points)) A description enclosed in a bracket, and written in italics, indicates a
non-verbal activity. For example ((points at picture)).
- A dash indicates a sharp cut-off of the prior word or sound.
::: Colons indicate that the speaker has stretched the preceding sound or
letter. The more colons the greater the extent of the stretching.
! Exclamation marks indicate an animated or emphatic tone.
() Empty parentheses indicate the presence of an unclear fragment of
tape.
(guess) The words within a single bracket indicate the transcriber’s best guess
at an unclear utterance.
. A full stop indicates a stopping fall in tone. It does not necessarily
indicate the end of a sentence.
, A comma indicates a ‘continuing’ intonation.
? A question mark indicates a rising inflection. It does not necessarily
indicate a question.
"# Pointed arrows indicate a marked falling or rising intonational shift.
They are placed immediately before the onset of the shift.
under Underlined fragments indicate speaker emphasis.
CAPITALS Words in capitals mark a section of speech noticeably louder than that
surrounding it.
88 Degree signs are used to indicate that the talk they encompass is
spoken noticeably quieter than the surrounding talk.
>< ‘More than’ and ‘less than’ signs indicate that the talk they encompass
was produced noticeably quicker than the surrounding talk.
! Arrows in the left margin point to specific parts of an extract discussed
in the text.
NOTES
1 The transcription conventions used in are based on the system originally
this article (as shown in the Appendix) developed by Gail Jefferson, as
J. RADFORD 43
reported in Atkinson and Heritage with the comprehension of syntax,

(1984: ix–xvi). morphology and word meanings
2 ‘Expressive’ language difficulties may (Conti-Ramsden and Botting 1999).
include some or all of the following: 3 ‘Pragmatic’ language difficulties are
problems with expressive syntax, mor- defined as problems in using language
phology, articulation, phonology and appropriately in a conversational con-
word finding difficulties. ‘Receptive’ text despite well-formed syntax
language difficulties entail problems (Bishop 2000).
REFERENCES

Alexander, R. 2001. Culture and Pedagogy: Dockrell, J., G. Lindsay, B. Letchford, and
International Comparisons in Primary Education. C. Mackie. 2006. ‘Educational provision for
Oxford, UK/Malden, USA: Blackwell. children with specific speech and language dif-
Alexander, R. 2006. Towards Dialogic Teaching: ficulties: perspectives of speech and language
Rethinking Classroom Talk. 3rd edn. Dialogos, therapy service managers,’ International
UK: University of Cambridge. Journal of Language and Communication
Atkinson, J. and J. Heritage (eds). 1984. Disorders 41/4: 423–40.
Structures of Social Action: Studies in Conversation Drew, P. 1997. ‘ ‘‘Open’’ class repair initiators
Analysis. Cambridge: Cambridge University in response to sequential sources of trouble
Press. in conversation,’ Journal of Pragmatics 28/1:
Bishop, D. V. M. 2000. ‘Pragmatic language 69–101.
impairment: a correlate of SLI, a distinct sub- Edwards, A. D. and D. P. G. Westgate. 1994.
group or part of the autistic continuum?’ Investigating Classroom Talk. 2nd edn. London:
in D. V. M. Bishop and L. B. Leonard (eds): Falmer Press.
Speech and Language Impairments in Children: Gardner, H. 2005. ‘A comparison of a mother
Causes, Characteristics and Outcomes. East Sussex, and a therapist working on child speech’
UK: Psychology Press. in K. Richards and P. Seedhouse (eds):
Bloch, S. 2005. ‘Co-constructing meaning in Applying Conversation Analysis. Basingstoke:
acquired speech disorders: word and letter repe- Palgrave Macmillan.
tition in the construction of turns’ in K. Richards Jefferson, G. 1987. ‘On exposed and embedded
and P. Seedhouse (eds): Applying Conversation correction in conversation’ in G. Button and
Analysis. Basingstoke: Palgrave Macmillan. J. Lee (eds): Talk and Social Organization.
Button, G. and N. Casey. 1984. ‘Generating topic: Clevedon: Multilingual Matters.
the use of topic initial elicitors’ in J. M. Atkinson Koshik, I. 2002. ‘Designedly incomplete
and J. Heritage (eds): Structures of Social Action: utterances: a pedagogical practice for
Studies in Conversation Analysis. Cambridge: eliciting knowledge displays in error correction
Cambridge University Press. sequences,’ Research on Language and Social
Button, G. and N. Casey. 1985. ‘Topic nomina- Interaction 35/3: 277–309.
tion and topic pursuit,’ Human Studies 8: 3–55. Law, J., G. Lindsay, N. Peacey,
Cazden, C. B. 2001. Classroom Discourse: The M. Gascoigne, N. Soloff, J. Radford,
Language of Teaching and Learning. 2nd edn. S. Band, and L. Fitzgerald. 2000. Provision
Portsmouth: Heinemann. for Children with Speech and Language Needs in
Conti-Ramsden, G. and N. Botting. 1999. England and Wales. London: DfEE.
‘Classification of children with specific lan- Leonard, L. 2000. Children with Specific Language
guage impairment,’ Journal of Speech, Language Impairment. Cambridge, MA: MIT Press.
and Hearing Research 42: 1195–204. Loewen, S. 2004. ‘Uptake in incidental focus on
Dockrell, J. and D. Messer. 1999. Children’s form in meaning-focused ESL lessons,’
Language and Communication Difficulties: Language Learning 54/1: 153–88.
Understanding, Identification and Intervention. Lyster, R. 1998. ‘Negotiation of form, recasts
London: Cassell. and explicit correction in relation to error
types and learner repair in immersion class- Ridley, J., J. Radford, and M. Mahon. 2002.
rooms,’ Language Learning 48: 183–218. ‘How do teachers manage topic and repair?’
Macbeth, D. 2004. ‘The relevance of repair for Child Language Teaching and Therapy 18/1: 43–58.
classroom correction,’ Language in Society 33: Sadler, J. and K. Mogford-Bevan. 1997.
703–36. ‘Teacher talk with children with language dis-
Mehan, H. 1979. ‘ ‘‘What time is it Denise?’’: orders: four case studies 2,’ Child Language
asking known information questions in class- Teaching and Therapy 11: 37–58.
room discourse,’ Theory into Practice 18/4: Schegloff, E. A. 1972. ‘Notes on a conversa-
285–94. tional practice,’ in D. Sudnow (ed.): Studies in
Mercer, N. 1995. The Guided Construction of Social Interaction. New York: Free Press.
Knowledge: Talk Amongst Teachers and Learners. Schegloff, E. A. 1997. ‘Practices and actions:
Clevedon: Multilingual Matters. Boundary cases of other-initiated repair,’
Mercer, N. 2000. Words and Minds: How We Use Discourse Processes 23: 499–545.
Language to Think Together. London: Routledge. Schegloff, E. A., G. Jefferson, and H. Sacks.

Mercer, N. and K. Littleton. 2007. Dialogue and 1977. ‘The preference for self-correction in the
the Development of Children’s Thinking: A organization of repair in conversation,’
Sociocultural Approach. London: Routledge. Language 53: 361–82.
Mroz, M., F. Smith, and F. Hardman. 2000. Seedhouse, P. 2004. The Interactional Architecture
‘The discourse of the literacy hour,’ Cambridge of the Language Classroom. Malden, MA:
Journal of Education 30/3: 379–90. Blackwell.
Nystrand, M. 1997. Opening Dialogue: Understand- Sinclair, J. M. and R. M. Coulthard. 1975.
ing the Dynamics of Language and Learning in the Towards an Analysis of Discourse: The English
English Classroom. New York: Teachers College Used by Teachers and Pupils. London: Oxford
Press. University Press.
Pomerantz, A. 1988. ‘Offering a Smith, F., F. Hardman, K. Wall, and M. Mroz.
candidate answer,’ Communication Monographs 2004. ‘Interactive whole class teaching in the
55: 360–73. national Literacy and Numeracy Strategies,’
Radford, J. and C. Tarplee. 2000. ‘The manage- British Educational Research Journal 30/3:
ment of conversational topic by a ten-year-old 395–411.
child with pragmatic difficulty,’ Clinical Skidmore, D. 2000. ‘From pedagogical dialogue
Linguistics and Phonetics 14/5: 387–403. to dialogical pedagogy,’ Language and Education
Radford, J. and J. Ireson. 2006. ‘Developing 14/4: 283–96.
language skills through dialogic discourse,’ Wells, G. 1993. ‘Re-evaluating the IRF
AFASIC Abstract Spring 1: 3. Sequence: a proposal for the articulation
Radford, J., J. Ireson, and M. Mahon. 2006. of theories of activity and discourse for
‘Triadic dialogue in oral communication tasks: the analysis of teaching and learning in
what are the implications for language learn- the classroom,’ Linguistics and Education 5:
ing?,’ Language and Education 20/3: 191–210. 1–37.
Style Shifts among Japanese Learners

before and after Study Abroad in Japan:
Becoming Active Social Agents in Japanese
NORIKO IWASAKI
School of Oriental and African Studies, University of London

Previous studies on L2 Japanese sojourners often reported that learners
overuse the plain style or haphazardly mix the plain and polite styles upon
return. These styles, which are often associated with formal or informal con-
texts, also index complex social and situational meanings, and native speakers
are reported to shift their styles to create desired contexts. In order to better
understand L2 development of the use of the plain and polite styles during
study abroad, the current study examined the use of the polite/plain styles
and style shifts among five English-speaking male students who studied in
Japan for one academic year by comparing their performances both quantita-
tively and qualitatively in oral proficiency interviews before and after they
studied abroad. Upon return, three predominantly used the polite style talking
to the interviewer (their former teacher), while two primarily used the plain
style. Though the quantitative analysis may lead one to conclude that these
two students regressed in their pragmatic competence, the qualitative analysis
revealed that all five learners gained some understanding of social meanings
of the plain and polite styles and became more active social agents who make
decisions to shift the styles.
INTRODUCTION
Societal interactions are structured by sociocultural norms of communities
and are not easy to replicate in foreign language classrooms. Study abroad
experiences provide second language (L2) learners with opportunities to
observe first-hand how the language is used, to use the language themselves,
and to observe how their utterances and performances unfold in subsequent
interactions. Hence, studying abroad may facilitate acquisition of the aspects of
language that are the most intimately associated with social norms and situa-
tions, such as the use of formal and informal styles.
Yet, previous studies show that, though learners gain sociolinguistic
knowledge while studying abroad, their knowledge and performance diverge
from target norms. Barron (2006) found that Irish learners of German
switched between the two address forms of ‘you’ (‘intimate/simple’ du and
‘polite/distant’ Sie) haphazardly within a turn after a 10-month stay in
Germany. Regan (1995) also found that French learners overuse informal
ne deletion in negatives after one academic year in France, which she
46 STYLE SHIFTS AMONG JAPANESE LEARNERS
attributes to insufficient input in formal contexts. L2 Japanese speakers

are also reported to overuse the informal style after studying abroad, presum-
ably due to the influence of informal interactions (Marriott 1993, 1995).
Therefore, ‘the recurrent finding was that students learn a more informal
style during study abroad than at home in the classroom’ (Kasper and Rose
2002: 230).
Interestingly, Kinginger and Farrell (2004) found that some advanced
learners of French deliberately choose not to conform to the norms of address
forms of ‘you’ (intimate tu and polite/distant vous) despite their knowledge.
Learners sometimes feel that they do not need to conform to the norm
(Barron 2003), or that the norm conflicts with their identity (Dufon 2000;

Siegal 1995). Thus, the learning of formal/informal styles is not merely a
matter of acquiring the forms and associating them with certain contextual
features (e.g. the addressee’s social status). Rather, L2 learners learn to make
their own choices as to which forms to use based on their understanding of
the forms’ social meanings.
Acquiring ways to convey an appropriate demeanor is vital in learning
Japanese as an L2. Japanese has no neutral forms of politeness, and speakers
cannot easily opt out of choosing the level of politeness—though there are
some avoidance strategies speakers employ, such as the use of incomplete
sentences (Cook 2006). Finite verbs and copulas are marked to indicate
some acknowledgment of social and contextual factors as examples (1a and
b) and (2a and b) indicate. The forms in (1a) and (2a) are generally used in
informal situations and/or with an addressee with whom the speaker feels
close, while (1b) and (2b) are used in more formal situations, most likely
with a socially and psychologically distant addressee. The latter is the polite
form, which is also referred to as des(u)/mas(u) form, or ‘addressee honorifics’
because it acknowledges distance and formality towards the addressee.1
(1) a. Iku. ‘(I) will go.’
b. Iki-masu. ‘(I) will go’.
(2) a. Kiree da. ‘(it) is pretty.’
b. Kiree desu. ‘(it) is pretty.’
The choice between polite and plain forms needs to be made constantly with
the consideration of various social and situational factors. Japanese textbooks
generally first introduce the polite verb forms –mas(u) and copula des(u). This is
presumably due to the ease of acquisition of polite forms and to the presumed
desirability to err on the side of formality rather than informality. While the
use of plain forms requires the learning of rather cumbersome morphological
rules for negative and past formations [e.g. iku (non-past), ika-nai (negative),
itta (past)] for each verb, the polite form formation is consistent across all verbs
N. IWASAKI 47
[e.g. iki-masu (non-past), iki-masen (negative), iki-masita (past)]. Furthermore,

the consequence of using the polite form in informal situations is generally
assumed to be not as adverse as the opposite in which case the speaker may
sound rude.2
Learning the appropriate use of the styles was found to be the major obstacle
for Australian secondary school students even after a year in Japan. Many
overused the plain form (Hashimoto 1993; Marriott and Enomoto 1995;
Marriott 1993, 1995). Most of the students in Marriott’s (1995) study used
the polite style before going to Japan, but upon return they predominantly
used the plain style or haphazardly mixed them. Interestingly, some of the
students used a greater proportion of polite forms during role plays and picture

descriptions than in interviews with a university instructor, which Marriot
attributed to their higher awareness of language forms.
Difficulty of acquiring appropriate styles is also reported for L2 Japanese
learners who are college/university students or older. Atsuzawa-Windley
and Noguchi (1995) found that university students who had previously stud-
ied in Japan for various lengths of time did not perform well in aspects related
to politeness, overusing the informal plain style. Siegal’s (1995, 1996) in-depth
study of four adult women studying in Japan demonstrated the complexity
involved in the acquisition of sociolinguistic competence—involving the inter-
play of race, gender, and social status—which sometimes led to their resistance
of using polite expressions.
These previous studies show that studying abroad in Japan often leads to
learners’ overuse of plain forms. However, these studies might have neglected
the social/situational meanings of the styles by assuming that language style is
bound to certain contextual features, such as the social status of the interlo-
cutor. Recent work on style shifts among native Japanese speakers demon-
strates that they do not always observe socially agreed-upon rules. Rather,
they actively create speech styles and shift between them during interactions
with the same addressee.
Polite and plain style and style shift

What has thus far been called polite style is the style that utilizes polite forms
of verbs –mas(u) or copula des(u), which express deference to addressees. The
distinction between the polite and plain styles is traditionally viewed as a
formal versus informal distinction based on social and situational factors
(e.g. social status, presence/absence of intimacy) and is thus explained to L2
Japanese learners.
However, the choice of these two styles is not simply a matter of observing
social norms based on discernment of the contextual feature. Instead, polite
and plain expressions index social meanings (i.e. the nature of relations and
identities), such as in-group/out-group orientation (Makino 1983, 2002) and
distance/proximity to the self and social role (Cook 1996). The choice between
the two depends on ‘beliefs about who should speak deferentially and formally
to whom, and under what circumstances’ (Okamoto 1999: 58), and the two
styles are often mixed. In predominantly polite discourse, a speaker may
switch to the plain style to state summaries, facts, convictions, feelings (e.g.
Makino 1983, 2002), or to provide background information that is subordinate
to the main idea (Ikuta 1983; Maynard 1991). Japanese speakers also express
empathy (Ikuta 1983), spontaneous assertion of the speaker’s thought (Cook
1996), soliloquy, emotional, exclamatory reactions or closeness/friendliness
(Okamoto 1999) by switching to the plain style.
Okamoto (1999), for example, found that in an interaction between a pro-
fessor and a former student the student occasionally used the plain style for
exclamatory, emotional, or soliloquy-like expressions despite their social status

and age difference. The professor also mixed the styles to maintain a degree of
formality and to show friendliness. Likewise, Cook (2006) found that, in aca-
demic consultation sessions, a student shifted to the plain style to index his
conviction to himself or to co-construct an idea with the professor by supplying
the continuation of the professor’s utterance.
The current study shares Okamoto’s (1999) and Cook’s (2006) theoretical
perspective of ‘indexicality’ that the plain/polite forms (indirectly) index social
meanings (and thus its use is a social action) and that Japanese speakers
manipulate speech styles ‘in order to create a desired context, in particular,
preferred interpersonal relations and identities’ (Okamoto 1999: 70), and
‘every move that the speaker makes is his or her choice’ (Cook 2006: 288).
Native speakers vary in their beliefs and understanding of politeness and in
their preferred modes of presenting their social roles, and they use this knowl-
edge to actively create speech styles.
Hence, the development of the use of the plain and polite forms involves not
only an understanding of their respective social meanings but also the making
of spontaneous choices of styles with the consideration of various contextual
and situational factors. Study abroad experiences give L2 learners opportu-
nities to interact with various interlocutors, and they may acquire speech
styles through the process of language socialization by becoming active mem-
bers in the society (e.g. Schieffelin and Ochs 1986).
THE CURRENT STUDY

This study examines the use of polite and plain styles among five male uni-
versity students from the United States, comparing their use of these styles
before and after they studied abroad for a year. This may represent the circum-
stances in which politeness may not be highly valued by the participants
because of the emphasis on informality in North American values (e.g.
Althen 1988) and of a general cross-cultural expectation for women to be
more polite than men (e.g. Brown and Levinson 1987; Okamoto 2002).
By employing the oral proficiency interview (OPI), which complies with the
guidelines of the American Council on Teaching of Foreign Languages
(ACTFL), this study first assesses the participants’ gains in overall proficiency,
N. IWASAKI 49
and then investigates how their use of the plain/polite styles changed as a
result of studying abroad. In particular, it focuses on how L2 Japanese learners
switch between the two styles before and after studying abroad.
Participants
Five male students, referred to by pseudonyms as Alan, Henry, Peter, Sam, and
Greg, from a northeastern US state university voluntarily participated. Their
ages at the time of departure ranged from 19 to 21. One participant, Greg, had
studied Japanese for 1 year (6 hours of instruction per week for two semesters)
before studying abroad, and the other four had completed 2 years of Japanese

instruction (6 hours of instruction per week for four semesters). They attended
four different universities in Japan. Henry studied in Aichi, Peter in Ibaraki,
Sam in Tokyo, and Alan and Greg studied at the same institution in Hyogo.
Procedures and analyses

Before studying abroad, the students participated in OPIs conducted by a cer-
tified OPI tester, who was the primary Japanese teacher of Henry, Peter, Sam,
and Alan in the two semesters preceding their study abroad. In the fall after
they returned, they participated in OPIs again. The tapes and the interviewer’s
ratings of the post-study-abroad OPIs were sent to ACTFL to receive official
ratings to avoid potential biases of the tester who had knowledge of the partic-
ipants’ pre-study-abroad performances. The learners’ gains in overall oral pro-
ficiency were assessed by these OPI ratings. Because the criteria for OPI ratings
do not include the use of appropriate styles (except for Superior), gains in these
ratings mostly indicate the learners’ gains in other aspects of their performance
(functions/tasks, content, text types, accuracy). The OPIs before and after
study abroad are referred to as pre-OPI and post-OPI below.
The OPIs were transcribed and analyzed both quantitatively and qualitatively.
The appropriate base style for an interviewee to use in such interviews is the
polite style, especially because the interviewer is a former teacher, who is clearly
older than the participants. Role plays are part of the OPI protocol, but the types
of role plays varied depending on the proficiency levels of the participants. For
interviewees approaching Superior-level, the Japanese OPI requires formal and
informal role plays to elicit both styles. Thus, the most proficient speakers of the
five, Peter and Sam, were given role plays for both styles.
In the quantitative analysis, predicates produced in OPIs, excluding the role
play segments, were coded for the forms used: the polite form, plain form,
omission of predicate/copula (in the case of nominal predicates). The propor-
tions of each type were computed. The sentence-final matrix predicates and
coordinate/subordinate predicates in which polite forms can be used without
sounding excessively polite were considered.3 Further, these non-matrix
clauses were broken down into two types to aid in the analysis. This is because
whether and to what extent the plain form in non-matrix clauses has social
meanings that contrast with the polite form depends on the degree of subor-
dination of the clauses in question (Makino 1983; Cook 2001). The plain/polite
forms used in the sentence-final matrix predicates primarily determine the
level of politeness. The plain forms in highly independent clauses (ending
with ga ‘but’, kedo ‘but’, and kara ‘because’) index some informality, but
they may not necessarily express informality in clauses possessing marginal
independence (ending with node ‘because’ and si ‘and’). For the sentence-final
nominal predicates, such as yo-nensei da ‘I am a senior’, the omission of the
copula da as in yo-nensei 6 0 is common in spoken Japanese. There are also
instances of omission of entire predicates such as in zenzen ‘not at all’. These
omissions are tallied as predicate/copula omissions; they are equivalent to the

plain forms in degree of politeness/formality.
In the qualitative analysis, the students’ style shifts in sentence-final matrix
predicates were examined (leaving the examination of non-matrix clauses to
future studies due to space limitations). The current analysis primarily focuses
on the interaction with the interviewer taking her authentic role (university
faculty), but performances in the role plays are also discussed. Note, however,
these styles may index multiple social meanings, and their meanings may be
occasionally ambiguous (Okamoto 1999). This is because the functions dis-
cussed above are not mutually exclusive. Thus, multiple plausible functions
realized by the form are noted, though they may not be the only possibilities.
FINDINGS
Gains in overall oral proficiency
The participants’ OPI ratings were rather high even before studying abroad,
most likely because they were highly motivated, enough to opt to study
abroad. Four improved their proficiency by 1 or 2 sub-levels as indicated in
Table 1. Henry, whose proficiency was at the Intermediate-Mid in pre-OPI,
remained at the same level. Interestingly, despite the fact that OPI criteria do
not include the appropriate speech styles for Novice to Advanced levels, those
who rated higher demonstrated more appropriate and/or active use of styles
and shifts. For example, as discussed below in the quantitative analyses, those
who were rated higher (Alan, Sam, and Peter) appropriately chose the polite
style as the base style. Moreover, as discussed in the qualitative analysis, Peter,
who was rated the highest, actively shifted styles in a target-like manner rather
than simply conforming to the appropriate polite base style.
Quantitative analysis: proportions of use of the polite

and plain styles
Table 2 shows the types of predicates each participant used. Table 3 shows
breakdowns of the plain forms into three types of clauses. In terms of the
proportion of the polite forms, the five participants did not differ greatly in
N. IWASAKI 51
Table 1: Gains in OPI ratings

Pseudonyms Length of prior Before study After study
instruction (years) abroad abroad
Alan 2 Intermediate-High Advanced-Low

Henry 2 Intermediate-Mid Intermediate-Mid
Peter 2 Advanced-Low Advanced-High
Sam 2 Intermediate-High Advanced-Mid
Greg 1 Intermediate-Low Intermediate-Mid

Table 2: Types of predicates produced
Pre-study abroad OPI Post-study abroad OPI
Polite Plain Copula Repair Total Polite Plain Copula Repair Total
n (%) n (%) omission n (%) n (%) n (%) omission n (%) n (%)
n (%) n (%)
Greg 74 (79) 9 (10) 8 (9) 3 (3) 94 13 (10) 96 (73) 23 (17) 0 (0) 132
Henry 73 (72) 16 (16) 8 (8) 5 (5) 102 31 (20) 103 (67) 20 (13) 1 (1) 155
Alan 128 (87) 14 (10) 1 (1) 4 (3) 147 172 (86) 21 (11) 5 (3) 1 (1) 199
Sam 107 (84) 16 (13) 2 (2) 2 (2) 127 106 (85) 17 (14) 2 (2) 0 (0) 125
Peter 134 (80) 27 (16) 4 (2) 2 (1) 167 127 (70) 44 (24) 10 (5) 1 (1) 182
Table 3: Clauses where the plain forms were used

Pre-study abroad OPI Post-study abroad OPI
Matrix kara, kedo, node, si Total Matrix kara, kedo, node, si Total
clauses ga clauses clauses clauses ga clauses clauses
Greg 0 0 9 9 83 12 0 95
Henry 5 11 0 16 67 36 0 103
Alan 1 9 4 14 10 9 2 21
Sam 0 8 8 16 3 2 12 17
Peter 3 15 9 27 17 13 14 44
pre-OPIs: Greg 79 per cent, Henry 72 per cent, Alan 87 per cent, Sam 84 per
cent, and Peter 80 per cent. All participants used the polite forms as their base
style. Upon return, Alan, Sam, and Peter primarily maintained the polite style
(86 per cent, 85 per cent, and 70 per cent), but Henry and Greg used the plain
forms as their base style. Their proportions of polite forms were only 10 per
cent and 20 per cent in their post-OPIs.
In pre-OPI, Greg mostly used polite forms (74/94), and his plain forms were
limited to nine uses in node and si clauses (which does not necessarily index
informality), eight copula/predicate omissions (five of which were moo iti-do?
‘once again?’ requesting repetition), and three repairs. He repaired his use of
the plain form to the polite form as shown in excerpt (3) line 2. In what

follows, plain forms are underlined and polite forms are bolded. (‘I’ refers to
Interviewer; pseudonym initials refer to participants.)
(3) (Greg, Pre-OPI)
1 I: De, ano, hima-na zikan-mo arimasu ka.
and well free time-also have-NONPAST-AH Q
‘And do you also have free time?’
2→ G: Hima-no toki, aa, video geemu-o suru, aa, o suru n desu.
free when uh video game-ACC play uh ACC N COP
‘When I have free time, uh I play ... uh it is that I play video games.’
Upon return, Greg used only 13 polite forms in 132 predicates, including
eight uses of the epistemic modal marker desyoo with rising intonation, three
uses of the polite form of the verb omoimasu ‘(I) think’. Greg did not use daroo,
the informal equivalent of desyoo. In fact, Cook (2007) does not consider these
two forms as polite and plain counterparts because desyo(o) is used in most
informal situations as well. Taking this into consideration, Greg used only
five polite forms including three instances of omoimasu in post-OPI.
Likewise, Henry also primarily spoke in the polite style in pre-OPI, with
a slightly higher tendency to use plain forms than Greg. He used 73 polite
forms in 102 predicates. He made five repairs, suggesting a conscious effort
to speak in the polite style while resisting the inclination to use the plain style
as shown in an exchange about a student organization he belongs to in excerpt
(4) line 3.
(4) (Henry, Pre-OPI)
1 I: Membaa-wa takusan iru n desu ka.
member-TOP many exist-NONPAST N COP-AH Q
‘Are there many members?’

N. IWASAKI 53
2 H: Membaa? Ano, [inaudible]-no hito-ni, ano, nan-nin-kurai, a, zyuu-go-nin-gurai
member well -GEN person-DAT well how-many-CL 15-CL-about
ano, takusan aru, ano takusan hito zyanai to omou kedo,
well many exist well many people COP-NEG QT think but
‘Member? Well, [inaudible] people, well, how many, about 15, uh, well, there are
many...well, I don’t think there are many, but,’
3 taisetu-na hito da kara,

important people COP-NONPAST because
→ totemo omosiroi, omosiroi yo, omosiroi desu.
very interesting very interesting-SFP(assert) interesting COP-AH
‘Because they are important people, it (the club) is very interesting, interesting, it
is very interesting.’
In post-OPI, his polite forms dramatically decreased. He used only 31 polite

forms in 155 predicates, including one desyo, and he no longer corrected him-
self to be polite. Instead, he once adjusted a polite form to the plain form (his
base style in post-OPI) as in excerpt (5) line 4.
(5) (Henry, Post-OPI)
1 I: Sotugyoo-site kara donna sigoto-ga sitai n desu ka.
graduate-GER after what-kind work-NOM want-to-do N COP Q
‘After graduation, what kind of job do you want to do?’
2 H: (laugh) zense, zensen, wakarimasen.
[fragment] at-all know-NEG-AH
‘I don’t know at all.’
3 I: zenzen wakarimasen ka.
at-all know-NEG-AH Q
‘You don’t know at all?’
4→ H: Honto ni, ano, kangaete imasen, kangaete nai.
really uh think-GER has-NEG-AH think-GER NEG
‘Really, well, I haven’t thought, haven’t thought about it.’

Henry’s mix of the styles appeared to be haphazard, but the qualitative

analysis below reveals that his style shift is not entirely random. Like Greg,
he had a slight tendency to use the polite forms for expressions of epistemic
stance (omou ‘think’, wakaru ‘understand’, and kanziru ‘feel’). These verbs
accounted for 6 of 19 verbs expressed in the polite forms out of 31 polite
forms, though he also used these verbs in the plain forms.
Unlike Greg and Henry, Alan and Sam spoke in the polite style both in
pre- and post-OPIs. Their plain forms were mostly in non-matrix clauses:
Alan used plain forms 13/14 (non-matrix/of all plain forms) in pre-OPI
and 11/21 in post-OPI, and Sam 16/16 in pre-OPI and 14/17 in post-OPI.
Peter, who received the highest OPI ratings both in pre- and post-OPIs,

also mostly spoke in the polite style in both OPIs. Interestingly, however,
his use of polite forms decreased in post-OPI from 80 per cent to 70 per
cent, and his plain forms in matrix predicates increased from 3 to 17
instances. The question, then, is whether he shifted the styles in ways
similar to how native speakers do in ways that reflect the social meanings
of the forms.
Qualitative analysis
The quantitative analysis above showed that Greg and Henry’s choice of
base style was no longer the polite style in post-OPIs, while Alan and
Sam predominantly maintained the polite style. Peter increased his use of
plain forms in his predominantly polite style. Their style shifts are exam-
ined below. Though it is likely that the interviewer’s choice of style would
influence the interviewee’s style choice, the current interviewer consis-
tently used polite forms and only used one-word utterances for clarification/
confirmation and plain forms in role plays (when playing a friend or a
child role).
Greg and Henry

In pre-OPI, Greg consistently used polite forms, and all plain forms were
in –si/node clauses. The copula was occasionally omitted when he requested
or provided clarification. However, there was one shift in a role play to
invite a friend to go to a movie. He spoke in the polite style, but switched
to the plain style in a soliloquy-like, exclamatory utterance in excerpt (6)
line 3.
(6) (Greg, Pre-OPI)
1 G: A, kyoo-no yoru-wa hima desu ka.
ah today-GEN night-TOP free COP Q
‘Uh, are you free tonight?’

N. IWASAKI 55
2 I: Hai, zikan arimasu yo. Kyoo-no yoru dattara.
yes time have-NONPAST-AH SFP(assert) today-GEN night COP-CONDITIONAL
‘Yes, I have time—tonight.’
3→ G: um. yokatta. um Supaidaa-man-o mitai desu ka.
um good-PAST um Spiderman-ACC watch-want COP-AH Q
‘Great! Uh, do you want to see Spiderman?’

Though this is an effective shift, it is unclear whether he had acquired the
meaning of an in-group self-directed shift for soliloquy/exclamatory utterance
because this was the only instance of shift. Possibly, he learned the sequence
yokatta! as an equivalent to the English ‘Great!’.
Upon return, his base style was the plain style, and his switches from the
plain to polite style appeared to be rather haphazard. He used only two polite
forms, other than eight instances of desyo(o) and three omoimasu ‘I think’. The
use of the polite form for the verb omou is not lexically fixed for Greg as he also
used the plain form omou six times in post-OPI. The remaining two instances
used the copula desu and -masu as shown in excerpts (7) and (8), lines 5 and 4,
respectively.
(7) (Greg, Post-OPI)
1 I: Gureggu-san-wa, ano, ima daigaku-no nan-nen-sei na n desu ka.
name-TOP well now university-GEN what-year-student COP N COP-NONPAST Q
‘Greg-san, uh, which year are you at the university now?’
2 G: Etto, ima, yo-nensei-gurai da to omou.
well now 4th-year-about COP QT think
‘Well, now, I think I am about a senior.’
3 Hontoni roku-nensei-gurai kedo...
really 6th-year-about but
‘....Actually, about the sixth year, but...’
4 I: Yo-nensei-gurai na n desu ka.
4th-year-about COP N COP-NONPAST Q
‘You are about the fourth year?’

5 → G: Un. Roku-nen-gurai benkyoo-site ita ga...ima, etto, sono, ima yo-nensei desu.
yeah 6-year-about study-GER AUX but now uh well now 4th-year COP-NONPAST AH
‘Yeah. I studied about six years, but... Now, uh, well, now I am a senior.’
Greg predominantly used plain forms in this sequence as well, and his moti-
vation for the abrupt switch to the polite form in line 5 does not seem to utilize
the social meanings of the polite form, such as deference or presentation of
self. However, when a situation clearly calls for expressions of deference, he
uses the polite form. In excerpt (8), he asks the interviewer a question to

request something from her, when she gave Greg an opportunity to ask ques-
tions as part of OPI protocol for Novice-High- to Intermediate-level candidates.
1 I: Etto, ano, ryuugaku-no koto-toka takusan Gureggu-san-ni situmon-simasita kedo,
uh well study-abroad-GEN about-such many name-DAT question-PAST-AH but
Gureggu-san, watasi-ni situmon-wa arimasu ka.
name me-DAT question-TOP have-NONPAST-AH Q SFP
‘I asked you a lot of questions about your study abroad, but do you have any
questions for me?’
2 Nan-demo ii desu yo.
what-ever good COP-NONPAST AH SFP(assert)
‘Any questions will be alright.’
3 G: Etto, Alan-wa ikkyuu-o uketai kara, etto, kyooryoo*-o sagasite iru.
well Alan-TOP level-one-ACC take-want because uh textbook-ACC look-for AUX
‘Well, because Alan wants to take the level 1 proficiency exam, he is looking for
[textbooks].’(*Greg probably meant ‘textbook’ [kyookasyo])
4 → Etto, ikkyuu-no nooryoku-siken motte imasu ka.
well level-one-GEN proficiency-exam have Q
‘Do you have a level 1 exam?’
5 I: Hai, mae-no hurui no-wa motte imasu kedo.
yes past-GEN old one-TOP possess-NONPAST-AH but
‘Yes, I have the old versions.’

N. IWASAKI 57
6 G: Un, sore-demo ii.
yeah that-COP-GER-also good-NONPAST
‘That will be fine.’
Unlike his other utterances, in which he primarily talks about himself

(his hobby, experiences in Japan, and his future goal), the agent of the verb
motu ‘possess’ in line 4 is the interviewer.
Moreover, when he performed a role play to ask his teacher to arrange a
make-up exam, he consistently used the polite style as in excerpt (9). First, he
carefully chose a very polite expression for apology and quickly corrected his

unintended use of a plain form to the polite form, and continued to use the
polite style in line 2.
1→ G: Etto, sensei, moosiwake arimasen ga,
uh teacher excuse have-NEG-AH but
tyotto siken-ni maniawanakatta, aemasen desita.
little exam for make-it-on-time-NEG-PAST POTENTIAL-NEG-PAST
‘Uh, teacher, I am very sorry. I did not, could not make it to the exam.’
2 → Tesuto-ni iku toki, watasi-no kuruma-wa kowarete simaimasita.
test-to go when my car-TOP break-GER end-up-PAST-AH
‘When I was going to the exam, my car broke down.’
3 I: Demo siken-no mae-ni renraku-ga arimasen desita nee...
but exam-GEN before-at contact-NOM exist-PAST-AH SFP
‘But you did not notify me in advance...’
4 → G: Unn, etto, zikan-ga arimasen desita.
Umm well time-NOM have-NEG-PAST-AH
‘Umm, uh, I did not have time.’
5 → Etoo, soreni, eto, keitai-o motte imasen kara, etto, renraku-dekimasen desita.
well also uh cell-phone-ACC possess-NEG because well contact-POTENTIAL-NEG-
PAST
‘Well, besides, uh, because I do not have a cell phone, well, I could not contact
you.’
His utterances in the role play clearly indicate that he can use the polite
style consistently in both matrix and kara clauses if he chooses to do so.
He uses the polite style in situations that call for expressing deference. This
is in contrast to his use of polite forms in pre-OPI in which he may have been
passively observing the social norms that he learned in classes. Though Greg’s
base style no longer conforms to the norm, he was actively making choices as
to which style to use.
Like Greg, Henry primarily used the polite style in pre-OPI, and changed his
base style to the plain style in post-OPI. In pre-OPI, he used only five plain
forms in matrix predicates, most of which can be considered similar to native
speakers’ shifts: situmoi-nikui by which he probably meant setumei-sinikui ‘it is

difficult to explain’ as a soliloquy-like utterance and empathic expressions, as
shown in excerpt (10) line 3. Prior to this sequence, Henry said that in his club
there are many people who are important to him.
(10) (Henry, Pre-OPI).
1 I: Hitori erande, donna hito ka osiete kudasai.
one-person select-GER what-kind person Q tell-imperative-AH
‘Then could you select one and tell me what kind of person s/he is?’
2 H: Eto, supeingo-ga hanasemasu. Yoku hanasemasu.
well Spanish-NOM speak-POTENTIAL-AH well speak-POTENTIAL-AH
‘Well, (she) can speak Spanish. (She) can speak very well.’
3 → Sorekara, kao-ga kawa, kawaii.
also face-NOM [fragment] cute-NONPAST
‘Also, (her) face is cut...cute.’
Henry’s use of the plain form kawaii ‘(her face) is cute’ in excerpt (10) line 3
is an effective shift to express an emphatic, emotional reaction to what is
being described. His pre-OPI suggests that, before studying abroad, he under-
stood that the appropriate base style to address a teacher/interviewer is the
polite style and that some style mixes are possible.
In post-OPI, however, he used the plain style as the base style. He shifted to
the polite styles 31 times, including one instance of desyo and six instances of
expressions of epistemic stances. Some of his shifts to the polite style appeared
to be rather haphazard as shown in his descriptions of how he found a house to
rent in excerpt (11). Henry first corrected his plain style utterance to polite in
line 3, but abruptly switched back to the plain form in line 4. It is unclear what
prompted this switch.
N. IWASAKI 59
1 I: Dooyatte, sono sumu tokoro-o sagasitari, issyoni sumu hito-o sagasita n desu ka.
how uh live place-ACC look-for-such together live people-ACC look-for N COP-
NONPAST-AH Q
‘How did you look for the place to live and people to live with?’
2→ H: ano staffu membaa-ga sono uti-ga aiteru tte, i, itte ta kara,

that staff member-NOM that house-NOM vacant-that said because,
3 sono ie-wa kurabu-no miitingu-de, sono ie-ga aiterutte itte masita kara,
that house-TOP club-NOM meeting-at that house-NOM is-vacant-AH because
4 sagasu hituyoo-ga nakatta.
search need-NOM exist-NEG-PAST
‘Because that staff member was saying that the house was vacant, because they
were saying that the house is vacant at the club meeting, there was no need to
look for it.’
However, closer examination reveals that some shifts can be explained. In

excerpt (12), where Henry describes his roommates, the information provided
by the plain forms of hiku ‘play’ in line 3 is background information subordi-
nate to the main idea (Makino 1983; Maynard 1991).
(12) (Henry, Post-OPI).
1 I: Yuniiku-na hitotati desita ka.
unique people COP-NONPAST-AH Q
‘Were they (roommates) unique people?’
2 H: Aa, un, ma, minna, ongaku su ongaku, uh, gakki-o, gakki, hikemasita, minna.
well uh everyone music..uh musical-instrument-ACC play-POTENTIAL-PAST-AH
everyone
‘Yes, uh, well, everyone lik..played music, music, musical instruments,
everyone.’
3→ Hutari-ga gitaa hikeru... Hitori-ga piano hikeru, hiketa kara,
two-NOM guitar play-POTENTIAL one-NOM piano play-POTENTIAL play-POTENTIAL-
PAST because
minna issyo-ni yoku onngaku-o kiitari, ongaku-o tuku, tukuttari simasita.
everyone together often music-ACC listen, music-ACC compose did-AH
‘Two can play the guitar, and one can play could play the piano and so we often

did such activities as listening to music together and comp, composing music.’
Moreover, in post-OPI, he consistently switched to the polite style when

asking the interviewer questions, such as in excerpts (13) and (14), lines 2
and 4.
I: de, uti-no sigoto-wa minna de, ano, kootai de sita n desu ka.
and house-GEN work-TOP all by well turn do-PAST N COP-NONPAST-AH Q
‘Did you take turns doing house chores?’
H: Kootai tte nan desu ka.
(Taking-turns) that what COP-AH Q
‘What is kootai?’
1 I: Nanika Henrii-san-kara watasi-ni situmon-sitai koto-ga arimasu ka.
something name-from me-DAT want-to-ask thing-NOM have-NONPAST-AH Q
‘Are there anything that you would like to ask me about?
2 → H: Ima kariforunia ni sunde imasu ka.
now California in live-GER AUX-AH Q
‘Are you living in California now?’
3 I: Hai.
‘Yes.’
N. IWASAKI 61
4→ H: Sono daigaku-wa doo desu ka.
that university-TOP how COP-AH Q
‘How is the university?’
Though some of Henry’s shifts seem haphazard, he speaks in the polite style
when he asked the interviewer questions. Indeed, if he had produced the
questions as in excerpts (13) and (14) in the plain style, he would have
sounded severely inappropriate. Moreover, like Greg, Henry appropriately
switched to the polite style when he performed a role play. He needed to

call an apartment manager to request a window repair, and he mostly main-
tained the polite style as in excerpt (15), lines 1 and 3.
1→ H: Ano, watasi-no mado-ga kowaremasita.
uh my window-nom break-AH-PAST
‘Uh..my window broke.’
2 I: Aa, soo desu ka.
ah so COP-NONPAST-AH Q
‘Is that so?’
3→ H: Ano, kodomo-ga yakyuu-o yarinagara, ano, sono beesubooru-ga,
well child-NOM baseball-ACC play-while, well that baseball-NOM
ano, watasi-no mado-o butukete, kowaremasita.
uh my window-ACC hit-GER break-AH-PAST
‘Well, while a child was playing baseball, his ball hit my window, and it broke.’
4 Doo sureba ii desu ka.
what do-CONDITIONAL good COP-AH-NONPAST Q
‘What would be good to do?’
5 I: Anoo, donnahuu-ni kowarete iru n desu ka.
well what-way-in break-GER AUX N COP-NONPAST-AH Q
‘Well, how is it broken?’
6 H: Akerarenai si, mado-no gurasu-ga, moo, nanka, un,
open-POTENTIAL-NEG and window-GEN glass-NOM already somehow um
mado-no gurasu-ga, ko. to-iu-ka, uun, nai.

