Encyclopedia Of Language and Linguistics (2 Volume Set) 2011009057

Citation preview

LANGUAGES AND LINGUISTICS

ENCYCLOPEDIA OF LANGUAGE AND LINGUISTICS (2 VOLUME SET)

No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.

LANGUAGES AND LINGUISTICS Additional books in this series can be found on Nova‘s website under the Series tab.

Additional E-books in this series can be found on Nova‘s website under the E-books tab.

LANGUAGES AND LINGUISTICS

ENCYCLOPEDIA OF LANGUAGE AND LINGUISTICS (2 VOLUME SET)

KIMBERLY L. MEINKEN EDITOR

Nova Science Publishers, Inc. New York

Copyright © 2012 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers‘ use of, or reliance upon, this material. Any parts of this book based on government reports are so indicated and copyright is claimed for those parts to the extent applicable to compilations of such works. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS. Additional color graphics may be available in the e-book version of this book.

Library of Congress Cataloging-in-Publication Data Encyclopedia of language and linguistics (2 volume set) / editors: Kimberly L. Meinken. p. cm. Includes index. ISBN:  (eBook) 1. Language and languages--Encyclopedias. 2. Linguistics--Encyclopedias. I. Meinken, Kimberly L. P29.E4823 2011 403--dc22 2011009057

Published by Nova Science Publishers, Inc. †New York

CONTENTS Volume 1 Preface Chapter 1

Chapter 2

Chapter 3

Chapter 4

Chapter 5

Chapter 6

ix As the Call, so the Echo: Proverbs as Cues to Cultural Knowledge and Language Change Kristyl Kepley

1

Specifying Coordination An Investigation into the Syntax of Dislocation, Extraposition and Parenthesis Mark de Vries

37

Learner Perception and Strategies for Pragmatic Acquisition: A Glimpse into Online Learning Materials Julie M. Sykes and Andrew D. Cohen

99

"Best Success Through Language Loss?" An Incipient Austrian Sociolinguistic Study on Open Questions in Education and Migration Research Katharina Brizic

135

A Contribution to the Study of Russian Euphemistic Youth Slang Brian Cooper

159

What Do We Learn on Language Acquisition from Williams Syndrome? Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

175

Chapter 7

The Need for Explicit Inferential Methods in Linguistics Kent Johnson

Chapter 8

‗Doing‘ Disciplinary Power: Manager-Subordinate Interaction in a Staff Meeting Jonathon Clifton

191

207

vi Chapter 9

Chapter 10

Chapter 11

Contents Simulation Tasks: Can EFL Learners Interact as Effectively with Each Other as they Can with Native Speakers? Yasuo Nakatani

225

The Effect of Text Difficulty of Reading Comprehension Tests on Cognitive and Metacognitive Strategy Use Mohammad Hossein Keshavarz and Hassan Rouhi

245

Cultural and Linguistic Influence on Developmental Neural Basis of Theory of Mind and Self-construal: Whorfian Hypothesis Revisited Chiyoko Kobayashi Frank

Chapter 12

Language Processing in Bimodal Bilinguals Anthony Shook and Viorica Marian

Chapter 13

Psycholinguistic Abilities and Phonological Working Memory in Bilingual Children with Specific Language Impairment: A Cross-Cultural Study Dolors Girbau

265 299

329

Chapter 14

Psycholinguistic Challenges in Processing the Arabic Language Raphiq Ibrahim

345

Chapter 15

Neurocognitive Aspects of Processing Arabic and Hebrew Raphiq Ibrahim

367

Chapter 16

Visual Word Access in Monolinguals and Bilinguals in English and Spanish John Evar Strid and James R. Booth

387

Volume 2 Chapter 17

Vowels in Semitic Alphabet Languages Raphiq Ibrahim

Chapter 18

Methodological Issues in Research on Bilingualism and Multilingualism Lilian Cristine Scherer, Rochele Paz Fonseca and Ana Inés Ansaldo

411

431

Chapter 19

Bilingualism and Hispanic American Intelligence Test Scores Philip G. Gasquoine, Aracely Cavazos, Juan Cantu and Amy A. Weimer

445

Chapter 20

Language Development through a Bilingual Lens Eswen Fava and Rachel Hull

465

Chapter 21

Improving Reading Skills for ESL Learners Using SoundSpel Michael D. Young, Michelle L. Wilson and Alice F. Healy

479

Chapter 22

A Novel Transliteration Approach in an English-Arabic Cross Language Information Retrieval System Ghita Amor-Tijani and Abdelghani Bellaachia

493

Contents Chapter 23

Methods for Cross-Language Information Retrieval Kazuaki Kishida

Chapter 24

Language and Thought: Linguistic Influence on Developmental Neural Basis of ‗Theory of Mind‘ Chiyoko Kobayashi

Chapter 25

Chapter 26

Chapter 27

Chapter 28

Chapter 29

Chapter 30

Chapter 31

Chapter 32

Chapter 33

Chapter 34

Is a Story the Same as the Sum of its Parts? The Deferential Effects of Context and Isolated Word Training on Reading Fluency Sandra Lyn Martin-Chang

vii 507

555

583

Children‘s Language and Literacy Learning during a Science Concept Inquiry Ni Chang

605

Reading Impairment in Childhood: Overview of the Electrophysiological Correlates of Developmental Dyslexia Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

665

Interaction between Flexible Cognition and Language Comprehension in Children with and without Language Impairment Klara Marton Ready for Tomorrow‘s International World: Content and Language Integrated Learning (CLIL) Providing Children with Good First and Foreign Language Skills Tuula Merisuo-Storm Language Development in Special Populations: The Case of Selective Mutism Matilda E. Nowakowski, Louis A. Schmidt, Charles E. Cunningham, and Angela E. McHolm

697

723

741

Early Communicative Channel in Childhood: Development and Impairment Gianluca Esposito and Paola Venuti

757

Does Language Experience Influence the Acquisition of Lexicon? Insights from the Intellectual Disability Field Bruno Facon and Thérèse Bollengier

773

Late Talkers: Children with Developmental Dysphasia (Longitudinal Follow-Up) Olga Dlouha

793

Early Middle Ear Effusion and Central Auditory Processing at Age Seven Dale L. Johnson, Constance D. Baldwin and David P. McCormick

817

viii Chapter 35

Chapter 36

Index

Contents Multilingual Specific Language Impairment (SLI): Future Directions for Research Tanja Rinker and Steffi Sachse

827

Longitudinal Study between the Apparent Pain Insensibility and the Non Verbal Communication and Symbolisation Disorders in Autism Guillaume Bronsard, Olivier Bonnot, Geneviève Haag and Sylvie Tordjman

833

841

PREFACE This book discusses current research in the field of language and linguistics. Topics presented include syntax of dislocation, extraposition and parenthesis; learner perception and strategies for pragmatic acquisition; the need for explicit inferential methods in linguistics; language processing in biomodal bilinguals; visual word access in monolinguals and bilinguals; reading impairment in childhood; children's language and literacy; language development in special populations and the early communicative channel in childhood. Chapter 1 - Proverbs have been touted as a part of the memes of a culture: within a community, wise, creative and intelligent behavior is likely to be noted and appreciated; in turn, the acknowledgement of this behavior's success is frequently preserved in the form of a proverb which functions as part of a community's store of knowledge and their commonly shared values. Like some other parts of speech, proverbs are quasi-fixed linguistic utterances whose meanings have evolved over time and space. And, whereas, many people feel confident relating the meaning of a proverb, they often feel uncomfortable explaining how the proverb's meaning relates to its pure linguistic form. This is because proverbs, like other parts of speech, change over time; and this makes proverbs an interesting space in which to investigate language analysis and interpretation. This chapter discusses the conceptual processes that underlie proverb meaning and how proverb usage impacts cultural understandings; it also seeks to engage in the discourse of language change by investigating how varying degrees of use and diverging interpretations of proverb meanings across generations ultimately affect their fitness. The paper concludes with the results of a study involving proverb interpretation among participants of three age groups. The study focused on how successful each generational group would be at interpreting traditional and modern proverbs. How—in this high-paced, information-flooded environment—did younger people interpret the proverbs of their parents and grandparents? Additionally, would the older generation be able to accurately interpret the proverbs of their grandchildren? And would the middle generation appear as a bridge between the two groups? Chapter 2 - Coordination is a syntactic construction with a varying semantics. I argue that besides additive, disjunctive and more specialized kinds of coordination, there is another main type, namely one that can be characterized as specification. Grammatical configurations involving specifying coordination can be found at diverse levels of sentence structure. This provides an new perspective on a range of syntactic phenomena, such as apposition, parenthesis, left- and right-dislocation, and extraposition. Certainly, all these constructions are

x

Kimberly L. Meinken

different, and they have been studied before in isolation – but they also have properties in common. I propose a basic typology in terms of four independent factors: restrictiveness, anchoring, backgrounding, and place of attachment. A common problem for the analysis of non-restrictive material is that it is a priori unclear how to incorporate such phrases and clauses in the host structure, since the syntactic relationship between the two cannot be expressed in terms of regular subordination, which is an automatic result of syntactic merger according to the standard view. Furthermore, several types of phrases or clauses can show up either in a dislocated position or as an intervening parenthesis, which is also unexpected from a traditional perspective. These problems can be solved by means of the particular mode of construal discussed in this chapter, specifying coordination, which is then divided into two kinds: i) restrictive specifying coordination, which can be used to explain extraposition, ii) non-restrictive (parenthetical) specifying coordination, which triggers a secondary message (a proposition if the parenthesis is anchored). Thus, it is shown that superficially very different syntactic constructions make use of the same abstract mechanisms. Chapter 3 - Research on second language (L2) pragmatics indicates that explicit instruction helps learners improve their pragmatic performance and that interlanguage pragmatic (ILP) development is an important component of communicative competence. A pedagogical model for pragmatics needs to address the true complexity of pragmatic development. Moreover, an informed approach to L2 pragmatic instruction calls for the inclusion of multiple knowledge areas and a strategy-based understanding. CALL technologies play an increasingly important role in ensuring that pragmatics instruction is comprehensive in nature. Likewise, an inclusive model would incorporate strategies for the learning and performance of pragmatics. The intention is to assist learners in gaining the necessary skills to tackle L2 pragmatic complexities in a variety of contexts. Moreover, innovative technologies have the potential not only to aid in the creation of effective online materials for learning pragmatics, but also to provide a valuable environment for interaction. At this point, little research has empirically analyzed the use of an online, strategy-based model for learning L2 pragmatics. Even less has specifically targeted learners‘ perceptions of this space. In order to arrive at a greater understanding of these issues, this paper compares learner perception data from two research projects which have utilized CALL materials to enhance learners‘ strategies for L2 pragmatics. The first study was an in-depth, qualitative research project addressing the use and effectiveness of the first-of-its-kind learner self-access website in L2 Spanish pragmatics with an extensive strategy overlay. The second was an extensive study which analyzed the use of the first synthetic immersive environment (a 3dimensional gaming space designed with a specific educational outcome) used as an instructional space for learning Spanish pragmatics, specifically requests and apologies. This chapter compares learners‘ perceptions of their own strategy development within these two unique, online contexts, and utilizes patterns found in qualitative interview data to provide insights into how learners view both types of CALL environments for pragmatic development. The overarching goal is to consider potential roles that each of these mediated contexts could play in a strategies-based approach to L2 pragmatics. Future application of the findings to materials design and creation, research, and pedagogy will also be presented.

Preface

xi

Chapter 4 - Which languages are actually transmitted in immigrant families? This is a crucial point in the present fierce European debate on family-language use and immigrant pupils‘ linguistic/educational success. Quantitative sociology often arrives at the conclusion that maintaining immigrant-family languages is counterproductive to children‘s educational success. Qualitative linguistics, in contrast, mainly reveals a positive maintenance effect on children‘s school (language) proficiency. The incipient project deals with the question of which of these approaches is justified and how such inconsistencies could emerge. It furthermore investigates why some groups, such as the Turkish immigrants, regularly score so low in regard to educational as well as linguistic success, whether this has anything to do with their language use, as often assumed, and what new answers to these issues would mean for the present fierce educational policy debates in many European and other immigration societies. From a preceding explorative study and a new explanatory model the following hypotheses have emerged: a. The interdisciplinary inconsistencies could be caused by the continuing wide gap between qualitative and quantitative data-collection methods, which will thus in the present project be overcome as far as possible by interdisciplinary methods as well as theory. b. The methods in turn have to be highly sensitive when actual familial language-use patterns (and thus also stigmatised languages or even language loss) are at stake. c. However, when language proficiency, use, transmission and loss appear as collective phenomena, they can only result from high societal pressure on a whole group, and thus from social inequality – here: language-political inequality. The sending as well as the receiving societies‘ education and language policies have paramount influence on how people construct (or deconstruct), show (or hide) and transmit (or give up) their languages and identities. d. If language transmission/ shift can be uncovered in its many facets, it should be possible to examine the overall hypothesis stating: family-language loss does not facilitate school language acquisition, just because language acquisition is a holistic process and highly depends on the quality of parental input. If parents, in contrast, transmit the language(s) they feel "at home" with, children score higher in both family and school languages. The incorporation of several disciplines concerned (sociology, socio-/psycholinguistics, research on language shift etc.) is intended to make the present approach innovative. Thus, for methodology and theory, but also for educational policy practice in sending and receiving societies ample output can be expected. Chapter 5 - After an introductory discussion of the difficulties of distinguishing between slang and euphemism, an examination is presented of some sexual and scatological aspects of Russian euphemistic youth slang under these headings: Pregnancy and abortion; Contraception; Copulation; Erection; Expletives; Group sex; Homosexuality; Masturbation and menstruation; Oral sex and orgasm; Penis; Prostitute; Virgin and vulva/vagina; Backside and anal sex; and Lavatorial matters and vomiting. The conclusion is reached that slang is mainly a young persons‘ phenomenon and that, though many slang locutions die, a few become established, which justifies the study of slang as one of many sources by which a living language enriches itself and continues to develop. Slang is concerned to an appreciable extent with sex and scatology. Even in the case of euphemistic slang words, the euphemized substitute may itself be contaminated by association with the taboo word and come to be considered unacceptable or not quite decent and therefore to be avoided. Chapter 6 - Williams syndrome (WS) is a rare genetic disorder involving the deletion of approximately 27 genes on the locus 7q11.23. Work over the last twenty years has focused on