Window-GEN glass-NOM [fragment] rather umm exist-NEG-NONPAST
Gurasu-ga nai, ima.
glass-NOM exist-NEG-NONPAST now
‘I can’t open it, and the window glass, well, the window glass is bro.., or rather, is
gone now’.
In excerpt (15) line 6, Henry returned to the plain form in the -si clause,
akerarenai, and in the matrix predicate nai. Otherwise, he mostly used the

polite style in this role play.
Thus far, we have qualitatively examined the choice of styles by Greg and
Henry, whose post-OPI base style was the plain style, diverging from the native
norm. Striking similarities were found between the two. They both spoke in
the polite style when expressing deference is highly needed—in asking for
information and making requests.
Alan and Sam

The quantitative analyses revealed that Alan and Sam were the most
consistent in their use of polite forms. In his pre-OPI Alan used the plain
form only once in a matrix clause as in excerpt (16) line 3, but the predi-
cate in question is more likely an embedded predicate in a disfluent sequence.
Prior to excerpt (16), Alan stated that after practice one can do amazing things
in juggling.
(16) (Alan, Pre-OPI)
1 I: huun, sugoi koto tte donna koto desu ka.
hmm amazing thing QT what-kind thing COP-NONPAST-AH Q
‘Hmmm, what sort of things are those ‘amazing things’?
2 A: Anoo..aa...booru-o, koo, takusan, aa, jaguru-suru toka,
well ah ball-ACC this-way many ah juggle such-as
‘Well..uh, like..juggling many balls, and..’
3 → aa, tatoeba, ima, watasi-wa, oo, itu..itutu? itutu-no booru-o jaguru-suru
ah for-example now I-top [fragment]...five five -GEN ball-ACC juggle
‘Uh, for example, now, I juggle fiv..five? five balls’
4 A: ano, rensyuu simasu.
well practice-do-NONPAST-AH
‘...well, I do practice.’
N. IWASAKI 63
His disfluency made the plain form suru at the end of line 3 look like a
sentence-final verb; however, it is likely that instead he was saying itutu-no
booru-o jaguru-suru rensyuu simasu ‘I practice juggling five balls’. Put simply,
Alan probably did not shift styles in matrix predicates at all in pre-OPI.
In post-OPI, Alan did use plain forms in matrix predicates (10 instances).
His shifts resemble native speakers’ shifts; many of them are soliloquy-like.
He is thinking to himself, using such expressions as doo ieba ii ‘how shall I
say it’, and . . . tte ieba ii kana ‘that’s probably how I should describe it’, or
evaluating past events while recalling them. In excerpt (17), he first informed
the interviewer that his travel was amazing in the polite style in line 3,
and recalling his experience, he makes an exclamatory remark as to how far

it was in the plain style in line 5.
(17) (Alan, Post-OPI)
1 I: De, donogurai-no aida ryokoo-sita n desu ka.
and how-long-GEN duration travel-PAST N COP-NONPAST-AH Q
‘So, how long did you travel?’
2 A: A, ikka..kurai, ikkagetu-gurai, toriaezu, ano, nihonkai-made, aru..arukimasita.
ah [fragment]..about one-month-about first well Japan-Sea-to walk-PAST-AH
‘Ah, about one, one month, first I wal..walked to the Japan Sea.’
3 Are-wa sugokatta desu.
that-TOP amazing-PAST COP-AH
‘That was amazing.’
4 I: Sugoi desu nee.
amazing COP-NONPAST-AH SFP(exclamation)
‘That’s amazing.’
5→ A: un, de, tyotto tookatta.
yeah and a bit far-PAST
‘Yeah, and that was a bit far.’
Like Alan, in pre-OPI Sam predominantly used the polite style and did not
shift to plain forms in matrix clauses. Though there was one instance of the
plain form in a matrix-like clause shown in excerpt (18) line 2, this sequence
seems to be incomplete. He is describing his audition to become a music major
at his university.
(18) (Sam, Pre-OPI)
1 I: Oodisyon-to iu no-wa donna koto-o sita n desu ka.
audition-QT say N-QT what-kind thing-ACC do-PAST N COP-NONPAST-AH Q
‘What kinds of things did you do in the audition?’
2 → S: Eto, senkoo, unn, ongaku-no senkoo-o saseru tame-ni, eto
well major um music-GEN major-ACC do-CAUSATVE-NONPAST purpose-of uh

sensei-no mae-ni, zibun-no, zibun-no, a, itiban tokui-na, aa, gakki-o hiite, aa,
teacher-GEN front-in self-GEN self-GEN best good musical-instrument-ACC play-
GER ah
sono ato-de, sensei-ga saseru, aa, saseru, aa...
that later-at teacher-NOM do-CAUSATIVE-NONPAST ah do-CAUSATIVE-NONPAST AH
‘Well, in order to major...let students major, (they make them) to play their own,
own, favorite musical instrument in front of the teacher, and after that, the teacher
let.. let.. ah...’
3 S: Ha....hairu ka kimemasu.
[fragment] enter whether decide-NONPAST-AH
‘... decides whether the student gets in.’
In contrast, in post-OPI, Sam did shift his style three times when he was
talking to himself with exclamation or emotion as in Oo sore-wa muzukasii ‘oh,
that’s difficult (to explain)’, itiban tanosikatta ‘it was the most enjoyable’, and
logical consequence that reflects his thought process when stating his opinion
as in excerpt (19) line 2. Prior to excerpt (19), the interviewer mildly chal-
lenged Sam’s opinion in favor of a recent change from seniority-based to
merit-based employment in Japan, stating that while some people get pay
raises, some are laid off due to this merit system.
(19) (Sam, Post-OPI)
1 S: Eeto, warui syoku-nin-wa kaiko-site ii n zyanai desu ka.. to omoimasu kedo.
well bad employee-TOP layoff-GER good COP-NEG-AH Q COMP think-NONPAST-AH
but
N. IWASAKI 65
2 → Eto, kaisya-ni warui eikyoo-o oyobosu kamo-sirenai.
well company-on bad influence-ACC give may-NONPAST
‘Well, isn’t it okay to layoff bad employees? I think... Otherwise, it might affect
the company negatively.’
In line 2, he spontaneously states his thought using the plain form. Alterna-
tively, the segment expressed in the plain form may be an increment. In other
words, background reasoning for his stance (i.e. it is alright to layoff those who
are incapable), which could have been stated earlier (e.g. before kaiko-site mo ii n

zya nai desu ka), may have been added after the final predicate, omoimasu kedo.
Sam was given a role play that called for the plain style in pre- and post-
OPIs. His performances in the role plays revealed his understanding of the
social meanings of both styles. In pre-OPI, he was instructed to talk to a
friend about a car accident he had in his friend’s car. Though he primarily
used the plain forms, he shifted the plain forms as in excerpt (20).
(20) (Sam, Pre-OPI)
1 S: A, konnitiwa, aa, watasi-wa tyoto..mondai-ga aru n de..da.
ah hello ah I-TOP little problem-NOM exist COP-GER..COP-NONPAST
‘Ah, hello, uh, I hav, have a problem.’
2 I: Mondai?
‘Problem?’
3 S: unto Kyoko-san-no kuruma-o unto, unten-site iru toki, kootuu ziko-ga atta n da.
um [name]-GEN car-ACC um drive-GER AUX when traffic accident-NOM occur COP-
NONPAST
‘Um, when I was driving your car, I had a car accident.’
4 Watasi-no sei dewa nakute, hoka-no untensyu-wa yopparatte ita to omoimasu.
I-GEN fault COP-NEG other-GEN driver drunk was COMP think-NONPAST-AH
‘It wasn’t my fault, I think the other driver was drunk.’
5 Eto, aa, bampaa-ga aa bampaa-wa tyotto kowarete ita.
Uh ah bumper-NOM ah bumper-TOP a little broken was
‘Uh..uh, the bumper, bumper was a little broken’
6 Sore igai, aa, sore igai wa daizyoobu de, da.
that other ah other-TOP all-right COP-GER COP-NONPAST
‘But other, other than that, all right, it is all right.

7 Eto, aa, watasi-ga naosu, aa, naosu okane-o aa harawasete kudasai.
uh ah I-NOM repair ah repair money-ACC ah pay-CAUSATIVE IMPERATIVE-AH
‘Uh..please let me pay the money to repair it.’
Sam was using both the plain forms atta n da and daizyoobu da and the polite
forms omoimasu and harawasete kudasai in this role play. Though not all native
speakers necessarily switch styles in the same way, this is the kind of occasion
in which native speakers manipulate the styles to sound apologetic. Sam’s
performance in pre-OPI indicates that he understood the general social mean-

ings of both the polite and plain styles prior to studying abroad, as he appro-
priately switched from the plain style to the polite in his role play. In post-OPI,
he also switched appropriately from the polite to plain as discussed above, and in
a role play where he talked to a visitor’s 4-year-old child in the plain style to be
friendly and approachable.
In pre-OPIs, Alan and Sam (except for Sam’s role play) were passive observers of
the alleged social norms and only used the polite style. Upon return, they volun-
tarily switched from the polite to the plain styles, which cautious speakers would
choose to do only if they have the knowledge of social meanings of the shifts.
Peter
Peter increased the use of plain forms in matrix predicates from 2 to 17
instances. The qualitative analysis revealed that most of his shifts resemble
reported native speakers’ strategic shifts. In pre-OPI, he largely maintained
the polite style and shifted to the plain style only twice. One is a soliloquy-
like expression sore-wa muzukasii ‘It’s difficult (to explain)’, and the other is to
state his stance by choosing from nominalized propositions that the interviewer
listed—whether his teacher, who adopted the Suzuki method of piano instruc-
tion, valued ‘being creative’ or ‘imitating teachers’, as in excerpt (21).
(21) (Peter, Pre-OPI)
1 I: De, sono Suzuki-sensei-no hoohoo, toiuka, Suzuki Program-dewa,
and that Suzuki-teacher-GEN method rather Suzuki Program-in-TOP
dotira-o daizi-ni suru n desu ka, kurieitibu-na koto to,
which-ACC value-COP do N COP-NONPAST-AH Q, creative N
sensei-to onazi yoo-ni, ma, sensei-no mane-o suru koto to.
and teacher-with same like well teacher-GEN imitation-ACC do N and
‘So, in Suzuki-sensei’s method, or Suzuki Program, which one do they value
more, being creative or doing the same.. imitating the teacher?’

N. IWASAKI 67
2 P: Sensei-no mane-o suru, motiron.
Teacher-GEN imitation-ACC do-NONPAST of-course
‘Imitate the teacher, of course.’
Peter more actively used style shifts in post-OPI; 12 of 17 plain forms in

matrix predicates were in self-directed soliloquy-like utterances—wondering,
recalling, or expressing emotions, such as soo kana ‘I wonder’, a musume-san-ga
ita ‘ah, he had a daughter’, and omosirokatta ‘that was fun’. He often concluded
his turn with plain forms, which served as summary statements, often recalling

his experiences with emotion or internal thoughts, such as in excerpt (22). His
uses of the plain forms are congruent with the functions of style shifts discussed
in the previous studies (e.g. Maynard 1991; Okamoto 1999; Makino 2002; Cook
2006). Peter, in response to the interviewer’s follow-up explanation concerning
his kendo club activities, concluded his turn with a plain-form predicate.
(22) (Peter, Post-OPI)
1 I: Dooiu koto-ni nareru koto-ga dekinakatta n desu ka, hazime wa.
what-kind things-to accustom N-NOM can-NEG-PAST COP-NONPAST-AH Q, first-TOP
‘What kinds of things could you not get used to, at first?’
2 P: hazime-wa, ma, sugoi, sonnani karada-o tukau-no-wa narete inakatta si,
first-TOP uh really that-much body-ACC use-N-TOP accustom-GER AUX-NEG-PAST
and
ato-wa, iroina sinpaigoto-ga atta kara, ma, nanka, hakkiri ienai n desu kedo,
rest-TOP various worry-NOM have-PAST because, well clearly say-NEG N COP-
NONPAST-AH but
zentaitekini zikan-ga kakatta ki-ga suru.
overall time-NOM take-PAST feeling-nom do-NONPAST
‘At first, I hadn’t been used to using the body that much, and I had various things
to worry about. I cannot clearly say..., but overall, I feel that it took a long time
(for me to get used to the club).’
In three instances, information provided in plain form predicates offer sub-

ordinate information (Ikuta 1983; Maynard 1991) as in excerpt (23), in which
he is explaining his part-time job.
(23) (Peter, Post-OPI)
1 I: Huun, donna baito site iru n desu ka.
Hmm what-kind part-time-job do-GER AUX-NONPAST N COP-NONPAST-AH Q
‘Hmm, what kinds of part-time jobs are you doing?’
2 P: Mazu-wa kyampasu-no [name of cafeteria]-de hataraite iru n...
first campus-GEN -LOC work-GER AUX that...

sandoicchi tukuttari suru.
sandwich make-such-as do
‘First, I work in the campus cafeteria... like, making sandwiches.’
He also performed a role play in which he talked to a little child using the
plain style in a friendly approachable way. Overall, Peter’s shifts both in pre-
and post-OPIs are quite appropriate. Importantly, he became more active in
shifting styles upon return, demonstrating his capacity to play a more active
role in interactions in Japanese society.
SUMMARY AND CONCLUSION

The quantitative analysis showed that even though all five participants spoke
in the polite style prior to studying abroad, upon return, Greg and Henry’s
styles diverged from the norm. They clearly overused the plain forms, con-
firming the reported tendency for some L2 sojourners to use the informal style
upon return. The other three continued to use the polite style as their base
style, though Peter increased the use of plain forms. If the quantitative analysis
alone were considered, one might conclude that Greg and Henry (and possibly
Peter) did not gain pragmatic competence and regressed in their use of the
polite/plain styles while studying abroad.
However, the qualitative analyses revealed that all participants gained some
understanding of social meanings of the styles and became more active lan-
guage users by choosing and shifting the styles. Rather than merely observing
the assumed native speakers’ norms presumably taught to them, they appear to
be actively making choices as to which style to use in different contexts. Alan,
Sam, and Peter, who appropriately chose the polite style as their base style in
post-OPIs, occasionally switched to the plain style. Peter, whose OPI rating was
the highest, most actively shifted the styles. Their switches to the plain form
index utterances that are close to their ‘selves’ (their self-directed thoughts,
recalls, and emotions) and/or organize information (subordinate information,
summary statements). Their moderate and appropriate style shifts also have the
effect of not sounding too distant, resulting in a degree of friendliness.
N. IWASAKI 69
On the other hand, both Greg and Henry chose to use the plain style as the
base style in post-OPIs—though this choice may be considered inappropriate.
Crucially, both Greg and Henry’s choice is not due to their inability to use the
polite style. When the situations called for an expression of deference, they
demonstrated their ability to consistently use the polite style, which reflects
their understanding of the social/situational meanings of the polite style.
Moreover, only some of their shifts to the polite style seem haphazard. It is
evident from the qualitative analyses that all participants in this study learned
through their study-abroad experiences that Japanese speakers make choices,
and they themselves also learned to choose between styles—though there is
considerable variability among individuals as to whether and to what extent

their choices resemble those of native speakers.
The conclusion above is not without limitation. In this study, the ways in
which the participants use the plain/polite styles in post-OPIs, as compared
with pre-OPIs, demonstrate that they became more active users of the style
shifts that reflect their social meanings, but they do not tell us the processes
by which they learned to become such active social agents. We can only infer
that they learned the social meanings of these forms through the active process
of language socialization while in Japan. We need to await future studies that
examine the sojourners’ socialization during their stay abroad or their retro-
spective views about their experiences and language use while abroad in order
to uncover the intricate relationship between their language choice, experi-
ences, and perceptions, which is no doubt related to their sense of identities.
L2 Japanese learners are generally introduced to the polite forms first and
are given abundant practice using them in the classroom. Moreover, both
linguists and textbook writers traditionally assert that Japanese speakers
observe the socially agreed-upon rules. As a result, they may simplistically
associate the polite forms with formal contexts and the plain forms with infor-
mal contexts, which are often only equated with interactions with close
friends, without truly understanding the social meanings of the styles.
During their stay in Japan, however, as they socialize and interact with
Japanese speakers, they are likely to be encouraged or pressured to use the
plain style by their Japanese peers and/or host families. They then may realize
that the plain style is not bound to certain contextual features (e.g. talk with
close friends), but that the form has social meanings which one employs
moment-to-moment to index intimacy/friendliness and self-directed utter-
ances to create a desired context.
If socialization in the target language community is the key for L2 learners to
acquire social meanings associated with different styles or address forms, then
the question arises as to whether formal instruction in the home country can
play any role in L2 learning of such social meanings. Though normative asso-
ciation between the form and context may be useful for the sake of simplicity,
some discussion of the social meanings of the forms and their flexibility and
variation among Japanese speakers that reflect the meanings may prove help-
ful for students.
ACKNOWLEDGEMENTS
I would like to express my sincere gratitude to Janet S. Shibamoto Smith, Gabriele Kasper,
Jane Zuengler, and the anonymous reviewers for their helpful and constructive comments,
and to Tammy Gales for thorough and thoughtful proofreading/editing.
APPENDIX
Abbreviations used in the excerpts
Particles
TOP topic marker
NOM nominative

ACC accusative
GEN genitive
DAT dative
INSTR instrumental
QT quotative
SFP sentence final particle
Others
AH address honorific
AUX auxiliary verb
COP copula
GER gerund
NEG negative
Q question marker
N nominalizer
CL classifier
NOTES
1 Japanese also has ‘referent honorifics’ interactions with their peers where
through which the speaker expresses sounding distant may be negatively
deference towards the referent viewed.
or humbleness (when referring to 3 In highly dependent clauses, the use
his/her own action). These are realized of the polite-form predicates
by verb choice. results in hyper-politeness. For exam-
2 This assumption may be question- ple, ittara ‘if I go’ can be expressed as
able when considering the possibility ikimasitara, which sounds excessively
that students primarily participate in polite.
REFERENCES
Althen, G. 1988. American Ways: A Guide for Atsuzawa-Windley, S. and S. Noguchi. 1995.
Foreigners in the United States. Yarmouth, ‘Effects of in-country experience on the acqui-
Maine: Intercultural Press. sition of oral communication skills in
N. IWASAKI 71
Japanese,’ Australian Review of Applied Makino, S. 1983. ‘Speaker/listener-orientation

Linguistics 12: 83–98. and formality marking in Japanese,’ Gengo
Barron, A. 2003. Acquisition in Interlanguage Kenkyuu 84: 126–45.
Pragmatics: Learning How to do Things with Makino, S. 2002. ‘When does communication
Words in a Study Abroad Context. Amsterdam: turn mentally inward? A case study of Japanese
John Benjamins. formal-to-informal switching,’ in N. M. Akatsuka
Barron, A. 2006. ‘Learning to say ‘you’ in and S. Strauss (eds): Japanese/Korean Linguistics 10.
German: the acquisition of sociolinguistic com- Stanford, CA: CSLI, pp. 121–35.
petence in a study abroad context’ Marriott, H. 1993. ‘Acquiring sociolinguistic
in M. A. Dufon and E. Churchill (eds): competence: Australian secondary students in
Language Learners in Study Abroad Contexts. Japan,’ Journal of Asian Pacific Communication
Clevedon: Multilingual Matters. 4/4: 167–92.
Brown, P. and S. C. Levinson. 1987. Politeness: Marriott, H. 1995. ‘The acquisition of politeness
Some Universals in Language Usage. vol. 4. patterns by exchange students in Japan,’

Cambridge, UK: Cambridge University Press. in B. Freed (ed.): Second Language Acquisition
Cook, H. M. 1996. ‘The Japanese verbal suffixes as in a Study Abroad Context. Amsterdam:
indicators of distance and proximity’ Benjamins, pp. 197–224.
in R. Dirven, R. W. Langacker, and J. R. Taylor Marriott, H. and S. Enomoto. 1995. ‘Secondary
(eds): The Construal of Space in Language and exchanges with Japan: Exploring students’
Thought. Berlin: Mouton de Gruyter, pp. 3–27. exchanges and gains,’ Australian Review of
Cook, H. M. 2001. ‘Why can’t learners of JFL Applied Linguistics 12: 64–82.
distinguish polite from impolite speech Maynard, S. K. 1991. ‘Pragmatics of discourse
styles?’ in K. R. Rose and G. Kasper (eds): modality: a case of ‘da’ and ‘desu/masu’ forms
Pragmatics in Language Teaching. Cambridge: in Japanese,’ Journal of Pragmatics 15: 551–82.
Cambridge University Press, pp. 80–102. Okamoto, S. 1999. ‘Situated politeness: manip-
Cook, H. M. 2006. ‘Japanese politeness as ulating honorific and non-honorific
an interactional achievement: Academic con- expressions in Japanese conversations,’
sultation sessions in Japanese universities,’ Pragmatics 9/1: 51–74.
Multilingua 25: 269–91. Okamoto, S. 2002. ‘Ideology and social mean-
Cook, H. M. 2007. ‘The Role of Desho(o) in JFL ings: rethinking the relationship between lan-
Learners’ Language Socialization’. Paper pre- guage, politeness, and gender,’ in S. Benor,
sented at the CLIC Symposium entitled, M. Rose, D. Sharma, J. Sweetland, and
‘Language Socialization, Interaction and Q. Zhang (eds): Gender Practices in Language.
Culture,’ UCLA, February 23–24. Stanford: Center for the Study of Language
Dufon, M. A. 2000. ‘The acquisition of linguistic and Information, pp. 91–113.
politeness in Indonesian by sojourners in nat- Regan, V. 1995. ‘The acquisition of sociolin-
uralistic interactions. PhD thesis. University of guistic native speech norms: effects of a year
Hawai’i Manoa, 1999,’ Dissertation Abstracts abroad on second language learners of French,’
International-A 60/11: 3985. in B. Freed (ed.): Second Language Acquisition
Hashimoto, H. 1993. ‘Language acquisition of in a Study Abroad Context. Amsterdam: John
an exchange student within the homestay Benjamins, pp. 245–67.
environment,’ Journal of Asian Pacific Schieffelin, B. B. and E. Ochs. 1986. ‘Language
Communication 4: 209–24. socialization,’ Annual Review of Anthropology 15:
Ikuta, S. 1983. ‘Speech level shift and conversa- 163–91.
tional strategy in Japanese discourses,’ Siegal, M. 1995. ‘Individual differences
Language Science 5: 37–53. and study abroad: women learning
Kasper, G. and K. R. Rose. 2002. Pragmatic Japanese in Japan,’ in B. Freed (ed.): Second
Development in a Second Language. Oxford, UK: Language Acquisition in a Study Abroad Context.
Blackwell Publishing. Amsterdam: John Benjamins, pp. 225–44.
Kinginger, C. and K. Farrell. 2004. ‘Assessing Siegal, M. 1996. ‘The role of learner subjectivity
development of meta-pragmatic awareness in in second language sociolinguistic competency:
study abroad,’ Frontiers: The Interdisciplinary western woman learning Japanese,’ Applied
Journal of Study Abroad 10: 19–42. Linguistics 17/3: 356–82.
doi:10.1093/applin/amp011 Advance Access published on 9 April 2009
The Relationship between Applied

Linguistic Research and Language Policy
for Bilingual Education1
DAVID CASSELS JOHNSON

Washington State University

Currently, restrictive-language policies seem to threaten bilingual education
throughout the USA. Anti-bilingual education initiatives have passed easily in
California, Arizona, and Massachusetts, while one was closely defeated in
Colorado, and federal education policy has re-invigorated the focus on English
education for English language learners, while concomitantly obfuscating the
possibility of native language maintenance and developmental bilingual educa-
tion. This is the educational landscape within which bilingual education
researchers, educators, and students must face the formidable challenge of pre-
serving educational choice and bilingual education. Thus, substantive research is
needed on how bilingual educators navigate this challenging ideological and
policy landscape. Based on an ethnographic study of bilingual education lan-
guage policy, this article takes up this challenge by focusing on how beliefs about
Applied Linguistics research influence the interpretation and appropriation of
federal language policy in one US school district. The results have implications
for the relationship between the Applied Linguistic research community and
language policy processes.
TOWARDS MULTI-LAYERED RESEARCH IN

LANGUAGE POLICY
The field of language planning and policy (LPP) has progressed through a
number of phases [see review in Ricento (2000)], engendering multiple frame-
works which attempt to describe the process of national language planning/
policy development (Fishman 1979; Haugen 1983). While enumerating some
of the possible (and perhaps preferable) steps in and goals of language plan-
ning, these earlier frameworks have been criticized for focusing on top-down
policy making and ignoring the sociopolitical contexts in which language plan-
ning occurs [see review in Ricento (2000)].
Reacting to these so-called positivistic approaches, critical research empha-
sizes how the state can use language policy to perpetuate systems of social
inequality (Tollefson 1991, 2002, 2006; Ricento 1998; Wiley 2002;
Pennycook 2006). Tollefson (2006) articulates the aims of critical language
policy (CLP): (i) it is critical of traditional apolitical LPP approaches and instead
‘acknowledges that policies often create and sustain various forms of social
inequality, and that policy-makers usually promote the interests of dominant
D. C. JOHNSON 73
social groups’ (Tollefson 2006: 42); (ii) it seeks to develop more democratic
policies which reduce inequality and promote the maintenance of minority
languages; and (iii) it is influenced by critical theory.
CLP scholarship has helped illuminate how language policies can be ideolog-
ical, and presents a rich picture of language policy development as one aspect
amongst many socio-political processes which may perpetuate social inequal-
ity, but it has also been criticized for underestimating the power of
human agency (Ricento and Hornberger 1996) and not capturing the processes
of language policy development (Davis 1999). As others have noted
(Pennycook 2002; Hornberger and Johnson 2007), an (over)emphasis on the
hegemonic power of language policies delimits the agentive role of local edu-

cators during implementation. Ricento and Hornberger (1996) introduce the
metaphor of an onion to evoke the multiple layers through which language
policy develops and emphasize the language policy power of the teacher exer-
cised through pedagogical decisions. They argue that LPP research has unsuc-
cessfully accounted for activity in all layers. Similarly, Ricento (2000) points
out that language policy research has tended to fall short of fully accounting
for precisely how micro-level interaction relates to the macro-levels of social
organization.
Still, more recent work has examined implementation. Proposition 227 was
a voter-approved measure that restricted the development and maintenance of
bilingual education programs in California. After Proposition 227 was enacted,
bilingual programs were reduced (Gándara 2000), but Garcı́a and Curry-
Rodriguez (2000) note that teachers responded to the law in different
ways—from defiance to acceptance. Reflecting this complexity, ethnographic
portraits of California education, post-Proposition 227, reveal different find-
ings. Baltodano (2004) finds that the formerly pro-bilingual education parents
internalized the English-only ideology in Proposition 227, thus succumbing
to its hegemonic influence. Valdez (2001) and Stritikus (2002), on the other
hand, discuss the agentive role that teachers played in implementation, some-
times shaping the English-only focus of Proposition 227 to meet the needs of
their linguistically diverse students.
Some anthropological (Levinson and Sutton 2001; Gale and Densmore
2003) and sociological work (Ball 2006) on educational policy foregrounds
the agentive role of local educators in policy processes. For example,
Levinson and Sutton (2001) propose a sociocultural approach to educational
policy which recognizes the power in authorized policy and involves research-
ing official policy formation, but also emphasizes the need to consider policy
appropriation2 when the ‘temporarily reified text is circulated across the various
institutional contexts, where it may be applied, interpreted, and/or contested
by a multiplicity of actors’ (Levinson and Sutton 2001: 2). They hope that
traditional and clear-cut divisions between policy formation and implementa-
tion will be replaced with conceptualizations of policy making as a process
which stretches across time and contexts (Ball 2006).
74 APPLIED LINGUISTIC RESEARCH AND LANGUAGE POLICY
The fate of bilingual education relies on language policy at multiple levels

from national policies [like No Child Left Behind (NCLB)] to classroom poli-
cies, and everything in between. Yet, most LPP research has traditionally
adopted a ‘top-down’ approach in that it examines either policy language
and/or the historical and sociopolitical development of policies without look-
ing at how such language is interpreted and appropriated. While ethnographic
studies of educational policy (Walford 2003) and educational language policy
(Stritikus 2002; Skilton-Sylvester 2003; Canagarajah 2006) are emerging,
Levinson and Sutton (2001) note that there is still a paucity of ethnographi-
cally informed data on educational policy and, similarly, Blommaert (1996)
and others (Davis 1999; Hornberger and Johnson 2007) argue that the field

of language policy needs more ethnographic studies which illuminate local
language policy processes.3 As well, little work has been done which includes
both critical analyses of policy discourse and ethnographically captures the
multiple layers of language policy creation, interpretation, and appropriation,
or, in Ricento and Hornberger’s (1996) terms, illuminates the various layers
of the LPP onion.
Thus, this article focuses on the appropriation of one policy—Title III of
NCLB—in the School District of Philadelphia (SDP) and, more specifically,
I focus on a theme which emerged during ethnographic field work: the role
of Applied Linguistics research and researchers in the creation, interpretation,
and appropriation of language policy. The following research question guides
this article: how does Applied Linguistics research shape the interpretation
and appropriation of Title III of NCLB?
METHOD
The ethnography of language policy
This study is an ethnography of language policy which ‘include[s] textual
and historical analyses of policy texts but must be based in an ethnographic
understanding of some local context’ (Hornberger and Johnson 2007). The
results reported herein are part of a larger 3-year (2002–5) multi-sited ethno-
graphic study of bilingual education policy and program development in the
SDP (Johnson 2007). Ethnographic data collection emerged out of a series of
action-oriented research projects on language policy and bilingual program
development with bilingual education teachers, administrators, and outside
researchers. These projects engendered participant observation and field note
collection in many different contexts within the SDP, including the following,
which are of interest for this article: (i) a series of meetings attended by
teachers and administrators during the development of the SDP language
policy, entitled ‘Policy for English Language Learners’ (2005); (ii) the central
administrative office in charge of language policy which operated under three
different names during data collection but is currently called the Office
of Language, Culture, and the Arts (OLCA); and (iii) bilingual teacher
D. C. JOHNSON 75
professional development meetings. For the sake of data triangulation, I con-

ducted multiple formal and informal interviews with key teachers and admin-
istrators, including recorded interviews with four administrators and four
bilingual teachers and I recorded naturally occurring conversation at language
policy meetings. I also analyzed (and contributed to) multiple drafts of the
formal SDP language policy as well as informal policy texts (SDP language
education literature and web site texts).
In order to contextualize what was happening in the SDP, the ethnographic
data were then compared with: (i) ‘top-down’ policy texts, including Title VII
of the Elementary and Secondary Education Act (ESEA), commonly known
as the Bilingual Education Act (BEA), and the multiple drafts of the policy

which took its place, Title III of NCLB; (ii) the political discourse which accom-
panied the creation, interpretation, and appropriation of Title III, including
Senate and House debate over NCLB (found in the legislative record), political
literature released by politicians, and web site information from the USA
and Pennsylvania Departments of Education; and (iii) interviews with one
Federal Department of Education administrator and one Pennsylvania
Department of Education Administrator.
Language policy discourse analysis

Along with ethnography, I use intertextual discourse analysis to analyze the
connections between the various layers of policy discourse. Bakhtin (1986)
proposes that both the texts we write and the speech we create (discourse)
are filled with the echoes of previous speakers and writers. These echoes, or
intertextual connections, imbue texts and discourse with dialogic overtones and
multiple meanings and any interpretation of a (potentially multi-voiced) text
or discourse requires an understanding of these intertextual connections.
Lemke notes that an intertextual approach requires any given text to be
analyzed ‘in the context of and against the background of other texts and
discourses’ (Lemke 1995: 10).
For this study, the object of analysis is policy discourse which includes spoken
interaction (e.g. policy meetings, congressional debate, and interviews) and
writing (e.g. language policy language). Contained within this policy discourse
are policy texts which are a part and product of the discourse and typically take
the form of a language policy in the traditional sense, i.e. the language policy
text. For example, Title III is the product of congressional debate and, there-
fore, a piece of Title III text is both a policy text and a product/part of policy
discourse. What is referred to as a text, then, is simply the reproduction of some
policy discourse and text for the purpose of analysis in this article. Following
Gee’s (1990) distinction of little d and big D discourse, (policy) Discourse refers to
larger discourses [after Foucault (1978)] which influence behavior and inter-
action and tend to normalize particular ways of speaking and being. The pur-
pose of intertextual discourse analysis of language policy is to analyze policy
text and discourse ‘in the context of and against the background of other texts
and discourses’ to illuminate policy Discourse and establish links between the
various layers of language policy.
Specifically for this article, I focus on circulating ideas about ‘research’ as
they influence bilingual education policy. Each text, collected during ethno-
graphic fieldwork, is analyzed in light of features within the layers of context
which may be intertextually linked. Explanation of the context in which
each text was collected is supported by field notes and participant observation.
Intertextual connections are then analyzed between micro-level language
policy discourse and ‘top-down’ or macro-level language policy Discourse.
The goal here is to trace the intertextual links between federal and local
language policy discourse by focusing on a particular theme: the role of

Applied Linguistics research and/or researchers in language policy creation,
interpretation, and appropriation.
FEDERAL LANGUAGE POLICY

This section deals with the role of research in the development of Title III
and in the resulting policy language. An analysis is offered of early drafts
of Title III and the discourse (legislative debate) which led to its enactment.
I look at how ‘bilingual education’ is defined and the requirement that
all language education programs should have scientifically based research
support.
Defining bilingual education

Beginning in 1968 when it was first enacted, Title VII of the ESEA, otherwise
known as the BEA, was the preeminent federal educational language policy
which governed how federal money was administered to language education
programs in the USA. At the time of its inception, the goals of bilingual educa-
tion were vague and definitions and typologies were scarce (Crawford 1998;
Ricento 1998; Wiley and Wright 2004); those on the political left and right in
the US supported what seemed to be an effective way for transitioning ELLs
into English language classrooms. This ambiguity in goals affected each install-
ment of the act when it was revised four more times, but in 1994 there was a
definitive shift in the policy language toward valuing bilingualism as a resource
(Wiese and Garcia 2001) and promoting both transitional and developmental4
bilingual education (in which native languages are developed). Yet, in 2002,
Title VII was effectively replaced by Title III of the newly named ‘No Child Left
Behind Act’ (NCLB). With its vigorous attention to English language acquisi-
tion for ELLs, Title III has fomented concern that developmental bilingual
education will be phased out and transitional or English-only pedagogical
approaches phased in (Wiley and Wright 2004).
While critics of Title III decry the dominant focus on English language edu-
cation and concomitant obfuscation of bilingual education, early drafts of the
policy were even more restrictive. House Resolution 1 (HR 1) was introduced
D. C. JOHNSON 77
in the House of Representatives by John Boehner (Republican-Ohio) and

began with ‘findings’ which make a conspicuous shift away from the language
in the BEA:
(a) FINDINGS – The congress finds as follows: (1) English is the common
language of the United States and every citizen and other person
residing in the United States should have a command of the English
language in order to develop their full potential [HR 1, Title III,
Sec. 3102]
HR 1 places English at the fore, both in education and in society, because it is
necessary for developing one’s full potential.
This nationalistically monolingual sentiment is accompanied by a three-year

time limit on ELL participation in bilingual education:
[Purpose:] (3) to assist local educational agencies to . . . prepare
limited English proficient students, including recent immigrant
students, to enter all-English instructional settings within 3 years
[emphasis added, HR1, Title III, Sec. 3102].
HR 1 might have restricted bilingual education to three-year transitional pro-
grams and concomitantly outlawed developmental bilingual education.
However, when HR 1 was sent to the Senate for debate, Senators such as
Jack Reed (Democrat-Rhode Island) argued that 3 years was an arbitrary
time limit and students may need more time before they transition into all
English instructional settings (22 June 2001, Congressional Record).
Therefore, in the Senate, the ‘findings’ and 3-year time limit were struck
from the policy and, when it was sent back to the House, legislators like
Silvestre Reyes (Democrat-Texas), who championed late-exit bilingual educa-
tion throughout the legislative process, celebrated the new version of the bill
as a victory for bilingual education. On the House floor, Reyes exclaimed:
I am proud to support the conference report on HR 1 . . . bilingual
programs are important to limited English proficiency children
because they build on native language proficiency to make
the transition to all-English academic instruction . . . Opponents of
bilingual education favored placing a three year time limit on
how long students can be enrolled in bilingual education regardless
of what level of English proficiency they reach . . . The compromise
bill gives students the flexibility to remain enrolled in bilingual
education as long as is appropriate. [Congressional Record,
20 December 2001, emphasis added]
Reyes identifies ‘bilingual education’ as a pedagogical method which builds on
native language proficiency to transition students to all-English academic
instruction. He does not believe that time restrictions should be placed on
this transition but he considers bilingual education to be, by definition, transi-
tional, and not a method for native language maintenance and, thus, his
definition does not include developmental programs.
While bilingual education supporters, like Reyes (and Ted Kennedy,

Democrat-Massachusetts), saw Title III as a victory for bilingual education,
opponents of bilingual education also felt victorious. In a press release,
Boehner, who had introduced HR 1, proclaimed:
As a result of the No Child Left Behind Act (H.R. 1), the bipartisan
education reform legislation signed in January by President
Bush, bilingual education programs across the country are being
transformed to give new tools to parents and to focus on helping
limited English proficient (LEP) children learn English. [News
from the committee on Education and the Workforce, 17 October
2002]

Boehner’s press release suggests that Title III will transform bilingual education
programs such that they focus on English and, presumably, not mother tongue
education. This, according to HR 1 (which Boehner introduced to Congress),
is necessary for ELLs to develop their full potential, and the quicker they can
transition to English, the better.
Both political allies (like Reyes and Kennedy) and political opponents
(like Boehner) celebrated Title III as a victory; as well, even political allies of
bilingual education perpetuated the Discourse that bilingual education was
necessarily transitional, a notion that is reflected in the final policy text. This
policy Discourse ignores applied linguistic research which (i) compares multi-
ple bilingual program typologies (Hornberger 1991; Garcı́a and Baker 2007)
and (ii) demonstrates the effectiveness of developmental bilingual education
relative to transitional and ESL-focused programs (Thomas and Collier 1997,
2002; Lindholm-Leary 2001; Krashen and McField 2005; Rolstad et al. 2005;
August and Shanahan 2006; Genesee et al. 2006).
‘Scientifically based research’

Along with the focus on English, Title III includes the provision that any
education program must be based on ‘scientifically based research’, repeating
the phrase 119 times. For example, in the beginning of Title III, one of the
‘Purposes’ sets the tone for the rest of Title III:
The purposes of [Part A of Title III] are to . . . provide State agencies
and local agencies with the flexibility to implement language
instructional educational programs, based on scientifically-based
research on teaching limited English proficient children, that the
agencies believe to be the most effective for teaching English
[Title III, Part A, Sec. 3102, 9].
This excerpt suggests that local educators must choose language programs
which are supported by scientifically based research; yet it also poses an inter-
pretive dilemma by simultaneously insisting on local flexibility (another
salient theme coursing through NCLB) while placing at least one, and perhaps
two, restrictions on local choice: (i) any program must be based on
D. C. JOHNSON 79
scientifically based research; and (ii) the agencies must believe in the effective-
ness of their chosen program. What is believe referring to—beliefs about the
most effective programs or beliefs about scientifically based research? Both
interpretations are plausible based on the policy text alone.
In order to ascertain how the US Department of Education interprets
these policies, I interviewed the director of the State Consolidated Grant
Division5 (Brinda Sea6) in the Office of English Language Acquisition to
discover what, if any, educational programs were supported or promoted
by the Department of Education. She immediately and unequivocally
responded that they do not promote or prefer any particular method and
are, in fact, prohibited from doing so: ‘We stay completely out of it’ (telephone

conversation, 24 May 2006). She stressed that it was up to the states and
schools to choose particular pedagogical programs for ELLs as long as the
chosen programs are ‘research-based’ and added that as long as they provide
support for their programs in the form of research, they are eligible to receive
Title III funds.
While NCLB does outline the parameters of scientifically based research (see
Title IX, A-B7), the US Department of Education provides little guidance for
local school districts8 regarding which programs the research supports or what
even counts as scientifically based research. Brinda Sea neither advocates
particular bodies of research nor language programs which such research
might support, and asserts that the US Department of Education cannot
prescribe particular pedagogical approaches.9 Not only does federal policy
discourse offer no clear answers, but also federal policy texts, like Title III,
can be confusing since they necessarily contain multiple voices and are
an intertextual mixture of new and old policy language. Of the multiple
authors of Title III, whose intentions carry the most power? Further,
how are such policies—whose authors have conflicting intentions—
interpreted and appropriated in school districts around the USA? How
do educators make sense of these multi-voiced texts?
THE SDP
The SDP is an example of a local education agency responsible for interpret-
ing and appropriating Title III. The fifth largest school district in the USA, the
SDP has around 200,000 students, 14,000 of whom receive ESOL/bilingual
program support services (representing more than a third of the ELL popula-
tion for the entire state of Pennsylvania). While almost 50% of the ELLs
are Spanish speaking, the SDP also serves large populations of Khmer,
Vietnamese, Chinese, and Russian speakers.
The SDP was a forerunner in bilingual education policy and has a history
of supporting bilingual education programs—shortly after Title VII competitive
grant funding became available under the BEA in 1968, pilot programs were
implemented. Since then, bilingual programs have included Russian, Chinese,
Khmer, and Spanish programs. Today, one central administrative office
oversees all ESOL/bilingual programs. OLCA is the primary interpretative con-

duit for federal language policy and the administrators therein are the primary
arbiters for interpretation and appropriation of educational language policies—
including Titles VII and III.
The data reported in this study arose from my close work with OLCA
on various projects including the SDP language policy which began with
what was called a ‘language policy retreat’ (March, 2003) and ended with its
ratification at a school board meeting (June, 2004). I acted as a participant–
observer during the production of the policy and I also helped edit the policy
document at various stages in its development. However, while this gave me
an insider’s perspective of how a local policy might be created, the minor

additions I made to the document were largely ignored in the end.
OLCA, in a sense, facilitates the intertextual links between federal and local
language policy, yet the nature of this facilitation varies depending on who is
doing the interpreting and, here, I focus on the various ways that applied
linguistic research and researchers influenced the interpretation and appro-
priation of Title III. Specifically, I look at how two OLCA administrators
and one outside applied linguistic researcher (and SDP consultant) influenced
language policy in the SDP during the development of the SDP language
policy.
Reflexive appropriation of applied linguistics research

In the fall of 2002, OLCA recruited Eve Island, a Title VII consultant to the
SDP and well-known and respected applied linguistic researcher, to begin
developing an official language policy for the entire school district.
Beginning in the spring of 2003, Eve organized and facilitated a series of
policy meetings which culminated in a 2-day language policy retreat at
which around 35 school district employees, teachers, and principals gathered
to flesh out what they would like to see in a language policy.
Eve was a human conduit and intertextual bridge between the Applied
Linguistics research community and the SDP practitioners. Eve would portray
the bilingual teachers’ struggle to maintain their bilingual programs as part of
a larger bilingual education movement going on throughout the USA. For
example, before the language policy retreat, Eve sent an inquiry to the
TESOL Advocacy Listserv, requesting advice for writing a school district
language policy; while she did not receive specific suggestions, she did receive
words of encouragement which she then announced at the beginning of the
language policy retreat. She reported that Donna Christian said, ‘I’ve never
seen anything like this!’ and Richard Ruiz said that he ‘loved this!’ (field
notes, 13 March 2003). In this way, well-known applied linguists were posi-
tioned as supporters of bilingual program development and the language
policy initiative within the SDP.
Eve not only brought words of encouragement to the language policy
retreat, but she also relied heavily on David Corson’s book Language Policy in
D. C. JOHNSON 81
Schools: A Resource for Teachers and Administrators (1999) (which proposes

theoretical models of school district language policy development) as a foun-
dation for the development of the SDP policy. She used the book as a step-
by-step manual for constructing the policy and she perpetrated the modus
operandi of Corson’s policy theories. For example, Corson adamantly argues
for school language policy creation which draws upon the knowledge and
resources of many different actors, not just a few key administrators, and
in doing so can resist ‘unfair aspects of reproduction and . . . help soften
social injustices’ (Corson 1999: 25). At the language policy retreat, Eve and
I discussed the ramifications of Corson’s perspective on the SDP language
policy:

Eve refers to creating spaces or the emancipation potential within a
language policy for challenging dominant ideologies and question-
ing how linguistic capital is allotted and this is really important in
Corson’s book. She hopes his theories about the power of a school
language policy to be emancipatory manifests in the SDP [field
notes, 14 March 2003].
Eve herself anthropomorphized the philosophy espoused in Corson’s book by
facilitating negotiation between teachers and administrators and empowering
teachers in the language policy development process. Eve challenged tradi-
tional divisions of language policy ‘creators’ and ‘implementers’.
Eve cited a community of applied linguistic researchers to promote and
maintain local Discourses about the benefits of developmental bilingual edu-
cation and she portrayed larger societal Discourses outside the SDP—as
espoused by Donna Christian and Richard Ruiz—as supportive of local
policy discourse. In this way, Eve challenged monolingual Discourses, so
prevalent within and without US educational language policy, and instead
portrayed applied linguistic research and researchers which support develop-
mental bilingual education as a dominant Discourse.
Applied linguistics research as support

During the development of the SDP language policy, I interviewed another
champion of developmental bilingual programs—the Chief Officer of OLCA,
Emily Dixon-Marquez. Emily helped author both the Title III application
and Title VII grants and was also an integral member of the community that
developed and drafted the SDP language policy. Her influence, however, tran-
scends the time limits of this study as she first started working in the district
in 1973, shortly after Title VII monies were made available under the BEA.
In a recorded interview, she reflects on the history of US bilingual education
language policy, beginning with the BEA:
David: How do you interpret the parameters that were set in the
Bilingual Education Act?
Emily: Well, I think that whoever set them probably wasn’t basing
it on research because there was none – so in terms of federal grants
3 and 5 years were always convenient numbers.
David: What are the 3 and 5 years?
Emily: Well, transitional bilingual programs were initially 3 years
and then out in the (mainstream) – and that’s how the legislation
read.
David: So in that title VII grant has continued until NCLB?
Emily: But it’s really evolved – it’s become an enrichment model
whereas before it was just remediation or a deficit type

approach . . . it’s much much much better . . . I think it started it
was like three million – and it’s now into the hundreds of millions
[recorded interview, 11 April 2003].
Emily characterizes the growth of language policy in the USA as developing
and incorporating a better sense of the research because, in part, there was no
body of bilingual education research when Title VII was developed. According
to Emily, US education policy has grown alongside and been buttressed
by applied linguistic research—as the research trends have grown away
from remediation or deficit approaches, and increasingly supported
enrichment models of language education, so has policy, leading right up to
NCLB.
I then asked Emily about NCLB’s requirement that all educational programs
be supported by scientifically based research and how this influenced their
Title III application. Emily said that they referred to the results reported in
Thomas and Collier (1997, 2002) suggesting that two-way (or dual language)
and one-way developmental bilingual education are the two most effective
program models for facilitating content knowledge, acquisition of English,
and maintenance of students’ mother tongues. Based on these studies,
OLCA chose to develop ‘dual language programs’. I asked her why these pro-
grams were acceptable under Title III:
Emily: [W]e preceded NCLB in terms of models and so forth – we’ve
been doing dual language, that’s not new to us, so it hasn’t been
enlightening in that sense. We’re glad that it’s an acceptable model
that they subscribe to.
David: Why is [dual language] an acceptable model?
Emily: I think probably because of the Thomas and Collier study –
which the government is comfortable with and they think has some
integrity to it – and it’s research based . . . plus I think this country is
finally coming to the realization – and I’m probably being really
optimistic, and maybe naı̈ve here – but coming to the realization
that we need to learn more about languages and cultures [recorded
interview, 11 April 2003].
D. C. JOHNSON 83
Emily asserts that NCLB is simply catching up with what they already knew
in the SDP—developmental bilingual education is an effective and research-
based language education model. Further, Emily senses a shift in the language
ideological landscape in the USA—we are coming to the realization that we
need to promote learning about various languages and cultures.
For Emily, the advancement of knowledge about languages and bilingual
education, supported by applied linguistic research, has engendered bilingual
educational policies which are increasingly sensitive to these shifts. She inter-
prets Title III as being at the forefront of this national ideological shift and thus
it embodies an enrichment model of bilingual education. Yet, as she argues,
the SDP preceded this national trend toward enrichment models of bilingual

education and thus NCLB has not been enlightening in that sense.
In our interview, Emily was generally very optimistic about Title III and its
effects on bilingual education in the SDP. Along with the policy text support
for dual language education, she expressed gratitude that under Title III, unlike
Title VII, money was no longer distributed according to a competitive grant
and the per pupil entitlement was more per pupil. For Emily, this was ‘won-
derful news’ (recorded interview, 11 April 2003). Emily’s interpretation of
Title III stands in contrast to its creators, like Reyes, who conceptualize ‘bilin-
gual education’ as necessarily transitional. This is not because Emily was una-
ware of the English-focused flavor of NCLB (as the author of the Title III
application, she was well aware of its requirements), but she still interpreted
it as supporting the dual language programs she intended to implement with
Title III money.
Referencing research in local language policy

Within the discourse community responsible for the SDP language policy,
Emily and Eve helped perpetuate the belief that developmental bilingual
education was superior to transitional programs. Inspired by this belief and
buttressed by applied linguistic research (Thomas and Collier 1997, 2002)
and researchers (e.g. Donna Christian, Richard Ruiz, and David Corson), it
appeared as though a policy which outright resisted and/or ignored Title III
would develop. However, initial policy language filled with idealism even-
tually gave way to pragmatism, and, while the initial community of policy
developers included a more egalitarian mix of teachers and administrators
from multiple levels of institutional authority, OLCA eventually took control
of the drafting process.
The developers of the policy, including Eve, and especially Emily, were
cognizant of the language in Title III and drafted a policy which appropriates,
ignores, and/or alters Title III text. The first shift is a conspicuous snub of
the ‘scientifically-based’ language in Title III in favor of:
High quality programs for ELL’s provide optimum conditions
for English Language acquisition, and therefore, the District com-
mits to . . . programs for ELLs using sound instructional practices
as shown by research in the field of second language acquisition

[III, 1a].
Note the immediate nod to English language acquisition and the district’s
assertion that they are committed to programs which ensure this, thus appro-
priating Title III’s English-focused discourse. However, using second language
acquisition research to engender sound instructional practices is conspicuously
different than Title III’s assertion that programs need to be based on ‘scientif-
ically based research’.
Then, unlike Title III, the policy defines bilingual education:
A bilingual education program develops and maintains first
language literacy as well as literacy in the second language

[Section III, Part 2b].
This text defines ‘bilingual education’ as a method which develops and
maintains first language literacy which is not a goal of transitional programs
and stands in contrast to Title III’s assertion that the goal of bilingual education
is for ‘limited English speakers to enter all English-instructional settings’.
Thus, definitions of bilingual education as necessarily transitional in federal
policy text and discourse are resisted (or ignored) in favor of language
which allows for developmental programs. Then, unlike Title III, the SDP
policy makes explicit claims about what language programs the ‘research’
supports:
The district recognizes that . . . research indicates that strong first
language development serves as an effective foundation for
second language acquisition [III, 3b].
This allusion to the interdependence hypothesis (Cummins 1981) asserts that
English acquisition is aided by first language development, and presumably,
pedagogy in which first languages are strongly developed.
The SDP language policy text makes a few important discursive shifts away
from, while concomitantly maintaining, some of Title III’s English-only focus.
While SDP language policy text appropriates some of the salient English-
focused language from Title III, it also creates implementational space for
developing various types of bilingual programs because it makes specific
claims about what the research supports—strong first language develop-
ment—and defines bilingual education as a program which develops first
languages.
Applied linguistic research as doctrine

The remainder of this article focuses on shifts in bilingual education policy
that took place between the Fall of 2003 and Spring of 2005, illustrated
by changing definitions of ‘good’ bilingual education and shifting interpreta-
tions of Applied Linguistics research. What is missing in this study, and what
might be expected at this point, is a description of the implementation of
D. C. JOHNSON 85
the SDP language policy. In large part, there was no implementation and
thus it existed, and continues to exist, as an interesting remnant on
the OLCA web site, but has no real power. Its lack of power is, in part,
due to a shift in pedagogical philosophy and bilingual education policy
at OLCA.
In the fall of 2003, OLCA acquired a new director of ESOL/bilingual pro-
grams, Lucı́a Sanchez, whose beliefs about applied linguistic research were
different than Emily’s and Eve’s and influenced language policy accordingly.
Emily Dixon-Marquez and Lucı́a Sanchez are an interesting comparative jux-
taposition of how language (education) ideologies influence interpretation and
appropriation of top-down language policies like Title III. Lucı́a started work-

ing at OLCA as Emily, who quit OLCA in 2004, was leaving, and as the
current director of ESOL/bilingual programs, Lucı́a has been the most influen-
tial interpreter and appropriator of federal and state language policy since
2004.
Before coming to Philadelphia, Lucı́a worked at the Pennsylvania
Department of Education where she reviewed and approved the Title III
plan submitted by Emily. Thus, when Lucı́a started working at OLCA, she
adopted the responsibility of overseeing implementation of Title III monies
which she herself had approved! Yet, Lucia’s beliefs about bilingual education
differ from Emily’s and stand in contrast to the additive bilingual beliefs
circulating through OLCA and the development of the SDP language policy.
I had a chance to interview Lucı́a and ask her about programmatic changes
that were underway:
When you talk about bilingual education, in many research based
models, it is uh – considered a way to help children, or assist
students who are second language learners, to acquire English –
The problem we have seen historically with bilingual education is
that – somehow even students who are born . . . in the United States
have been placed in bilingual education programs just . . . because,
even though they’re native speakers of English, they come in with
some gaps with their linguistic development due to maybe speaking
another language at home or even hearing another language at
home [recorded interview, 13 June 2005].
Like Silvestre Reyes, what Lucı́a refers to as ‘bilingual education’ is a
programmatic model, the primary goal of which is acquisition of English for
ELLs. Bilingual education in her usage is specifically intended for non-native
English speakers. Speaking or hearing a non-English language at home may
induce gaps in some English-speaking students’ English language develop-
ment, but those students, as English speakers, are still not good candidates
for bilingual education in her view. Lucı́a’s beliefs about research influence
her conceptualization of what ‘bilingual education’ is and who it is for; it does
not include developmental programs and is not for English speakers born in
the USA.
Lucı́a has promoted transitional bilingual programs which exit ELLs to

English-medium classrooms by middle school. Her interpretation of the
research accommodates (or vice versa) her interpretation of the intentions
of Title III:
Title III was created to improve English language acquisition
programs by increasing the services or creating situations where
the students would be getting supplemental services to move
them into English language acquisition situations [recorded inter-
view, 13 June 2005].
Lucia interprets the goals of ‘bilingual education’ and Title III as the same—
eventual transition of ELLs into mainstream classrooms. She does not feel

that the ELL students in Philadelphia are well served by maintenance programs
and for these students, the preferred model is transitional (with all bilingual
education students entering mainstream classrooms by middle school) with
the option of heritage language classes to continue biliteracy development.
Throughout ethnographic data collection, bilingual education meetings
were held which involved policy and program development and teacher train-
ing. During one such meeting, led by Lucı́a, a couple of the teachers expressed
concern that the programmatic changes being implemented were a clandestine
attempt to get rid of bilingual education and mainstream ELLs, a claim which
Lucı́a adamantly rejected. After Lucı́a introduced the transitional program, one
of the teachers challenged her decision:
Teacher: Who or where did the decision make – come from to [tran-
sition students]?
Lucia: Because, because, number one, we looked at all the programs
that are effective based on Krashen’s research – and the beginning
of Title III of the No Child Left Behind Act, which is long,
and there’s nothing we can do to change that [tape recorded,
1 December 2005].
Lucı́a’s beliefs about the research, and here she alludes to Stephen Krashen’s
research, are used to justify her interpretation of Title III as restrictively focused
on English language development. By conferring language policy decisions
to Stephen Krashen and Title III, Lucı́a deflects responsibility for her decision
to shift SDP language policy to outside experts and outside language policy,
both of which rigidly dictate a transitional bilingual language policy and
‘there’s nothing they can do to change that’.
Later in the meeting, she emphasizes this point by saying:
Everyone knows about Stephen Krashen – he’s a linguist that has
devoted most of his research to education, but he’s a linguist. He’s
a scientist that studies different linguistic patterns but he really – we
heard about the silent period through Krashen, we heard about
comprehensible input, that’s Krashen. We heard about the lower-
ing the affective filter, that’s Krashen, error correction, that’s
D. C. JOHNSON 87
Krashen, so all of that is good research that we all as language

teachers need to know. And he said, he is the expert, and he said
that, yes, you can introduce English right away – Yes it is important
that we know what the research says [tape recorded, 1 December
2005].
In these two texts, Lucı́a clarifies her beliefs about what role ‘research’ should
play in language policy decision making. While Eve refers outside research and
researchers in an attempt to enhance teacher agency, and Emily interprets the
research as supportive of local choice and developmental bilingual education,
Lucı́a positions the research, here embodied by Krashen, as setting rigid stan-
dards to which SDP language policy, and the teachers, must adapt. She

enhances Krashen’s standing as the expert by emphasizing his position as a
linguist and a scientist; in other words, Lucı́a is relaying Krashen’s expertise to
the teachers who should not blame the messenger for the message. The tea-
chers are stripped of their expertise and agency in making language policy
decisions—since they are neither linguists nor scientists (nor experts according
to Lucı́a)—and Lucı́a deflects the responsibility of SDP language policy decision
making to Krashen and Title III.
DISCUSSION
At this point, I would like to return to the research question proposed earlier:
how does applied linguistic research shape the interpretation and appropria-
tion of Title III in the SDP?
1 Ignoring the research: despite the preponderance of research supporting
the relative ‘effectiveness’ of bilingual programs, administrators in the US
Department of Education and federal lawmakers seem largely oblivious,
or at least do not subscribe, to particular studies or bodies of research.
Federal policy discourse perpetuates the limiting definition of ‘bilingual
education’ as a means for transitioning ELLs into English-medium class-
rooms. Obscured in this federal discourse about bilingual education is the
notion that it can be a means for fostering bilingualism and biliteracy for
both native and non-native English speakers alike; that it is a means for
utilizing languages as (cultural, educational, or economic) resources; or
that ELLs have a right to literacy in their mother tongues. Thus, federal
policy discourse and the resultant policy text (Title III) help perpetuate
the popular Discourse that bilingual education is, by definition,
transitional.
2 Vague demands for research support: still, in the US Department of
Education, the views about what the research supports are not clear
and/or not enforced when administering Title III monies, and the admin-
istrator interviewed for this study insisted that they would ‘stay out of
it’ as long as the Title III grant applicants provide (any) research support
for their chosen programs. While the definition of ‘scientifically based
research’ in Title IX is (somewhat) clear, this administrator did not give a

strict definition nor make claims about which programs are best, and she
is directly responsible for administering Title III monies, thus leaving
implementational space for creative interpretation of what ‘scientifically-
based research’ is and which programs it supports.
3 Including researchers in the local discourse community: applied linguistic
research is appropriated in the SDP in varying ways. Eve Island, the
Title VII consultant and applied linguistic researcher, appropriated
research (including her own) and voices from an outside community of
bilingual education advocates and applied linguistic scholars (like Richard
Ruiz) to contextualize and support educator efforts to develop develop-

mental bilingual policy and pedagogy. Positioning the research(ers)
as allies, and the SDP practitioners as experts, empowered teachers to
become active agents in policy making. Eve served as a human conduit
and intertextual link between a community of Applied Linguistics
researchers and SDP educators; she positioned the policy developer’s
initiative ‘within the context of and against the background of’ (Lemke
1995: 10) academic discourse which supports both developmental bilin-
gual education and developing school district policy in an egalitarian
way that appropriates a diversity of teacher and administrator voices.
Concomitantly, she challenged traditional divisions between policy ‘crea-
tors’ and ‘implementers’ and exemplified how policies which reduce
inequality and promote the maintenance of minority languages can be
developed more democratically (Tollefson 2006).
4 Applied linguistic research as support: the Titles VII and III grant writer,
Emily Dixon-Marquez, viewed the requirement that programs must be
based on scientifically based research as beneficial because the research
she knew—especially the Thomas and Collier studies (1997, 2002)—sup-
ported the programs she wanted to promote in the SDP. Within the
ideological space created at the language policy retreat and maintained
by key players, developmental bilingual programs were positioned as
superior to transitional programs and the Thomas and Collier studies
(1997, 2002) were appropriated as allies. Further, Emily interprets
Title III as being supportive, in its language and funding, of dual language
programs.
5 (Mis)appropriating Applied Linguistics research as a mandate: while Emily
(and Eve) referred outside research and researchers in an attempt to
enhance teacher agency, the new head of ESOL/bilingual programs,
Lucı́a Sanchez, positioned the research, here embodied by linguists like
Stephen Krashen, as setting rigid standards (i.e. transition into English) to
which SDP language policy, and teachers, must adapt. Lucı́a’s interpreta-
tion of Title III was filtered through her definition of ‘bilingual education’
as necessarily transitional and she cited Krashen (in her words, a linguist
and a scientist) as an expert who supports this transition. The expertise of
the teachers was diminished in favor of outside experts whose research
D. C. JOHNSON 89
placed restrictions on SDP language policy by demanding transitional

programs, which is an inaccurate interpretation of Krashen’s research
(Krashen 1996; Krashen and McField 2005). Lucı́a reified popular defini-
tions of bilingual education (as promoted by federal policy discourse)
as necessarily transitional and, further, limited the role of teachers in
language policy processes to that of ‘implementer’.
Cutting edge theoretical developments portray language policy as a potentially
hegemonic mechanism through which dominant ideologies about
language (education) are promoted (Pennycook 2006; Tollefson 2006). The
results of this paper suggest that emphasis should be placed on the word
‘potentially’ since it is shown that while language policies can have this

effect, the effects of policies are channeled through human agents in local
educational contexts and, thus, policy appropriation can be varied and unpre-
dictable. While Emily and Eve resisted the English-focused discourse in Title
III, Lucı́a appropriated it and promoted state-sponsored definitions of ‘bilingual
education’ over definitions cited in applied linguistic research. Both Emily and
Lucı́a advocate bilingual education but their beliefs about the best type of
bilingual education, and what the research says, colored their interpretations
of Title III. Using the same Title III policy language and money as support, they
implemented ESOL/bilingual programs in very different ways.
Emily’s interpretation that Title III is as flexible as it claims, and her beliefs
about bilingual education research, created and supported ideological and
implementational spaces for additive bilingualism. She and Eve fostered an
ideological space for bilingual language policy and program development in
which developmental programs were favored and an egalitarian discourse
community of administrators and teachers was encouraged to engage with
language policy processes. The educational transformation relies on Lucia’s
interpretation of Title III as rigidly English-dominant and her own beliefs
about language education research. In order for the effects of Title III to be
truly monolingual, at least in Philadelphia, administrators must allow them-
selves to be conscripted by its monolingual discourse.
CONCLUSION
It is difficult enough to pinpoint intentions in a single-authored text but
a language policy has multiple authors with varied intentions. As Bakhtin
(1986) argues, there are dialogic overtones and multiple meanings within
any given text. Title III is no different. Still, while a plurality of readings and
interpretations of national language policies like Title III of NCLB are possible
(and a reality in this study), policy Discourse constrains interpretation
and implementation possibilities—some interpretations will be privileged,
especially those sanctioned in federal discourse, while others may be obfus-
cated or discredited (Ball 2006). The production of Title III was characterized
by limiting definitions of bilingual education as necessarily transitional and
this influenced the resulting policy text and some interpretations in the SDP.
One administrator’s appropriation of Title III suggests that her interpretation
was constrained by the English-focused and transitional bilingual education
discourse; therefore, Title III put limits on what was educationally feasible.
Still, other bilingual educators in the SDP were neither swayed by the
English-focused discourse, nor the limiting definitions of bilingual education
so prevalent in Title III, and used applied linguistic research to support their
interpretation.
It is up to local education agents to either expand or restrict their bilingual
programs using Title III monies and it has gone both ways in the SDP. During
the first half of ethnographic data collection, Title III was interpreted as

flexible. In this ideological space of bilingual language policy and program
development, Title III was portrayed as a real windfall for SDP bilingual
programs and its flexibility was emphasized to teachers and other stakeholders.
A shift in personnel, however, led to different interpretations of how Title III
monies should be used, marked by beliefs that Title III’s focus on English was
a mandate for transitional programs.
As well, it is up to applied linguists to become engaged in local language
policy processes. This article shows that bilingual language policy implementa-
tion is an agentive process which can be influenced by applied linguistic
research and researchers. Because they are positioned as experts, at least in
Philadelphia, researchers can become members of the discourse community
which shapes the appropriation of language policy. In order to resist the pen-
dulum swing away from native language maintenance, researchers will need
to team with local educators—and especially administrators—to foster local
ideological spaces in which developmental bilingual education is championed.
And this is possible. SDP educators make choices and these choices influence
the implementation of Title III. Such choices are constrained by the language
policies which tend to set boundaries on what is allowed and/or what is
considered ‘normal’, but the line of power does not flow linearly from the
pen of the policy’s signer to the choices of the teacher. The negotiation at
each institutional level creates the opportunity for reinterpretations and
policy manipulation. Local educators are not helplessly caught in the ebb
and flow of shifting ideologies in language policies—they help develop, main-
tain, and change that flow.
TRANSCRIPTION NOTES
() transcription doubt
(?) unclear utterance
... ellipsis
- short pause
Italics emphatic stress
D. C. JOHNSON 91
NOTES
1 The study upon which this is based 5 This department is responsible ‘for the
won the 2008 National Association administration of new formula grants
of Bilingual Education dissertation and for providing technical assistance
competition. to State and Local educational
2 Levinson and Sutton (2001) prefer the agencies’, http://www.ed.gov/about/
term appropriation to implementation offices/list/oela/aboutus.html
which they feel implicitly ratifies a 6 All names are pseudonyms.
top-down perspective. I agree and 7 This passage might also be subject to
adopt this term. varying interpretations (Johnson
3 While other ethnographic work deals 2007).

with language policy (Hornberger 8 A school district is an administrative
1988; Davis 1998; Freeman 1998), unit which serves the schools in one
the object of analysis in this body of or more cities. There is one school dis-
work is not policy itself. trict which serves Philadelphia but it is
4 Developmental or additive bilingual divided into 11 regions.
education can be contrasted with sub- 9 Similar assertions were made by repre-
tractive bilingual education and can sentatives from the Pennsylvania
include one- and two-way/dual lan- Department of Education.
guage programs.
REFERENCES
August, D. and T. Shanahan. (eds) 2006. Crawford, J. 1998. ‘Language politics in the
Developing Literacy in Second-language Learners: U.S.A.: the paradox of bilingual education,’
Report of the National Literacy Panel on Social Justice 25/3: 50–69.
Language-minority Children and Youth. Cummins, J. 1981. ‘The role of primary lan-
Lawrence Erlbaum and Center for Applied guage development in promoting educational
Linguistics. success for language minority students in
Bakhtin, M. 1986. Speech Genres and Other Late California State Department of Education’ in
Essays. Texas University Press. Schooling and Language Minority Students: A
Ball, S. J. 2006. Education Policy and Social Class: Theoretical Framework. Evaluation, Dissemina-
The Selected Works of Stephen J. Ball. Taylor & tion and Assessment Center California State
Francis. University.
Baltodano, M. P. 2004. ‘Latino immigrant par- Davis, K. A. 1994. Language Planning in
ents and the hegemony of Proposition 227,’ Multilingual Contexts: Policies, Communities, and
Latino Studies 2: 246–53. Schools in Luxembourg. John Benjamins
Blommaert, J. 1996. ‘Language planning as a Publishing Company.
discourse on language and society: The linguis- Davis, K. A. 1999. ‘The sociopolitical dynamics
tic ideology of a scholarly tradition,’ Language of indigenous language maintenance and loss:
Problems and Language Planning 20/3, 199–222. a framework for language policy and planning’
Canagarajah, S. 2006. ‘Ethnographic methods in in T. Huebner and K. A. Davis (eds):
language policy’ in T. Ricento (ed.): An Sociopolitical Perspectives on Language Policy and
Introduction to Language Policy: Theory and Planning in the USA. John Benjamins
Method. Malden: Blackwell Publishing. Publishing Company.
Corson, D. 1999. Language Policy in Schools: A Fairclough, N. 2001. ‘The discourse of new
Resource for Teachers and Administrators. labour: Critical discourse analysis’
Lawrence Erlbaum. in M. Wetherell, S. Taylor, and S. J. Yates
(eds): Discourse as Data: A Guide for Analysis. Krashen, S. 1996. Under Attack: The Case against
Milton Keynes: The Open University. Bilingual Education. Language Education
Fishman, J. S. 1979. ‘Bilingual education, lan- Associates.
guage planning and English,’ English World- Krashen, S. and G. McField. 2005. ‘What
Wide 1/1: 11–24. works? Reviewing the latest evidence on
Foucault, M. 1978. The History of Sexuality: An bilingual education,’ Language Learner 1/2:
Introduction, Vol. 1. Random House. 7–10, 34.
Freeman, R. D. 1998. Bilingual Education and Lemke, J. L. 1995. Textual Politics: Discourse and
Social Change. Multilingual Matters. Social Dynamics. London: Taylor & Francis.
Gale, T. and K. Densmore. 2003. Engaging Levinson, B. A. U. and M. Sutton. (eds) 2001.
Teachers: Towards a Radical Democratic Agenda Policy as Practice: Toward a Sociocultural Analysis
for Schooling. Milton Keynes: Open University of Educational Policy. Ablex Publishing.
Press. Lindholm-Leary, K. J. 2001. Dual Language

Gándara, P. 2000. ‘In the aftermath of the Education. Clevedon: Multilingual Matters.
storm: English learners in the post-227 era,’ Pennycook, A. 2002. ‘Language policy and
Bilingual Research Journal 24/1: 2. docile bodies: Hong Kong and governmental-
Garcı́a, E. E. and J. E. Curry-Rodriguez. 2000. ity’ in J. W. Tollefson (ed.): Language Policy in
‘The education of limited English proficient stu- Education: Critical Issues. Lawrence Erlbaum.
dents in California schools: an assessment of the Pennycook, A. 2006. ‘Postmodernism in lan-
influence of proposition 227 in selected districts guage policy’ in T. Ricento (ed.): An
and schools,’ Bilingual Research Journal 24/1: 2. Introduction to Language Policy: Theory and
Garcia, O. and C. Baker. (eds) 2007. Bilingual Method. Blackwell Publishing.
Education: An Introductory Reader. Multilingual Ricento, T. 1998. ‘National language policy
Matters. in the United States’ in T. Ricento and
Gee, J. P. 1990. Social Linguistics and Literacies. B. Burnaby (eds): Language and Politics in the
Falmer. United States and Canada: Myths and Realities.
Genesee, F., K. Lindholm-Leary, W. Lawrence Erlbaum.
M. Saunders, and D. Christian. 2006. Ricento, T. 2000. ‘Historical and theoretical per-
Education English Language Learners: A Synthesis spectives in language policy and planning,’
of Research Evidence. Cambridge: Cambridge Journal of Sociolinguistics 4/2: 196–213.
University Press. Ricento, T. and N. H. Hornberger. 1996.
Haugen, E. 1983. ‘The implementation of corpus ‘Unpeeling the onion: language planning and
planning: theory and practice’ in J. Cobarrubias policy and the ELT professional,’ TESOL
and J. A. Fishman (eds): Progress in Language Quarterly 30/3: 401–27.
Planning. Mouton de Gruyter. Rolstad, K., K. Mahoney, and G. V. Glass.
Hornberger, N. 1988. Bilingual Education and 2005. ‘The big picture: a meta-analysis of
Language Maintenance. Foris Publications. program effectiveness research on English
Hornberger, N. H. 1991. ‘Extending enrichment Language Learners,’ Educational Policy 19/4,
bilingual education: revisiting typologies and 572–94.
redirecting policy I’ in O. Garcı́a (ed.): Skilton-Sylvester, E. 2003. ‘Legal discourse and
Bilingual Education: Focusschrift in Honor of decisions, teacher policymaking and the multi-
Joshua A. Fishman. Volume 1. John Benjamins. lingual classroom: constraining and supporting
Hornberger, N. H. and D. C. Johnson. 2007. Khmer/English biliteracy in the United States,’
‘Slicing the onion ethnographically: layers and International Journal of Bilingual Education and
spaces in multilingual language education Bilingualism 6/3,4: 168–84.
policy and practice,’ TESOL Quarterly (Special Stritikus, T. 2002. Immigrant Children and
Issue: Language Policies and TESOL: Perspectives the Politics of English-only. LFB Scholarly
from Practice) 41/3: 509–532. Publishing LLC.
Johnson, D. C. 2007. Language Policy within Thomas, W. P. and V. P. Collier. 1997. School
and without the School District of Philadelphia. Effectiveness for Language Minority Children.
Unpublished dissertation. University of National Clearinghouse for Bilingual
Pennsylvannia. Education.
D. C. JOHNSON 93
Thomas, W. P. and V. P. Collier. 2002. A Walford, G. (ed.) 2003. Investigating Educational

National Study of School Effectiveness for Policy through Ethnography. JAI.
Language Minority Students’ Long-term Academic Wiese, A. M. and E. E. Garcia. 2001. ‘The bilin-
Achievement. Center for Research on gual education act: language minority stu-
Education, Diversity, and Excellence. dents and US Federal Educational Policy,’
Tollefson, J. W. 1991. Planning Language, International Journal of Bilingual Education and
Planning Inequality: Language Policy in the Bilingualism 4/4, 229–48.
Community. Longman. Wiley, T. G. 2002. ‘Assessing language rights
Tollefson, J. W. (ed.) 2002. Language Policies in in education: a brief history of the U.S.
Education: Critical Issues. Lawrence Erlbaum. context’ in J. W. Tollefson (ed.): Language
Tollefson, J. W. 2006. ‘Critical theory in lan- Policies in Education: Critical Issues. Lawrence
guage policy’ in T. Ricento (ed.): An Erlbaum.
Introduction to Language Policy: Theory and Wiley, T. G. and W. E. Wright. 2004. ‘Against
Method. Malden: Blackwell Publishing.

the undertow: language-minority education
Valdez, E. O. 2001. ‘Winning the battle, losing policy and politics in the ‘‘age of accountabil-
the war: bilingual teachers and post- ity’’,’ Educational Policy 18/2: 142–68.
Proposition 227,’ The Urban Review 33: 237–53.
Seizure, Fit or Attack? The Use of

Diagnostic Labels by Patients with
Epileptic or Non-epileptic Seizures
1
LEENDERT PLUG, 2BASIL SHARRACK, and
2
MARKUS REUBER
1
Department of Linguistics and Phonetics, University of Leeds, UK and
2
Academic Neurology Unit, University of Sheffield, UK