xii

Kimberly L. Meinken

delineating the distinctive neurocognitive profile that characterizes this group. The first studies identified a dissociation between cognition and language demonstrating that the verbal performance of individuals with WS was markedly better than their performance in nonverbal areas. More recent studies have shown dissociations within language (e.g., relatively good phonological memory, impaired grammatical morphology) and within non-linguistic visuo-spatial skills (good face recognition, impaired block design). At the same time with data from larger groups of individuals with WS, it has become apparent that there is heterogeneity in performance within the WS group across cognitive domains, languages and cultures. By looking at heterogeneity in language in WS from different language communities and approaching the phenomenon from multiple perspectives, we can begin to identify factors mediating their performance. Thus, we can have a better understanding of underlying processes involved in the language profile in a disorder such as WS and of the role of external and internal factors on language acquisition and development. Chapter 7 - Like every discipline, theorizing in linguistics crucially involves drawing (often tentative) conclusions about the relation between the available evidence and the theories under consideration. At many points in their research, linguists must decide whether the available evidence (theoretical, empirical, etc.) supports a given theory well enough to accept it, possibly over various competing theories. Surprisingly, however, virtually no attention has been paid to this aspect of linguistic theorizing, which contrasts with the standard practices of the social, behavioral and psychological sciences. In the latter fields, whole subareas – including dedicated faculty positions, journals, societies, etc. – are devoted to the quantitative, frequently statistical, details of drawing inferences from data. But in linguistics, theoretical inferences typically have the form of informal, verbal, holistic judgments made by professional linguists. I begin by characterizing the issue of linguistic inferences, showing why the current methods are cause for concern. The psychological literature on professional judgment strongly suggests that professional judgments aren‘t as accurate as the experts believe, and that they are inferior to more explicit methods. In fact, linguistic inferences are particularly (although not uniquely) suspect in this respect. I then consider several responses linguists have made to the criticism of their reliance on expert judgment. These replies, I argue, do little to mitigate the concerns about how evidence is compiled and assessed in linguistics. Chapter 8 - This paper attempts to provide a space for dialogue between macro-historical Foucauldian analysis of power and conversation analysis‘ fine-grained technical analysis of talk-in-interaction. Arguing that such a dialogue is possible through the use of bridging as a research tool that focuses on areas of complementarity between macro and micro analysis, this paper shows how category bound access to powerful discourse resources (notably the first part of question and answer adjacency pairs) during a staff meeting allows the management to exercise disciplinary power over their subordinates. In this way, the Foucauldian notion of disciplinary power can be seen as an emergent feature of oriented-to pre-allocated access to the first part of question and answer adjacency pairs. Chapter 9 - This study asked whether a simulation interactive task could facilitate EFL (English as a foreign language) learners‘ conversation in actual classrooms. A restaurant simulation task was designed for Japanese college students in order to investigate whether learners‘ interaction with other learners can be meaningful compared with their interaction with native English speakers (NSs). The database was derived from videotapes of five student-student (S-S) conversations and five native English speaker-student (N-S)

Preface

xiii

conversations collected from the observation of the task performance. Transcripts were made and analysed by focusing on the number of turns, production rate and the quality and quantity of negotiation of meaning. The results indicate that although students in S-S use less sophisticated utterances, they can produce more language and use more turns than in N-S conversations. They frequently negotiate meaning and flexibly expand their interaction with peers. Their conversation is not the same as that of NSs but they could make use of opportunities to develop their target language communication skills in EFL contexts. Thus the two-way required information exchange tasks which facilitate each participant‘s personalised outcome can create spontaneous interaction for negotiation in classroom contexts. Chapter 10 - This chapter presents a research study whose aim was to investigate whether high-school students use different cognitive and metacognitive strategies while doing two reading comprehension tests which were different in terms of difficulty of the reading texts. A total of 57 senior high school students majoring in math and experimental science participated in this study. To reach the goals of the research, two reading comprehension tests were developed each of which contained 25 multiple-choice items, but they were different in terms of the difficulty level of the texts as estimated by Flesch readability formula and expert judgements. After the administration of the Nelson Test, the students were divided into high, mid, and low-proficiency groups. Next, the students were asked to answer the easy reading comprehension test and were instructed to consider the strategies they have used during the test to fill out the cognitive-metacognitive questionnaire. Two weeks later, the subjects were required to answer the difficult reading comprehension test task and fill out the cognitivemetacognitive questionnaire again. The results suggested that (1) the subjects resorted more to cognitive and metacognitive strategies in the easy reading comprehension tests compared to the difficult ones, (2) the cognitive and metacognitive strategy use did not differ across the proficiency level of the subjects, (3) the correlation coefficient between cognitive and metacognitive strategies and reading comprehension test performance was statistically significant (both in the easy and difficult tests), and (4) there was no interaction between cognitive and metacognitive test-taking strategies, language proficiency and text difficulty. The results are interpreted to have implications for EFL/ESL teachers and syllabus designers. Chapter 11 - Recent cross-cultural research on various cognitive functions found clear cultural influence on how we perceive our world. Likewise, neuroimaging research has found significant influence of culture / language in theory of mind (ToM) – ability to understand mental states of others – and self-construal (which is related to ToM) in the neural level. In this chapter, cross-cultural and brain imaging research on ToM and related social cognition are selectively reviewed. I discuss the roles of medial prefrontal cortex (mPFC) and temporoparietal junction (TPJ) that have been consistently implicated in ToM and perspective-taking (or distinguishing ―self‖ from ―other‖). These structures may be particularly important for culturally unique ways of social cognition related to ToM and self-construal. Next, I briefly review current developmental theories of ToM, and discuss whether or not the recent findings from neuroimaging studies of ToM in children support these theories. Functional relationship between ―language regions‖ and ToM will also be reviewed along with these discussions. Lastly, I present two models of culture / language-dependent and independent ToM development. Chapter 12 - Recent research suggests differences between bimodal bilinguals, who are fluent in a spoken and a signed language, and unimodal bilinguals, who are fluent in two