We present an analysis of the use of diagnostic labels such as seizure, attack, fit,
and blackout by patients who experience seizures. While previous research on
patients’ preferences for diagnostic terminology has relied on questionnaires, we
assess patients’ own preferences and their responses to a doctor’s use of different
labels through the qualitative and quantitative analysis of doctor–patient inter-
actions in a realistic clinical setting. We also examine whether two sub-groups
of patients—those with epileptic seizures and those with (psychogenic) non-
epileptic seizures—show different behaviours in this respect. Our findings
suggest first that patients make fine lexical distinctions between the various
diagnostic labels they use to describe their seizure experiences; secondly, that
patients play an active role in the development and application of labels for
their medical complaint; and thirdly, that attention to patients’ lexical choices
and interactive use or avoidance of labels can be relevant for the differential
diagnosis of seizures.
INTRODUCTION
It is widely acknowledged that while lexical choice plays a crucial role in
human verbal interaction (Hakulinen and Selting 2005), establishing the
meaning of alternative lexical items is far from straightforward, especially if
contextual factors are taken into account (Fischer 1998; Xiao and McEnery
2006; Norén and Linell 2007). In this article, we explore the function of lexical
choice and the meanings of a number of related lexical items in the context
of a specific type of medical interaction. We focus on the use of diagnostic
labels—that is, lexical items that refer to an illness or the symptom of an
illness—in interactions between a doctor and patients with recurrent seizures.
The choice of diagnostic labels is significant for many reasons. For patients,
the labels which describe their illness are a core aspect of their ‘illness repre-
sentations’ (Leventhal et al. 1992; Horne 1999; Hagger and Orbell 2003).
A range of studies have confirmed that ‘the manner in which individuals
perceive their illnesses is likely to impact on many aspects of their experience,
including the likelihood of seeking help, the particular nature of the help
being sought, the degree of adherence to the treatment prescribed, and the
L. PLUG, B. SHARRACK, and M. REUBER 95
likelihood of response to such treatment’ (Manber et al. 2003: 335). For exam-
ple, patients who view their depression as caused by a ‘chemical imbalance’
are unlikely to engage in or respond to psychotherapy, even if this treatment
is the most suitable from the physician’s point of view (Manber et al. 2003:
336). Associations between patients’ illness representations and treatment
outcomes have been found in diverse clinical scenarios including Irritable
Bowel Syndrome and Chronic Fatigue Syndrome (Edwards et al. 2001;
Rutter and Rutter 2007), cardiac events (Petrie et al. 2002; Lau-Walker
2004), and epileptic and non-epileptic seizures (NES) (Kemp et al. 1999;
Goldstein et al. 2004; Green et al. 2004).
For the doctor, it is important to adopt diagnostic labels which are clear,

but which do not carry unwanted connotations. Misunderstandings are
particularly likely in the area of medically unexplained symptoms, or ‘psycho-
somatic disorders’, which include NES. NES superficially resemble epileptic
seizures, but are not caused by abnormal electrical activity in the brain. NES
are an involuntary response to distressing situations, physical, or emotional
stimuli (Reuber and Elger 2003). Most patients with NES have a previous
history of traumatic experiences or face difficult dilemmas in their current
lives. However, they typically fail to recognize the relevance of psychological
or emotional factors for their seizures (Reuber and Grünewald 2007). They
may therefore resist labels that associate their disorder more closely with a
psychological problem and prefer labels which imply a physical aetiology,
such as epilepsy. However, the clear differentiation of epilepsy and NES
is very important to doctors because the two seizure disorders are treated
differently. Whereas the first line of treatment for epileptic seizures involves
the use of antiepileptic drugs, the treatment of choice for NES is psychotherapy
(Reuber and Elger 2003; Reuber et al. 2005). Patients with NES who are
inappropriately labelled as having epilepsy, or who fail to understand that
they do not have epilepsy, are at risk of receiving unnecessary and potentially
harmful medical treatments, and are unlikely to improve.
Unfortunately, there is considerable uncertainty amongst doctors about
the most appropriate diagnostic label for NES. One paper identified 19 different
terms in the recent medical literature (Scull 1997). Whereas some terms give
patients the impression that the doctor does not understand the cause of the
attacks, others suggest that seizures are ‘put on’ or ‘all in the mind’: in fact, it
has been shown that terms such as ‘hysterical seizure’ and ‘pseudoseizure’ are
likely to offend patients (Stone et al. 2003). The implications for compliance
with proposed treatment are obvious. For the doctor–patient relationship
to work effectively, ‘diagnostic labels have to be not only helpful to doctors
but also acceptable to patients’ (Stone et al. 2002: 1449; see also Page and
Wessely 2003).
Whereas the medical literature on diagnostic labels tends to discuss the
advantages and drawbacks of particular terms from the doctor’s point of
view, this article will focus on patients’ use of labels for their condition or
their main symptom. Our study is based on one-to-one conversations between
96 SEIZURE, FIT OR ATTACK?
a doctor and 21 patients with epileptic or NES. The study is a part of the
project Listening to people with seizures at the University of Sheffield, UK,
which set out to improve the differential diagnosis of seizure disorders by
analysing the communicative behaviour of patients with epileptic and NES
(Schwabe et al. 2007). This project was inspired by a range of sociolinguistic
studies carried out at the Bethel Epilepsy Centre and the University of Bielefeld
in Germany (Schwabe et al. 2008).
Our article seeks to answer two questions. First, we wanted to determine
how patients use diagnostic labels for their seizures and which labels they
prefer. Stone et al. (2003) employed a questionnaire to address the second
half of this question. They asked neurology outpatients a range of variants

of the question ‘If you had blackouts, your tests were normal and the doctor
said you had pseudoseizures, would that be suggesting that you were putting
it on?’ In contrast, our study takes a corpus-based approach (Biber et al.
1994), relying solely on the analysis of actual doctor–patient interactions.
This methodology allows us to assess patients’ preferences directly: that is,
we can gain insight into patients’ actual preferred use of diagnostic labels,
rather than their reported preference in hypothetical scenarios. The sociolin-
guistic methods applied in Listening to people with seizures are strongly informed
by work in Conversation Analysis (Drew et al. 2001; Schegloff et al. 2002),
which has a long history of investigating issues of lexical choice and labelling
in realistic interactional settings: see for example the work on membership
categorization by Sacks (1972, 1992) and Schegloff (2007a, 2007b).
Secondly, we set out to explore whether patients with epilepsy and patients
with NES differ in their preferences or use of diagnostic labels. Although epi-
leptic and NES look similar to an external observer, the results of the German
studies as well as the preliminary findings of the Listening to people with seizures
project show that patients with these aetiologically distinct seizure disorders
describe their seizures very differently (Schwabe et al. 2007; Plug et al. in
press). Broadly speaking, patients with epilepsy readily focus on the seizure
experience and on the description of individual episodes, and volunteer a lot
of information about how they feel during their seizures. Patients with NES,
on the other hand, are more likely to talk about the impact of the seizures
on their lives and about the failure of previous treatment; they need to be
prompted to focus on how they feel during seizures, and generally provide less
detailed information about individual seizure episodes than patients with
epilepsy.
In view of the superficial similarities between the manifestations of epileptic
and NES it is perhaps not surprising that the differential diagnosis represents
a serious challenge for doctors, who have to base their diagnosis on seizure
descriptions in most cases. Tests carried out in between seizures are only
modestly useful in this setting. Seizure recordings are impossible if seizures
are infrequent. These difficulties may explain why most patients eventually
diagnosed with NES on the basis of the ‘gold standard’ investigation (the
synchronous recording of a typical seizure with video and EEG
(electroencephalography)) have carried an inaccurate diagnosis of epilepsy

for several years (Reuber et al. 2002). Any additional differential diagnostic
pointers, for instance from patients’ interactive use or selection of labels, could
therefore be very useful in clinical practice.
METHODOLOGY
This study is based on the analysis of 21 first clinical encounters between
a doctor (M. R., a consultant neurologist) and patients with seizures. The
clinical interviews were conducted between August 2005 and July 2007. All
patients had been admitted to the neurology ward at the Royal Hallamshire

Hospital in Sheffield, UK, for 48 hours of video-EEG monitoring because
their admitting neurologist was uncertain whether they had epileptic or
NES. All patients had seizures with impairment of consciousness while being
monitored. All diagnoses were confirmed by the video-EEG recording of
a seizure which was considered typical of the habitual attacks by the patient
and a seizure witness. Eight patients were found to have epilepsy, 13 NES.
Pseudonyms were used to protect participants’ identity. The study was
approved by the South Sheffield Ethics Review Committee and all patients
gave written informed consent for their consultations to be recorded and
analysed.
The consultations were audio- and video-recorded using pre-installed
monitoring equipment. Each consultation lasted between 20 and 35 minutes
and followed a semi-standardized interview procedure which encourages
the doctor to adopt an unusually passive but very attentive role, and which
allows patients to develop their own communication agenda (Schwabe et al.
2007; Plug and Reuber in press). One aspect of the interview procedure of
particular importance for this study is that the doctor opens the consultation
by asking the patient about his/her expectations of the current hospital visit—
with no mention of seizures or any other diagnostic label. At the beginning
of the interview, it is therefore left to the patient how he/she refers to the
seizure episodes. Moreover, the doctor does not present a diagnosis during
the consultation. Therefore, while the doctor’s lexical choices may influence
the patient’s, the doctor cannot be said to impose a terminological frame of
reference: this is largely left to the patient.
All 21 consultations were transcribed following standard conversation-
analytic conventions. All nouns referring to the patients’ seizures—such as
seizure, fit, attack, blackout—were identified and subjected to further qualitative
and quantitative analysis. In particular, we were interested in establishing
whether different labels were used synonymously or whether differences in
meaning could be observed. We were also interested in how the doctor’s use
of a particular label affects the patient’s lexical choice in the immediately
subsequent speaking turns. Quantitative differences in usage between the
doctor, patients with epilepsy and patients with NES were analysed using
the Mann–Whitney U-test for independent groups (ordinal variables) and
Pearson’s Chi-square test (categorical variables). Two-tailed p-values <0.05 are

reported as significant.
It is worth highlighting that a number of factors that may be expected to
influence patients’ communicative behaviour, including lexical choice, were
controlled for in this study. As indicated above, all consultations were led
by the same doctor, who did not have an established relationship with any
of the patients and followed the same interview procedure in each case.
Therefore, for the purpose of our analysis ‘the doctor’ can be treated as
a relatively homogeneous category. Personal details were taken for each
patient, so that any effects of gender, age, and duration of medical treatment
could be investigated. In addition, all patients were assessed for their

general linguistic competence using the Graded Naming Test (McKenna and
Warrington 1983; Warrington 1997) for vocabulary and the Test for Reception
of Grammar Version 2 (Bishop 2003) for grammar. On the other hand, we
did not classify our patients in terms of social, geographical, or educational
background, so that ‘the patient’ remains a relatively heterogeneous category.
DIAGNOSTIC LABELS AND OVERALL USAGE PATTERNS

Table 1 provides an overview of the diagnostic labels identified in the 21
consultations. The diagnostic label seizure was used most frequently, both by
patients and by the doctor (255 instances in total). The term seizure was also
used by the greatest number of patients (16 out of 21) and it was used by the
doctor in most interviews (16 of 21).
The most commonly used alternative labels were fit, attack, and blackout;
together, these occurred 240 times in our data. A similar number of patients
used the terms fit and attack (12 versus 11, difference not significant); and
considering that 40 of the 66 usages of attack were attributable to a single
Table 1: Overview of diagnostic labels used by patients and doctor

Diagnostic label Patients’ usage Doctor’s usage
N Npatients N Nconsultations
Seizure 132 16 123 16

Fit 41 12 6 3
Attack 66 11 99 17
Blackout 22 4 6 5
Absence, Grand mal, Petit mal, 10 3 2 1
Partial seizure, Tonic clonic (seizure),
Dizzy do, Funny do, (Funny) turn,
Reaction, Chin thing, (Chin) episode,
Blank spell, Collapse
patient, the number of instances in which patients used the terms fit and attack
were also rather similar.
In the cases of fit and attack, there were interesting differences in usage
between doctor and patients. Patients used fit much more commonly than
the doctor. The doctor on the other hand showed a marked preference for
the label attack. Whereas he used attack 99 times in 17 consultations, he only
used fit 6 times in 3 consultations. This means that there were several
consultations in which a patient used the label fit repeatedly, but the doctor
chose not to adopt the label. Conversely, there were a number of encounters
in which the doctor used attack repeatedly, but the patient persisted in their
use of alternative labels. The differences between patients’ and doctor’s usage

were statistically significant, both for fit (usage: 41 versus 6, p = 0.006; number
of patients/consultations: 12 versus 3, p = 0.004) and attack (usage: 66 versus
99, p = 0.02; number of patients/consultations: 11 versus 17, p = 0.05). Blackout
was used sparingly by doctor and patients alike, and additional diagnostic
labels, including epilepsy-specific terms such as grand mal or partial seizure,
as well as relatively colloquial labels such dizzy do or funny turn, were used
10 or fewer times by patients and/or doctor.
To explore to what extent the described patterns of usage reflect patients’
own preferences, rather than alignment with the doctor, we checked in how
many consultations each of the diagnostic labels was first used by the patient
(patient-initiated usage) and in how many the patient only used a label after
the doctor had introduced it (doctor-initiated usage). We only found a small
number of examples (six patients, two labels) in which patients altered their
terminology after the doctor had first used a particular term. One of the 16
patients who used the label seizure only did so after it was first mentioned by
the doctor, whereas 15 out of our 21 patients self-initiated the use of seizure
as a diagnostic label. All 12 patients who used fit self-initiated its usage.
However, five out of the 11 patients who used attack did so only after the
doctor had introduced the term.
This offers further support for the suggestion that seizure is the most popular
diagnostic label in this patient group; that fit is relatively popular among
patients although it is used sparingly by the doctor; and that attack is used
frequently by the doctor while only a small number of patients use the label
without prompting.
DIFFERENCES IN MEANING BETWEEN DIAGNOSTIC LABELS

The quantitative differences reported above were replicated when we focused
on those encounters in which the patient used at least two different labels. For
example, 10 patients used both seizure and fit. In these 10 encounters, fit was
used a mean of three times per consultation, while seizure was used eight times.
Taken together, these quantitative observations suggest that while at first sight
the terms listed in Table 1 would appear to be synonymous, they may in fact
have different meanings in the context of the interactions we are dealing
with—if not at the level of denotation, perhaps in terms of their connotations

or the collocational patterns in which they participate. To assess whether this
is the case, we investigated the contexts in which the labels were employed
in our consultations in greater detail.
This qualitative investigation revealed patterns of usage which suggest
that there are indeed differences in meaning between individual labels.
In what follows, we outline the crucial patterns and propose an account
along two lines. First, we suggest that there are differences in the degree
of specificity of reference between certain labels. Secondly, we suggest
that the patients’ lexical choice is governed at least in part by the
perceived appropriateness of certain labels in medical versus lay registers.

We discuss the former differences in meaning in terms of differences at
the level of denotation, and the latter in terms of differences at the level of
connotation.
Differences at the level of denotation

Many of the patients’ usages of fit and blackout suggest that these labels refer
to a specific type of paroxysmal event, or a phase within a wider event trajec-
tory, while the most popular diagnostic label, seizure, has a more general and
inclusive denotation. Relevant examples of the use of fit are given below.
(1) a. ‘and then I fell down with the fit’ (Carl).
b. ‘that’s about fifteen minutes before I go into the really deep fit, you know’
(Sue).
c. ‘others go into a more, er a bigger fit kind of thing’ (David).
d. ‘I kept having these like fits, just collapsing and having fits, like grand mal
symptoms, jerking and losing consciousness’ (Chris).
e. ‘as I say I thought epilepsy was someone thrashing about having a fit on the
floor’ (Samantha).
Carl and Sue both use seizure repeatedly to refer to events that involve a
period of altered consciousness, followed by a period of complete unconscious-
ness and loss of physical control, and finally a gradual return to normality.
Their uses of fit in (1a) and (1b) suggest that Carl reserves this label for
the period of complete unconsciousness, while Sue uses it to refer to the
middle phase of her more serious seizures. Similarly, David describes several
types of paroxysm—with repeated use of seizure—including episodes in which
he loses control of his body, but remains conscious throughout. He uses fit
to refer to seizures in which he loses consciousness completely, as seen in (1c).
Chris’ use of fit in (1c) suggests that in addition to loss of consciousness, ‘fits’
involve involuntary movement—or ‘jerking’. Chris distinguishes his ‘fits’ from
‘blackouts’, to which we turn below. The same is apparent from Samantha’s
description of someone ‘thrashing about having a fit’. In Samantha’s case,
she does not describe her own seizures as ‘fits’, and she does not mention
involuntary movement as a prominent symptom.
The clearest evidence that (like fit) blackout is a more specific diagnostic label
than seizure is provided by Chris; in particular in the fragment given in (2).
(2) Doctor: well can I (.) take you back to your first (1.0) seizure (1.6)
Chris: which one
Doctor: well you know you’ve come here because of these blackouts (0.3)
what about the first one you can remember (0.3) what can you tell
me about that (1.6)
Chris: the blackout or the (.) fit cos i’m having like (.) two different types at
the moment
Chris’ subsequent usage of blackout suggests that both ‘fits’ and ‘blackouts’

involve complete loss of consciousness and possibly a limp collapse, but only
‘fits’ involve involuntary movements. This observation is confirmed by several
other uses of blackout. Additional relevant examples are given in (3).
(3) a. ‘the blackouts started being different [. . .] you know, just dropping’ (Chris).
b. ‘it was more like a blackout than anything else that one [. . .] nothing hap-
pened to me while I was gone if you like, I was just gone’ (Alastair).
c. ‘because when I’m having the blackouts I could just be stood there’ (Alastair).
d. ‘to me they were just blackouts, but not to everybody else [. . .] I just thought,
one minute I can remember something, next minute I can’t’ (Sue).
In (3a), Chris indicates that what he calls ‘blackouts’ are seizures that
involve him ‘just dropping’—but not ‘jerking’. While Alastair also describes
seizures in which he loses control over his body, he suggests that in some
‘nothing happens’, and he remains stationary while his mind has ‘gone’ (3b
and c). Sue’s use of blackouts in (3d) highlights a possible mismatch between
the diagnostic label that the patient may assign to a seizure-related event
on the basis of his/her own experience, and that of onlookers. Sue indicates
that she started experiencing brief periods of unconsciousness and loss
of memory. She was not aware of her physical state during these periods;
therefore she classified as ‘blackouts’ what to those around her looked
like ‘fits’.
The usage of attack by the patients (as well as the doctor) suggests that it is
synonymous with seizure; that is, it is a diagnostic label with a more general
denotation than fit and blackout. Relevant examples are given in (4).
(4) a. ‘it’s been ver- very – they’ve been very disruptive to my life, the, the er, the
seizures, the attacks I’ve been having’ (Zack)
b. ‘I had another, erm, I had another, er, seizure, but there again, I d- I don’t
know what happened with these attacks’ (Jack)
c. ‘I was having a fit at the time [. . .] everything was going off-scale in that
attack’ (Betty)
d. ‘it must be difficult to remember individual seizures when it’s – they’re such
long way back, but maybe, maybe you can tell me about the last attack, what
– just explain exactly what, what, what happened in, in the last seizure that
you had’ (Doctor to Sandra)
e. ‘so you, you said you’re scared after seizures [. . .] and how do you feel in the
attack?’ (Doctor to Laura)
Both Zack and Jack use seizure and attack within one speaking turn, with no
indication that the two labels differ in meaning (4a and b). Betty’s use of both
fit and attack in (4c) is consistent with the idea that the former is a sub-type
of the latter: she describes one particular ‘attack’ as involving her ‘having a
fit’. The doctor uses attack and seizure interchangeably on several occasions,
as seen in (4d and e).
Most of the less frequent diagnostic labels, in particular the epilepsy-specific
terms such as absence, partial seizure, grand mal, and petit mal refer to particular
types of seizures, but we have too few occurrences of these labels in our data

to be able to report significant generalizations regarding their contexts of use.
Our observations with regard to seizure, fit, blackout, and attack can be summar-
ized as in (5): our patients recurrently use fit and blackout to refer to a specific
sub-type of seizure or to a specific phase in a wider seizure trajectory, while
reserving seizure and, sparingly, attack to refer more generically and inclusively
to seizure-related events.
(5) seizure, attack GENERAL DENOTATION
fit blackout SPECIFIC DENOTATION
Differences at the level of connotation

In addition, we observed usage patterns which can be understood in terms
of a register-related difference between seizure on the one hand and fit and
blackout on the other. In particular, while seizure appears to have a more
general denotation than fit and blackout, its usage by some of our patients
suggests that it also has the connotation of ‘medical term’, while the more
specific fit and blackout are at the same time more ‘colloquial’, representing
the patient’s personal experience rather than a medical diagnosis. Again, we
illustrate the crucial usage patterns below.
First, we note the occurrence of instances of self-repair from fit to seizure.
Examples are given in (6).
(6) a. ‘in one respect the fi- seizures were better’ (Barbara)
b. ‘I could have it, have a fit – seizure whatever you will call them’ (Peter)
In (6a) Barbara starts saying ‘fits’, but cuts off its production in favour of
‘seizures’. She uses seizure as a diagnostic label throughout the rest of her
consultation, and does not use fit or blackout. Peter does produce ‘fit’, but
repairs it to ‘seizure’ immediately (6b). His accompanying remark ‘whatever
you will call them’ is suggestive of a concern with what is the ‘correct’
terminology to use in the setting of the consultation. In fact, one way of

understanding examples such as those in (6) is in terms of an orientation by
the patient to seizure as a medical term which can be used freely in interactions
with a doctor, and to fit as a term that is potentially inappropriate. The fact
that the doctor uses fit only six times in the 21 consultations is indeed con-
sistent with it being a dispreferred diagnostic label from the doctor’s point
of view.
Perhaps more strikingly, over half of the instances of fit and blackout in
our data occur either in the context of reported speech by or directed at a
lay person, or in the context of explicit reporting of personal experience,
rather than general description or diagnosis. Examples of these two contexts

are given in (7) and (8) below.
(7) a. ‘and he said I were, like, having, like, a fit, in my bed, jerking and everything’
(Chris)
b. ‘right mate, you had a fit’ (Peter)
c. ‘he’s scared I’m going to have a fit in the middle of the street’ (Betty)
d. ‘I just say hold on, I’m gonna have a fit’ (Sandra)
e. ‘he says, you’ve got a – have you – do you suffer from seizures, I says yeah’
(Tallulah)
f. ‘they said that the erm, seizures I’m having aren’t as bad while I’ve been on
medication’ (Betty)
In (7a) Chris reports the observations of his brother using fit. The fragment
both confirms that for Chris, ‘fits’ involve involuntary movement, and suggests
that fit is a label that a lay person would use. This is confirmed by Peter and
Betty’s reports of a third person’s use of fit (7b and c); on both occasions the
third person is an acquaintance, not a doctor. Sandra’s formulation of her
habitual warning to friends before a seizure (7d) is consistent with an orienta-
tion to fit as an appropriate term to use in interaction with non-experts.
Interestingly, Sandra self-initiates the use of seizure and uses it consistently
throughout the rest of the interview, in which the context of reported
speech does not reoccur. In our data, seizure is found in the context of reported
speech only when the speech is that of a doctor, as is the cases in (7e and f) in
which Tallulah and Betty describe particular hospital visits.
We have already seen the example of Sue indicating that when she first
started having seizures, she thought she was having ‘blackouts’, while other
people used alternative labels (3d). It is in this context—that of the patient’s
reporting of first ‘realising the illness’ (Halkowski 2006)—that we find more
uses of fit and blackout than seizure. Here, patients are explicitly reporting their
own experience before they received a medical diagnosis. Additional examples
of instances of fit and blackout in this context are given in (8).
(8) a. ‘when I first started having fits’ (Betty)
b. ‘and that’s when the blackouts started’ (Chris)
c. ‘and I’d maybe have a cluster of seizures, but I didn’t know they were seizures
at the time’ (Samantha)
d. ‘and I went into a seizure, but I didn’t know’ (Tallulah)

In (8a) Betty uses the label fit when referring to her very first seizures, while
in the rest of the interview she repeatedly uses seizure. Chris similarly uses
blackout—as well as fit—when referring to his first experience. The fragments
in (8c and d) illustrate the recurrent retrospective application of the diagnostic
label seizure to early, pre-diagnosis episodes: that is, the patient reports having
had a seizure but not knowing that it was one at the time. This strongly
suggests that the label seizure is one whose use is licensed by a medical diag-
nosis; not one that is readily used outside a medical context or in narratives
of patients’ first symptom discovery. Fit and blackout, on the other hand, are
used in exactly these contexts.

The usage patterns illustrated in this section confirm that seizure on the
one hand and fit and blackout on the other have different contextual distribu-
tions. One possible explanation for this is that there are register-related
differences in meaning between these diagnostic labels. In the context of our
interviews, patients appear to orient to seizure as a medical diagnostic label,
whose use is appropriate in interaction with doctors and licensed by a medical
diagnosis. They orient to fit and blackout as labels whose use is more appro-
priate in lay registers and which can be used to describe personal, pre-diagnosis
experiences of seizure-related events. This account is summarized in (9).
(9) seizure MEDICAL REGISTER
fit blackout LAY REGISTER
RESISTANCE TO THE USE OF DIAGNOSTIC LABELS

Having described which diagnostic labels patients use most commonly and
explored a number of usage patterns which suggest meaning differences
between some of these, we now focus on those diagnostic labels which patients
do not commonly use, or which they adopt with evidence of a certain amount
of hesitation or resistance. First of all, several patients seemed to show a general
resistance to the use of diagnostic labels for their seizures. This was evident, in
particular in the opening phases of the consultations. As indicated above, the
doctor started each consultation with an open inquiry regarding the patient’s
expectations of the current hospital visit, without overt reference to seizures.
All patients referred to seizures in their response to this inquiry, but eight out
of 21 initially used a pronoun rather than any of the diagnostic labels we have
discussed so far. A clear example is given in (10).
(10) Doctor: what were your expectations when you (0.3) when you came here
(0.3)
Alastair: erm (0.3) I was just hoping that we could find (1.0) what it was (0.3)
that was causing them?
Doctor: mmm
Alastair: hopefully lead to something that would stop me having them so I
could continue having a normal life
Doctor: mmm
Alastair: you know cos er (.) it’s been very disrupted since it started
Doctor: it’s been very disrupted
Alastair: yeah (.) well cos I’ve been off (.) of work on (.) at work off work
(several lines omitted; no diagnostic label)
they come and go so I can’t really (2.2) risk being in a car
Doctor: mmm

Alastair: which is a shame so as soon as er (2.2) something’s found out (.) I
can go back to being er (0.3) normal
Doctor: mmm
(3.6)
Alastair: yeah (0.3) so that’s where they’re up to (1.4) well that’s my
expectation
Doctor: mmm
Here, Alastair formulates an elaborate response to the doctor’s opening
inquiry in which he refers to his seizures several times using they/them (in
bold), and to his disorder using it (in bold). It is only after a further inquiry
by the doctor that he reverts to a diagnostic label—in this case blackout.
Evidence for general resistance against the use of diagnostic labels can
also be found at other points during the consultation. Some patients attempt
to avoid using labels in the description of particular episodes, even if
several labels have already been mentioned earlier on in the consultation.
Others accompany their use of labels with comments that downplay their
commitment to them. We have already seen an example of the latter
in (6b), in which Peter accompanies his self-repair from fit to seizure
with the comment ‘whatever you will call them’. Additional examples are
given in (11).
(11) a. ‘and then one of these (0.6) things started’ (Trudie)
b. ‘I just want the fits to s- whatever they are to stop completely’ (Betty)
c. ‘during the seizure or whatever it is I’ve had’ (Betty)
In (11a), Trudie hesitates at the point when the use of a diagnostic label
is relevant. After a pause, she opts for ‘things’—avoiding the use of a specific
label. In (11b and c), Betty uses both fit and seizure with the indication that
she does not know what types of seizures she’s having—and, by extension,
that she does not know whether her use of diagnostic labels is ‘correct’.
The examples in (10) and (11) suggest that patients may feel they lack the
authority to provide a diagnostic label for their experiences. One strategy is,
then, to avoid using any label, at least until the doctor suggests one; another
is to make it clear to the doctor that the use of a given diagnostic label is
open to correction.
We have already mentioned patients’ resistance to the use of the label attack,
which the doctor used frequently as a synonym of seizure: only 11 out of our
21 patients used it. While one of these eleven patients used the label more
than 40 times during the consultation, most others used it sparingly. As seen
above, six patients chose not to adopt the label despite the doctor’s repeated
use. Illustrative fragments are given in (12) and (13). In both, the doctor’s
inquiry contains attack, but in their responses the patients choose to use an
alternative label.
(12) Doctor: so you (.) came in here to learn about the small
Ken: yeah
Doctor: attacks

(0.2)
Ken: I’ve had one while I’ve been here
(1.7)
Doctor: [mmm
Ken: [but they’re just (0.9) they only last for a couple of seconds (.)
there’s a slight sort of partial seizure (0.3) it’s er (1.3) it sort of
doesn’t develop into a full seizure
(13) Doctor: so that’s in the attacks you had that feeling

Samantha: yeah (.) yeah
Doctor: mmm
Samantha: erm (.) but up until this point I’d never ever had a full blown
seizure where I sort of lost consciousness (.) I could carry on
going about my business whilstever this were going on in my
head
It is perhaps surprising that the most popular label overall, seizure, was also
the one that was most frequently and overtly resisted. First of all, recall that
16 out of 21 patients used this label at some point during the consultation.
This means that five patients did not, although the doctor used the label
repeatedly in two of these cases. We have already seen Chris’ use of fit and
blackout in direct response to the doctor’s inquiry about his first ‘seizure’ in (2)
above. In fact, Chris avoids the use of seizure throughout the consultation,
while he uses fit and blackout repeatedly. Similarly, Steve avoids seizure despite
the doctor’s repeated usage, in this case in favour of several labels including
funny turn and funny do.
Several other patients use seizure only in the context of reporting a previous
medical diagnosis, rather than their own experience of seizures—using
alternative labels in all other contexts. For example, we have seen two such
uses of seizure by Tallulah in (7e) and (8d) above. Elsewhere in the consulta-
tion, she overtly resists the adoption of seizure, despite its use by the doctor,
as seen in (14).
(14) Doctor: is this related to (.) to the seizures er er not waking up from a
seizure or just not (.) waking up
Tallulah: not waking up from (0.3) a sei- er having a fit
Here, Tallulah’s response to the doctor’s inquiry takes the form of a repetition
of part of the inquiry: ‘not waking up from a seizure’. However, Tallulah
hesitates before ‘a seizure’ and subsequently cuts off its production in favour
of ‘having a fit’. Her self-repair, which constitutes a marked deviation from
the part-turn repetition format and a marked refusal to align with the doctor
in terms of lexical choice, makes it very clear that the adoption of the diag-
nostic label seizure is problematic for her.
In addition, the hesitations and hedging remarks illustrated in (11) above
most frequently involve the use of the label seizure. Some more examples are
given in (15).

(15) a. ‘I‘ve never had any problem with er (0.8) seizures or anything’ (Trudie)
b. ‘because there’s – I seem to have erm two different sorts of (0.9) seizures
happening’ (Pat)
c. ‘just really to find out what the problem is or what’s causing (.).hhh erm
(0.3) the seizures’ (Sandra)
If patients’ general resistance to using diagnostic labels is due to their perceived
lack of authority on the issue of what their seizures should be called, then their
specific resistance to the use of seizure may be explained in terms of the
‘medical’ connotations of this particular label. We have suggested above that
while fit and blackout are lay terms which a patient can use freely without
implications of self-diagnosis, seizure is a more formal medical diagnostic
label. As such, its use may imply a degree of command and understanding
of seizure-related medical terminology. While about half of our patients use
seizure without evidence of hesitation or resistance—and in fact a few appear
to adopt the term precisely because they feel the context of the medical
consultation occasions it, as seen in (6) above—some display an orientation
to seizure as medical vocabulary to which they themselves do not have
straightforward right of access. The latter patients leave it to the doctor
to decide on an ‘official’ diagnostic label that best covers what they describe
as their experience.
RELEVANCE FOR THE DIFFERENTIAL DIAGNOSIS OF

PATIENTS WITH SEIZURES
At this point, we turn to our second research question: do patients with
epilepsy and patients with NES differ in their usage preferences or responses
to certain diagnostic labels used by the doctor? So far we have discussed our
patient group as a whole; however, when we compare the two clinical sub-
groups striking differences emerge.
The clearest difference relates to the usage of seizure as a diagnostic label—
or the resistance to using seizure described in the previous section. When we
compare the frequency of usage between the two sub-groups, we see that
76 out of the 132 instances of seizure were produced by patients with epilepsy
and 56 by patients with NES. This means that the mean number of instances
per patient was 9.5 for the epilepsy sub-group and 4.3 for the NES sub-group.
First, we can therefore state that patients with epilepsy used seizure more
frequently than patients with NES. When we do the same comparison for fit
and blackout we find the reverse pattern. Of the 41 instances of fit, 31 were
produced by patients with NES (2.4 instances per patient) and 10 by patients
with epilepsy (1.3 instances per patient). All 22 instances of blackout were
produced by patients with NES. That is, patients with epilepsy preferred seizure
over fit and blackout, and patients with NES used fit and blackout but appeared
to disprefer the use of seizure. The pattern is summarized in Figure 1. A Mann–
Whitney U-test reveals that the difference for seizure is statistically significant

(mean 9.5 versus 4.3, p = 0.034), although the differences for fit and blackout
are not.
Table 2 provides a more detailed view of the usage of and resistance to
the label seizure among patients in the two sub-groups. Of the eight patients
with epilepsy, all used seizure in the course of the consultation, and all self-
initiated its usage. Of the 13 patients with NES, five did not use seizure at all,
and six did not self-initiate its usage. Chi-square tests reveal that in both cases
the difference between the two sub-groups is significant (no use of seizure:
2 = 4.04, p = 0.04; no self-initiation: 2 = 5.17, p = 0.02). Of the five patients
who use seizure with evidence of resistance to the label, four have NES and
one has epilepsy (difference not statistically significant). Together, these figures
mean that only three out of 13 patients with NES use the label seizure without
priming and without reservation, as opposed to all but one of eight patients
with epilepsy. A Chi-square test confirms that this overall difference between
the two sub-groups is significant (2 = 8.24, p = 0.004).
Of course, given the relatively small numbers on which our comparisons
are based, we cannot at this point conclude that the observed differences
between the two sub-groups of patients are robust and generalizable
beyond our patient group. Nevertheless, the statistical significance of several
of our comparisons does suggest that the differential patterns we observe are
unlikely to be due to chance alone. It may be noted that the factors sex,
age, and duration of medical treatment showed no significant effects in the
same comparisons. Moreover, there is no observable difference between
the two patient groups in terms of general linguistic competence: all patients
scored average or above in both the Graded Naming Test and the Test for
Reception of Grammar Version 2.
While our second research question focuses on the patients’ rather than
the doctor’s lexical choices and communicative behaviour, it is worth noting
that in our data, the doctor used the label attack more often in consultations
with patients eventually found to have NES than in those diagnosed as
having epilepsy (6.5 times per conversation versus 2.0 times, p = 0.019).
Given the findings we have just reported, we may now have an explanation
for this: the doctor may have resorted to the use of attack when he observed
a degree of resistance to the label seizure on the patient’s part. This would
10
Epilepsy
9
NES
8
7
Instances per patient

2
0
Seizure Fit Blackout
Diagnostic label
Figure 1: Bar chart showing mean instances per patient for seizure, fit, and
blackout in epilepsy and NES sub-groups
Table 2: Resistance to the diagnostic label seizure, comparing the epilepsy

and NES sub-groups
Disorder N No use of No self-initiation Use of seizure Any evidence
seizure (%) of use of seizure (%) with evidence of of resistance
resistance (%) to seizure (%)
Epilepsy 8 0 (0) 0 (0) 1 (13) 1 (13)

NES 13 5 (38) 6 (46) 4 (31) 10 (77)
mean that patients’ terminological preferences directly influence the doctor’s

choice of labels.
Finally, while patients with epilepsy were less likely to show resistance
to the label seizure and to use fit and blackout instead, they were also more
likely to adopt epilepsy-specific terminology when describing their seizure
experience. The diagnostic labels absence, partial seizure, tonic clonic (seizure),
grand mal, and petit mal occur almost exclusively in consultations with patients
with epilepsy. This may seem a trivial observation, but it is not: each of our
patients’ diagnosis was unclear at the time of the consultation we recorded,
and all patients who turned out to have NES had previously been diagnosed
with epilepsy—and had therefore been exposed to epilepsy-related diagnostic

terminology. Nevertheless, the patient groups behaved differently in terms
of their willingness to adopt this terminology. Patients with epilepsy readily
adopted diagnostic labels which patients with NES may have seen as medical
vocabulary items which were somehow less appropriate for them to use than
the lay terms fit and blackout.
DISCUSSION AND CONCLUSION

We set out to address two questions: first, which diagnostic labels do patients
prefer to refer to their seizures; and secondly, do patients with epilepsy and

patients with NES differ in their usage preferences or responses to the choice
of certain diagnostic labels by the doctor? With respect to the first question,
our study shows that seizure is a particularly popular diagnostic label, while
attack is dispreferred and resisted by many patients. Further studies are neces-
sary to establish whether this resistance is a general feature of patients with
seizures, or whether it is particular to our sample of patients. Unfortunately,
the label was not included in the set evaluated by Stone et al. (2003), and we
have little insight into patients’ interpretations of this label from other sources.
In addition, it remains to be established how common the relative preference
of the label attack is amongst doctors. The label occurs in the context of the
medical term Non-Epileptic Attack Disorder (NEAD), which is certainly widely
used in the professional literature. If our findings are replicated by studies
in larger patient populations, doctors may need to reconsider whether it is
beneficial to use a term in this setting which is not readily acceptable to
patients.
We also found that fit and blackout are relatively popular, but that they
are not synonymous with seizure: patients recurrently use fit and blackout
to refer to particular sub-types of seizure or phases within a wider seizure
trajectory, and use the terms in the context of lay description of their experi-
ence. In contrast, seizure is used in contexts that favour established medical
vocabulary. We have suggested that it is particularly the register difference
between seizure on the one hand and fit and blackout on the other that explains
why some patients show resistance to the use of seizure, while such resistance
is much less common for fit and blackout. Especially when talking about
their own experience, patients may feel they do not have access to medical
vocabulary—or at least that their access must be licensed by the doctor in the
course of the consultation. They may also feel that by using medical vocabu-
lary, they would be claiming a degree of authority over their disorder with
which they do not feel comfortable. Again, further studies are needed to
confirm whether the differential use and treatment of seizure on the one
hand and fit and blackout on the other is generalizable beyond our sample of
patients.
As it stands, our proposed meanings of these labels account for a number
of observed usage patterns in our consultations. Further studies on a wider
patient population may reveal patterns for which our proposals cannot
account, and will provide a stronger empirical basis for the statement of the
lexical meanings of the various diagnostic labels used by doctors and patients.
Moreover, our analysis somewhat undervalues the fact that parameters of
lexical choice may shift during the course of an interaction (Hakulinen and
Selting 2005; Norén and Linell 2007). While we have discussed in some detail
several contexts in which individual labels are recurrently used and have
commented on the extent to which the doctor’s usage of terminology influ-
ences the patient’s and vice versa, our observations only scratch the surface
of the sequential-interactional dimension of lexical choice. We must leave this
dimension as an area for further research.

With respect to the second question, we noted several differences in the way
patients with epilepsy and patients with NES use diagnostic labels. Patients
with NES showed a tendency to avoid the use of labels such as seizure or
attack, and to prefer labels such as fit and blackout. Patients with epilepsy did
not show significant resistance to seizure, and several used epilepsy-specific
medical terminology themselves. We have indicated that the factors sex,
age, duration of treatment, and general linguistic competence do not help
to explain this difference between the two patient groups. Other factors may
be significant, however: as pointed out above, we did not control for the
social, geographical, and educational background of the patients, let alone
their sociolinguistic competence, sensitivity to register or state of mind at
the time of the consultation. Again, more research is needed to assess the
robustness of our current findings.
Notwithstanding these reservations, it is interesting to consider the possibil-
ity that the observed difference between the two patient groups is related
to the distinct ways in which patients with epilepsy and NES communicate
about their seizures. As suggested above, sociolinguistic research on a large
sample of German patients has shown that while patients with epilepsy readily
volunteer information about subjective seizure symptoms and particular
seizure episodes, the communication behaviour of patients with NES in
relation to their seizures is characterized by ‘focusing resistance’. They tend
to offer brief and superficial seizure accounts and put more emphasis on the
situations in which seizures occur or on the negative impact seizures have
had on their lives. When asked about specific seizure episodes, such as their
first, last or worst seizure, they often say that they cannot remember, or pro-
duce general accounts of their seizures rather than descriptions of particular
events (Schwabe et al. 2008).
At first sight, it is difficult to make sense of the relationship between
the usage patterns reported in this article and the wider communicative differ-
ences between patients with epilepsy and NES. If patients with epilepsy spend
more of their consultations talking about their personal seizure experiences,
one would expect them to use fit and blackout more frequently than seizure.
Similarly, if patients use the less generic and inclusive fit and blackout as part
of an attempt to downgrade the seriousness of their disorder in communicating
with the doctor, one would not expect patients with NES to use these
terms more frequently than patients with epilepsy: it is patients with NES
who tend to stress the negative impact of the seizures on their lives. Patients
with epilepsy are more likely to communicate that they are actively engaged
in minimizing the impact of the seizures on their daily lives, and that they
are coping as well as they can. We suggest that it may be this higher level of
engagement with the process of diagnosis and treatment which makes patients
with epilepsy more likely to adopt vocabulary which we have characterized as
‘medical’ in consultations than patients with NES. The latter patients’ general
resistance to engaging in a detailed discussion of their seizure experience
would seem compatible with a resistance to the use of this type of vocabulary,

and a tendency instead to resort to more colloquial formulations. Provided
that our findings are generalizable to a larger patient population, this would
be an interesting hypothesis for further research.
While the clinical relevance of our findings remains to be established, espe-
cially given their basis in careful post hoc analysis rather on-line observation
during consultations, they confirm that while healthcare professionals play
a significant role in shaping patients’ views, patients take an active part in
the negotiation of diagnoses and the use of diagnostic labels (Gerhardt
1989). Patients are capable of translating differences in seizure manifestations
into lexical distinctions, and may resist the use of certain labels or actively
encourage the doctor to apply particular labels to their experiences. Despite
suggestions in early labelling theory (Hagan 1973; Mercer 1973), labelling
is therefore not a one-way process from doctor to patient, but a complex,
interactional negotiation with doctors and patients as participants with
equal stakes (Maynard 1992; Gill and Maynard 1995; Peräkylä 1998).
REFERENCES
Biber, D., S. Conrad, and R. Reppen. 1994. Gerhardt, U. 1989. Ideas About Illness. An
‘Corpus-based approaches to issues in Intellectual and Political History of Medical
applied linguistics,’ Applied Linguistics 15: Sociology. Macmillan.
169–89. Gill, V. T. and D. W. Maynard. 1995. ‘On
Bishop, D. V. M. 2003. The Test for Reception ‘‘labelling’’ in actual interaction: delivering
of Grammar, version 2 (TROG-2). Psychological and receiving diagnoses of developmental dis-
Corporation. abilities,’ Social Problems 42: 11–37.
Drew, P., J. Chatwin, and S. Collins. 2001. Goldstein, L. H., A. Deale, S. Mitchell-
‘Conversation analysis: A method for research O’Malley, B. K. Toone, and J. D. C. Mellers.
into interactions between patients and health- 2004. ‘An evaluation of cognitive behavioral
care professionals,’ Health Expectations 4: 58–70. therapy as a treatment for dissociative seizures,’
Edwards, R., R. Suresh, S. Lynch, Cognitive Behavioral Neurology 17: 41–49.
P. Clarkson, and P. Stanley. 2001. ‘Illness Green, A., S. Payne, and R. Barnitt. 2004.
perceptions and mood in chronic fatigue ‘Illness representations among people with
syndrome,’ Journal of Psychosomatic Research nonepileptic seizures attending a neuro-
50: 65–68. psychiatry clinic: A qualitative study based
Fischer, K. 1998. From Cognitive Semantics on the selfregulation model,’ Seizure 13:
to Lexical Pragmatics. Mouton de Gruyter. 331–39.
Hagan, J. 1973. ‘Labeling and deviance: A case in the doctor–patient encounter,’ Journal of
study in the sociology of the interesting,’ Social the Royal Society of Medicine 96: 223–27.
Problems 20: 447–58. Peräkylä, A. 1998. ‘Authority and accountabil-
Hagger, M. S. and S. Orbell. 2003. ‘A meta- ity: The delivery of diagnosis in primary health
analytic review of the common-sense model care,’ Social Psychology Quarterly 61: 301–20.
of illness representations,’ Psychological Health Petrie, K. J., L. D. Cameron, C. J. Ellis,
18: 141–84. D. Buick, and J. Weinman. 2002. ‘Changing
Hakulinen, A. and M. Selting. (eds) 2005. illness perception after myocardial infarction:
Syntax and Lexis in Conversation. John An early intervention randomized controlled
Benjamins. trial,’ Psychosomatic Medicine 64: 580–86.
Halkowski, T. 2006. ‘Realising the illness: Plug, L. and M. Reuber. 2009. ‘Making the
Patients’ narratives of symptom discovery’ diagnosis in patients with blackouts: It’s all in
in J. Heritage and D. Maynard (eds): the history,’ Practical Neurology 9: 4–15.

Communication in Medical Care. Cambridge Plug, L., B. Sharrack, and M. Reuber. 2009.
University Press, pp. 86–114. ‘Conversation analysis can help to distinguish
Horne, R. 1999. ‘Patients’ beliefs about treat- between epilepsy and non-epileptic seizure dis-
ment: The hidden determinant of treatment orders: A case comparison,’ Seizure 18: 43–50.
outcome?’ Journal of Psychosomatic Research 47: Reuber, M. and C. E. Elger. 2003. ‘Psychogenic
491–95. nonepileptic seizures: Review and update,’
Kemp, S., S. Morley, and E. Anderson. 1999. Epilepsy and Behavior 4: 205–16.
‘Coping with epilepsy: Do illness representa- Reuber, M., G. Fernández, J. Bauer,
tions play a role?’ British Journal of Clinical C. Helmstaedter, and E. Elger. 2002.
Psychology 38: 43–58. ‘Diagnostic delay in the diagnosis of
Lau-Walker, M. 2004. ‘Relationship between ill- psychogenic non-epileptic seizures,’ Neurology
ness representation and self-efficacy,’ Journal of 53: 493–95.
Advanced Nursing 48: 216–25. Reuber, M. and R. Grünewald. 2007. ‘The first
Leventhal, H., M. Diefenbach, and seizure: Is it epilepsy?’ in C. P. Panayiotopoulos
E. A. Leventhal. 1992. ‘Illness cognitions: (ed.): Newly Identified Epileptic Seizures:
Using common sense to understand treatment Diagnosis, Procedures and Management.
adherence and affect cognition interactions,’ Medicinae, pp. 145–58.
Cognitive Therapy Research 16: 143–63. Reuber, M., S. Howlett, and S. Kemp. 2005.
Manber, R., A. S. Chambers, S. K. Hitt, ‘Psychologic treatment for patients with psy-
C. McGahuey, P. Delgado, and chogenic nonepileptic seizures,’ Expert Opinion
J. J. B. Allen. 2003. ‘Patients’ perception of in Neurotherapeutics 5: 737–52.
their depressive illness,’ Journal of Psychiatric Rutter, C. L. and D. R. Rutter. 2007.
Research 37: 335–43. ‘Longitudinal analysis of the illness representa-
Maynard, D. W. 1992. ‘On clinicians co-impli- tion model in patients with Irritable Bowel
cating recipients’ perspective in the delivery of Syndrome,’ Journal of Health Psychology 12:
diagnostic news’ in P. Drew and J. Heritage 141–48.
(eds): Talk at Work. Cambridge University Sacks, H. 1972. ‘On the analyzability of stories
Press, pp. 303–30. by children’ in J. J. Gumperz and D. Hymes
McKenna, P. and E. K. Warrington. 1983. (eds): Directions in Sociolinguistics: The
The Graded Naming Test. NFER-Nelson. Ethnography of Communication. Holt, Rinehart,
Mercer, J. R. 1973. Labeling the Mentally and Winston, pp. 325–45.
Retarded: Clinical and Social System Perspective on Sacks, H. 1992. Lecures on Conversation. vol. 2.
Mental Retardation. University of California Blackwell.
Press. Schegloff, E. A. 2007a. ‘Categories in action:
Norén, K. and P. Linell. 2007. ‘Meaning poten- Person-reference and membership categoriza-
tials and the interaction between lexis and tion,’ Discourse Studies 9: 433–61.
contexts: An empirical substantiation,’ Schegloff, E. A. 2007b. ‘A tutorial on membership
Pragmatics 17: 387–416. categoization,’ Journal of Pragmatics 39: 462–82.
Page, L. A. and S. Wessely. 2003. ‘Medically Schegloff, E. A., I. Koshik, S. Jacoby, and
unexplained symptoms: exacerbating factors D. Olsher. 2002. ‘Conversation analysis and
applied linguistics,’ Annual Review of Applied 2003. ‘What should we call pseudoseizures?
Linguistics 22: 3–31. The patient’s perspective,’ Seizure 12: 568–72.
Schwabe, M., S. J. Howell, and M. Reuber. Stone, J., W. Wojcik, D. Durrance, A. Carson,
2007. ‘Differential diagnosis of seizure disor- S. Lewis, L. MacKenzie, C. P. Warlow, and
ders: A conversation analytic approach,’ Social M. Sharpe. 2002. ‘What should we say
Science and Medicine 65: 712–24. to patients with symptoms unexplained by
Schwabe, M., M. Reuber, M. Schöndienst, disease? The ‘‘number needed to offend’’,’
and E. Gülich. 2008. ‘Listening to people British Medical Journal 325: 1449–50.
with seizures: How can Conversation analysis Warrington, E. K. 1997. ‘The graded naming
help in the differential diagnosis of seizure dis- test: A restandardisation,’ Neuropsychological
orders?’ Communication and Medicine 5: 59–72. Rehabilitation 7: 143–46.
Scull, D. A. 1997. ‘Pseudoseizures or non- Xiao, R. and T. McEnery. 2006. ‘Collocation,
epileptic seizures (NES): 15 synonyms,’ semantic prosody, and near synonymy: A
Journal of Neurology, Neurosurgery and

cross-linguistic perspective,’ Applied Linguistics
Psychiatry 62: 200. 27: 103–29.
Stone, J., K. Campbell, N. Sharma,
A. Carson, C. P. Warlow, and M. Sharpe.
Applied Linguistics: 31/1: 115–135 ß Oxford University Press 2009
English in Advertising: Generic

Intertextuality in a Globalizing
Media Environment
AN H. KUPPENS
University of Antwerp, Belgium

Across the globe, the use of English is a popular advertising technique. The ever
expanding body of studies on this topic has revealed a number of explanations
for the use of English in the advertising. It can be related to the larger marketing
strategy of a campaign, to the cultural connotations English carries, or English
can be used for creative-linguistic reasons. The current article, however, will
present an analysis of four examples of advertisements in which English is
used for reasons that have not been discussed in the scholarly literature so
far. More specifically, in these advertisements, which intertextually refer to
a range of British and American media genres, specific registers of English
are used to mark the generic intertextuality of the ads. The analysis, I believe,
sheds new light on the use of English in the media, and more particularly on
issues such as viewers’ agency and linguistic superiority.
INTRODUCTION: ENGLISH IN ADVERTISING

In the globalization of English, mass media plays an important role (Phillipson
and Skutnabb-Kangas 1997; Hjarvard 2004). The worldwide spread of English
not only occurs through the worldwide export of North American, British, and
Australian English-language media products, but also in multilingual ‘media
idioms’ (Jacquemet 2005) such as chat language, news broadcasting language,
and the language of cell phone messages. However, empirical studies of the use
of English in such media idioms are still quite sparse. One notable exception
is the research of English in advertising, which, after the publication of a few
early studies (e.g. Haarmann 1984, 1989; Masavisut et al. 1986; Bhatia 1987;
Takashi 1990; Mueller 1992; Cheshire and Moser 1994; Myers 1994) has seen
a striking expansion of research during the last decade (e.g. Griffin 1997, 2001;
Martin 1998, 2002a, 2002b, 2007; Meinhof 1998; Gerritsen et al. 1999,
2000, 2007; Kelly-Holmes 2000, 2005; Piller 2000, 2001, 2003; Friedrich
2002; Ovesdotter Alm 2003; Bhatia and Ritchie 2004; Ustinova and
Bhatia 2005; Baumgardner 2006; Lee 2006; Ustinova 2006; Krishnasamy
2007). Whether they target Asian, European, or Latin-American consumers,
advertisers seem to regard the use of English words, sentences, and even entire
texts as an efficient strategy to sell brands and products to consumers.
These different studies have revealed a number of explanations for the use
of English in advertising, which can be categorized in three groups. A first set
116 ENGLISH IN ADVERTISING
of reasons pertains to the larger marketing strategy of a campaign. For instance,

brands sometimes choose to use the same campaign or slogan worldwide in
order to have a globally consistent marketing strategy and brand image. Some
brands also choose to use the same advertisement in different countries in
order to cut costs. In either case, English serves as the ‘lingua franca’ that
is understood—or that at least sounds familiar—in different countries
(e.g. Ovesdotter Alm 2003; Baumgardner 2006; Ustinova 2006). Within a
single country as well, English can be used to address different language
groups. In complex multilingual countries such as Switzerland, for instance,
English is sometimes used because it is regarded as a comparatively neutral
language, as it is not embroiled in the tensions between local languages

(Cheshire and Moser 1994).
Secondly, there are also creative-linguistic reasons for the use of English in
advertising. English words are, for instance, often used in the case of a ‘lexical
gap’, i.e. when there is no accurate equivalent for a word or expression in
the host language (e.g. Masavisut et al. 1986; Gerritsen et al. 2000), or when
the equivalent in the host language is considered taboo (Takashi 1990).
English words are also popular with advertisers because they are shorter
than the equivalents in the host language (Friedrich 2002), and because
they (are believed to) attract the attention of the consumer (e.g. Bhatia
1987; Martin 2002a, 2007; Ustinova and Bhatia 2005; Krishnasamy 2007).
The use of a foreign language also expands advertisers’ possibilities for linguis-
tic creativity. For instance, it opens the door to bilingual word puns (e.g. ‘Koe
and the Gang’ in a Flemish ad) and bilingual rhyming (e.g. ‘Trentenaire
On Air’ in a French ad; Bhatia and Ritchie 2004: 539) in advertisements
(e.g. Myers 1994; Martin 2002b, 2007; Ustinova and Bhatia 2005).
A third set of reasons for the use of English in advertising pertains to
the cultural connotations that English carries. Here, English is used because
the values or stereotypes associated with it (e.g. internationalism, modernity,
and Britishness) are assumed to reflect positively on the product (e.g. Cheshire
and Moser 1994; Martin 2002a; Ustinova 2006) or to appeal to the ‘implied
reader’ of the advertisement (e.g. Piller 2001; Lee 2006). In such advertise-
ments, foreign languages are not used for their communicative value, but for
their symbolic value (Kelly-Holmes 2000). Indeed, several studies report that
for advertising purposes, consumers need not even understand the foreign
language that is used, as long as they recognize the connotations that it is
associated with. Very telling in this respect is that Martin (2002a), Piller
(2000), and Masavisut et al. (1986) report advertisers’ practices to even use
‘invented’ or ‘nonsensical’ English in advertisements, i.e. meaningless words
or sentences that only sound English—and can thus activate certain values
with the consumers.
When it comes to the connotations that are associated with different for-
eign languages, English seems to constitute a rather exceptional case. While
other foreign languages derive their connotational values primarily from
the countries in which they are the dominant language (e.g. German technical
A. H. KUPPENS 117
sophistication; French culinary superiority), English also has ‘meaning, use

and significance independent of the countries in which it is spoken’ (Kelly-
Holmes 2000: 76). In this respect, English is a ‘bicultural’ language (Cheshire
and Moser 1994), as it symbolizes values that are stereotypically associated
with the USA (e.g. freedom) or Great Britain (e.g. class and traditionalism),
as well as ‘general’ values such as youth, prestige, modernity, globalization,
cosmopolitanism, and internationalism. The associations in the latter case
seem to indicate that English is a ‘neutral’ and ‘transparent’ language, ‘tied
to no particular social, political, economic or religious system, [belonging] to
everyone or to no one’ (Wardhaugh 1987: 14–15). However, the line between
‘American’ or ‘British’ values on the one hand, and ‘general’ values on the

other, is probably far more blurry than it seems. For instance, the association
of English with the ‘general’ notion modernity, might be strongly related to
the stereotype of the USA as the modern country par excellence. Furthermore,
the association of English with values such as quality, progress, reliability,
credibility, and prestige, reveals that implicit notions of American (or, more
generally, Western) culture as ‘superior’ underlie the use of English in adver-
tising. Advertisers often even use English to market locally produced products
to local consumers, in the hope that they will think they are produced
abroad (Masavisut et al. 1986; Ovesdotter Alm 2003). ‘People still think
that what comes from abroad is better, and if it’s in English – it’s even
better’, an Ecuadorian advertiser states in Ovesdotter Alm (2003: 152).
In the same study, another advertiser says: ‘[U]sually we relate English with
North American and European and, therefore, with quality, with well-made,
with technology, with research, with design, with something that is good.
American is good. American equals good quality’ (p. 151). In such cases,
the use of English amounts to an uncritical celebration of the language and
its associated culture(s).
The high incidence of English in advertising is often seen as an example
of ‘linguistic imperialism’ (Phillipson 1992): The US dominant position in
the production of commodity culture enables it to force English onto other,
seemingly defenseless cultures. English thus becomes ‘the language of global
mass consumer culture’ (Featherstone 1995: 8). The use of English in adver-
tising, and in media more generally, is often a prime target for opponents
of American cultural imperialism. For instance, the French ‘Loi Toubon’,
which restricts the use of foreign languages in the media, was defended
by its proponents as being an instrument against ‘Americanization’ (Machill
1997: 494).
As becomes obvious from the summary above, the existing literature has
provided insights into quite a range of explanations for the use of foreign
languages in advertising. However, a sample of 746 advertisements that
was collected in Flanders (the Dutch-speaking part of Belgium) contained
a number of advertisements for which the use of a foreign language could
not be fully explained by any of the reasons discussed above. These 746
advertisements were collected over a 1-year period (in September 2004,
March 2005, and September 2005) from 10 different Flemish television

stations (commercial as well as public broadcasting stations). English was by
far the foreign language which was used most often in these advertisements:
not less than 45 per cent of the advertisements contain English words. If we
include the use of English in product names (e.g. ‘Head and Shoulders for
Men’; ‘Electrabel Do-My-Care’), this percentage even rises to 65 per cent.
French was the second most popular language, featuring in 4 per cent of the
ads (14 per cent if French product names are included), followed by Italian
with 1 per cent (2 per cent if Italian product names are included).
In 25 out of the 746 advertisements, a foreign language seemed to be
used for a reason that has, to my knowledge, not been discussed so far in

the literature. I propose that in these advertisements, which intertextually
refer to a range of (media) genres, the foreign language functions as a
‘linguistic cue’ to the intended intertextual references that are made. In
22 of the 25 advertisements, the foreign language was English. In this
article, four examples of this type of advertisements will be discussed. Three
are taken from the sample collection of 746 Flemish advertisements, the
fourth is an advertisement recorded from Dutch television. These four adver-
tisements were chosen because they include intertextual references to an
interesting range of media genres (e.g. fiction and nonfiction, British and
American) and provide examples of intergeneric as well as intrageneric
intertextuality. The analysis, I believe, sheds new light on the use of English
in advertisements, and more particularly on the issues such as viewers’
agency and linguistic superiority. First, the next section will discuss the role
of intertextuality in advertising.
INTERTEXTUALITY IN ADVERTISING
In a sense, every media text is intertextual (Barthes 1975; Bakhtin 1986;
Fiske 1987). Producers do not create their texts in a vacuum: other media
texts inevitably permeate their work. Similarly, when readers consume
media texts, their decoding is influenced by other texts they are familiar
with. In this view, intertextuality refers to ‘the fundamental and inescapable
interdependence of all textual meaning upon the structures of meaning
proposed by other texts’ (Gray 2005: 3–4).
However, one should distinguish this type of ‘inescapable intertextuality’
(Fiske 1987: 115) from the ‘intertextual intent’ that some texts have. Such
texts ‘aim themselves at other texts and genres, and [. . .] want us to read
them through other texts or genres’ (Gray 2005: 4). In this type of intertex-
tuality, which Hitchon and Jura (1997) have called ‘postmodern intertextual-
ity’, the meaning of a new text is almost entirely dependent on the reader’s
knowledge of the ‘referenced’ text(s). Withalm (2003), for instance, discusses
a number of advertisements that intertextually refer to movies. The meaning
and/or humor of these ads can only be understood and appreciated if viewers
are familiar with movies such as Hitchcock’s classics Psycho (1960) or The Birds
A. H. KUPPENS 119
(1963), or box office hit Titanic (1997). Besides such intergeneric intertextual
references (i.e. references to media genres other than advertising), advertise-
ments can also contain intrageneric references, such as references to other
particular advertisements, or to more general advertising clichés, codes, or
routinized strategies (Cook 2001). In both inter- and intrageneric intertextual-
ity, pastiche, parody, and humor often feature prominently.
The advent of postmodern intertextual advertising in the 1980s attempted
to appeal to a generation of critical, media-literate, and skeptical viewers
(Goldman and Papson 1994; Leiss et al. 2005). ‘After nearly forty years
of watching TV ads, viewers had grown too acclimatized to advertising’s
routinized messages and reading rules. [. . .] Savvy, media-literate viewers

now present advertisers with a challenge. Bored and fatigued, these viewers
restlessly flip around the channels in search for something that will momen-
tarily arrest their attention and fascination’ (Goldman and Papson 1994: 24).
The creativity, humor, and reflexivity that are typical of intertextual advertise-
ments, constitute an exciting way of appealing to advertising-literate viewers
who ‘see through’ classic advertising strategies. If viewers recognize the
intertextual references, the advertisement may function as ‘a source of ego
enhancement’. By positioning the viewer as the holder of the necessary
cultural capital, the advertiser ‘appears to speak to the viewer as a peer’
(Goldman and Papson 1994: 43). Furthermore, the enjoyment and pleasure
that viewers derive from such advertisements, potentially contribute to
positive attitudes towards the ad, and hence to the brand (Hitchon and
Jura 1997). Also, Stam et al. (1992: 203) point out with regard to intergeneric
intertextuality that ‘the self-referential humor signals to the spectator that
the commercial is not to be taken seriously, and this more relaxed state of
expectation renders the viewer more permeable to the commercial message’.
Proctor et al. (2002) even state that the vagueness of such advertisements
makes the audience vulnerable to manipulation and confusion.
O’Donohoe’s (1997a, 1997b) study with Scottish 18- to 24-year-olds has
pointed out that young adults are remarkably sophisticated in their consump-
tion of intertextual advertisements. Still, the use of intertextual references in
advertising is a very risky business, as its success relies almost entirely on what
authors have variably referred to as the viewer’s cultural literacy (Hitchon and
Jura 1997), intertextual competence (Gray 2005: 30), genre literacy (Gray 2005:
32), popular culture capital (Fiske 1992), or media literacy (Goldman and Papson
1994). As McCracken (1986: 75) points out, the viewer is always ‘the final
author’ of an advertisement. If viewers miss the familiarity with another media
genre or text that is necessary to understand an advertisement, they may
become very irritated with it (Hitchon and Jura 1997).
Transcultural intertextuality
The process of cultural and media globalization has dramatically increased
the resources for intertextual references. As Hesmondhalgh (2002: 178)
points out: ‘Cultural texts originated in one country are increasingly seen,
heard, and so on in other countries. Because of this increasing flow of cultural
texts, audiences and symbol creators can, in many places, draw on texts
from many other different places. Texts, genres and even technologies (such
as musical instruments) will be often reinterpreted and adapted by symbol
creators in other contexts.’ Similarly, Regev (2007: 126) considers the ‘world
culture’ that globalization has created as ‘a bank of visual, sonic and textual
stylistic elements and techniques of expression, from which every local pro-
ducer at the national level can draw materials for her own use’. In Appadurai’s
(1996: 35) terminology, mediascapes provide ‘large and complex repertoires
of meanings, narratives, and ethnoscapes to viewers throughout the world’.

Van Elteren (1996: 69) focuses on American popular culture, which he
regards as ‘one big self service store’ or ‘superculture’, in that it is ‘a reservoir
of cultural elements from which one may borrow as much and in as many
ways as one wishes’. Van Elteren’s phrasing is quite euphemistic: ‘one may
borrow’, he states, but the fact is that the dominance of American cultural
products in global media flows narrows down the repertoire of cultures to
borrow from. Nevertheless, those who borrow from the American cultural
repertoire are free to creatively employ their borrowings in a myriad of
ways, including in ways that invert, ridicule, or oppose American culture.
In the next section, I will discuss four intertextual advertisements
which ingeniously employ viewers’ increasing familiarity with foreign media
genres. I will argue that English plays a vital role in anchoring the intertex-
tuality of the advertisements.
ANALYSIS: FOUR ADVERTISEMENTS, FOUR GENRES,

FOUR ENGLISHES
In the current article, I will discuss a sample of four advertisements in which
English is used to mark intertextual references to American or British media
genres. The first three advertisements I will discuss were produced by Belgian
advertising agencies, and broadcast in Flanders, the Dutch-speaking part
of Belgium. They advertise products that are only or predominantly relevant
to Belgian consumers. In other words, despite the fact that they are entirely
or almost entirely in English, these ads are not intended for an international
audience.
The fourth advertisement, however, was produced by a Canadian agency
and targeted an English-speaking audience, but was later (partly) adopted to
advertise a different product to a Dutch audience. This fourth advertisement
was included in the analysis because it shows that there can be other reasons
for not translating originally English-language advertisements than lower costs
or the demand for worldwide ‘consistency’.
The use of English in the four advertisements will be examined through
a qualitative content analysis. Thereby, not only the linguistic, but also the
A. H. KUPPENS 121
visual, auditory, and narrative characteristics of the ads will be taken into
account. This holistic approach is required in order to fully grasp the inter-
textual references made in the advertisements. As advertisers’ possibilities
of making ‘global’ generic intertextual references are greatly affected by
their audiences’ genre literacy, and as this literacy is primarily dependent on
these audiences familiarity with foreign media, the next section will give a
short description of Flemish and Dutch audiences’ access to English-language
media.
The setting: languages and media in Flanders and

the Netherlands

When it comes to the position of English, Flanders and the Netherlands are
quite similar. Flanders is the Dutch-speaking part of the officially trilingual
Belgium. The other official languages are French, which is spoken by about
40 per cent of inhabitants, and German, which is spoken by less than 1 per
cent. Belgium is made up of four language communities: Flanders is the
officially monolingual Dutch part in the north, Wallonia is the officially mono-
lingual French part in the south, Brussels is the officially bilingual (French and
Dutch) metropolitan district, and German is spoken by a small community
in the east of Belgium. In addition to these official languages, about 70 other
languages are spoken in families living in Flanders (De Houwer 2007: 414).
From an educational point of view, French is the second language in Flanders:
children generally start learning French in the fifth year of primary education
(at the of age 10 or 11 years), while other languages such as English and
German are only taught from secondary education on. However, culturally
speaking, English is far more important in Flanders than French. While the
Flemish are avid consumers of English-language television programs, movies,
and music, French language media products have only a minor role in
the Flemish media landscape: they are broadcast far less frequently, and
when they are, they reach far smaller audiences than do English-language
products. In Flanders, foreign language television programs and movies
are (with the exception of some children’s programs and movies) always
broadcast with Dutch subtitles, in contrast to Wallonia, where foreign pro-
grams and movies are generally provided with French voice-over. Most
Belgians also have access to a range of English-language television channels,
such as BBC1, BBC2, BBC World, CNBC, and CNN. The intensive contact
the Flemish audience has with English through popular media explains why
English is the most popular foreign language with youngsters in Flanders, and
why Flemish 18-year-olds are on average more proficient in English than
in French, despite the fact that they have taken a significantly larger
number of French than English classes during their educational careers
(Housen et al. 2001). Because of the popularity of English-language television
programs and movies, the Flemish can be expected to be familiar with a wide
range of American, British, and (to a lesser extent) Australian media genres
(including, but not limited to: American, British, and/or Australian soaps,
drama series, sitcoms, documentaries, reality television, talk shows, whodun-
its, animated series, and science fiction).
When it comes to the access to, and popularity of English-language media,
the situation in the Netherlands is comparable with that in Flanders. However,
in the Netherlands, English is the first foreign language children learn in
school (at about the age of 10 years); French and German are only taught in
secondary education.
Advertisement 1: advertisement for the Belgian

telecommunications company Belgacom

The first advertisement I will discuss promotes Belgacom, Belgium’s leading
telecommunications company. The advertisement featured in a campaign
with the slogan ‘Soms blijf je beter waar je bent’ (‘Sometimes it’s better to
stay where you are’), which aimed at convincing clients to stay with former
public enterprise and monopoly holder Belgacom, rather than switching over
to the new companies that had recently entered the telecommunications
market.
The advertisement starts with the image of a field of corn, baking in the
sun against a background of chirping crickets (Figure 1A). While the image
switches to a close-up of a frog (Figure 1B), a male voice-over with a British
accent says, ‘When temperatures rise to fifty degrees centigrade, the river frog,
also known as the statue frog, avoids any physical activity.’ The advertisement
switches again to an image of the corn, and then to two different close-ups
Figure 1: (A–F) Intergeneric intertextuality in an advertisement for the

Belgian telecommunications company Belgacom. This advertisement parodies
the style of a British wildlife documentary. (Images reproduced from the
Belgacom commercial ‘Frog’ with permission from Manoëlle Ballaux)
A. H. KUPPENS 123
of the frog (Figure 1C and D). Meanwhile, the voice-over continues, ‘This
particular specimen hasn’t moved an inch for almost half a day now.’ Next,
we see a long-shot image of the frog jumping onto the road, ending up under-
neath the wheels of a passing car (Figure 1E). The voice-over says, ‘Oh’,
and the advertisement shows the image of a flattened out frog, who
croaks ‘Bollocks’. ‘Well’, a voice-over says, in Dutch, ‘Sometimes it’s better
to stay where you are. With Belgacom, for instance’ (my translation).
Simultaneously, the slogan ‘Sometimes it’s better to stay where you are’
appears on screen in Dutch (Figure 1F). Next, the Belgacom logo is shown.
This advertisement is a textbook example of intergeneric intertextuality, as
it intertextually refers to the genre of British wildlife documentary. Several

cues are used to temporarily convince the viewer that (s)he is watching
a documentary on a frog species, rather than an advertisement. There are,
for instance, the background sounds (chirping crickets, the croaking of the
frog), the visual look of the advertisement, the editing choices (cutting
between long shots of the habitat of the animal and close-ups of the animal),
and the tone of the voice-over (authoritative, at a slow pace, with well-
considered pauses). More importantly, the use of English could be seen as a
crucial marker of the intertextuality of the ad. Indeed, the adult, male, British
English speaking voice functions as a reference to British wildlife documen-
taries in the tradition of David Attenborough.
Interestingly, besides just providing the viewer with a translation of the
English audiotape, the Dutch subtitles also seem to be contributing to indexing
the advertisement as a British wildlife documentary, as these are generally also
used in the British documentaries that are broadcast in Flanders.
Advertisement 2: advertisement for the Flemish radio

station Studio Brussel
The second advertisement featured in a campaign for Studio Brussel, the youth
radio station of the Flemish public broadcaster VRT. The campaign centered
around a fake international sports event called ‘The Music Games’, in which
different national teams compete in such ludicrous disciplines as 20 meter moon-
walk, guitar throwing, long singing, and breakdance curling. All four television
advertisements are entirely in English (only the product name, Studio Brussel,
is in Dutch). Here, I will discuss the advertisement ‘Breakdance Curling’.
The advertisement begins with images of a giant sports center behind
a ‘Music Games’ logo (Figure 2A). An American voice-over says ‘Welcome
back to the music games. We’re about to see a new event called break-
dance curling.’ As the advertisement switches to images of the curling sheet
inside the sports center, the voice-over continues: ‘Well there’s a ton weight
on that young man’s shoulders, and I’m sure glad that I’m not out there.’ The
advertisement then shows the Swedish curling team being presented to
the audience (Figure 2B), and a second male voice states, ‘So are we, Dave,
so are we.’ A third male voice comments, ‘Okay, India leading. Here comes

Figure 2: (A–F) Intergeneric intertextuality in an advertisement for the
Flemish radio station Studio Brussel. This advertisement parodies
American sportscaster talk. (Images reproduced from the Studio Brussel com-
mercial ‘Breakdance Curling’ with permission from Peter Claes; ß VRT)
Sweden.’ As the Swedish ‘athlete’ starts spinning on his back on the ice
(Figure 2C), the second voice continues, ‘And he’s off! Look at him spin,
burning up that ice . . . Sweep, com’on sweep it up! Sweep!’ The third voice
adds, ‘This really looks good Steve.’ As the Swedish athlete nears the target
(Figure 2D), the second voice says, ‘Look out, India, look out!’ The third voice
yells, ‘And it’s good! Sweden wins. What a game!’ Finally, both sportscasters
(i.e. the ‘second voice’ and ‘third voice’) are shown (Figure 2E), with the left
one chuckling, ‘Now look who’s going home with the gold. But now first a
word from our sponsor.’ While the Studio Brussel logo appears on the screen
(Figure 2F), the sportscaster on the right-hand side—seemingly unaware that
he is still on air—tells his colleague on the left-hand side, ‘I’m freezing my ass
off’; the left sportscaster responds with ‘I can’t feel my toes’.
This ad, too, is an example of intergeneric intertextuality. More specifically,
the advertisement presents a parody of American sports commentary, thereby
adopting several features of the genre. For instance, the advertisement follows
the general structure of a sports event broadcast: Starting with the announce-
ment of the event and an image of the venue (Figure 2A), then switching
to the presentation of the performing team (Figure 2B), followed by the
actual performance (Figure 2C and D), and finishing with the commentators’
wrap-up (Figure 2E and F). The advertisement also adopts the typical ‘dialogic
presentation’, with one announcer who narrates the play-by-play (e.g. ‘Okay,
India leading. Here comes Sweden’, ‘And it’s good! Sweden wins. What
a game’) and one commentator providing the color commentary (e.g. ‘Look
at him spin, burning up that ice . . . Sweep, com’on sweep it up! Sweep!’)
(Ferguson 1983: 156; Hansen 1999). The fact that the two presenters
are actually shown to the viewer further marks the advertisement as an
A. H. KUPPENS 125
American sports event broadcast (as this is rarely the case in Flemish broad-
casts). Examples of visual features that refer to a sports event broadcast are
the event logo (Figure 2A–F), the display with information on the competitors
(Figure 2B), and the sportscasters’ sartorial choices (big headphones, wearing
a warm casual coat over a tuxedo; Figure 2E). Also very significant is the use
of typical elements of the language of sports commentary such as updates
of the course of the game (‘India leading’), the use of specific routines
(e.g. ‘What a game!’, ‘And he’s off!’, ‘And it’s good!’), some syntactic features
of sports commentary (copula deletion in ‘India [is] leading’; the use of and
for starting a new statement (as in ‘And he’s off!’, ‘And it’s good!’)), and the
sportscasters’ rising intonation as the game nears its climax (Ferguson 1983;

Delin 2000). Again, it seems that a very specific register of (in this case,
American) English is used to mark the ‘intended’ intertextuality of the adver-
tisement. Moreover, the American sportscasters’ register is not only adopted,
but also mocked, most noticeably when the euphemistic cliché ‘a word from
our sponsor’ is used (rather than ‘commercials’). This kind of lampooning
also takes place at the end of the ad, when both sportscasters make the not
uncommon mistake of making comments on air that are not intended to
be heard by the viewers (‘I’m freezing my ass off’, ‘I can’t feel my toes’).
Advertisement 3: public service advertisement against

teenage pregnancy
The third advertisement is a public service advertisement that warns against
teenage pregnancy. The advertisement was created by a Flemish advertising
agency and was jointly commissioned by the federal Public Health Department,
the Belgian National Institute for Health and Invalidity Insurance (Riziv/Inami),
and several health insurance companies. The advertisement was broadcast
on Flemish television and cinema screens in 2005, and was accompanied
by a web site on teenage pregnancy.
At the outset, the advertisement seems to promote a new video game
called ‘Teenage Mum’. The advertisement depicts the animation style of
The Sims video games (and advertisements), while a hyperactive voice-over
states, ‘Get ready for teenage mum, the ultimate gaming experience. Test
your teenage motherhood skills in twenty-four exciting levels. Change diapers
around the clock. Try to study while your baby keeps you up all night. Look
after your child, but don’t miss phone calls from your friends.’ Meanwhile,
the advertisement shows the different ‘levels’ of the video game, accompanied
by head-up displays such as the indication ‘press start’ (Figure 3B), a progress
bar (Figure 3E) which shows the teenage mum’s study progress, and the
indication ‘Friend Lost’ (Figure 3F) which appears when the teenage
mum fails to pick up the phone as she is chasing her crawling baby.
The voice-over continues, ‘Your ultimate challenge: Not having a nervous
breakdown.’ Next, the background turns black, the hyperactive music stops,
and the exhausted teenage mum character falls down to her knees and

Figure 3: (A–I) Intra- and intergeneric intertextuality in ‘Teenage Mum’,
a public awareness advertisement against teenage pregnancy. This advertise-
ment intertextually refers to the popular The Sims video games, as well
as to the hyperactive style of ‘typically American’ advertisements. (Images
reproduced from the Laura commercial ‘Teenage Mum’ with permission
from Bruno Ruebens, Manager of Marketing, Nationaal Verbond van
Socialistische Mutualiteiten, Sint-Jansstraat 32, 1000 Brussels)
sighs (Figure 3H). Finally, white letters against the black background read
(originally in Dutch): ‘Being a teenage mum is not a game. Use contraception’
(Figure 3I) and ‘The pill is now cheaper if you are under 21. Info on laura.be’
(my translation).
This advertisement combines both intrageneric and intergeneric intertex-
tuality. Intergenerically, the commercial appropriates the animation style of
the popular The Sims video games. The setting of the ‘Teenage Mum’ game
(everyday life, ‘real world’ rather than fantasy world), the visual look of
the characters and their environment, the erratic physical movements of the
characters, and the several head-up displays that are shown, are all highly
similar to the The Sims video games. However, the style of the advertisements
for The Sims games is generally not as hyperactive as the Teenage Mum
advertisement. On the contrary, such advertisements adopt a rather ‘laidback’
style, with minimal use of voice-over. The Teenage Mum advertisement,
however, adopts the style of a different type of commercial, namely that
of energetic commercials, in which a voice-over tries to mention as many
of the products’ exciting features as is possible in the constrained duration of
an advertisement. This short, hard sell, informational and rather irritating style
A. H. KUPPENS 127
is typical of American advertising (Nevett 1992). The adoption of this adver-

tising style constitutes the intrageneric intertextuality of the advertisement.
Much of the humor and irony of the advertisement is situated in the contrast
between the excited tone of the voice-over, and the miserable situation of
the Teenage Mum character. The intrageneric intertextuality of the advertise-
ment explains, I believe, why the first part of the advertisement is in English:
the American English that is used marks the intertextual reference the adver-
tisement makes to the ‘typically American’ advertising style. Indeed, as soon
as the intertextual nature of the advertisement is revealed, the advertisement
switches to Dutch (Figure 3F).
This advertisement exemplifies many of the advantages that intertextual

advertising has, especially when aiming at a young audience. The reflexivity
which the advertisement achieves in parodying a specific advertising style
allows the advertiser to ‘speak to the viewer as a peer’ (Goldman and
Papson 1994: 43). At the same time, the deconstruction of that advertising
style enables the advertiser to distance themselves from ‘commercial advertis-
ing’, thus reinforcing the informational nature of the ad. Hence, this adver-
tisement cleverly manages to enjoy the visual attractiveness and appeal of
the video game—not coincidentally The Sims, a game that is especially appeal-
ing to female youth (Nutt and Railton 2003)—without sacrificing its credibility
as a public awareness ad.
Advertisement 4: advertisement for the bathroom cleaning

product Cif Power Cream Spray
The fourth advertisement was broadcast in the Netherlands in 2006. In
contrast with the advertisements discussed above, this advertisement was
originally produced for an English-speaking audience. Originally created by
a Canadian advertising agency for the Unilever product ‘Vim Cream’, this
advertisement was later slightly adapted (i.e. supplied with subtitles and
a different product name, tag line, and slogan) and used for promoting a
different Unilever product (namely ‘Cif Power Cream Spray’).
The advertisement starts with a close-up image of a worn-out woman in
an orange overall. As melancholic music starts to play, a young, sad-looking
girl puts her hand on the glass window that separates her from the woman
(Figure 4A). As their hands meet on the glass window (Figure 4B), the girl
asks, ‘When are you gonna get out of here?’ The woman responds sadly: ‘In
a while . . . I gotta get back’ (Figure 4C). As we see the girl fight back her
emotions, the advertisement suddenly switches to a long shot of the woman
and the girl, who turn out to be located at either sides of a glass shower
door (Figure 4D). As the woman starts cleaning the bath tub, the girl puts
both hands on the shower door, and desperately yells, ‘I love you momma.’
The woman emotionally responds with ‘I love you too baby!’ (Figure 4E). The
advertisement then shows the cleaning spray that is advertised (Figure 4F),
and a voice-over states (originally in Dutch): ‘Thorough cleaning does not

Figure 4: (A–F) Intergeneric intertextuality in an advertisement for Cif Power
Cream. This advertisement borrows several elements of the style and structure
of a Hollywood movie or television drama prison scene. (Images reproduced
from the Cif Power Cream Spray commercial with permission from Unilever
Nederland BV, located in Rotterdam, the Netherlands)
need to be a punishment. For Cif introduces Cif Power Cream Sprays. The
power of Cif, in a handy spray’ (my translation).
This fourth advertisement can be interpreted as intertextually referring to
the genre of American television and Hollywood drama, and more specifically
to movie scenes of prisoners being visited by family members. Again, several
visual, auditory, and narrative cues are used to temporarily make viewers
believe that they are watching a scene from a Hollywood movie or television
drama, rather than a commercial for cleaning products. An important audi-
tory reference to the genre is the melancholic background music. The inter-
textual nature of the advertisement is further visually enhanced by its
grayish look, by the orange jumpsuit the mother is wearing (typical prison
dress), and by the editing choices that where made (e.g. close-ups of emo-
tion-ridden faces, shot-reverse-shots of the interactants). The most conspicu-
ous intertextual references are, however, the several clichés of Hollywood
scenes of prison visits that are used in the ad. The most prominent
examples of such references are the hands of the mother/prisoner and
the child/visitor touching through the glass (see especially the close-up in
Figure 4B), the mother’s/prisoner’s utterance ‘I gotta get back’ (in movies
generally preceded by a prison ward’s request to leave the visiting room),
and the over-the-top emotions that lie in the facial expressions of both inter-
actants (e.g. Figure 4E) and in utterances such as ‘I love you momma!’ and
‘I love you too, baby!’
A. H. KUPPENS 129
Again, the intertextual nature of the advertisement can explain why

English is the dominant language in this advertisement. While the first part
of the advertisement (Figure 4A–E) is identical to the Canadian original, the
second part differs highly. In the Canadian ad, the tag line read ‘Don’t spend
your life cleaning’, while the adaptation that was broadcast in the Netherlands
reads (originally in Dutch): ‘Thorough cleaning does not need to be a punish-
ment.’ The product slogan in the Canadian advertisement was ‘Cleans
the tough stuff. Easily’, while the Dutch version, which advertises an entirely
different cleaning product, says ‘The power of Cif in a handy spray.’ So,
obviously, the use of English in this advertisement cannot be explained by
a desire for a globally consistent marketing strategy and/or brand image.