xiv

Kimberly L. Meinken

spoken languages, in regard to the architecture and processing patterns within the bilingual language system. In this chapter, we discuss ways in which sign languages are represented and processed and examine recent research on bimodal bilingualism. It is suggested that sign languages display processing characteristics similar to spoken languages, such as the existence of a sign counterpart to phonological priming and the existence of a visual-spatial loop analogous to a phonological loop in working memory. Given the similarities between spoken and signed languages, we consider how they may interact in bimodal bilinguals, whose two languages differ in modality. Specifically, we consider the way in which bimodal bilingual studies may inform current knowledge of the bilingual language processing system, with a particular focus on top-down influences, and the fast integration of information from separate modalities. Research from studies looking at both production and perception suggests that bimodal bilinguals, like unimodal bilinguals, process their languages in parallel, with simultaneous access to both lexical and morphosyntactic elements. However, given the lack of overlap at the phonological level (the presumed initial locus of parallel activation in unimodal studies) in bimodal bilinguals‘ two languages, we conclude that there are key differences in processing patterns and architecture between unimodal and bimodal language systems. The differences and similarities between unimodal and bimodal bilinguals are placed in the context of current models of bilingual language processing, which are evaluated on the basis of their ability to explain the patterns observed in bimodal bilingual studies. We propose ways in which current models of bilingual language processing may be altered in order to accommodate results from bimodal bilingualism. We conclude that bimodal bilingualism can inform the development of models of bilingual language processing, and provide unique insights into the interactive nature of the bilingual language system in general. Chapter 13 - A variety of concepts/types of bilingualism and bilingual programs in U.S.A./Europe are presented. The cognitive benefits of bilingual education across several languages are reviewed. Diagnosis and intervention issues in bilingual children with Specific Language Impairment (SLI) / Typical Language Development (TLD) are discussed, including some behavioral and neurophysiology findings concerning language processes. A cross-cultural study was done by comparing children from U.S.A. (with SLI/TLD) and children from Spain (with SLI/TLD), who were involved in a larger project (Girbau & Schwartz, 2007, 2008). Forty-four sequential bilingual children (7;6-10;11 years old), with L1 = Spanish and L2 = English/Catalan, participated. The psycholinguistic abilities in any bilingual group with TLD were significantly higher than in any bilingual group with SLI (Spanish-English/Spanish-Catalan). The similarities of the cross-cultural profiles are discussed. Only children with TLD from Spain produced significantly more correct non-words (in the Spanish Non-word Repetition Task) than children with TLD from U.S.A. (who were exposed to English phonetics). This cross-cultural difference was not found for children with SLI; they all performed poorly in U.S.A. and Spain. The Spanish task was a good marker for SLI in both countries. Our results support the phonological working memory deficit associated with SLI, which appears to be independent of the particular bilingual background. The English Non-word Repetition Task was not sensitive in identifying SLI in these Hispanic unbalanced bilinguals, since English was their L2; their phonotactic representations in L1 seem to determine their performance on the task. The Spanish non-word repetition accuracy correlated significantly with the Auditory Association subtest from the Spanish ITPA, in children with SLI/TLD and in Spain/U.S.A.

Preface

xv

The task also correlated significantly with the Grammatical Integration subtest for children with SLI and in Spain/U.S.A. Both subtests involve auditory working memory, but the second one has also some visual support through pictures. Implications of the results for the crosscultural identification of SLI in bilinguals are discussed. Chapter 14 - The 2006 PISA (Program for International Student Assessment ) report of worldwide scholastic achievements showed that about 50% of Israeli Arabic students were found to exhibit the lowest reading achievement scores in the PISA tests (level 1 and below) as compared to the other participating groups. Also, the MEITZAV national testing program in Israel (2001–2002) showed an achievement gap in language skills (reading and reading comprehension) between Arab students and Jewish students in the school systems. This gap was larger than those found in the other areas tested (mathematics, science and technology, and English). The aim of this chapter is to explore the cognitive basis of these difficulties, specifically the diglossic situation in Arabic. Furthermore, the chapter discusses the unique features of the Arabic language that might contribute to the inhibition and slowness of reading acquisition and might even hinder the acquisition of basic academic skills. Finally, a model with a comprehensive basis (cognitive and neurocognitive) will be built in order to explain the complex linguistic situation of beginning Arabic learners. This chapter is concerned with the cognitive evidence bearing on the nature of the mechanisms of language processing in Arabic which has critical linguistic characteristics and a diglossic factor. Additionally, other aspects, including a neurofunctional perspective, will be discussed. Chapter 15 - The aim of this chapter is to explore the neurocognitive basis of the difficulties that the Arabic-Hebrew bilingual encounters in processing the Arabic language as a result of the diglossic situation in Arabic (spoken Arabic and Modern Standard, or Literary Arabic). Furthermore, the chapter discusses the unique features of the Arabic language that might contribute to the inhibition and slowness of reading acquisition and might even hinder the acquisition of basic academic skills. In the first section, two case studies of ArabicHebrew aphasic patients (M.H. and M.M.) are presented, with different disturbances in the two languages, Arabic (L1) and Hebrew (L2). They exhibited a complementary pattern of severe impairment of either L1 (Arabic) or L2 (Hebrew) constituting a double dissociation. These results suggested that the principles governing the organization of lexical representations in the brain are not similar for the two languages. The second section focuses on the functional architecture of reading in Hebrew and in Arabic. The effects of characteristics of Arabic and Hebrew as Semitic languages on hemispheric functioning were systematically examined. These patterns are compared with the modal findings in the literature, which are usually based on English. Also, the effects of the absence of almost all vowel information, the orthographies of the two languages, and their non-concatenative morphological structure were investigated. It was shown that when languages make different types of demands upon the cognitive system, interhemispheric interaction is dynamic and is suited to these demands. In that regard, both Arabic and Hebrew require a higher level of interhemispheric interaction than does English. Chapter 16 - This research examined if visual word access varies according to language and bilingual status by comparing Spanish and English, priming two syllable CVCV words with bilingual children and monolingual children. The results suggest that lexical access in English is based on a larger phonological sub-lexical unit because of greater report of a unit bigger than the syllable, among both bilingual and monolingual subjects. In contrast, only

xvi

Kimberly L. Meinken

weak evidence suggested that Spanish lexical access was based on the syllable because of greater report of that unit for bilinguals and monolinguals. Finally, monolingual or bilingual status of the reader did not have influence on English lexical access; however, Spanish bilinguals were influenced by the acquisition of English, suggesting that orthographically opaque languages can have an effect on transparent languages or that immersion, language dominance and literacy experience can influence reading in the other language. Chapter 17 - While several alphabetic systems are in use, this chapter focuses on the two Semitic alphabet languages—Arabic and Hebrew—especially in the orthography of the two languages. Semitic scripts are unique in that short vowels are represented as diacritics on consonant letters. The unique characteristics of Arabic and Hebrew orthographies make them unique for investigations among Latin alphabets or even one among the other. Hebrew and Arabic are both read from right to left. Chapter 18 - Research on language processing in the brain has increased and has become prominent in the past decades, fostered mainly by the adoption of neuroimaging techniques. For instance, language studies on the lesion paradigm, with which neuropsychology was born, have considerably improved by the use of refined neuroimaging techniques. Some of the techniques mainly used in research on language processing are functional magnetic resonance imaging (fMRI), positron emission topography (PET), magnetoencephalography (MEG), electroencephalography (EEG), and functional near-infrared spectroscopy (fNIRS), among others, which have allowed researchers to acquire in vivo brain images of the order of seconds or even milliseconds. The availability of these increasingly refined techniques, allied to the research interest in understanding language processing and its related cognitive components has led to the emergence of a new research field, named Neuropsycholinguistics. In this field, a growing topic of study is bi/multilingualism and its neural correlates. Neuroimaging research on bi/multilingualism has investigated both comprehension and production, by adopting a variety of experimental designs and criteria for participants‘ recruitment. This diversity, although informative, has represented a challenge for drawing general conclusions from the data. The term ―bilingual‖ has been defined in distinct ways, depending on the researcher´s concept on the degree of language mastering considered to be necessary for a person to be characterized as such. The definition here to be adopted is the one proposed by Grosjean, according to which the term ‗bilingual‘ refers to an individual who uses two or more languages or dialects in his or her everyday life, regardless of the context of use. Whenever this definition is taken into account, it is possible to conceive that more than half of the world´s population can be considered bilingual. Nowadays, it is common also to choose the word ―multilingual‖ to characterize individuals who speak more than two languages. This word will be used in this sense throughout the chapter, although the word ―bilingual‖ is preferably used by some authors. In this scenario of multilingualism, understanding how languages coexist in the brain is of relevant importance, since this type of knowledge may support studies on first and second language acquisition and neuropsychological rehabilitation of language and communication impairment, and language teaching, among others. Moreover, it may contribute to enhance our understanding about the processing of executive functions in cognitive neuropsychology, such as the switching component, which has been investigated in bilingual participants. Concerning research on bi/multilingualism, a rigorous methodological control needs to be taken by researchers in order to guarantee data validity and reliability, since several co-