Rather, I believe it is the intertextuality of the advertisement which explains
why the Dutch version of the advertisement did not use Dutch voice-over
translation (dubbing), despite the fact that this is an often used strategy
when English-language advertisements (especially those for cleaning products
and detergents) are adapted for a Dutch-speaking audience. If the advertise-
ment had been dubbed in Dutch, the intertextual reference to Hollywood
drama would be largely lost, for American movies are never dubbed when
they are broadcast in the Netherlands, but always subtitled. In fact, advertising
is one of the very few genres in which dubbing is used in the Netherlands.
In this sense, again, the subtitles in the advertisement seem to do more than
just providing a translation; they also enhance the intertextuality of the ad.
A Dutch-dubbed version of the advertisement would probably be less readily
mistaken by viewers for a scene from a Hollywood movie or television drama;
hence, the intertextual nature, as well as the humorous climax of the adver-
tisement (i.e. when the ‘prisoner’ is revealed to be a cleaning housewife) could
get ‘lost in translation’. Also, the orange jumpsuit the prisoner/mother is wear-
ing would no longer make any sense, for orange suits are typically worn by
American prisoners, not by Dutch(-speaking) prisoners.
DISCUSSION
In this article, I discussed four television advertisements in which English is
the dominant language, despite the fact that they target Dutch-speaking
audiences. In these advertisements, the choice for English cannot be explained
by the motivations that have thus far been described in the scholarly literature:
the advertisements are not part of an international campaign, and English
is not used for purely creative-linguistic reasons (e.g. for linguistic puns,
to fill a lexical gap), nor for the cultural connotations it potentially reflects
on the product or its potential users. Rather, in these ads, specific varieties and
registers of English seem to be used because of their association with the genre
to which the advertisement intertextually refers. Advertisers thereby draw on
viewers’ familiarity with a range of American and British media genres. This
observation has interesting implications for our views of foreign English in
advertising and of the global spread of English through popular culture more
generally.
First of all, the analysis shows that, when English is used in advertisements,
this does not necessarily entail an uncritical celebration of that language and
the cultural values it is associated with. The current scholarly studies have
demonstrated that English is often used because it is supposed to imbue the
advertised product with the cultural connotations it carries, or because it is
expected to appeal to a specific target audience that values those connotations.
These connotations can be culturally linked to a specific English-speaking
country (e.g. ‘British class’), or they can be more general (e.g. ‘modernity’).
In the four advertisements discussed here, however, while the connotations

English is used for are obviously linked to either the USA or Great Britain,
these connotations are not used because they are assumed to reflect on the
product or on its users. For instance, in the first advertisement I discussed,
British English is not used because of the reputation the British hold with
regard to the product that is advertised (i.e. telecommunications), but for
their identification with the genre the advertisement intertextually refers
to (i.e. wildlife documentary). Likewise, in the other three advertisements
under study, American English is not used because the USA is highly regarded
in the respective fields of contraception, radio stations, and cleaning products,
but for the identification of American English with hyperactive hard-sell
advertisements, dialogic sports commentary, and syrupy Hollywood drama.
The distinction between both motivations for using English is crucial, as
both involve an entirely different appreciation of the English language. In
the advertisements described in the existing scholarly literature so far,
English generally enjoyed the status of a superior language: The language of
(American) high quality, (British) class, of prestige, modernity, globalization,
cosmopolitanism, internationalism, etcetera, in short: the language of ‘all
things good’. The advertisements that were discussed in the current article,
however, do not involve such an uncritical celebration of the English lan-
guage. On the contrary: English seems to be used in these advertisements
because it is the language of specific media genres—genres, however, that
are parodied and mocked, and whose conventions are subverted.
In her analysis of French print and television advertisements, Piller (2001)
has observed that English is often used at the end of an advertisement, and
in important parts of the advertisement such as the slogan. When it is placed
in such dominant parts of the ad, Piller claims, English is vested with the
meaning of authority, authenticity, and truth. In the advertisements discussed
in the current article, however, the opposite is the case. Here, English is used
only in the beginning of the advertisements, while Dutch plays a dominant
role at the end of the advertisements, more specifically when the intertextual
nature of the advertisement is revealed, as well as the advertised product, the
slogan, and other important information. In this sense, English enjoys quite
A. H. KUPPENS 131
a ‘subordinated’ position in the ads: it is used for parodying and mocking the
genres it is associated with, it is used for entertaining the viewer, but when
the ads come to revealing their ‘real’ message, English is quickly disposed of.
Secondly, the advertisements discussed here demand a different, more
active role from their readers than advertisements in which English is used
because of the connotational values it evokes. The latter type of advertisements
assumes an essentially passive role for their readers, as they draw on readers’
more or less subconscious associations between English and certain cultural
connotations or stereotypes. In this sense, consumers are regarded as gullible
and uncritical: it does not really matter what an advertiser says about the
product; as long as he says it in English, consumers will rise to the bait (as

is especially evident in advertisers’ use of ‘nonsensical English’ mentioned
earlier). The advertisements discussed here, however, draw on readers’ cul-
tural and media literacy, and on their ability and creativity to link certain
visual, auditory, narrative, and linguistic cues to specific media genres. One
could almost state that while the first type of advertising works better
when the reader lacks a certain amount of media literacy (more specifically,
advertising literacy)—for it depends partly on the reader’s unawareness of
the fact that English is used to activate certain connotations—the second
type can only be successful when a certain degree of media literacy is
present—for the reader has to be familiar with the intertextual references
that are made in order to understand the advertisement.
Thirdly, the analysis presented here challenges monolithic notions of the
spread of English. In the advertisements under study, a range of different
registers of English is used to intertextually refer to an array of different
media genres. Obviously, analyses that see the popularity of English in adver-
tising worldwide merely as evidence for the idea that English is ‘the language
of global mass consumer culture’ lack the complexity that is needed for
a thorough understanding of the phenomenon.
It is, of course, no coincidence that American and British popular culture
are by far the foreign cultures most referred to in Flemish advertisements,
and hence, that English is the prevailing foreign language in such advertise-
ments. The dominant position that American and British products hold in
the Flemish media landscape is thus reflected in the intertextual references
that are made. Very relevant in this respect is that in the French-speaking
part of Belgium, where English-language television programs and movies are
transmitted with French voice-over instead of with subtitles, the Teenage
Mum advertisement and the Belgacom advertisement were broadcast with
French audio. Indeed, English would not serve as a ‘linguistic cue’ to the inter-
textual nature in the advertisements, as Anglo-Saxon programs are not broad-
cast in English there. In this sense, the advertisements analyzed here provide
a clear demonstration of the idea that the dominant position of Anglo-Saxon
in global media flows reinforces the global spread and popularity of English.
However, we should be very careful in our understanding of such processes,

and avoid lapsing into an overstated economic determinism. As stated earlier,
my analysis shows that the adoption of English by advertisers does not equal
admiringly and respectfully embracing the language. Rather, the ways English
is used in the advertisements analyzed here express a quite complex and
reflexive appreciation of the language. Thus, while the global flows of
Anglo-Saxon media products might enhance the use of English in Flemish
media products such as advertisements, one should be wary of making auto-
matic assumptions about those who adopt the language.
In summary, the advertisements I discussed here differ on a number of
points from advertisements which use English because of its cultural connota-

tions. However, these different motivations also share some similarities. Most
importantly, transcultural intertextuality is also the condition which enables
the cultural-connotational use of English to be successful. Here, I refer not to
the ‘intended intertextuality’ displayed by the four ads I discussed, but rather
to what Fiske (1987: 115) called inescapable intertextuality: the idea that as we
create and decode texts, we inevitably rely on other texts we consumed.
Advertisements which use English for the cultural connotations the language
is believed to reflect on the product, build on the associations viewers hold
between English and certain connotations, and these associations have been
largely constructed through (the advertisers’ and the viewers’) consumption of
other media texts (Kelly-Holmes 2005). The same ‘large and complex reper-
toires of meanings, narratives, and ethnoscapes’ (Appadurai 1996) the adver-
tisements discussed here rely on to make their intended intertextual references
work, are the necessary condition for the associations advertisers and viewers
hold between American English and freedom, or between British English and
class.
Finally, the analysis reported here also shows the importance of a holistic
approach to the analysis of advertisements, and media products more gener-
ally. In a simple analysis of the language of the advertisements which does not
take into account their visual and narrative aspects, the link between the use
of English and the intertextual references made in the advertisements might
have gone unnoticed. Language use is always a cultural practice, and so, the
study of advertising language must involve an analysis of the broader cultural
meaning of the advertisements.
ACKNOWLEDGEMENTS
The author would like to thank Philippe Meers, Annick De Houwer, Jelle Mast and the
anonymous reviewers for their insightful feedback on an earlier version of this article. Many
thanks also to Steven Malliet for his suggestions with respect to gaming jargon, and to
Manoëlle Ballaux (Belgacom), Christ Lannoy and Karel Vinck (Duval Guillaume Antwerp),
Erik Liebe (Unilever Rotterdam), Bruno Ruebens (NVSM) and Geert Van Hoeymissen (VRT) for
their kind permission to publish images from their commercials.
Conflict of interest statement. None declared.
A. H. KUPPENS 133
REFERENCES
Appadurai, A. 1996. Modernity at Large. Cultural Gerritsen, M., H. Korzilius, F. van Meurs, and
Dimensions of Globalization. University of I. Gijsbers. 2000. ‘English in Dutch commer-
Minnesota Press. cials: Not understood and not appreciated,’
Baumgardner, R. J. 2006. ‘The appeal Journal of Advertising Research 40/4: 17–31.
of English in Mexican commerce,’ World Gerritsen, M., C. Nickerson, A. Van Hooft,
Englishes 25/2: 251–66. F. van Meurs, U. Nederstigt, M. Starren,
Bakhtin, M. M. 1986. ‘The Problem of Speech and R. Crijns. 2007. ‘English in product
Genres.’ In C. Emerson and M. Holquist (eds): advertisements in Belgium, France, Germany,
Speech Genres and Other Late Essays. University the Netherlands and Spain,’ World Englishes
of Texas Press, pp. 60–102. 26/3: 291–315.
Barthes, R. 1975. S/Z. Cape. Goldman, R. and S. Papson. 1994. ‘Advertising

Bhatia, T. K. 1987. ‘English in advertising: in the age of hypersignification,’ Theory, Culture
Multiple mixing and media,’ World Englishes & Society 11/3: 23–53.
6/1: 33–48. Gray, J. 2005. Watching with The Simpsons.
Bhatia, T. K. and W. C. Ritchie. 2004. Television, Parody, and Intertextuality. Routledge.
‘Bilingualism in the global media and advertis- Griffin, J. L. 1997. ‘Global English invades
ing’ in Ritchie (ed.): The Handbook of Poland,’ English Today 13/2: 34–41.
Biligualism. Blackwell, pp. 513–46. Griffin, J. L. 2001. ‘Global English infiltrates
Cheshire, J. and L.-M. Moser. 1994. ‘English as Bulgaria,’ English Today 17/4: 54–60.
a cultural symbol: The case of advertisements Haarmann, H. 1984. ‘The role of ethnocultural
in French-speaking Switzerland,’ Journal of stereotypes and foreign languages in Japanese
Multilingual and Multicultural Development 15/6: commercials,’ International Journal of the
451–69. Sociology of Language 50/1: 101–21.
Cook, G. 2001. The Discourse of Advertising Haarmann, H. 1989. Symbolic Values of Foreign
2nd edn. Routledge. Language Use. From the Japanese Case to a General
De Houwer, A. 2007. ‘Parental language input Sociolinguistic Perspective. Mouton de Gruyter.
patterns and children’s bilingual use,’ Applied Hansen, A. D. 1999. ‘Narrating the game:
Psycholinguistics 28/3: 411–24. achieving and coordinating partisanship in
Delin, J. 2000. The Language of Everyday Life. An real time,’ Research on Language and Social
Introduction. Sage. Interaction 32/3: 269–302.
Featherstone, M. 1995. Undoing Culture. Hesmondhalgh, D. 2002. The Cultural Industries.
Globalization, Postmodernism and Identity. Sage. Sage.
Ferguson, C. A. 1983. ‘Sports announcer talk: Hitchon, J. C. and J. O. Jura. 1997.
Syntactic aspects of register variation,’ ‘Allegorically speaking: Intertextuality of the
Language in Society 12/2: 153–72. postmodern culture and its impact on print
Fiske, J. 1987. Television Culture. Methuen. and television advertising,’ Communication
Fiske, J. 1992. ‘The cultural economy of Studies 48/2: 142–58.
fandom’ in Lewis (ed.): The Adoring Audience. Hjarvard, S. 2004. ‘The globalization of lan-
Fan Culture and Popular Media. Routledge, guage. How the media contribute to the
pp. 30–49. spread of English and the emergence of media-
Friedrich, P. 2002. ‘English in advertising and lects,’ Nordicom Review 25/1–2: 75–97.
brand naming: sociolinguistic considerations Housen, A., S. Janssens, and M. Pierrard.
and the case of Brazil,’ English Today 18/3: 2001. Frans en Engels als Vreemde Talen in
21–8. Vlaamse Scholen. Een Vergelijkend Onderzoek.
Gerritsen, M., H. Korzilius, F. van Meurs, and VUB Press. [French and English as Foreign
I. Gijsbers. 1999. ‘Engels in Commercials Languages in Flemish Schools. A Comparative
op de Nederlandse Televisie. Frequentie, Study.]
Uitspraak, Attitude en Begrip,’ Communicatiewe- Jacquemet, M. 2005. ‘Transidiomatic practices:
tenschap 27/2: 167–86. [English in Commercials Language and power in the age of globaliza-
on Dutch Television. Frequency, Pronuncia- tion,’ Language and Communication 25/3:
tion, Attitude and Comprehension.] 257–77.
Kelly-Holmes, H. 2000. ‘Bier, Parfum, Kaas. and implications,’ Journal of Advertising 21/4:
Language Fetish in European advertising,’ 61–71.
European Journal of Cultural Studies 3/1: 67–82. Nutt, D. and D. Railton. 2003. ‘The Sims: Real
Kelly-Holmes, H. 2005. Advertising as Multilingual life as genre,’ Information, Communication &
Communication. Palgrave Macmillan. Society 6/4: 577–92.
Krishnasamy, K. 2007. ‘English in Tamil: The O’Donohoe, S. 1997a. ‘Leaky boundaries.
language of advertising,’ English Today 23/3–4: Intertextuality and young adult experiences
40–49. of advertising’ In M. Nava, A. Blake, I.
Lee, J. S. 2006. ‘Linguistic constructions of MacRury, and B. Richards (eds): Buy
modernity: English mixing in Korean televi- This Book. Studies in Advertising and
sion commercials,’ Language in Society 35/1: Consumption. Routledge, pp. 257–75.
59–91. O’Donohoe, S. 1997b. ‘Raiding the postmodern
Leiss, W., S. Kline, S. Jhally, and J. Botterill. pantry. Advertising intertextuality and the

2005. Social Communication in Advertising. young adult audience,’ European Journal of
Consumption in the Mediated Marketplace. Marketing 31/3–4: 234–53.
Routledge. Ovesdotter Alm, C. 2003. ‘English in the
Machill, M. 1997. ‘Background to French lan- Ecuadorian commercial context,’ World
guage policy and its impact on the media,’ Englishes 22/2: 143–58.
European Journal of Communication 12/4: Phillipson, R. 1992. Linguistic Imperialism.
479–509. Oxford University Press.
Martin, E. 1998. ‘The use of English in written Phillipson, R. and T. Skutnabb-Kangas. 1997.
French advertising: A study of code-switching, ‘Linguistic human rights and English in
code-mixing, and borrowing in a commercial Europe,’ World Englishes 16/1: 27–43.
context,’ Studies in the Linguistic Sciences 28/1: Piller, I. 2000. ‘Multilingualism and the modes
159–84. of TV advertising’ In F. Ungerer (ed.): English
Martin, E. 2002a. ‘Cultural images and different Media Texts Past and Present. Language
varieties of English in French television com- and Textual Structure. John Benjamins,
mercials,’ English Today 18/4: 8–20. pp. 263–79.
Martin, E. 2002b. ‘Mixing English in French Piller, I. 2001. ‘Identity constructions in multi-
advertising,’ World Englishes 21/3: 375–402. lingual advertising,’ Language in Society 30/2:
Martin, E. 2007. ‘ ‘Frenglish’ for sale: Multilingual 153–86.
discourses for addressing today’s global Piller, I. 2003. ‘Advertising as a site of language
consumer,’ World Englishes 26/2: 170–88. contact,’ Annual Review of Applied Linguistics
Masavisut, N., M. Sukwiwat, and 23/1: 170–83.
S. Wongmontha. 1986. ‘The power of the Proctor, S., T. Proctor, and I. Papasolomou-
English language in Thai media,’ World Doukakis. 2002. ‘A post-modern perspective
Englishes 5/2–3: 197–207. on advertisements and their analysis,’
McCracken, G. 1986. ‘Culture and consumption: Journal of Marketing Communications 8/1:
A theoretical account of the structure and 31–44.
movement of the cultural meaning of Regev, M. 2007. ‘Cultural uniqueness and
consumer goods,’ Journal of Consumer Research aesthetic cosmopolitanism,’ European Journal
13/1: 71–84. of Social Theory 10/1: 123–38.
Meinhof, U. H. 1998. Language Learning in the Stam, R., R. Burgoyne, and S. Flitterman-
Age of Satellite Television. Oxford University Lewis. 1992. New Vocabularies in Film
Press. Semiotics. Structuralism, Post-Structuralism and
Mueller, B. 1992. ‘Standardization vs. specializa- Beyond. Routledge.
tion: An examination of westernization in Takashi, K. 1990. ‘A sociolinguistic analysis of
Japanese advertising,’ Journal of Advertising English borrowings in Japanese advertising
Research 32/1: 15–24. texts,’ World Englishes 9/3: 327–41.
Myers, G. 1994. Words in Ads. Edward Arnold. Ustinova, I. P. 2006. ‘English and emerging
Nevett, T. 1992. ‘Differences between American advertising in Russia,’ World Englishes 25/2:
and British television advertising: Explanations 267–77.
A. H. KUPPENS 135
Ustinova, I. P. and T. K. Bhatia. 2005. Wardhaugh, R. 1987. Languages in

‘Convergence of English in Russian TV Competition: Dominance, Diversity and Decline.
commercials,’ World Englishes 24/4: Blackwell.
495–508. Withalm, G. 2003. ‘Commercial intertextuality’
van Elteren, M. 1996. ‘Conceptualizing the In S. Petrelli and P. Calefato (eds): Logica,
impact of US popular culture globally,’ The Dialogica, Ideologica. I Segni Tra Funzionalitá ed
Journal of Popular Culture 30/1: 47–89. Eccedenza. Mimesis, pp. 425–36.

Applied Linguistics: 31/1: 136–155 ß Oxford University Press 2009
doi:10.1093/applin/amp015 Advance Access published on 9 May 2009
A Subject–Object Asymmetry in
the Comprehension of wh-Questions
by Korean Learners of English
JIN-HWA LEE
Chung-Ang University, Seoul, Korea

Previous studies on English as a second language (L2) argue for the relative ease
of object wh-questions based on the finding that L2 learners are more accurate
and faster in judging the grammaticality of object wh-questions than that
of subject wh-questions in English. This article re-examines this claim by inves-
tigating L2 learners’ comprehension of long-distance wh-questions at different
stages of English acquisition. A total of 113 Korean-speaking learners of English
with different years of English instruction participated in a picture-based
comprehension task. Contrary to previous studies, the results of the present
study point toward a strong preference for subject wh-questions to object
wh-questions. The learners were more accurate and improved faster in subject
wh-questions than in object wh-questions. In addition, they showed a strong
tendency to interpret object wh-questions as subject wh-questions. These results
are in line with distance-based accounts of processing complexity. Subject
wh-questions are easier to process because the distance between the wh-word
and the gap is shorter and therefore poses less burden on working memory in
subject wh-questions than in object wh-questions.
INTRODUCTION
Wh-questions have been extensively investigated in the second language (L2)
acquisition literature, mostly within the Chomsky’s (1981) Principles and
Parameters framework (Bley-Vroman et al. 1988; Schachter 1989, 1990;
Johnson and Newport 1991; White 1992). Under the assumption that
wh-movement is constrained by Universal Grammar (specifically, Subjacency
(Chomsky 1977)), and that this constraint is instantiated in some languages but
not in others, the central question of the studies was whether L2 learners are
able to access the constraint when their first language (L1) does not instantiate
it. L2 learners’ ability to accept grammatical questions and reject ungrammatical
ones in L2 has been interpreted as evidence for UG availability in L2 acquisition.
More recently, however, a different perspective has been brought to
research on wh-questions. Researchers tried to explain L2 learners’ perfor-
mance in terms of sentence processing or its relationship with the grammar
(Schachter and Yip 1990; Juffs and Harrington 1995; White and Juffs 1998;
Juffs 2005). It was Schachter and Yip who first drew researchers’ attention to
the processing of wh-movement. They found that Chinese and Korean learners
J.-H. LEE 137
of English, and even native speakers of English had more difficulty in accepting
grammatical wh-questions when the subject of an embedded clause was
extracted as in (1) than when the object was extracted as in (2).
(1) Whoi did Cathy believe ei met her friend? (subject extraction)
(2) Whoi did Cathy believe Tom met ei? (object extraction)
This asymmetry is not explainable by a grammar deficit because both types

of questions are equally grammatical in English. Researchers have therefore
attributed these differences to sentence processing. It has been suggested
that an additional processing load is induced for subject extraction by back-
tracking (Schachter and Yip 1990) or gap reanalysis (Juffs and Harrington

1995), as will be detailed later.
Not all the parsing-based accounts predict the relative difficulty of subject
extraction, however. A distance-based account, which has been empirically
supported by the L1 parsing literature, yields the opposite prediction on the
processing of wh-questions. According to this account, the accuracy of the
processing of a filler-gap dependency such as wh-questions or relative clauses
decreases as more material is interpolated between the filler and its gap
(O’Grady 1997; Gibson 2000; McElree 2000; McElree et al. 2002). This
is because more material induces more demands on memory resources and
makes the retrieval of wh-filler at the gap site more difficult. In the sentences
above, the distance between the filler who and its gap is longer in object extrac-
tion in (2) than in subject extraction in (1). Therefore, object extraction in (2)
would be more difficult to process.
With these opposing predictions for a subject-object asymmetry in the
processing of English wh-questions, this study reexamines the claim that L2
learners have more difficulty with subject extraction than object extraction
by carefully reviewing the previous studies. The findings of psycholinguistic
research suggest that sentence processing is affected by various nonsyntactic
factors such as the animacy of Noun Phrases (NPs) (Traxler et al. 2002), refer-
ential properties of NPs (Warren and Gibson 2002), and the plausibility of a
gap interpretation (Traxler and Pickering 1996; Williams et al. 2001). Arguing
that the potential influence of these factors was not properly taken into con-
sideration in the previous studies, this study investigates whether the pre-
viously observed subject–object asymmetry in grammaticality judgements
pertaining to wh-questions is replicable (i) under a stricter experimental con-
trol, (ii) in a different type of task (i.e. listening comprehension), and (iii) with
learners at earlier stages of English acquisition.
REEXAMINATION OF PREVIOUS STUDIES ON THE SUBJECT–

OBJECT ASYMMETRY IN WH-QUESTIONS
Schachter and Yip (1990) found that Chinese and Korean L2 learners of
English were less accurate in judging grammatical subject extraction than
138 SUBJECT–OBJECT ASYMMETRY IN L2
grammatical object extraction. While their mean scores for object wh-questions
ranged from 5.1 to 6.9 out of 9 depending on the number of clauses
over which the wh-word had been extracted, their mean scores for subject
wh-questions ranged from 2.8 to 5.1. More interestingly, native English
speakers showed a similar pattern.
Juffs and Harrington (1995) not only replicated Schachter and Yip’s (1990)
findings but also observed that advanced adult Chinese L2 learners of English
were slower in judging subject extraction than object extraction from infiniti-
val clauses. In addition, using a moving window technique which measures
word-by-word reading time, they found a greater slowdown at the point
where the learners encountered the subject extraction site than the object

extraction site in finite clauses. Juffs and Harrington interpreted this result
to indicate that Chinese learners had greater difficulty in reanalyzing sub-
ject gaps than object gaps. Similar results were reported by Juffs (2005) with
Japanese and Spanish learners of English. White and Juffs (1998) further
showed that this asymmetry pattern in long-distance wh-questions is not
affected by learning environments. In their study, Chinese learners of
English, regardless of whether they were studying in ESL (English as a
Second Language) or EFL (English as a Foreign Language) settings, were less
accurate and slower in judging subject extraction.
The observed asymmetry in favor of long-distance object wh-questions
among adult L2 learners in the above studies cannot be attributed to problems
in grammatical competence per se for two reasons. First, the same pattern
of asymmetry was observed among not only L2 learners but also adult
native English speakers who have fully developed grammatical competence.
Second, the same L2 learners who showed the asymmetry in grammatical
long-distance questions could correctly reject ungrammatical long-distance
wh-questions at the same level as native speakers, indicating that they
had grammars constrained by the Subjacency Condition. Accordingly, the
researchers tried to explain the observed asymmetry in relation to processing
complexities. Consider sentences (1) and (2) again.
(1) Whoi did Cathy believe ei met her friend? (subject extraction)
(2) Whoi did Cathy believe Tom met ei? (object extraction)
Schachter and Yip (1990) attributed the preference for object extraction over
subject extraction to the general nature of left-to-right processing in human
sentence processing and the Minimal Attachment principle (Frazier and Fodor
1978). Minimal Attachment states that the parser attempts to integrate an
incoming item into the currently constructed phrase structure using the min-
imal number of syntactic nodes. To apply the Minimal Attachment principle
to the sentences above, the parser initially posits an empty category (linked to
the wh-word) as the object of believe for both sentences because this leads to
a simpler structure than projecting an embedded clause as its complement.
J.-H. LEE 139
As parsing proceeds in sentence (1), the verb met calls for a subject, which
leads the parser to ‘backtrack’ to posit a subject gap that is linked to the
wh-word that had previously been associated with the potential object gap
in the matrix clause. According to Schachter and Yip, it is this backtracking
that makes subject extraction more difficult to process because it is against
the left-to-right nature of human processing and therefore demands an
extra processing effort. On the other hand, no backtracking occurs in object
extraction in (2). When the parser encounters Tom, it only has to embed the
NP node inside the following clause. As parsing proceeds, the verb met calls for
an object, and the initial object gap in the matrix clause falls in the postverbal
position. This process goes from left to right without any backtracking.

Juffs and Harrington (1995) explain the asymmetry in terms of the reana-
lysis of the gap, drawing on the principle of Generalized Theta Attachment.
(3) Generalized Theta Attachment (GTA)
Every principle of the Syntax attempts to be maximally satisfied at every point during
processing. (Pritchett 1992: 138)
The parser initially interprets who as the object of believe and thus posits
a matrix object trace in both sentences. However, as it encounters met in
the subject wh-question in (1), the initial matrix object trace must be reana-
lyzed as a subject trace in the embedded clause. This reanalysis gives rise
to concomitant changes in theta role assigner (believe ! met), theta role
(internal ! external), and Case assigned (accusative ! nominative). On the
other hand, in the object wh-question in (2), the parser has to reanalyze
the trace as an embedded object rather than a matrix object. This process
requires a change only in theta role assigner (believe ! met). The type of
theta role and Case that are assigned remains the same. According to Juffs
and Harrington (1995), since subject extraction involves more changes in
theta role and Case, it is harder to parse than object extraction for not only
L2 learners but also native speakers of English.
The review of the aforementioned studies suggests that subject wh-questions
are more difficult to process than object wh-questions and that the locus of
the difficulty is the extraction site. However, this conclusion is perhaps pre-
mature since the previous studies in particular suffer from nontrivial design
flaws in terms of test sentences. For instance, Schachter and Yip (1990) used
different verbs and lexical items for subject and object questions (e.g. Who
did you say John suspects fell in love with Sue? versus Who did the President say
he thinks he’ll appoint as ambassador?), thereby leaving open the possibility
that lexical knowledge plays a role in the observed asymmetry pattern. The
animacy of wh-words and NPs is another hidden factor which may have
affected the result in the study. L1 studies suggest that animate NPs are
more likely to be associated with an agent or subject, whereas inanimate
ones are more likely to be associated with the object (Ervin-Tripp 1970;
Tyack and Ingram 1977; Traxler et al. 2002). In Schachter and Yip’s study,
two-thirds of stimuli included inanimate or at least nonhuman wh-words
(e.g. what or which), which are more favorable to object extractions. Therefore,
it is possible that L2 learners’ preference of object wh-questions to subject wh-
questions in the study reflects the effect of animacy rather than the effect of
syntactic configuration.
Unlike Schachter and Yip (1990), Juffs and Harrington (1995) matched
lexical items, but their stimuli are still biased either against subject
wh-questions or in favor of object wh-questions for other reasons. Referential
properties of NPs are one of the concerns. According to Warren and Gibson
(2002), the less accessible the referent of an NP is in the discourse, the more
resources are required to find or construct it in sentence processing. In this
regard, they argue that pronouns whose referents are already present in

the discourse context require the fewest resources, while indefinite NPs
which must be constructed in the discourse model require the most resources,
with proper nouns and definite NPs falling in between. In Juffs and
Harrington’s study, many of the NPs in the stimuli were short proper nouns
or pronouns, which are considered to require a small or moderate quantity
of resources to access. While this may have reduced the potential processing
load in both types of questions, the reduced processing load may have been
particularly advantageous in the case of object extraction, where the distance
between an extracted element and its gap is longer and therefore integrating
cost is higher. In addition, some object wh-questions were pragmatically more
acceptable than their subject counterparts (e.g. Who do the police want to arrest?
versus Who does Ann want to arrest the thief?). It is also possible that object
questions ending with a stranded preposition (e.g. What does the man think
the car crashed into?) may have provided a more salient cue to the extraction
site than their subject counterpart (e.g. What does the man think crushed into
the car?), given that a sentence-final position is more salient than a sentence-
medial position.1 On the other hand, the use of present tense in the embedded
clause in some sentences may have added more computational load on subject
wh-questions by requiring long-distance agreement (e.g. Who did Ann say likes
her friend? versus Who did Jane say her friend likes?). Overall, the inclusion of
those sentences in stimuli may have affected the subject–object asymmetry in
a way to favor object wh-questions.
The hidden influence of lexical-semantic properties of the stimuli in
the previous studies is confirmed by Baek (2007). When he controlled for
possible effects of animacy of wh-words, referential properties of NPs, and
tense of embedded verbs in his replicate study of Juffs and Harrington
(1995), he obtained the opposite asymmetry pattern in grammaticality
judgements. Both native and Korean speakers of English performed better in
correctly accepting subject-who and subject-what questions than object-who
questions. However, the on-line reading data were still consistent with those
of Juffs and Harrington in that the Korean learners of English took longer
in processing embedded verbs of subject wh-questions than those of object
J.-H. LEE 141
wh-questions, indicating greater processing difficulty for the reanalysis of

the subject gap. Baek explained these contradictory results from grammatical-
ity judgements and on-line reading by claiming that the syntactic reanalysis
is not a determining criterion but one of several factors affecting the difficulty
involved in grammaticality judgements. He mainly attributed the L2 learners’
better performance on subject wh-questions to nonsyntactic factors such as
animacy of wh-words and referential properties of NPs.
The above reexamination of test sentences used in the previous studies
revealed various lexical-semantic biases in favor of object wh-questions.
Even without the problematic stimuli, however, the previous studies
including Baek (2007) have two major limitations. First, the fact that all of

the previous studies rely solely on grammaticality judgement tests (sometimes
with measuring reaction time) raises a question about the generalizability
of their findings. The possibility that the use of other tasks might lead to
different results cannot be excluded. Particularly relevant in this regard is
the fact that L1 studies which have adopted elicited production or comprehen-
sion tasks found exactly the opposite pattern of asymmetry, that is, a prefer-
ence for subject wh-questions over object wh-questions (Crain and Thornton
1991, 1998; Yoshinaga 1996). While this finding could be taken to show
that adult L2 learners are different from L1 children in processing English
long-distance wh-questions, it is equally possible that task type played
a role in the different results. To resolve this matter, it is essential to conduct
L2 studies that are comparable with L1 studies in their data-collection
methods.
Another limitation in the previous studies concerns the range of subjects
they investigated. They all tested only advanced-level L2 learners in compar-
ison with native speakers. It remains to be shown that whether the same
pattern of asymmetry will be found among L2 learners at earlier stages of
English acquisition and whether there is any change in the asymmetry pat-
tern along with the stages.
To fill the gap in the existing research on L2 long-distance wh-questions,
this study examined the comprehension of subject and object bi-clausal
wh-questions by Korean-speaking learners of English. In light of the method-
ological issues raised above, this study compared the listening comprehension
of subject and object wh-questions by Korean learners of English with different
years of English instruction. Three research questions were addressed:
1 Is the asymmetry observed in grammaticality judgement replicable in the
comprehension of English long-distance wh-questions by Korean learners
of English?
2 Is the previously observed asymmetry replicable in earlier stages of L2
English acquisition?
3 Is there any change in the asymmetry along with number of years of
English instruction?
THE EXPERIMENTAL STUDY

Participants
Four groups of Korean-speaking learners of English participated in this
study. The first three groups consisted of EFL students attending schools in
Seoul, Korea: 29 of the 9th graders, 39 of the 11th graders, and 29 college
students. The fourth ESL group consisted of 16 graduate students studying
at a US university. Table 1 contains a summary of data on the four groups
of participants.
All the graduate students belonging to the ESL group reported TOEFL scores
above 600. Our graduate group was comparable with the Chinese-speaking

learners of English in Juffs and Harrington (1995) in terms of age, age of
first exposure to English, and length of stay in the USA, although the use
of different proficiency tests (TOEFL and Michigan, respectively) precludes
ascertaining similar levels of proficiency across the two groups.
The three EFL groups started learning English in 7th grade and thus
differed in number of years of English instruction they had received as 3, 5,
and over 6 years, respectively. Although no general English proficiency test
scores were available for these groups, given that they learned English as part
of public schooling and that the college group was taking an intermediate-level
English class at the time of the experiment, it can be safely assumed that they
belonged to earlier stages of English acquisition than the subjects in Juffs and
Harrington’s (1995) study who had 16 years of exposure to English including
33.85 months of residence in the USA on average. Yet, possible individual
Table 1: Biographical data summary of participants

Group Age (years) Others
The present study

9th graders in 14–15
Korea (N = 29)
11th graders in 16–17
Korea (N = 39)
College students 18–25
in Korea (N = 29)
Graduate students 25–40 Age at first exposure—12.32 years
in the USA (N = 17) Average-31.09 Lengths in the USA—30.88 months
Test score (TOEFL)—634.20
Juffs and Harrington (1995)