Preface

xvii

occurring factors may influence results and their interpretation. If not controlled, these interferences may confound data analyses and theoretical discussions. The aim of this chapter is to discuss important methodological issues regarding research on neurologically preserved bilingual and multilingual populations. It includes the variables to be considered in methodological aspects ranging from individual characteristics to task design and presentation, among others. Chapter 19 - Hispanic Americans as a group score 0.5SD below White Americans on intelligence test measures (administered in English) that emphasize language processing, but score similarly when visual-perceptual/visual-motor processing is required. The reason for this language decrement is unknown. This chapter considers the possible contribution of bilingualism to this effect, as studies linking bilingualism and cognition (conducted with multiple ethnic groupings) have consistently shown a bilingual disadvantage compared to monolinguals on language processing tasks. Two data sets (older children and adults) of bilingual Hispanic American performance on various intelligence test measures administered in Spanish and English showed evidence of a visual-perceptual/visual-motor over language processing advantage of about 1SD. The size of the visual-perceptual/visual-motor over language advantage was similar in both languages suggesting it is bilingualism-related and not due to low English language proficiency. Bilingualism appears to be a potentially important factor in the Hispanic American language processing decrement seen on intelligence tests, although no direct study on the effect of this variable has yet been conducted. Chapter 20 - Bilingualism is a fertile resource for studying facets of language development and brain plasticity that may not be apparent in monolinguals. This chapter will summarize historical and contemporary findings in the literature, discuss methodological issues that influence their interpretation, and suggest future directions for examining the neural substrates of language development and the consequences of having two languages in one brain. Chapter 21 - This study examined the effects of using a revised, transparent spelling system SoundSpel, a phonetic reading tool, with learners of English as a Second Language. During 6 training sessions, 12 participants used unaltered material and 12 used SoundSpel texts, in parallel with standard English, when reading American elementary school material. They then answered multiple-choice comprehension questions. Both groups were pre-tested and post-tested on comprehension tests of similar elementary school material without SoundSpel. No group differences were found across tests or training (in quiz performance or reading time), suggesting no beneficial or harmful effects from using SoundSpel. A post hoc analysis suggested that SoundSpel would be most beneficial for students who learn to speak English before they learn to read it. Chapter 22 - One of the main issues facing Cross Language Information Retrieval (CLIR) is untranslatable words, i.e., words not found in dictionaries, which are usually referred to as Out Of Vocabulary (OOV) words. Bilingual dictionaries in general do not cover most proper nouns (e.g., names of places, people, countries, etc.), which constitute a large proportion of OOV words. As they are often primary keys in a query, their correct translation is often necessary to maintain a good retrieval performance. Because they are spelling variants of each other in most languages, an approximate string matching technique against the target database index is usually used to find the target language correspondents of the original query key. The n-gram technique has proven to be the most effective among other approximate

xviii

Kimberly L. Meinken

string matching techniques. A more complicated issue arises when the languages dealt with have different alphabets. The approach usually taken is transliteration. It is applied based on phonetic similarities between the languages involved. However, transliteration by itself cannot guarantee the exact spelling of the transliterated words as found in the document collection. There are a variety of ways that a transliterated word can be spelled despite conventions that might exist. The fact that there is no one correct way of spelling a transliterated word shows the need for a technique that is capable of generating the different spellings found in the document collection. In this study, we chose to combine both transliteration and the n-gram technique in an English-Arabic CLIR system, in which Arabic documents were searched using English queries. We evaluated the effectiveness of this approach and compared it with other transliteration approaches. Experimental results showed the retrieval improvement gained using our transliteration approach over other existing approaches. Chapter 23 - This article reviews technical methods for enhancing effectiveness of crosslanguage information retrieval (CLIR), in which target documents are written in different languages from that used for representing a search request. As the Internet has spread since the 1990s, the importance of CLIR has grown, and the research community of information retrieval has been tackling various CLIR problems. The purpose of this article is to overview exhaustively CLIR techniques developed in the research efforts. The following research issues on CLIR are covered: (1) strategies for matching the query and documents written in different languages, e.g., automatic translation or transliteration techniques, (2) techniques for solving the problem of translation ambiguity, (3) formal retrieval models for CLIR such as application of the language modeling, (4) methods for searching a multilingual document collection in which two or more languages are used for writing documents, etc. Chapter 24 - Theory of mind (ToM) – ability to perceive, interpret and predict behaviors or actions of others based on their underlying mental states – has been explored with a plethora of paradigms in normally and atypically developing children by many researchers. ToM is fundamental for human social interaction universally. Among ToM researchers it has long been debated whether or not language development constrains ToM development. Recent results from neurological studies suggest that some aspects of language (e.g., grammar) may function merely as a ‗co-opted‘ system, but other aspects of language (e.g., pragmatics and reading communicative intentions) may profoundly affect ToM throughout the development. Neuroimaging studies of ToM development are still scarce. However, results from a few studies that explored neural correlates of ToM and related socio-cognitive functions revealed age-related differences in ToM/social-cognition-specific brain activity in several language regions. Despite the increasing evidence that supports the developmental relationship between language and various cognitive capacities ToM, current developmental theories of ToM (except one) discount the linguistic effects on ToM. There are four mainstream theories of ToM development; 1) Modular ToM hypothesis, 2) Theory, theory of mind hypothesis, 3) Simulation ToM hypothesis, and 4) Linguistic determinism of ToM hypothesis. In the first part of the chapter, I will discuss both behavioral and neurological evidence that supports or negates the linguistic influence on ToM. In the second part of this chapter, I will discuss those four prominent theories of ToM development and whether or not evidence from the most current neuroimaging studies of ToM in children and adults (including ours) support these theories. Finally, based on the most current results, I will present a new model for the developmental mechanism of ToM.

Preface

xix

Chapter 25 - The first priority for novice readers is to understand that the letters they see in written language correspond to the sounds they hear in speech. Once children grasp the letter-to-sound pairings afforded to them by the alphabetic principle, they can begin to identify some of the written words that are contained in their spoken vocabularies. At this point, the question becomes whether reading practice should take place in stories (context) or in lists (isolation). Two camps have emerged with opposing views on this topic. On the one hand, investigators have shown increased long-term word recognition when children practice reading words in isolation. In contrast, it has also been suggested that reading words in context can act as a secondary self-teaching mechanism. Proponents of this view argue that the added value of context increases reading ability beyond that achieved by other methods of training. At present, then, the evidence on best training practices (context vs. isolation) is mixed. The studies discussed here begin to unify these two differing perspectives by elucidating the circumstances that mediate the utility of each type of training. In brief, during the five training studies reviewed, context training dominated isolated word training with regards to word acquisition. However, after words were learned, retention over an eight-day interval approached ceiling under both conditions. Finally, fluency transfer (reading words quickly and accurately in new texts or new lists) was maximized when the congruency between training and testing was high; words trained in isolation were read more fluently in novel lists, whereas words trained in context were read more fluently in novel stories. To conclude, the decision of which training method to use cannot be made independently from the task to be done after training has ceased. However, if the goal of instruction is to have students who excel during training, remember what they have learned, and who use their learning to read new texts fluently, the research plainly suggests that training in context is a more fruitful method of instruction than isolated word training. Chapter 26 - Using children‘s literature to assist in science inquiry and in knowledge building in other subject areas has been on the rise due to the benefits of supporting children‘s language and literacy learning. However, published research has reported a substantial variation in the methods of utilizing children‘s literature to support learning and language development. A deficiency in the association between reading and writing, and in the amount of time children are given to write has also been reported. ―Writing was not linked to reading, and students seemed to spend relatively little time composing‖ In some classrooms in the United States, science is separately taught and given relatively less time in comparison with English/Language Arts. In the real world, however, when it comes to learning science concepts, children do not often separate science from other subject areas that assist them in the inquiry, because the human brain naturally makes meaningful connections when new information is presented. ―One of the most important constants throughout human development is that all domains of development are interrelated. Development in one dimension influences and is influenced by development in other realms‖. Furthermore, Allington, et al also found ―. . . hardly any art or drama activities linked to the reading children were doing in any of the classrooms‖ (p. 10). Therefore, the purpose of this study was to examine what might provide a scaffolding to a child‘s language and literacy competencies when reading was closely connected to writing and when art was linked to reading in the process of acquiring a science concept. The underlying research question is ―What does an integrated curriculum provide to facilitate and how does an integrated

xx

Kimberly L. Meinken

curricular approach mediate children‘s language and literacy competencies when they are engaged in acquiring a science concept?‖ Chapter 27 - Electrophysiological measurements enabled us to greatly increase our knowledge on the most spread learning disability in childhood represented by the Developmental Dyslexia (DD). The present chapter reviews the most relevant studies, which used either Event Related Potentials (ERPs) or Electroencephalographic (EEG) bands to investigate reading disabilities in developmental age. Several studies are here described, which succeeded in showing processing abnormalities not only in dyslexics, but also in children genetically at risk of dyslexia, through the analysis of both the "classical" electrophysiological components (i.e., MMN, P300, N400) and earlier evoked potentials. The electrophysiological markers of neuronal dysfunctions found in these children, helped scientists to uncover the psychophysiological mechanisms chiefly involved in this language disorder. These are: deficits in speech sound processing and impairment in manipulating the phonological features of grapheme strings. A considerable advance in the field has been recently reached by using EEG bands, traditionally used for detecting group differences in resting state, but currently extended to the measure of cognitive activation obtained through different experimental tasks. Such studies revealed functional differences in both fast and slow EEG rhythms between dyslexics and controls. The use of two functionally distinct EEG rhythms, theta and beta bands, in line with several ERP results, supports the view that dyslexics' reading difficulties are related to a linguistic impairment which is represented essentially at phonological level. Similar conclusions were further supported by the analysis of delta rhythm - a functional marker of cortical inhibition - during the performance of different linguistic tasks. Group differences in delta activation points to a delay in dyslexics' brain maturation. The review of recent literature on electrophysiological correlates of DD, on one hand supports the view that ERPs, with their excellent time resolution, can represent an optimal tool for investigating language disorders mainly along time domain. On the other hand, EEG bands are able to show both functional and maturational aspects of dyslexics' brains, thus representing a distinct measurement with respect to ERPs. These two electrophysiological methods provide complementary information for a better understanding of the neural and cognitive mechanisms involved in DD. In conclusion, electrophysiological measures could represent an effective tool for making early diagnoses of dyslexia or for predictions of future reading problems in infants at risk, but they could also be valuable for testing the efficacy of rehabilitative trainings. Chapter 28 - The focus of this chapter is on the impact of various cognitive functions on language acquisition and language processing in different groups of children. Empirical data are presented on individual variations of language and cognition in typically developing children, children with specific language impairment, and in young adults with dyslexia. The interaction between flexible cognition (attention switching, working memory, and inhibition control) and language comprehension was examined in English-speaking and Hungarian participants. The findings suggest that working memory performance and language comprehension are strongly influenced by task complexity, age, and language structure. Working memory storage and processing are affected differently by stimulus complexity. Children with specific language impairment and young adults with dyslexia showed deficits in flexible cognition and language comprehension (oral and written). Children with SLI performed more poorly than their age-matched and language-matched peers in tasks that

Preface

xxi

measured working memory capacity and inhibition control. In addition to the group effects, participants with language impairment showed different errors and performance patterns than their peers. Chapter 29 - The study described in the article investigated the effects of bilingual teaching (Content and Language Integrated Learning, CLIL) on the development of children‘s literacy skills during their first six school years. In the CLIL classes 20–25 per cent of the instruction was given in a foreign language. The results of the study showed that the children in the CLIL classes learned to read and write their first language equally well than their peers in monolingual classes. In addition, bilingual teaching did not have a negative effect on the development of those children who started school with poor learning readiness either. After the two first study years the children in the CLIL classes were more fluent readers and had better reading comprehension and spelling skills than their peers in the other classes. After four school years it was obvious that the children‘s creative writing skills had also benefited from bilingual teaching. The students in the CLIL classes had learned to pay attention to languages, as well to their mother tongue as to other languages. Moreover, they had more positive attitudes towards reading, writing and foreign language learning. Especially the boys‘ attitudes proved to be more positive in the CLIL classes than in the other classes. After six study years the students in the CLIL classes had achieved significantly better first language spelling skills than the students in the other classes. They made significantly less spelling errors and understood significantly better different texts. Furthermore, they showed more proficiency in deriving the meaning of new words from the written context than the other students. Likewise, they succeeded significantly better in finding the most important facts of the non-fiction text and summarising the text than their peers in the other classes. It is worth mentioning that in the other classes the girls‘ skills were significantly better than the boys‘ skills but in CLIL classes the difference was not significant. The students in the CLIL classes enjoyed studying the foreign language and also studying through it. Consequently, their foreign language skills developed very well. Chapter 30 - Selective mutism is a disorder that is characterized by a lack of speech in specific unfamiliar situations or in the presence of unfamiliar individuals despite speaking normally in other situations. Although the disorder usually first appears in the preschool years, it is typically not diagnosed until the child is between 6.5 to 9 years of age. Given that the disorder manifests itself early in development, the child‘ limited speaking in select situations and its impact on language development warrants careful consideration. The purpose of the present chapter is to review the literature on language performance in children with selective mutism. The chapter is divided among three sections. In the first section, we provide an overview of selective mutism, including its definition, prevalence, diagnosis, classification, etiology and functional impairment. In section two, we review the research that has been conducted to date on language performance in children with selective mutism. We conclude the chapter in section three with a review of possible developmental pathways for the association between selective mutism and language performance as well as implications for future research. Chapter 31 - Crying, which represents the very first communicative channel infants can use to communicate with their environment, plays an important role in child development. Crying is a biological signal that alerts those in the caregiving environment about the needs and wants of the infant and motivates them to respond. Cries act to release specific and