Chinese-speaking Average-29.11 Age at first exposure—13.20 years
learners of English Lengths in the USA—33.85 months
in the USA (N = 26) Test score (Michigan)—65.35/80
J.-H. LEE 143
variation within groups requires caution in interpreting any between-group

differences in their performance on wh-questions. In addition, individual per-
formance needs to be investigated.
Method
The experimental study made use of an elicited comprehension task.
Participants were required to listen to recorded questions and choose answers,
based on accompanying pictures. This task was chosen for two reasons: to
explore whether the asymmetry previously observed in grammaticality judge-
ment tasks is replicable in other types of tasks, and to make the present study

comparable with L1 studies which have reported the opposite asymmetry
based on comprehension or production tasks. For the latter reason, test mate-
rials were adapted from Wilhelm and Hanna (1992), who investigated the
comprehension of mono-clausal wh-questions by L1 children.2
A total of 10 bi-clausal who-questions, five subject questions and five object
questions, were tested in the task.3 The complete list of test sentences is pro-
vided in the Appendix. Each subject question had an object question counter-
part with the same verbs, as illustrated in (4) and (5).
(4) Who did the girl think pinched the pig? (Subject wh-question)
(5) Who did the boy think the cow pinched? (Object wh-question)
Matrix clauses used one of three verbs, think, say, and believe and embedded
clauses used one of five verbs, push, pinch, pull, bite, and hit. All the verbs were
presented in the past tense forms. To control for the influence of referential
properties of NPs (Warren and Gibson 2002), all the NPs consisted of a definite
article the and a common noun. In addition, animals were used as the subject
and the object of embedded clauses, in order to avoid any possible animacy
bias (Ervin-Tripp 1970; Tyack and Ingram 1977; Traxler et al. 2002) and
the possible influence of plausibility constraint (Traxler and Pickering 1996;
Williams et al. 2001) by making the embedded clauses semantically reversible.
Each question was presented with a picture that presented information
relevant to the test questions. A sample picture that was used to test the subject
wh-question in (4) is presented in Figure 1.
As illustrated in Figure 1, each picture contains two animate agents, one of
which corresponds to the subject of the matrix clause. Each agent is associated
with a sub-picture presented as a ‘thought bubble’. Sub-pictures contained
three animals in different combinations and different orders, which represent
the content of the embedded clause—that is, what the agent believed, thought,
or said. Since the sub-pictures associated with each agent are different, a cor-
rect answer can be obtained only when learners correctly identify both the
subject of the matrix clause and the syntactic role of who in the embedded
clause. Thus, learners had to pay attention to both matrix and embedded
clauses to find an answer.4

Figure 1: Sample picture for a subject wh-question
Items were randomized so that the subject and object questions in a pair
did not occur consecutively. The questions were recorded by a female native
English speaker with a pause of 8 s between questions.
Procedure
Data from 9th graders, 11th graders, and college students were collected
in Seoul, Korea during regular English classes. For the 11th graders and the
college group, the researcher visited their classes and collected data with the
help of the teachers of the classes. In the case of the 9th graders, the English
teacher alone conducted the data collection. This difference should not have
had a significant influence on the results because the role of data-gatherer
in this study was limited to management of the session, e.g. distributing test
materials and playing the previously recorded tape.
Each student was given a booklet of pictures and an answer sheet. When
the teacher or the researcher played the tape, students were required to listen
to each wh-question and choose an answer based on an accompanying
picture. Each sentence was repeated twice. The task took approximately
10 min to complete. After the test, the students were given small gifts in appre-
ciation for their participation.
The graduate students were tested in Hawaii, either individually or in a small
group. The procedure was identical to the one that was employed for the EFL
groups except that the graduate group listened to each sentence only once and
J.-H. LEE 145
was asked to fill in an additional background questionnaire at the end of the

test, for the purpose of comparison with the Chinese-speaking learners of
English in Juffs and Harrington (1995).
RESULTS
Subject–object asymmetry in groups
The scores for the four groups in the form of mean correct responses to subject
and object questions are presented in Table 2.
In general, the participants performed better on subject wh-questions than

on object wh-questions: on average, they responded correctly to four out of
five subject wh-questions, compared with a mean score of just 2.49 for object
questions. The same asymmetry was observed in all the groups.
In order to check whether these differences within and between groups
were statistically significant, a General Linear Model analysis was conducted
with Group as a between-subjects factor (9th graders, 11th graders, college
students, and graduate students) and Question Type as a within-subjects
factor (subject and object questions). The results are summarized in Table 3.
There were significant main effects for Group (F (3, 109) = 22.170, p < .05)
and for Question Type (F (1, 111) = 64.282, p < .05) with considerably
Table 2: Mean accuracy scores for subject and object wh-questions by group
Subject wh-Q (k = 5) Object wh-Q (k = 5)
Group Mean SD Mean SD
9th graders (n = 29) 2.79 .90 2.10 1.11

11th graders (n = 39) 4.54 .85 2.41 1.80
College students (n = 29) 3.93 1.03 2.03 1.66
Graduate students (n = 16) 5.00 .00 4.19 1.11
Total (n = 113) 4.00 1.16 2.49 1.66
Table 3: Factorial ANOVA for group and question type

Source df SS MS F p 2
Between Group 3 103.120 34.373 22.170 .000 .379

Within Q Type 1 97.217 97.217 64.282 .000 .371
Interactions Group Q Type 3 23.269 7.756 5.129 .002 .124

p <.05
high effect sizes (2 = .379 for Group, 2 = .371 for Question Type). However,
the interaction of Group and Question Type was also significant
(F (3, 107) = 5.129, p < .05). This means that the effect of Question Type
was moderated by that of Group, as illustrated in Figure 2.
While all the groups were more accurate on subject wh-questions
than on object wh-questions, the difference was greater for 11th graders and
college students than the other two groups.
Since a significant Group effect was identified, two separate one-way
ANOVAs (Analysis of Variance) and post hoc analyses were conducted on
each of the comprehension of subject and object questions in order to locate
the differences among the four groups. Again, the Group effect was significant

in both subject wh-questions (F (3, 109) = 31.526, p < .05) and object wh-ques-
tions (F (3, 109) = 8.126, p < .05). Tukey HSD analyses showed that all the
between-group differences were statistically significant in subject wh-ques-
tions, except the one between 11th grader group and graduate student
group. On the other hand, in object wh-questions, only the graduate student
group was significantly different from the other EFL groups, who exhibited no
significant difference from each other.
The 9th grader group was generally poor at interpreting both types of
wh-questions, as shown by the low mean scores in Table 2 (2.79 for subject
extraction and 2.10 for object extraction), although they were still more accu-
rate on subject wh-questions than on object wh-questions. Compared with the
9th graders, the 11th grader group did far better on subject wh-questions (4.54)
to the extent that they were not distinguishable from the graduate group.
However, virtually no change occurred in object wh-questions (2.41). The
college group was not statistically different from the first two groups in their
performance on object wh-questions, while their performance on subject
wh-questions was inferior to that of the 11th grader group, although still
better than that of the 9th grader group. This seems to indicate that the college
group was not necessarily at a higher stage than the 11th grader group in their
9th
3
11th
College
2
Graduate
0
Su b -Q Ob j-Q
Figure 2: Comparison of subject and object questions in each group

J.-H. LEE 147
English acquisition,5 which cannot be confirmed due to the lack of general

English proficiency measures. However, considering English learning condi-
tions in Korea, it is highly possible that some college students experience loss
in their English proficiency. All the Korean EFL learners have to learn English
as an obligatory subject until 12th grade. Once they enter university, however,
they are relatively free to choose which English courses to take and how many
hours to devote to language study. While some students keep studying
English, others may discontinue it for several years and only resume it later.
In the latter case, their English proficiency cannot but diminish. Although the
college group in this study was taking the same English conversation class,
it is still possible that they had experienced some language loss due to likely

discontinuous and relatively sparse English instruction at universities. Given
this possible within-group variation, it is crucial to examine individual data,
as will be done in the following section.
Finally, the graduate group manifested statistically significantly better
performance on object questions. They were almost twice as accurate as the
other groups on object wh-questions (4.19). Nevertheless they still performed
better on subject wh-questions (5.0) than on object wh-questions. This asym-
metry is the opposite of the one reported in Juffs and Harrington (1995)
despite the fact that this group is comparable with their Chinese group in
terms of background profile.
Subject–object asymmetry within individuals

Given possible individual variations within each group, it is necessary to inves-
tigate individual participants’ data. Based on their scores for subject and
object questions, individual learners were classified into one of three patterns:
no asymmetry (i.e. the same scores on subject and object questions), a pref-
erence for subject questions (i.e. a higher score on subject questions), and
a preference for object questions (i.e. a higher score on object questions).
Table 4 present the number of learners exhibiting each pattern.
The majority of 9th graders, 11th graders, and college students belonged
to the subject preference pattern; that is, they performed better on subject
wh-questions than on object wh-questions (52, 72, and 72 per cent of the
Table 4: Number of learners for three response patterns

Group Sub = Obj (%) Sub > Obj (%) Sub < Obj (%)
9th graders (n = 29) 8 (28) 15 (52) 6 (21)

11th graders (n = 39) 8 (21) 28 (72) 3 (8)
College students (n = 29) 5 (17) 21 (72) 3 (10)
Graduate students (n = 16) 9 (56) 7 (44) 0 (0)
Total (n = 113) 30 (27) 71 (63) 12 (11)
group, respectively). A subject preference was still prevalent in the graduate

group (44 per cent).
On the other hand, out of a total of 113 participants, only 12 showed an
object preference. Half of these were 9th graders, for whom the possibility
of blind guessing cannot be completely excluded due to their low proficiency.
There were no instances of object preference among the graduate
students, the majority of whom (56 per cent) were balanced in interpreting
subject and object wh-questions, indicating that they had overcome the
initial strong preference for subject extraction observed in the 9th grader,
11th grader, and college groups.

Distribution of responses to subject and object questions
The distribution of participants’ responses to subject and object questions
was further analyzed to determine how the participants processed each type
of question. In the elicited comprehension task, the participants were given
three options to choose for each question. A learner’s choice of these options
reflects whether he or she interpreted a given question correctly or reversely,
or failed to interpret it at all. For instance, in Figure 1 accompanying the
question of who did the girl think pinched the pig?, the monkey is the correct
answer, the dog is an irrelevant item, and the cow is the response selected
if the learner misinterprets the subject question as an object question
(i.e. who did the girl think the pig pinched?). Table 5 presents the distribution
of learners’ responses.
All the nonresponses were found among the 9th graders, which may indi-
cate an inability to deal with long-distance wh-questions appropriately.
Relatively high percentages of responses to irrelevant items in this group,
as compared with the other three groups, also seem to support this interpreta-
tion. As expected from the mean accuracy scores that we examined earlier,
there were more correct answers to subject questions than to object questions
regardless of the group.
Table 5: Distribution of learners’ responses

Subject wh-Q (k = 5) Object wh-Q (k = 5)
Group C R IR NR Total C R IR NR Total
9th graders (n = 29) 81 38 20 6 145 61 66 14 4 145

11th graders (n = 39) 177 11 7 0 195 94 99 2 0 195
College students (n = 29) 114 19 12 0 145 59 81 5 0 145
Graduate students (n = 16) 80 0 0 0 80 67 10 3 0 80
Total (n = 113) 452 68 39 6 565 281 256 24 4 565
Note: C = correct response; R = reversal response; IR = irrelevant response; NR = no answer;

Total = the total number of responses (the number of participants the number of test items)
J.-H. LEE 149
Particularly noteworthy in Table 5 are learners’ reversals. Remember that

a reversal reveals that learners misinterpreted an object question as a subject
question, or vice versa. If learners frequently and consistently made reversal
errors only in one type of question, this provides another piece of evidence for
its relative difficulty compared with the other. Reversal errors in object ques-
tions were far more frequent than in subject questions. While a total of 256
object wh-questions (45 per cent of all the responses) were misinterpreted as
subject wh-questions, there were only 68 reversal errors (12 per cent of all the
responses) in subject wh-questions. This general pattern of more reversal errors
in object wh-questions than in subject wh-questions was observed in all groups.
Moreover, the frequency of these reversal errors in object wh-questions did

not decrease along with the years of English instruction in the three EFL groups.
DISCUSSION
This study has addressed the question of whether the asymmetry previously
observed in judgements of wh-question grammaticality is replicable in the L2
comprehension of English long-distance wh-questions by Korean learners of
English. Contrary to the previous finding, the results of the present study clearly
showed that subject extraction is easier to comprehend than object extraction.
Evidence for the relative ease of subject extraction was threefold. First, within-
group comparisons of mean accuracy scores for subject and object questions
revealed that the learners were more accurate on subject extraction than on
object extraction. Second, the majority of individual learners obtained higher
scores on subject questions than on object questions, while only a few learners
showed the opposite pattern. Finally, the learners consistently made far more
reversal errors in object wh-questions than in subject wh-questions, indicating
a strong tendency to interpret object wh-questions as subject wh-questions.
This study also showed that this asymmetry in favor of subject questions
was not restricted to advanced learners. The learners at earlier stages of English
acquisition also showed the better performance on subject wh-questions
than on object wh-questions, indicating that they all rely on the same sentence
processing strategies.
Overall, the learners’ performance improved along with the number of years
of English instruction, but at different stages depending on the question type.
While a significant change occurred on subject wh-questions in 11th graders,
a significant change on object wh-questions was observed only in the graduate
group. This suggests that subject wh-questions are easier to improve on than
are object wh-questions, adding another piece of evidence for the relative ease
of subject wh-questions.
Then, why do L2 learners find long-distance subject wh-questions easier to
process than object question counterparts? Given that Korean has no
wh-movement, at least at the surface level, there is no reason to expect
Korean-speaking learners of English to prefer one type of extraction pattern
to the other in English. Besides, since the same pattern of asymmetry was
observed in all the groups, despite differences in the length of instruction and
the type of learning environment (i.e. EFL versus ESL), the preference for the
subject pattern can apparently not be attributed to instruction effects.
Another possibility worth explaining is that the learners in this study have
been exposed to more subject questions than object questions. No corpus
data are available specifically for L2 learning, but a possible indication of the
general practice of native English speakers comes from Stromswold’s (1995)
study of mono-clausal wh-questions in parental speech to children. Based on
an analysis of the CHILDES transcripts of 12 children, Stromswold reports
that adult English speakers asked more subject questions (M = 63.1 per cent,
SD = 21.4 per cent) than object questions. Interestingly, however, there was

no significant correlation between the relative frequency of adult who ques-
tions and their acquisition: children who heard more subject who questions
than object questions did not necessarily acquire them at a younger age, lead-
ing Stromswold to suggest that ‘at best, the frequency in the input only
partially accounts for the order of acquisition of subject and object questions’
(p. 37).
Given the unlikelihood of L1, instruction, grammatical competence, and
arguably input effects, parsing difficulty seems to be the most plausible account
for the observed asymmetry. In particular, the finding of the present study fits
into distance-based accounts of processing complexity. It is widely agreed that
the difficulty of processing filler-gap structures such as wh-questions, relative
clauses, and clefts depends on the distance between the filler and the gap
(Carpenter et al. 1994; Frazier 1998; Hawkins 1999; O’Grady 2001). The
greater distance means more burden on working memory because the filler
has to be held until the parser encounters the gap. As more information is
processed between the filler and its gap, the likelihood increases that the rep-
resentation of the filler will decay and will be displaced from memory (McElree
2000).
The distance can be calculated in various ways. One approach is to measure
the linear distance between the filler and the gap by counting the number of
all the intervening words or alternatively, as put forward by Gibson (2000),
only the number of elements introducing new discourse referents (i.e. NPs and
main verbs). Another approach measures the structural distance, that is, the
depth of embedding of the gap by counting the nodes intervening between the
filler and the gap (Collins 1994; O’Grady 1997). Both linear and structural
approaches yield the same predictions for English wh-questions6—subject
wh-questions should be easier than object wh-questions.
(6) Whoi did the girl think ei pinched the pig? (Subject wh-question)
a. linear distance: four words, two words with discourse referents
b. structural distance: three nodes (S1, VP2, S2)
Who did [S2 the girl [VP2 think [S1 ___ [VP1 pinched the pig]]]?
(7) Whoi did the boy think the cow pinched ei? (Object wh-question)
J.-H. LEE 151
a. linear distance: seven words, four words with discourse referents

b. structural distance: four nodes (VP1, S1, VP2, S2)
Who did [S2 the boy [VP2 think [S1 the cow [VP1 pinched ___]]]]?
As can be seen in (6) and (7), the distance between who and its gap is
both linearly and structurally shorter in subject wh-questions than object
wh-questions. This, I suggest, is why the learners in this study performed
better on subject wh-questions than object wh-questions, although both
types of questions are equally grammatical.
Given that L1 acquisition studies also report the same preference for sub-
ject wh-questions in long-distance extraction (Crain and Thornton 1991, 1998;
Yoshinaga 1996), the above computational explanation offers a unified

account of the development of wh-questions in first and second language
acquisition. It also explains a parallel asymmetry in the L1 and L2 acquisition
of mono-clausal questions (for L1 acquisition, Ervin-Tripp 1970; Tyack and
Ingram 1977; Wilhelm and Hanna 1992; Yoshinaga 1996;7 Seidl and Hollich
2002, for L2 acquisition, Kim 1999; Lam 2001) as well as other gap-containing
structures such as relative clauses and cleft sentences (O’Grady 1999; O’Grady
et al. 2003).
To sum up, the findings of the present study contradict results reported in
the previous studies of adult L2 acquisition (Schachter and Yip 1990; Juffs
and Harrington 1995; White and Juffs 1998; Juffs 2005). One reviewer attri-
butes the discrepancy to different experimental settings. According to the
reviewer, while the stimulus sentences and procedure used in the previous
studies require reanalysis, which is argued to induce more processing difficulty
for subject wh-questions, the experimental setup used in this study eliminates
or greatly reduces the necessity for reanalysis. For instance, this study pre-
sented learners with pictures consisting of two animate agents and their
thought-bubbles. It is possible that these thought-bubbles indicated that
the main verb will take a clausal argument rather than the sentence-initial
wh-word, which circumvents the necessity of subsequent reanalysis. Based on
this, the reviewer suggests that this study, together with the findings from
previous research, shows that when the stimuli necessitate reanalysis, subject
wh-questions are more difficult than object wh-questions, but if the need for
reanalysis is eliminated, subject wh-questions become easier due to the shorter
distance between the filler and its gap. This interpretation, however, is hard
to sustain in light of the fact that Baek (2007) replicated Juffs and Harrington’s
study (1995) using the same experimental setup and stimuli requiring reana-
lysis of the subject gap, and yet found the same asymmetry as the present
study. Therefore, the need for reanalysis cannot account for the different
results of the present study and the previous studies.
The crucial difference of the present study compared with the previous
studies is that this study, as well as Baek’s (2007) study, controlled for various
nonsyntactic factors such as lexical items, animacy of NPs, and referential
properties of NPs, which may have exerted an unwanted influence on the

results in the previous studies in a way favorable to object wh-questions.
Interestingly, under a stricter experimental control, Baek observed that
not only Korean learners of English but also native English speakers performed
better on subject who-questions than on object who-questions. This indicates
that sentence processing is strongly influenced by various nonsyntactic
factors which need to be carefully considered in research on subject–object
asymmetry.
CONCLUSION

This study investigated Korean-speaking learners’ comprehension of
English long-distance wh-questions in relation to a subject–object asymmetry.
The results of the present study clearly show that Korean learners of
English find subject wh-questions easier to interpret than object wh-question,
in line with the processing accounts that hold that the distance between
a wh-word and its gap is a crucial determinant of computational difficulty.
On the other hand, the findings contradict those of previous adult L2
studies that rely on grammaticality judgement tests—a discrepancy that
may be attributable to the influence of various nonsyntactic factors. To identify
the relative roles of all the various factors affecting parsing complexity,
more studies involving carefully controlled stimuli and various measures
are needed.
ACKNOWLEDGEMENTS
I am grateful to William O’Grady and three anonymous reviewers for their comments and
suggestions on this paper. My thanks also go to the participants and to their teachers for
agreeing to participate in this experiment.
Conflict of interest statement. None declared.
APPENDIX: TEST ITEMS

Subject questions
1. Who did the grandmother think pushed the cow?
3. Who did the girl think pinched the pig?
4. Who did the boy say pulled the cat?
6. Who did the grandmother say bit the tiger?
10. Who did the boy believe hit the pig?
Object questions
2. Who did the boy say the cow bit?
5. Who did the dog believe the monkey hit?
J.-H. LEE 153
7. Who did the baby think the cat pushed?

8. Who did the girl say the cow pulled?
9. Who did the boy think the cow pinched?
NOTES
1 While it can be argued that two difficulty in answering object
adjacent tensed verbs in subject wh-questions.
wh-questions also may serve as a cue 3 There were no fillers in the task. As
to the extraction site, their sentence- one reviewer points out, it is possible
medial position is less salient than the that the participants used metalinguis-

sentence-final position of a stranded tic knowledge or conscious strategies.
preposition in object wh-questions. However, given that the task used in
Rather, Juffs (2005) considers two this study is a listening comprehension
adjacent tensed verbs in subject wh- task, not a grammaticality judgement
questions and object-extracted relative task, it is not unnatural that all the
clauses as the locus of processing diffi- test items are wh-questions, which
culty, although this claim does not coincide with the target structure of
sustain given Grodner and Gibson’s the present study. In addition, the
(2005) finding that object-extracted participants were given only 8 s to
relative clauses as in (2) were still answer each question and there were
more difficult to process than subject- only 10 items in total, which may
extracted ones as in (1) even when have allowed little time for metalin-
two verbs were separated by a prepo- guistic analysis of test items. If they
sitional phrase: did use metalinguistic knowledge or
(1) The reporter who sent the photo- conscious strategies, it may have
grapher to the editor hoped for a good been only at the later part of the
story. test. However, participants’ perfor-
(2) The reporter who the photogra- mance was not better on later items
pher sent to the editor hoped for a than earlier items, indicating that the
good story. participants did not succeed in detect-
2 It can be argued that the test materials ing the focus of the present study.
adapted from a child L1 study may 4 To ensure that learners have to inter-
look silly or even awkward to L2 lear- pret not only the embedded clause but
ners who are older and cognitively also the matrix clause is a very impor-
more mature. However, given that tant but nonetheless very often
this study involved four different age ignored requirement in the previous
groups, possibly with different studies on long-distance wh-questions.
English proficiency levels, test materi- For instance, Crain and Thornton
als should be manageable even for the (1991) prompted L1 children to use
youngest and the lowest proficiency the same matrix clause (i.e. what do
group. The materials in this study you think) repeatedly. One problem
seem to be appropriately challenging with using the same matrix clause
to the 9th and 11th grader groups. over and over is that as the elicitation
At the same time, they were not too task proceeds, the child might not
easy for adult learners as shown in the have to pay much attention to the
fact that the college group still had matrix clause because it became
formularized or routinized. If this does 6 The two approaches, however, gener-

happen, the learners’ task is no more ate different predictions in left-
different from producing or processing branching languages. For instance, in
mono-clausal wh-questions. For this relative clauses in Korean, a subject
reason, the information carried by gap is structurally closer to the head
the matrix clause also should contri- than is a direct object gap but is line-
bute to a correct interpretation of the arly more distant. Based on the finding
whole sentence. that English-speaking learners of L2
5 Individual learner data in Table 4 sup- Korean performed better on subject
port the similarity of the 11th grader relative clauses, O’Grady et al. (2003)
group and the college group. When suggest that structural distance is the
individual learners’ asymmetry was principal determinant of processing

examined, a similar distribution of difficulty.
the three asymmetry patterns, that is, 7 But see Stromswold (1995) for the
no asymmetry, a subject preference, opposite direction of asymmetry in
and an object preference, was found L1 acquisition and O’Grady (1999)
in the two groups. for its criticism.
REFERENCES
Baek, S. 2007. ‘Is subject extraction more Behavioral Determinants of Language
difficult than object extraction for second lan- Development. Erlbaum, pp. 321–37.
guage learners of English?’ Unpublished Crain, S. and R. Thornton. 1998. Investigating
manuscript, University of Illinois, Urbana- Universal Grammar: A Guide to Experiments on
Champaign. the Acquisition of Syntax and Semantics. MIT
Bley-Vroman, R., S. Felix, and G. Ioup. 1988. Press.
‘The accessibility of Universal Grammar in Ervin-Tripp, S. 1970. ‘Discourse agreement:
adult language learning,’ Second Language How children answer questions’ in J. Hayes
Research 4: 1–32. (ed.): Cognition and the Development of
Carpenter, P., M. Miyake, and M. Just. 1994. Language. Wiley, pp. 79–107.
‘Working memory constraints in com- Frazier, L. 1998. ‘Getting there (slowly),’ Journal
prehension: Evidence from individual differ- of Psycholinguistic Research 27: 123–46.
ences, aphasia, and aging’ in M. Gernsbacher Frazier, L. and J. D. Fodor. 1978. ‘The sausage
(ed.): Handbook of Psycholinguistics. Academic machine: A new two-stage parsing model,’
Press, pp. 1075–122. Cognition 6: 291–325.
Chomsky, N. 1977. ‘On wh-movement’ Gibson, E. 2000. ‘The dependency locality
in P. Culicover, T. Wasow, and A. Akmajian theory: A distance-based theory of linguistic
(eds): Formal Syntax. Academic Press, complexity’ in Y. Miyashita, A. Marantz, and
pp. 71–132. W. O’Neil (eds): Image, Language, Brain. MIT
Chomsky, N. 1981. Lectures on Government and Press, pp. 95–126.
Binding. Foris. Grodner, D. and E. Gibson. 2005.
Collins, C. 1994. ‘Economy of derivation and ‘Consequences of the serial nature of linguis-
the generalized proper binding condition,’ tics input for sentential complexity,’ Cognitive
Linguistic Inquiry 25: 45–61. Science 29: 261–90.
Crain, S. and R. Thornton. 1991. ‘Recharting Hawkins, R. 1999. ‘Processing complexity and
the course of language acquisition: Studies in filler-gap dependencies across grammars,’
elicited production’ in N. Krasnegor, Language 75: 244–85.
D. Rumbaugh, R. Schiefelbusch, and Johnson, J. and E. Newport. 1991. ‘Critical
M. Studdert-Kennedy (eds): Biological and period effects on universal properties of
J.-H. LEE 155
language: The status of subjacency in the Schachter, J. and V. Yip. 1990. ‘Grammaticality
acquisition of English as a second language,’ judgments: Why does anyone object to subject
Cognition 39: 215–58. extraction?,’ Studies in Second Language
Juffs, A. 2005. ‘The influence of first language Acquisition 12: 379–92.
on the processing of wh-movement in English Seidl, A. and G. Hollich. 2002. ‘Infants’ and
as a second language,’ Second Language Research toddlers’ comprehension of subject and object
21/2: 121–51. wh-questions,’ BULCD Proceedings 26: 596–607.
Juffs, A. and M. Harrington. 1995. ‘Parsing Stromswold, K. 1995. ‘The acquisition of sub-
effects in second language processing: Subject ject and object wh-questions,’ Language
and object asymmetries in wh-extraction,’ Acquisition 4: 5–48.
Studies in Second Language Acquisition 17: Tyack, D. and D. Ingram. 1977. ‘Children’s pro-
483–516. duction and comprehension of questions,’
Kim, S. 1999. ‘The subject-object asymmetry Journal of Child Language 4: 211–224.
in the acquisition of wh-questions by Korean

Traxler, M., R. Morris, and R. Seely. 2002.
learners of English,’ Journal of Pan-Pacific ‘Processing subject and object relative clauses:
Association of Applied Linguistics 3: 1–11. Evidence from eye movements,’ Journal of
Lam, Y. 2001. ‘L1 transfer in child L2 acquisition Memory and Language 47: 69–90.
of English wh-movement: Evidence from a Traxler, M. and M. Pickering. 1996.
Cantonese-speaking child.’ Paper presented ‘Plausibility and the processing of unbounded
in LOT-NWCL student conference, dependencies: An eye-tracking study,’ Journal
Universiteit Utrecht, The Netherlands.
of Memory and Language 35: 454–75.
McElree, B. 2000. ‘Sentence comprehension
Warren, T. and E. Gibson. 2002. ‘The influence
is mediated by content-addressable memory
of referential processing on sentence complex-
structures,’ Journal of Psycholinguistic Research
ity,’ Cognition 85: 79–112.
29: 11–123.
White, L. 1992. ‘Subjacency violations and
McElree, B., S. Foraker, and L. Dyer. 2002.
empty categories in L2 acquisition’
‘Memory structures that subserve sentence
in H. Goodluck and M. Rochemont (eds):
comprehension,’ Journal of Memory and
Island Constraints: Theory, Acquisition, and
Language 48: 67–91.
Processing. Kluwer, pp. 445–464.
O’Grady, W. 1997. Syntactic Development.
White, L and A. Juffs. 1998. ‘Constraints on
University of Chicago Press.
wh-movement in two different contexts of
O’Grady, W. 1999. ‘Toward a new nativism,’
nonnative language acquisition: Competence
Studies in Second Language Acquisition 21:
and processing’ in S. Flynn,
621–33.
O’Grady, W. 2001. ‘An Emergentist Approach G. Martohardjono, and W. O’Neil (eds): The
to Syntax.’ Unpublished manuscript, Generative Study of Second Language Acquisition.
University of Hawaii at Manoa. Erlbaum, pp. 111–29.
O’Grady, W., M. Lee, and M. Choo. 2003. Wilhelm, A. and K. Hanna. 1992. ‘On the
‘A subject-object asymmetry in the acquisition acquisition of wh-questions,’ Calgary Working
of relative clauses in Korean as a second Papers in Linguistics 15: 89–98.
language,’ Studies in Second Language Williams, J., P. Möbius, and C. Kim. 2001.
Acquisition 25: 433–48. ‘Native and non-native processing of English
Pritchett, B. L. 1992. Grammatical Competence wh-questions: Parsing strategies and plausibil-
and Parsing Performance. University of Chicago ity constraints,’ Applied Psycholinguistics 22:
Press. 509–40.
Schachter, J. 1989. ‘Testing a proposed univer- Yoshinaga, N. 1996. ‘Wh-Questions: A
sal’ in S. Gass and J. Schachter (eds): Linguistic Comparative Study of Their Form and
Perspectives on Second Language Acquisition. Acquisition in English and Japanese,’
Cambridge University Press, pp. 73–88. Unpublished doctoral dissertation, University
Schachter, J. 1990. ‘On the issue of complete- of Hawaii at Manoa.
ness in second language acquisition,’ Second
Language Research 6: 93–124.
REVIEWS
F. Christie and J. Martin (eds): LANGUAGE, KNOWLEDGE AND

PEDAGOGY: FUNCTIONAL LINGUISTIC AND SOCIOLOGICAL
PERSPECTIVES. Continuum, 2007.
What is knowledge? How is it organized? How is it distributed? What is the

relationship between language, social structure and the distribution of know-

ledge? And what difference does any of this make to education? Language,
Knowledge and Pedagogy takes on these challenging questions, in the process
seeking to forge a dialogue between Bernstein’s theoretical work in sociology
and Halliday’s theoretical work in linguistics. Putting together two such vast
and complex bodies of work is no mean feat. The contributors to this collec-
tion, generally experts in either Bernstein’s sociology of knowledge or
Halliday’s systemic functional linguistics achieve a commendable degree of
coherence: the volume has the dynamic feel of a deep and genuine dialogue.
Along the way, the dialogue explores the nature of academic disciplines and
the differences between them; the nature of education and its impact on the
distribution of knowledge; and the linguistic instantiation of some of
Bernstein’s theoretical ideas. Reviewing such a book presents several chal-
lenges. Few readers, including this one, will be well-versed in both
Bernstein and Halliday; and few will be familiar with more than a couple of
academic disciplines, so that evaluating some of the claims about the differ-
ences between these disciplines must rely to some degree on academic
common sense. Furthermore, the sheer scope of this ambitious project
makes it impossible to go into the many interesting avenues it presents.
I will, therefore, settle for sharing some of the insights that I found productive,
followed by some reflections on what the book itself suggests about the orga-
nization and distribution of knowledge as a case in point.
The book is organized into two main parts. The first focuses on more theo-
retical issues, such as the hierarchical nature of disciplinary knowledge. Maton
and Muller, in their chapter, explore the question of why Bernstein’s work is
focused on knowledge, or more particularly on the ‘transmission’ of know-
ledge. Part of their answer concerns the idea that knowledge is differently
valued, regulated and distributed. These differences are crucial in the repro-
duction of social structure, including, in particular, the structures of social
class. Furthermore, schools are a key site in which the differentially organized
transmission of knowledge, and hence the reproduction of social structure, is
played out. Hence in mathematics, my own subject background, mathematical
knowledge is highly valued, difficult to get, hierarchical and couched in a
particularly specialized language. And research in mathematics education
REVIEWS 157
suggests that schooling produces social class differences in mathematical

attainment. These class differences are explained in terms of working class
students not having access to the ‘elaborated code’ of school mathematics.
The role of language is apparent in both the organization and specification
of mathematical knowledge, and in learners’ access to this knowledge. And
the role of language is precisely where systemic linguistics has something to
offer, as Martin shows in the next chapter. In particular, he makes a link
between Halliday’s notion of grammatical metaphor and Bernstein’s vertical
discourse. One of the key linguistic features of a vertical discourse like mathe-
matics, is ‘a non-matching relation between grammar and semantics’ (p. 52),
such as through nominalization, in which processes or qualities appear as

nouns. In mathematics, words like factorization, calculation or equivalence
‘thingify’ (Martin’s word—an example of a noun realized as a process)
active processes, making the language of mathematics technical, specialized
and dehumanized. This process of ‘thingification’ (Martin’s nominalization)
also makes mathematical language more susceptible to unequal distribution:
middle class children learn to look behind the grammar at the meaning. The
relationship between horizontal and vertical discourses and grammatical meta-
phor underpins much of the subsequent thinking and analysis in the rest of the
book.
In the remaining chapters of the first part, Muller, Maton and Moore look at
different aspects of the organization of knowledge. Muller argues that insuffi-
cient attention to knowledge structures in school curricula leads to the invi-
sible pedagogies that disadvantage poor children. Maton introduces the idea of
knower structures to complement the more established one of knowledge
structures. He develops the example of a ‘classical education’ as being more
concerned with the cultivation of a certain kind of person than with any
particular knowledge. Furthermore, as his example is designed to show,
knower structures are related to social hierarchy: it is the classically educated
who continue to dominate the British ruling classes. Moore attempts to refute
the dreaded ‘anything goes’ relativism, engaging in an extended critique of the
argument that what makes something art is simply that someone sees it as art.
Moore’s chapter brings to the surface an undercurrent that runs throughout
the book: namely, a need for the possibility of progress, for some kinds of
knowledge to be better than others, for real, solid ground.
The second part of the book offers four discipline-specific analyses. Painter’s
chapter focuses on the language of early childhood. Through a case study of
one child, she demonstrates how most early child language can be character-
ized as horizontal discourse. Drawing on systemic functional linguistics, she
also traces the emergence of features of vertical discourse, such as grammatical
metaphor and the construction of taxonomies, identifying features of the
child’s interaction with his mother through which this emergence arises.
Christie and Macken-Horarik’s chapter is about subject English and the invi-
sible pedagogies they perceive as prevalent in English teaching and curricula.
They build their argument on a critique of prevailing theories of language and
158 REVIEWS
literacy, including cultural studies and new literacy studies. To address the
inequity that results from these invisible pedagogies, they propose the use of
functional linguistics as a ‘meta-language’ with which to compare different
approaches to subject English, and with which to construct alternative curri-
cula. Wignell’s chapter presents an analysis of the development of social
science as a vertical discourse through examination of texts by Hobbes,
Smith, Ricardo and Marx. In her chapter, O’Halloran presents a similar ana-
lysis of the development of mathematical and scientific discourse, highlighting
the different grammars of symbols and images and the process through which
these grammars, combined with that of natural language, combine to create
specialist forms of meaning. She brings her analysis up to date with a discus-

sion of how powerful modern software that allows the creation and manipula-
tion of both images and symbolic expressions is changing the nature of
meaning-making in mathematics and science. The book ends with a valuable
chapter in which the two editors and two of the principal contributors engage
in a written conversation of some of the themes of the book.
The book is valuable and challenging. It helped me to think about much of
my own work—the nature of mathematical discourse, the nature of mathe-
matics teaching and my own assumptions about what is effective in the teach-
ing and learning of mathematics. It made me think, for example, about what is
implicit in current trends towards a more discussion-based approach to mathe-
matics teaching, in which students work collectively to solve problems. Are
some students disadvantaged by this approach? What is the invisible peda-
gogy? In mathematics, unlike in subject English, there is an appeal to logic,
to correctness, and this leaves less room for ideology to enter. Muller argues
‘The condition for a teacher being an authoritative pedagogical agent is, at the
minimum, an internalized map of the conceptual structure of the subject,
acquired through disciplinary training’ (p. 83). In the case of mathematics,
the situation is more complex. While Muller’s condition is necessary, it is
not sufficient. Brilliant mathematicians can often make rather bad mathe-
matics teachers. The mathematical knowledge that teachers use is not, in
fact, the same as the mathematical knowledge that mathematicians use.
Aspects of the book made me uncomfortable. The prevalence of dichotomies
in Bernsteinian thinking is striking (vertical/horizontal; knowledge/knower;
restricted/elaborated; visible/invisible). For me, social life, including in mathe-
matics classrooms, is not so strongly binary; there is not much reflexivity in
this book. Similarly, I was uncomfortable with a discourse of education in
terms of knowledge (a noun) and transmission. Whereas the general challenge
to progressive education as perpetuating invisible pedagogies in new ways is
bracing and thought-provoking, the metaphor of transmission remains proble-
matically rigid and monolithic. A more psychologistic perspective on knowing
and the relationship between knower and known would enrich the dialogue.
Perhaps the sociologists and the linguists could helpfully be joined by socio-
cultural psychologists. Finally, the tone of the book sometimes comes across as
cliquey; the desire for verticality leads the sociologists to want everyone to be a
REVIEWS 159
Bernsteinian, and the linguists to want educationalists to all learn systemic

functional linguistics. In schooling, Maton and Muller advocate teaching ‘the
key to the code enabling success’ (p. 17); if, perhaps using systemic functional
linguistics, we can identify how language contributes to the unequal distribu-
tion of knowledge, and then we teach underprivileged students how to crack
this code, inequality will be reduced. If only it were that simple! My feeling is
that, even if these things came to pass, invisible pedagogies would remain.
Reviewed by Richard Barwell

University of Ottawa, Canada
doi:10.1093/applin/amp054 Advance Access published on 21 December 2009

S. Makoni and A. Pennycook (eds): DISINVENTING AND
RECONSTITUTING LANGUAGES. Multilingual Matters, 2007.
In his trenchant critique of structural linguistics, Bourdieu (1991) observes

that linguistics as an academic discipline has been far too pre-occupied with
idealist, abstracted approaches to the study of language—examining language
in isolation from the social and political conditions in which it is used. As he
comments ironically of this process: ‘bracketing out the social [. . .] allows lan-
guage or any other symbolic object to be treated like an end in itself, [this]
contributed considerably to the success of structural linguistics, for it endowed
the ‘‘pure’’ exercises that characterise a purely internal and formal analysis
with the charm of a game devoid of consequences’ (1991: 34). Or as Mey
observes, ‘linguistic models, no matter how innocent and theoretical they
may seem to be, not only have distinct economical, social and political pre-
suppositions, but also consequences [. . .] Linguistic (and other) inequalities
don’t cease to exist simply because their [. . .] causes are swept under the lin-
guistic rug’ (1985: 26).
Makoni and Pennycook would surely agree, given the central concerns of
their important and provocative edited collection. The editors argue convin-
cingly in their introduction that the idea of separate, enumerable ‘languages’,
arising out of structural linguistics, is a fiction or ‘invention’, constructed out of
the socio-historical and socio-political processes of nationalism and/or colonial-
ism. They proceed to outline how this process of invention operates, not only in
relation to the apparent abstract, formalized and distinct attributes of individual
‘languages’ but also the attendant ideology, or meta-discursive regime as they
term it, used to create/describe/delineate them. The latter ‘treats languages as
countable institutions, a view reinforced by the existence of grammars and
REVIEWS 159
Bernsteinian, and the linguists to want educationalists to all learn systemic

functional linguistics. In schooling, Maton and Muller advocate teaching ‘the
key to the code enabling success’ (p. 17); if, perhaps using systemic functional
linguistics, we can identify how language contributes to the unequal distribu-
tion of knowledge, and then we teach underprivileged students how to crack
this code, inequality will be reduced. If only it were that simple! My feeling is
that, even if these things came to pass, invisible pedagogies would remain.
Reviewed by Richard Barwell

University of Ottawa, Canada

S. Makoni and A. Pennycook (eds): DISINVENTING AND
RECONSTITUTING LANGUAGES. Multilingual Matters, 2007.
In his trenchant critique of structural linguistics, Bourdieu (1991) observes

that linguistics as an academic discipline has been far too pre-occupied with
idealist, abstracted approaches to the study of language—examining language
in isolation from the social and political conditions in which it is used. As he
comments ironically of this process: ‘bracketing out the social [. . .] allows lan-
guage or any other symbolic object to be treated like an end in itself, [this]
contributed considerably to the success of structural linguistics, for it endowed
the ‘‘pure’’ exercises that characterise a purely internal and formal analysis
with the charm of a game devoid of consequences’ (1991: 34). Or as Mey
observes, ‘linguistic models, no matter how innocent and theoretical they
may seem to be, not only have distinct economical, social and political pre-
suppositions, but also consequences [. . .] Linguistic (and other) inequalities
don’t cease to exist simply because their [. . .] causes are swept under the lin-
guistic rug’ (1985: 26).
Makoni and Pennycook would surely agree, given the central concerns of
their important and provocative edited collection. The editors argue convin-
cingly in their introduction that the idea of separate, enumerable ‘languages’,
arising out of structural linguistics, is a fiction or ‘invention’, constructed out of
the socio-historical and socio-political processes of nationalism and/or colonial-
ism. They proceed to outline how this process of invention operates, not only in
relation to the apparent abstract, formalized and distinct attributes of individual
‘languages’ but also the attendant ideology, or meta-discursive regime as they
term it, used to create/describe/delineate them. The latter ‘treats languages as
countable institutions, a view reinforced by the existence of grammars and
160 REVIEWS
dictionaries’ (p. 2), and one that is still widely employed in contemporary
national and/or post-colonial contexts. In the process, the much more complex
and fluid reality of multilingual speakers ‘on the ground’—heteroglossia, in a
Bakhtinian sense—is overlooked, even actively disguised/distorted.
Situating the construction/invention of languages within this wider political
context points to the second key concern of Makoni and Pennycook’s volume:
the very real material, often highly deleterious, consequences of this process of
language invention for individual multilingual speakers. Returning to
Bourdieu, such processes are clearly not a game devoid of consequences.
Rather, the construction of discrete, standardized languages, and related lan-
guage hierarchies, actively delimits the choices of speakers and undermines

their more variegated, fluid linguistic repertoires. For Makoni and Pennycook,
this includes not only national and colonial languages, linked to the formation
of nation-states, and English as the current lingua mundi, or world language,
linked to globalization. It also includes, more controversially, the (often recur-
sive) reconstruction of the so-called indigenous languages in colonial and
post-colonial contexts. As they argue, ‘these constructed [indigenous] lan-
guages were administratively assigned to colonized populations as mother
tongues and went on to form the basis of so-called mother tongue education
and vernacular literacy’ (p. 14, emphasis in original). Indeed, they conclude,
‘[e]pistemologically, one of the key rhetorical moves of colonialism was to
foster, then mask, the artificiality of indigenous languages [. . .], presenting
them as if they were a natural part of local contexts’ (p. 15).
From this, Makoni and Pennycook advocate for the active ‘disinvention’ of
all languages. In so doing, they give short shrift not only to structural linguis-
tics but also to pluralist approaches to socio-linguistics and related language
rights arguments in support of indigenous and other minoritized languages.
They argue that the latter strategies merely reproduce ‘the tropes of colonial
invention, overlooking the contested history of language inventions, and
ignoring the ‘collateral damage’ [. . .] that their embedded notions of language
may be perpetuating’ (p. 16). More controversially still, they suggest that these
strategies ‘continue to do damage to speech communities and deny those
[indigenous] people educational opportunities’ (p. 21).
While I do not agree with the latter arguments, for reasons I will return to
shortly, one of the key strengths of this volume is its sophisticated exploration
of a critical linguistics that does not reduce itself to the banality of ludic
post-modernism, where everything is relative and nothing is real. Rather,
each chapter highlights the contingent and constructed processes of language
formation, alongside a recognition that, once formed, particular language com-
munities reflect real communities of culture and power—a Durkheimian
‘social fact’, in effect—which clearly privilege some and disempower others.
Thus chapters by Heryanto, and Busch and Schick, explore the dialectic
between the emergence of standardized, written language varieties and the
role of nationalism and nation building, along with their social and political
effects, in relation to Indonesia, and Bosnia-Herzegovina, respectively.
REVIEWS 161
Pennycook examines the myths that pervade the role and influence of (stan-
dard) English as the current ‘international language’. Rather than achieving
social and economic mobility via learning English, as is often claimed: ‘[t]his
thing called English colludes with many of the pernicious processes of globa-
lisation, deludes many learners through the false promises it holds out for
social and material gain, and excludes many people by operating as an exclu-
sionary class dialect, favouring particular people, countries, cultures and forms
of knowledge’ (p. 101).
A similar scepticism is apparent in the chapters by Makoni and Mashiri, on
Africa, and de Souza, on Brazil, although their focus is primarily on so-called
indigenous languages. Both argue that the construction of standardized indig-

enous languages inevitably undermines and/or devalues the much wider lin-
guistic repertoires, particularly oral repertoires, of multilingual speakers.
Branson and Miller provide a comparable perspective in their fascinating dis-
cussion of sign languages and how they have been viewed and juxtaposed,
almost always negatively, in relation to spoken language varieties. Similarly,
Richardson examines the performative dimensions of hiphop/rap discourse, as
well as its links to resistance, the latter arising from its close historical and
political connections to African American Vernacular Discourses.
Thorne and Lantolf explore the implications of these wider arguments for
linguistic analysis itself and argue convincingly for a reconstructed approach
that dispenses with the language as form/communicative practices disjuncture
that still pervades structural linguistics. Rather, drawing on a Vygotskyian per-
spective, they call for a linguistics of communicative activity, an approach to
linguistic analysis based on ‘usage-based and meaning-centered characteriza-
tions of linguistically-mediated human activity’ (p. 171). Canagarajah con-
cludes the volume with some brief reflections on what a ‘disinvented’
approach to languages might subsequently entail, arguing that we must look
beyond modernism, which has been so central to constructing the various
binaries this volume sets itself against, and return to an analysis of
pre-colonial/pre-modern societies, ‘contexts of rampant multingualism and
inveterate hybridity’ (p. 238), which were subsequently suppressed by the
modernist demands of categorization, classification and codification. Again,
the methodological consequence of this is ‘that, rather than focusing on
rules and conventions [à la structural linguistics], we have to focus on strate-
gies of communication’ (p. 237).
But, interestingly, there is one area where an unnecessary binary persists in
this volume—that between indigenous language varieties and modernity.
Makoni and Pennycook argue (as do Makoni and Mashiri, even more tren-
chantly) that the construction of indigenous languages has had deleterious
effects comparable with those of the construction of national and/or colonial
languages: ‘the invention of languages has had particularly insidious conse-
quences for indigenous people, since the invention of the construct of
162 REVIEWS
indigenous peoples [. . .] produces [. . .] a need for identification with their

prenatural selves’ (p. 23). It is on the basis of this argument that they subse-
quently dismiss more pluralist socio-linguistic approaches to language policy
and related language rights arguments.
There are two problems with this position. The first has to do with social and
political context. Unlike their convincing analyses elsewhere, it is simply inac-
curate to suggest that the construction of indigenous languages is comparable
with that of national and/or colonial languages in terms of their (subsequent)
deleterious social and political effects. Rather, it is the latter which have invari-
ably delimited, pathologized and/or institutionally excluded the former, via
the historical processes of colonization, most usually. If indigenous language

varieties have been reconstituted in postcolonial contexts, as new languages of
power, it is almost always in relation to re-entering the public domain, usually
alongside other language varieties, and not as a primary basis for social exclu-
sion. It may be ‘the pluralisation of monolingualism’ (p. 22), but it is surely
better than nothing.
The second key issue that remains unresolved in relation to indigenous
language varieties is, ironically, the ongoing privileging of modernism in
their arguments. By assuming that indigenous language varieties remain
delimited to pre-modern, ‘pre-natural’ identities, fixing indigenous identities
and ‘consequently disqualifying socially embedded urbanized indigenous peo-
ples’ (p. 23) and related ‘urban vernaculars in Africa’ (p. 26), Makoni and
Pennycook end up reinforcing rather than, as elsewhere, deconstructing the
pre-modernity/modernity binary. The implication here seems clear—out with
the old (primordial/essentialist) indigenous language varieties, in with the new
(hybrid/fluid/dynamic) urban vernaculars.
And yet, this misses a key point. All cultural and linguistic identities, includ-
ing indigenous ones, inevitably change over time and no one is suggesting that
such language varieties are, or need be, frozen in time—indeed, they cannot.
The question is again rather a political one. Who has the locus of control to
influence language change—group members themselves, or other, more pow-
erful, cultural and linguistic groups? This is what language rights advocates,
whom Makoni and Pennycook too readily dismiss, highlight.
As the political theorist, Will Kymlicka, has observed, arguments for greater
minority rights (in this case, in relation to language varieties and their speak-
ers) are most often not based on some simplistic desire for cultural ‘purity’ or
for preserving an ‘authentic’ culture. Instead, at issue is the right ‘to maintain
one’s membership in a distinct culture, and to continue developing that cul-
ture [and associated language varieties] in the same (impure) way that the
members of majority cultures are able to develop theirs’ (1995: 105). Cultural
and linguistic change, adaptation and interaction are entirely consistent with
such a position, even if Makoni and Pennycook appear not to recognize this in
the context of indigenous language varieties.
REVIEWS 163
Be that as it may, this is an important, provocative, eloquent and timely

volume. It moves forward debates in critical linguistics and language policy in
significant ways and will shape discussions in these areas for some considerable
time to come.
Reviewed by Stephen May

University of Auckland, New Zealand
REFERENCES

Bourdieu, P. 1991. Language and Symbolic Power. Mey, J. 1985. Whose Language? A Study in
Polity Press. Linguistic Pragmatics. John Benjamins.
Kymlicka, W. 1995. Multicultural Citizenship: A
Liberal Theory of Minority Rights. Clarendon Press.
Alison Wray: FORMULAIC LANGUAGE: PUSHING THE BOUNDARIES.

Oxford University Press, 2008.
Formulaic Language: Pushing the Boundaries is the follow-up to Wray’s influential

2002 book, Formulaic Language and the Lexicon. The volume develops three
main ideas from the earlier work: (i) the mental lexicon comprises semanti-
cally unanalysed units of different sizes, from morphemes to fixed phrases,
including both fully fixed chunks and more flexible units permitting internal
variation; (ii) native speakers only analyse chunks of language into smaller
units if this is necessary to unpack its meaning or if they encounter paradig-
matic variations in the form; (iii) speakers use formulaic language to promote
their own interests by tightly constraining hearers’ interpretations of their
utterances.
Wray covers a wide range of topics in developing these ideas: from machine
translation to bugle calls via language learning and actors’ script memorization.
What unites these discussions is the overarching metaphor of ‘pushing the
boundaries’. In one sense, this refers to the nature of formulaic language as a
phenomenon that breaks down the boundaries between traditional linguistic
categories. In another, it refers to the need for cross-fertilization between areas
of investigation that have often been isolated from each other. Above all, how-
ever, the boundary metaphor refers to Wray’s strategy of investigating marginal
and extreme cases. Discussion is focused on what happens to communication in
situations where language is forced to operate ‘beyond its normal scope’ (p. 5),
and on the points where formulaicity crosses over into creativity, speech into
writing and linguistic into non-linguistic communication.
REVIEWS 163
Be that as it may, this is an important, provocative, eloquent and timely

volume. It moves forward debates in critical linguistics and language policy in
significant ways and will shape discussions in these areas for some considerable
time to come.
Reviewed by Stephen May

University of Auckland, New Zealand
REFERENCES

Bourdieu, P. 1991. Language and Symbolic Power. Mey, J. 1985. Whose Language? A Study in
Polity Press. Linguistic Pragmatics. John Benjamins.
Kymlicka, W. 1995. Multicultural Citizenship: A
Liberal Theory of Minority Rights. Clarendon Press.
Alison Wray: FORMULAIC LANGUAGE: PUSHING THE BOUNDARIES.

Oxford University Press, 2008.
Formulaic Language: Pushing the Boundaries is the follow-up to Wray’s influential

2002 book, Formulaic Language and the Lexicon. The volume develops three
main ideas from the earlier work: (i) the mental lexicon comprises semanti-
cally unanalysed units of different sizes, from morphemes to fixed phrases,
including both fully fixed chunks and more flexible units permitting internal
variation; (ii) native speakers only analyse chunks of language into smaller
units if this is necessary to unpack its meaning or if they encounter paradig-
matic variations in the form; (iii) speakers use formulaic language to promote
their own interests by tightly constraining hearers’ interpretations of their
utterances.
Wray covers a wide range of topics in developing these ideas: from machine
translation to bugle calls via language learning and actors’ script memorization.
What unites these discussions is the overarching metaphor of ‘pushing the
boundaries’. In one sense, this refers to the nature of formulaic language as a
phenomenon that breaks down the boundaries between traditional linguistic
categories. In another, it refers to the need for cross-fertilization between areas
of investigation that have often been isolated from each other. Above all, how-
ever, the boundary metaphor refers to Wray’s strategy of investigating marginal
and extreme cases. Discussion is focused on what happens to communication in
situations where language is forced to operate ‘beyond its normal scope’ (p. 5),
and on the points where formulaicity crosses over into creativity, speech into
writing and linguistic into non-linguistic communication.
164 REVIEWS
The book is divided into four sections. The first gives a general characteriza-
tion of formulaic language and summarizes the main claims from Wray (2002).
It also includes a review of the literature on the psycholinguistic processing of
formulae, a chapter discussing the role of formulaicity in oral epic poetry and
in the historical transition to literacy, and another discussing the differences
between speech and writing in terms of their use of formulaic language. These
discussions lead to a model of the ways in which the situational autonomy of a
text influences its level of formulaicity. The book’s second section starts by
setting out desiderata for a theory of formulaic language. Wray argues that a
comprehensive model needs to encompass and show the relationships
between three facets of formulaicity: description of what patterns exist; sys-

tematization of the (grammatical/semantic) principles underlying those pat-
terns; and patterns’ (social/psychological/biological) causes. She takes steps
towards such a theory by considering the extent to which various linguistic
approaches to the description and systematization of patterns might comple-
ment her earlier ‘cause’-oriented models. The second section also deals with
the issue of identifying formulaicity. After reviewing existing methods, Wray
describes what she calls a ‘diagnostic approach’ to formula identification, the
aim of which is not to determine whether particular stretches of language are
formulaic or not, but rather to raise researchers’ awareness of their own rea-
sons for thinking that particular stretches of language are formulaic.
Section three is a collection of short chapters, each describing a (previously
published) study into an aspect of formulaic language. The first looks at for-
mulaicity in machine translation, discussing a pilot system for translating
counter clerks’ spoken English into computer-generated sign-language for
deaf customers. The second discusses the use of pre-programmed formulas in
a speech generation program used by a woman with cerebral palsy. The third
and fourth chapters are concerned with foreign language learning. The former
describes a case study of a novice attempting in four days to learn through
formula memorization enough Welsh to deliver a cookery demonstration.
The latter describes an attempt to provide intermediate learners with formulaic
material to carry out an anticipated conversation. The fifth chapter relates a
court case in which the verdict rested on the interpretation of a particular piece
of formulaic language. The final chapter looks at actors’ memorization of
scripts and uses the case study of a TV comedy sketch to consider when
and why precise recall is required and what the effects of deviation from
form are.
The fourth section develops the book’s three main theses through discussion
of some key issues related to formulaic language. In the first chapter, Wray
argues for the idea that formulaic language is the ‘default option’ in language
production; that is, that language is handled holistically unless there is a spe-
cific reason for engaging in analysis. In the second chapter, she discusses what
determines the level of formulaicity in language. This question is addressed
from three different perspectives: how and why formulaicity might have
emerged in the early stages of the evolution of the human language capacity;
REVIEWS 165
the relationship between the amount of formulaicity in a language and the

amount of contact its speakers have with outsiders; and the effects on indivi-
dual communicative acts when different interlocutors have different levels of
formulaic knowledge or when they make deliberate choices to reject or adopt
the use of particular formulae. The third issue is the place of formulaic lan-
guage in adult second language learning. Situational differences between adult
and child learners that might make formula-based learning problematic for
adults are reviewed and the limitations of formula-based learning are dis-
cussed. Wray then suggests a possible teaching approach in which adults are
‘front-loaded’ with a suitably selected range of form-meaning pairings that
would both enable the learner to engage in interactions that would provide

further opportunities of learning and constrain their generalizations to native-
like patterns.
These themes are pursued further in the following chapter, which concerns
‘teaching’ language to computers. Wray argues that some of the persistent
problems in natural language processing and machine translation are similar
to those faced in adult learning. She believes that in both cases the difficulties
are rooted in an over-emphasis on equipping learners/computers with mini-
mal units and highly productive rule systems. She proposes instead an
approach that parallels that which she takes child learners to follow, with
computers using input as the basis for noticing recurrent sequences and pat-
terns, but engaging only in the minimum analysis necessary to derive mean-
ing, so that paradigmatic rules are not generalized beyond the variation
actually attested in the input.
The final issue concerns the extent to which formulaic language constrains
what we say and think. After discussing the trade-off between the advantages
of using memorized forms—for example, for maintaining fluency, idiomaticity,
or accuracy—and the costs, in terms of the efforts spent in memorization, and
of the loss of ownership over one’s utterances, and after briefly describing
work on the use of formulaic phrases to ensure doctrinal conformity during
China’s Cultural Revolution, Wray turns to the question of the extent to which
various intrinsically formulaic communication systems—a hypothetical holis-
tic protolanguage, the flag signals used in motor racing and at sea, and military
bugle calls—constrain thought and action, and the extent to which these sys-
tems are open to creative elaborations. She concludes that, since ideas need to
be unpacked and assimilated before they can be applied, formulaicity is unli-
kely ever to effectively constrain thought; however, in the extreme examples
discussed, it certainly can, and does, place constraints on creativity of
expression.
It will be clear from this review that Pushing the Boundaries covers a prodi-
gious range of topics. One slightly frustrating consequence of this breadth is
that discussion of individual issues is often rather brief; another is that chapters
sometimes seem to lack internal coherence. However, in an area where
researchers from different specializations have often been isolated from each
other, the ‘cross-border’ perspectives that such a broad sweep opens up, and
166 REVIEWS
the consequent possibilities for inter-disciplinary dialogue more than make up

for these niggles. The second language specialist, for example, is likely to find
new sources of insight here in the commonalities that are highlighted between
the task of the foreign language teacher and that of the computational linguist,
between the choices for self-expression faced by the language learner and by
individuals with language disorders, and between the issues facing students
learning idiomatic phrases and actors learning their lines.
Another notable strength of the book is its simultaneous commitment to
both application and theory. On the one hand, numerous concrete suggestions
are made for practical applications in language learning, natural language
processing, language therapy and more. On the other hand, serious attempts

are made to fit formulaic language into a broader conceptual framework.
Given the current under-theorization of much work in this area, this commit-
ment is very welcome. Particularly valuable in this regard is the series of
models which the book articulates. In many cases, data are rather thin on
the ground and these models may therefore be rather too speculative to con-
vince sceptics. However, by formulating her ‘best guess’ at what is happening
on the basis of the evidence available, Wray both provides us with a series of
clearly stated theses that future work will be able to interrogate and elaborate,
and highlights what is currently missing in the evidence base. It seems likely
that this will make Pushing the Boundaries a fruitful stimulus for research and
debate for several years to come.
Reviewed by Philip Durrant

Bilkent University, Ankara, Turkey
Ruth Wodak, Rudolf de Cillia, Martin Reisigl and Karin Liebhart: THE
DISCURSIVE CONSTRUCTION OF NATIONAL IDENTITY. SECOND
EDITION. Transl. by Angelika Hirsch, Richard Mitten and J. W. Unger.
Edinburgh University Press, 2009.
Writing a review of the second edition of The Discursive Construction of National

Identity has been an extremely difficult task. How to write an evaluation of an
iconic volume, praised already everywhere and by everyone in the academe?
Just one look at online book sellers brings ample examples of tributes, admi-
rations and recommendations of Wodak, de Cillia, Reisigl and Liebhart’s work.
A publishing history of the volume also speaks a lot about its credentials.
At first it was a German edition by Suhrkamp in 1998; this was followed by
an abbreviated and translated text for its first English edition by Edinburgh
University Press in 1999, reprinted in 2003 and transferred to digital print in
166 REVIEWS
the consequent possibilities for inter-disciplinary dialogue more than make up

for these niggles. The second language specialist, for example, is likely to find
new sources of insight here in the commonalities that are highlighted between
the task of the foreign language teacher and that of the computational linguist,
between the choices for self-expression faced by the language learner and by
individuals with language disorders, and between the issues facing students
learning idiomatic phrases and actors learning their lines.
Another notable strength of the book is its simultaneous commitment to
both application and theory. On the one hand, numerous concrete suggestions
are made for practical applications in language learning, natural language
processing, language therapy and more. On the other hand, serious attempts

are made to fit formulaic language into a broader conceptual framework.
Given the current under-theorization of much work in this area, this commit-
ment is very welcome. Particularly valuable in this regard is the series of
models which the book articulates. In many cases, data are rather thin on
the ground and these models may therefore be rather too speculative to con-
vince sceptics. However, by formulating her ‘best guess’ at what is happening
on the basis of the evidence available, Wray both provides us with a series of
clearly stated theses that future work will be able to interrogate and elaborate,
and highlights what is currently missing in the evidence base. It seems likely
that this will make Pushing the Boundaries a fruitful stimulus for research and
debate for several years to come.
Reviewed by Philip Durrant

Bilkent University, Ankara, Turkey
Ruth Wodak, Rudolf de Cillia, Martin Reisigl and Karin Liebhart: THE
DISCURSIVE CONSTRUCTION OF NATIONAL IDENTITY. SECOND
EDITION. Transl. by Angelika Hirsch, Richard Mitten and J. W. Unger.
Edinburgh University Press, 2009.
Writing a review of the second edition of The Discursive Construction of National

Identity has been an extremely difficult task. How to write an evaluation of an
iconic volume, praised already everywhere and by everyone in the academe?
Just one look at online book sellers brings ample examples of tributes, admi-
rations and recommendations of Wodak, de Cillia, Reisigl and Liebhart’s work.
A publishing history of the volume also speaks a lot about its credentials.
At first it was a German edition by Suhrkamp in 1998; this was followed by
an abbreviated and translated text for its first English edition by Edinburgh
University Press in 1999, reprinted in 2003 and transferred to digital print in
REVIEWS 167
2005. And finally an extended and revised 2009 second edition shows how
important this book was, is and almost certainly will remain for academics and
students of many disciplines within the broadly defined social sciences.
It is after all a book that inspires many. Its ever interesting topic of nation
and identity, focus on sameness and uniqueness in contrast to other work
within this field which prefers to concentrate on otherness, its stimulating
deployment of other important concepts from related disciplines, such as ‘ima-
gined community’ and habitus, to the study of discourse makes this book
widely appreciated and well regarded. By the same token, vigorous method-
ology of a discourse-historical approach developed in the Vienna School of
Discourse Analysis, immaculate examination of multi-genred data (i.e. com-

memorative speeches, media texts, focus group interviews, qualitative inter-
views) showing both discursive strategies and argumentation schemes but also
forms of realization, richness of linguistic extracts to illustrate authors’ point,
all these features turned this book into an exceptional example of a distin-
guished research monograph. The authors choose to focus on the discursive
data that was harvested around the celebratory and commemorative events of
1995, the year of the 50th anniversary of the Second Republic, also the year
of Austria’s accession to the European Union. Thematically, they concentrate
on topics of common past, present and future, common culture and territory as
well as on the concept of homo nationalis.
Apart from a dazzling red cover and an additional translator (J. Unger),
readers of the second edition will find one new chapter titled ‘The ‘‘Story’’
Continues: 1995–2008’ which brings the investigation up to date with a com-
parative analysis of similar discursive events, including the commemorative
celebration of 2005 in Austria. Austria’s second EU presidency in 2006 is also
examined, and so is the Austrian and European political crisis following a rise
of national populism in the country. The methodological principle of the
extended case method (Burawoy 2009) is deployed here. Clearly, this discur-
sive variation of a ‘focused revisit’ is organized around two aspects: ‘internal
processes within the field site itself and forces external to the field site’
(Burawoy 2003: 645). In their analysis, the authors take into consideration
significant socio-political developments in Austria, such as the impact of the
EU membership.
Having a high opinion of the book, I also wish to express some criticism. The
authors argue that ‘there is no such thing as one national identity in an essen-
tialist sense, but rather that different identities are discursively constructed
according to context, that is, to the degree of public exposure of a given utter-
ance, the setting, the topic addressed, and so on’ (pp. 186–7). And they con-
tinue: ‘Discourses in different political and media publics stand in reciprocal
relation to one another in much the same way as do discourses of powerful
elites and ‘‘everyday discourses’’. We have described the details of such dis-
cursive ‘‘exchanges’’ using the concepts of intertextuality and recontextualisa-
tion’ (p. 187). Indeed, while investigating developments in the discursive
construction of Austrian national identity, researchers do pay particular
168 REVIEWS
attention to political processes that influence and shape the discourses under
investigation. This seems to be slightly narrowing the method of ‘revisiting’ the
field. After all, the Vienna School sees discourse as inter alia ‘integrating var-
ious different positions and voices’ (Wodak 2009: 39); they postulate the need
to look closely into different contexts, one of which is ‘the extra-linguistic
social/sociological variable’ (Wodak 2009: 38). One can wonder then why
other types of data were not taken into consideration. I think here mainly of
the broad range of data from semi-public/semi-private settings such as internet
discussions. This is in my opinion an extremely interesting and broadening
source of texts which may reveal interesting aspects of the researched issue of
national identity. It is the expansion of Internet which has already transformed

and continues to transform our modes of communication, including areas of
activity in which we express opinion. It seems to me that, for example, readers’
and viewers’ letters have often been replaced by e-mails to the newspapers and
broadcasters. I would like to see some examples of this social and indeed dis-
cursive change incorporated into the extended edition of such a remarkable
book.
However, although I point at some minor shortcomings, I strongly recom-
mend this monograph to everyone who is interested in the topics of identity,
nation, politics and history. It is also worth mentioning some of the other
books that will be an interesting companion reading to the volume reviewed
here, thematically (e.g. Martin and Wodak 2003; Heer et al. 2008; Wodak
2009) and methodologically (e.g. Wodak and Meyer 2001; Weiss and Wodak
2003; Wodak and Krzy_z anowski 2008). Whether it is assessed together with its
companion volumes or as a solo act, The Discursive Construction of National
Identity continues to impress, inspire, encourage and motivate both young
and more experienced researchers of discourse and society.
Reviewed by Aleksandra Galasińska

University of Wolverhampton, UK
REFERENCES
Burawoy, M. 2003. ‘Revisits: an outline of a Weiss, G. and R. Wodak (eds). 2003. Critical
theory of reflexive ethnography,’ American Discourse Analysis. Theory and Interdisciplinarity.
Sociological Review 68/5: 645–79. Palgrave Macmillan.
Burawoy, M. 2009. The Extended Case Wodak, R. 2009. The Discourse of Politics in Action.
MethodUniversity of California Press. Politics as Usual. Palgrave Macmillan.
Heer, H., W. Manoschek, A. Pollak and Wodak, R. and M. Krzy_z anowski (eds). 2008.
R. Wodak (eds). 2008. The Discursive Construction Qualitative Discourse Analysis in the Social Sciences.
of History: Remembering the Wehrmacht’s War of Palgrave Macmillan.
Annihilation. Palgrave Macmillan. Wodak, R. and M. Meyer (eds). 2001 Methods
Martin, J. R. and R. Wodak (eds). 2003. of Critical Discourse Analysis. Sage.
Re/Reading the Past: Critical and Functional
Perspectives on Time and Value. John Benjamins.
doi:10.1093/applin/amq002
NOTES ON CONTRIBUTORS
Richard Barwell is an Associate Professor at the Faculty of Education, University

of Ottawa, Canada. His research is located in the intersection of applied lin-
guistics and mathematics education, with a particular focus on multilingual-
ism/bilingualism in the teaching and learning of mathematics. His research
interests include mathematics classroom discourse, mathematics learning
in multilingual settings, and the relationship between learning language
and learning curriculum content. His work has been published in peer-
reviewed journals in Applied Linguistics, Mathematics Education, and General

Education. Address for correspondence: Faculty of Education, University
of Ottawa, 145 J-J Lussier, Ottawa K1N 6N5, Canada.
<richard.barwell@uottawa.ca>
Philip Durrant is currently a Visiting Assistant Professor in the Graduate School

of Education at Bilkent University. His research interests include corpus lin-
guistics, English for Academic Purposes, and all aspects of formulaic language.
Address for correspondence: Graduate School of Education, G Binasi No. 259,
Bilkent University, 06800, Ankara, Turkey. <durrant.phil@googlemail.com>
Aleksandra Galasińska is a Senior Research Fellow in European Studies at the

University of Wolverhampton. Her main research interests focus on ethno-
graphic and discursive aspects of lived experience of post-communism and
post-enlargement migration, with journal publications in Narrative Inquiry,
Multilingua, Journal of Ethnic and Migration Studies, Ethnicities, Discourse &
Society, and Journal of Multicultural Discourses. She is co-editor of Discourse and
Transformation in Central and Eastern Europe (with Michal Krzy_zanowski,
Palgrave Macmillan, 2009). Currently, she is working on the edited
volume devoted to Polish discourses of transformation (with D. Galasinski,
Amsterdam/Philadelphia: John Benjamins, DAPSAC series, forthcoming).
Address for correspondence: School of Law, Social Sciences and
Communications, University of Wolverhampton, Wulfruna Street,
Wolverhampton WV1 1LY, UK. <a.galasinska@wlv.ac.uk>
Noriko Iwasaki is a Lecturer in Language Pedagogy in the Department of

Linguistics at the School of Oriental and African Studies (SOAS), University
of London. She received a PhD in Second Language Acquisition and Teaching
at the University of Arizona. Her research interests include study abroad
and language production processes. Recent publications include: ‘Assessing
progress towards the advanced level Japanese after a year abroad: Focus
on individual participants’ in Japanese Language and Literature and ‘Processes
in L2 Japanese sentence production’ in The Handbook of East Asian
Psycholinguistics (Cambridge University Press). Address for correspondence:
Department of Linguistics, School of Oriental and African Studies,
170 NOTES ON CONTRIBUTORS
University of London, Thornhaugh Street, Russell Square, London WC1H

0XG, UK. <ni3@soas.ac.uk>
David Cassels Johnson is an Assistant Professor of education at Washington

State University. His research interests revolve around the interaction between
language policy and educational opportunity. The study upon which this
article is based won the 2008 National Association of Bilingual Education
dissertation competition. <johnsondc@wsu.edu>
An Kuppens is a teaching assistant and a PhD candidate at the Department

of Communication Studies at the University of Antwerp. Her main research
interests are on the intersection between cultural globalization, the media,

and English. Recent publications include an article on African-American
and Jamaican English in subcultural niche media in Journal of Communication
Inquiry. Address for correspondence: Department of Communication Studies,
University of Antwerp, Sint-Jacobstraat 2, B-2000 Antwerp, Belgium.
<an.kuppens@ua.ac.be>
Jin-Hwa Lee earned a PhD in second language acquisition from the University
of Hawai’i in 2006 and is currently working in the Department of English
Education at Chung-Ang University, Seoul, Korea. Her main research interests
include second language acquisition and instruction, corpus-based L2 analysis
and teaching, task-based language teaching, and L2 programme development.
Address for correspondence: Department of English Education, 221 Heukseok-
dong, Dongjak-gu, Seoul 156-756, Korea. <jinhlee@cau.ac.kr>
Stephen May is a Professor of Education in the School of Critical Studies in

Education, Faculty of Education, University of Auckland, New Zealand. He
is the author of the award-winning Language and Minority Rights: Ethnicity,
Nationalism and the Politics of Language (Routledge, 2008). He is the founding
editor of the journal Ethnicities (Sage) and Associate Editor of Language
Policy (Springer). Address for correspondence: School of Critical Studies in
Education (CRSTIE), Faculty of Education, The University of Auckland,
Private Bag 92601, Symonds Street, Auckland 1150, New Zealand.
<s.may@auckland.ac.nz>
Leendert Plug is a Lecturer in the Department of Linguistics and Phonetics,

University of Leeds. He was previously Postdoctoral Research Associate at
the University of Sheffield (on the project ‘Listening to people with seizures’),
and has a PhD in Linguistics from the University of York. His teaching is in the
areas of phonetics and phonology, and he has research interests in phonetics,
phonology, conversation analysis, and the history of linguistics. Address
for correspondence: Department of Linguistics and Phonetics, School of
Modern Languages and Cultures, University of Leeds, Leeds LS2 9JT, UK.
<l.plug@leeds.ac.uk>
NOTES ON CONTRIBUTORS 171
Julie Radford is a course director of the Masters in Special and Inclusive

Education at the Institute of Education, University of London. Her main area
of research is the application of conversation analysis to exploring language
learning processes in classrooms. She has a special interest in the joint
construction of topic and repair. She has also published works that examine
the linguistic skills and abilities of children with specific speech and
language needs: for example, word finding and pragmatic difficulties.
Current work studies the use of gaze and gesture in storybook sharing.
Address for correspondence: Faculty of Children and Health, Institute of
Education, 25 Woburn Square, London WC1H 0AA, UK. <j.radford@ioe.ac.uk>
Markus Reuber is a Senior Clinical Lecturer at the University of Sheffield, UK,

and Honorary Consultant Neurologist at the Sheffield Teaching Hospitals
NHS Trust. He trained as a neurologist and epileptologists in Leeds, UK, and
Bonn, Germany. His clinical practice focuses on the diagnosis and treatment of
epileptic and non-epileptic seizure disorders. He has a research interest in
behavioural neurology, especially the aetiology and treatment of non-epileptic
seizures, co-morbidity in patients with epilepsy and doctor–patient communi-
cation. Address for correspondence: Academic Neurology Unit, University of
Sheffield, Royal Hallamshire Hospital, Glossop Road, Sheffield S10 2JF, UK.
<markus.reuber@sth.nhs.uk>
Basil Sharrack is an Honorary Senior Clinical Lecturer at the University of

Sheffield, UK, and a Consultant Neurologist at the Sheffield Teaching
Hospitals NHS Foundation Trust. He trained as a neurologist at Guy’s
Hospital and the National Hospital for Neurology and Neurosurgery in
London. His clinical practice focuses on the diagnosis and treatment of multi-
ple sclerosis and other CNS inflammatory conditions. He has a research interest
in neuro-immunology, neuro-ophthalmology, and behavioural neurology.
Address for correspondence: Academic Neurology Unit, University of Sheffield,
Royal Hallamshire Hospital, Glossop Road, Sheffield S10 2JF, UK.
<basil.sharrack@sth.nhs.uk>
Ling Shi is an Associate Professor in the Department of Language and Literacy

Education in the University of British Columbia. Her research has been
published in journals such as Written Communication, Language Awareness,
TESOL Quarterly, Language Testing, English for Specific Purposes, and Journal
of English for Academic purposes. Address for correspondence: 2034 Lower Mall
Road, UBC, Vancouver, B.C., Canada V6T 1Z2. <ling.shi@ubc.ca>

Applied Linguistics, Volume 31

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Applied Linguistics, Volume 31

Diunggah oleh

Hak Cipta:

Format Tersedia

APPLIN-31(1)Cover.

qxd 1/27/10 3:06 PM Page 1

ISSN 0142-6001 (PRINT)

Volume 31 Number 1 February 2010

Published in cooperation with

REVIEWS AND FORUM EDITOR

A subscription to Applied Linguistics comprises Oxford University Press, Great Clarendon

ISSN 0142-6001 (PRINT)

Volume 31 Number 1 February 2010

Published in cooperation with

NOTES ON CONTRIBUTORS 169

Textual Appropriation and Citing Behaviors

This article explores the citing behaviors of 16 undergraduates in a North

Inability to identify plagiarized texts

The present study

examined passages of students’ texts that contained citations. In the latter

Table 1: Participants’ profiles

Elmer Commerce 1 English 18 Born in North

Interviews and data analysis

theory . . . That’s why I kind of

FINDINGS AND DISCUSSION

Reasons mentioned more frequently for citing texts borrowed

Securing support for one’s writing

Table 3: Classification of reasons students gave for using source text,

Interpretations of source text

Reasons related to learning

Total 159 67 77 33 236 100 122

Learning ‘facts’ and ‘new information’ in various academic disciplines

Reasons mentioned more frequently for not citing

Interpreting source information as common knowledge/term

considered common knowledge included statements that ‘the industrial policy

Learning not to cite

in the reference list. As an example, Cary referred to her learning experiences

Claiming authorship of ideas and learning to construct knowledge

Quotations, paraphrases and summaries

Table 4: Citation types, format of textual borrowing, and students’

Cited Not Cited Not Cited

Limited identifications for formats of textual borrowing

Understanding the roles of summaries, quotes and paraphrases

Concern for plagiarism

description of the situation. . . . And I guess that’s sort of a border-

Practices of Other-Initiated Repair in the

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

In line 2, the teacher’s TIE at first appears to resemble an instruction, although it

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

PARTICIPANTS, DATA SET AND METHOD

qualified as speech and language therapists/pathologists, they did have

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Table 1: Children’s SSLDs

Note: 0 = none, 1 = mild, 2 = moderate, 3 = severe.

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Non-specific (‘general’) RIs

T’s question (line 1) would typically be heard by the child as a request

the child an unambiguous message that a repeat is needed. In fact, line 3

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Specific RIs (designedly incomplete utterances) that pinpoint

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

Although Tom offers the name of an animal, T withholds acceptance of this

very precisely by leading the child, at least syntactically, up to the location of

Downloaded from applij.oxfordjournals.org by guest on December 31, 2010

information that could be usable in the subsequent correction. Both examples