xxii

Kimberly L. Meinken

appropriate patterns of caregiving actions, thereby ensuring the infant‘s survival. The aim of this chapter is to review studies that have investigated expressions of distress, specifically crying, during early stages of development in infancy. Emphasized is the importance of cry evaluation in specific psychopathology, particularly Autism Spectrum Disorders (ASD). First the neurobiological activation that takes place during an episode of crying, both in the caregiver and the infant brain, is described. A description of the function of crying episodes and their evolution, starting from the fetal stage through childhood, follows. Finally, how anomalies in the expression of cry can bias the caregiver‘s perception of them, and how parents‘ reactions to pathological crying can be qualitatively different from their reactions to typical crying of same-aged children, is described. Such differences can confound parental attempts to share feelings and develop inter-subjectivity with their children Chapter 32 - Correlational, quasi-experimental and experimental research conducted with typical children has shown that language input strongly influences the development of vocabulary and syntax. Another approach to this issue, based on the pathological method, is reviewed in the present chapter. This approach consists in studying participants with intellectual disability because they present a ―natural‖ (i.e., unprovoked) dissociation between their general developmental level and their educational experience. Indeed, compared to intellectually average children of the same mental age, children and adolescents with intellectual disability have had more learning opportunities simply because, given their chronological age, they have lived longer. In particular, and all things being equal, they have participated more often in educational or leisure activities and have a far greater language experience, including more verbal interactions with their parents, grandparents, educators, teachers and all their other caregivers. This language experience related to age may well give them, among other things, an advantage with regard to the acquisition of additional vocabulary. Indeed, if interindividual differences in lexical development are caused, at least in part, by differences in language input, then CA-related language experience should give adolescents or children with intellectual disability a vocabulary advantage over typical children of comparable cognitive level. In the present literature review, the classical approaches (i.e., correlational, quasi-experimental and experimental) used to study the relationship between the language learning environment and the acquisition of lexicon are successively presented. The pathological method is then considered and illustrated with the aim of showing that the study of children and adolescents with intellectual disability might be an innovative way to examine the relationship between language experience and language development. Chapter 33 - This study reports longitudinal follow-up of 300 children with developmental dysphasia. Their expressive disturbances are a manifestation of their auditory decoding impairment. These children were investigated on battery of tests of spoken language, of phonological processing and of different audiometric procedures, in order to register all the information necessary to be able judge the speech-language development. The tests focused on diagnosis of central auditory disorder, confirmed the difficulties in association area in children with developmental dysphasia. In speech perception, the temporal processing is one of the functions necessary for the discrimination of phonemes, and of similar words. Our results confirmed long-term problems of children with developmental dysphasia with central auditory perception disorder. Chapter 34 - This study examined the relation of middle ear effusion MEE) in the first three years of life to central auditory processing at age seven. Ears were examined on a

Preface

xxiii

regular basis frequently in the home over the first three years and a count of days with bilateral MEE was used to quantify ear status. Central auditory processing was assessed with SCAN, Screening Test for Auditory Processing Disorders. With a sample of 76 children studied prospectively from birth there were significant correlations between MEE and all four SCAN measures. A possible link between SCAN and attention problems was also examined using the Test of Variables of Attention (TOVA). Three of 16 possible correlations were significant, suggesting a weak relation between attention and SCAN. Chapter 35 - Children with Specific Language Impairment (SLI) show a significant language deficit with no obvious accompanying condition such as mental retardation, neurological damage, or hearing impairment. While about 6-8% of all children are affected by SLI, it is therefore logical to assume that 6-8% of all multilingual children are affected by SLI as well. In order to diagnose SLI with certainty, SLI must be diagnosed in both languages. While for many languages diagnostic tools exist, there are a large number of languages for which appropriate tests have not yet been designed. Particularly in children with successive bilingual language development, inadequate acquisition of the second language may mask a clinically relevant language impairment or vice versa. While diagnosis is difficult, the issue of adequate language training or therapy respectively is just as complicated. This situation may be further exacerbated by a general uncertainty of professionals about the effects of multilingual intervention or parent advising regarding the choice of one or two languages. In addition, little is known about the neurophysiological bases and manifestations of multilingual SLI – making appropriate intervention all the more difficult. In this short communication, we will outline problems and questions with regard to diagnosis and intervention and address future areas of research in the field. Chapter 36 - Numerous clinical reports sustain the hypothesis of the pain insensitivity in autistic children. This insensitivity may be explained by the increase of the opioid activity (in particular the Bêta-endorphin). However, the rates of the plasma beta-endorphin concentrations in autistic children are contradictory. We have therefore studied in autism the relationship between the behavioural pain reactivity and the verbal and non verbal communication disorders. This study suggests that the apparent decrease of the behavioural pain reactivity in autistic children can be explained by a different way to express pain, specifically in relation with the communication and symbolisation disorders and other cognitive disorders (learning disabilities; own body image disorders; difficulties for mental representations of sensations and emotions; difficulties in understanding causes and effects relations), rather than a real endogenic analgesia.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 1

AS THE CALL, SO THE ECHO: PROVERBS AS CUES TO CULTURAL KNOWLEDGE AND LANGUAGE CHANGE Kristyl Kepley Florida Atlantic University, Department of Languages, Linguistics and Comparative Literature, Boca Raton, FL, US

Abstract Proverbs have been touted as a part of the memes of a culture: within a community, wise, creative and intelligent behavior is likely to be noted and appreciated; in turn, the acknowledgement of this behavior's success is frequently preserved in the form of a proverb which functions as part of a community's store of knowledge and their commonly shared values. Like some other parts of speech, proverbs are quasi-fixed linguistic utterances whose meanings have evolved over time and space. And, whereas, many people feel confident relating the meaning of a proverb, they often feel uncomfortable explaining how the proverb's meaning relates to its pure linguistic form. This is because proverbs, like other parts of speech, change over time; and this makes proverbs an interesting space in which to investigate language analysis and interpretation. This chapter discusses the conceptual processes that underlie proverb meaning and how proverb usage impacts cultural understandings; it also seeks to engage in the discourse of language change by investigating how varying degrees of use and diverging interpretations of proverb meanings across generations ultimately affect their fitness. The paper concludes with the results of a study involving proverb interpretation among participants of three age groups. The study focused on how successful each generational group would be at interpreting traditional and modern proverbs. How—in this high-paced, information-flooded environment—did younger people interpret the proverbs of their parents and grandparents? Additionally, would the older generation be able to accurately interpret the proverbs of their grandchildren? And would the middle generation appear as a bridge between the two groups?

2

Kristyl Kepley

Introduction ―I quote others only in order to better express myself.‖ –Michel De Montaigne.

As far as we know, every culture counts proverbs amongst its linguistic accoutrements. In fact, while these pithy pronouncements may be less common in the everyday speech of Americans, they serve as a frequent and foundational element of typical conversations in many vernaculars, the world over. Proverbs are born of folk wisdom, as stylized sayings that reveal—or define—that which is, or should be, the commonsense of daily life. They are metaphors with a specific purpose—abstract linguistic constructions used to communicate an implicit cultural model which the speakers share. Like metaphors, they are conceptual in nature; but, the proverb asserts its unique identity with its pragmatism: ―whether explicit or implicit, the evaluative assertions expressed in proverbs lend them directive force as recommendations for a desired course of action‖ (White 1987: 151). Of course, ordinary language routinely employs the conceptual and pragmatic functions of proverbs; but proverbs are unique in their peculiar form of indirectness, which acts to mitigate the judgmental quality that might otherwise be attributed to the speaker. Paremiologists1 have also pointed to the effectiveness of proverbs as an additional explanation for their prevalence; and, undoubtedly, there is truth in this proposition, since we must assume that specific linguistic constructions are only retained to the extent that they are effective at conveying information. Likewise, linguistic constructions only retain their effectiveness at conveying information if they are used with enough frequency within a community to pass intelligibly from one generation to the next in tact. There is no going back: once a proverb loses its foothold, it is nearly impossible for it to regain its position in the community‘s lexicon. One of the commentaries of post-modernism—an era we may or may not be living in, depending upon whom you ask—is that time and space have been compressed:2 the speeding up of global processes and technologies have meant that the world feels smaller and distances shorter; and, as a result, events in one place impact another place with never-before-seen speed. If this is the case, and certainly we can assume that to some degree this goes without saying, then we must acknowledge the possibility that language—like other Western, postmodern, life processes—must be subject to vaster influence and more rapid change than before. Hence, one might be called to question if the traditional proverbs familiar to American Culture will possess the same basic life-span as those proverbs that have gone before. Or, should we expect a more rapid turn-over from traditional proverbs to modern inventions of proverbial wisdom? It might seem that Western—and as addressed here, specifically, American—Culture is ripe for this eventuality. The linguistically conservative among us engage in a continuous chorus of warnings that our modern youth have been too quick to forget our sacred cultural and linguistic past. But, really, this is the cry of all generations in all times and places. So, we can all take a deep cleansing breath, knowing that the English language does not face immediate, impending doom. 1 2

Proverb scholars. From Greek: parameia = proverb (Mieder 2004). Cf. David Harvey (1990).

As the Call, so the Echo

3

Perhaps, post-modern technology is upping the ante a bit, but the game remains virtually the same. Thus, the guiding question of this study is not whether traditional proverbs are meeting an untimely end or not, but if traditional proverbs are losing their place within the cultural landscape because they no longer fulfill their role as efficient constructions of communicative effectiveness amongst a newer generation of speakers who do not possess a clear concept of their meaning. In asking this question, other questions naturally followed: when speakers are unfamiliar with a proverb, how do they go about attempting to understand it? Are older people more successful at this interpretation than younger people? Are younger speakers more adept at interpreting the proverbs from modern popular culture than older speakers? And, if so, is this because these proverbs speak more directly to cultural models familiar to younger speakers or is it simply because younger speakers have had more encounters with these particular proverbs than with the timeworn proverbs of previous generations? Clearly, these questions are but the start of a stimulating and needed investigation, but even these questions are too many to be answered in the space here allotted. Thus, the focus of this chapter will be on the unique character of proverbs, the underlying conceptual processes involved in their usage, and how an investigation of proverb interpretation may contribute to the field of language change. Finally, the results of a pilot study on proverbs that showed a significant difference in the accuracy of interpreting modern and traditional proverbs as a function of age is also discussed.

What Is a Proverb? ―A Proverb Is the Wisdom of Many and the Wit of One.‖ –Lord John Russell

The Bible is the source for dozens of our most common ones.3 The Greeks and Romans of Antiquity invoked them,4 and they have been found extensively in the works of Plato, Sophocles, Homer, Cicero, Horace, and Plutarch, among others. Be they borrowings from the Middle Ages5 or modern innovations,6 proverbs appear to be an enduring part of the human linguistic experience. When Aristotle felt compelled to define them, he focused on their venerability, pithiness and ease of quotation; but, dictionaries generally define a proverb as ‗a pithy saying in widespread use that expresses a basic truth‘ or as ―a condensed but memorable saying embodying some important fact of experience that is taken as true by many people.‖7 While these definitions are certainly true, they barely even begin to express the complexities and unique characteristics of proverbs. In 1932, the American paremiologist Bartlett Jere Whiting produced a definition which has become ―proverbial‖ amongst other paremiologists:

3

E.g. ―As you sow, so you reap‖ Galatians 6:7 and ―Do as you would be done by‖ Matthew 7:12. ―Make haste slowly,‖ ―Love is blind‖ and ―Barking dogs do not bite,‖ for example, can be traced back to classical times. 5 ―Strike while the iron is hot,‖ ―No rose without thorns,‖ ―All that glitters is not gold.‖ 6 From computer culture comes the proverb, ―garbage in, garbage out.‖ This American invention has been adopted by other cultural and linguistic groups in recent years (Mieder 2004:13). 7 Various sources 4

4

Kristyl Kepley A proverb is an expression which, owing its birth to the people, testifies to its origin in form and phrase. It expresses what is apparently a fundamental truth—that is, a truism—in homely language, often adorned, however, with alliteration and rhyme. It is usually short, but need not be; it is usually true, but need not be. Some proverbs have both a literal and figurative meaning, either of which makes perfect sense; but more often they have but one of the two. A proverb must be venerable; it must bear the sign of antiquity, and, since such signs may be counterfeited by a clever literary man, it should be attested in different places at different times. This last requirement we must often waive in dealing with very early literature, where the material at our disposal is incomplete (Whiting 1932: 302; as quoted by Mieder, 2004: 2).

This definition touches on many of the most unique characteristics of proverbs, but it is far too much of an eyeful to be either practical or informative. Yet, there is a general sense that people intuitively understand what a proverb is. Perhaps the definition of a proverb, then, is somewhat like the Supreme Court‘s ―proverbial‖ litmus test for defining pornography: we know it when we see it. In 1985, with this in mind, Wolfgang Mieder—one of the world‘s foremost paremiologists—developed the following definition of a proverb in response to this folk conception, based upon the most frequently cited characteristics of proverbs as mentioned by a random sampling of 55 Vermont citizens: A proverb is a short, generally known sentence of the folk which contains wisdom, truth, morals, and traditional views in a metaphorical, fixed, memorizeable form which is handed down from generation to generation (Mieder 1985: 119, as quoted in Mieder 2004:3).

Of course, this definition does little to put the matter to rest, though it does prove Mieder‘s point that there is a folk definition of proverb that is understood by most people. And, while this is certainly an accurate laymen‘s definition, how is it useful in distinguishing proverbs from clichés, proverbial phrases, slogans, Wellerisms,8 etc.? In pursuit of a more linguistically based definition, Norrick (1985: 69-71) developed an eleven-feature matrix that makes such a distinction. In his estimation, the following characteristics serve to distinguish proverbs from other speech genres: 1. Potential free conversational turn: basically, this emphasizes the freedom and fullness of the proverb—it does not rely upon a specific foregoing conversational turn for its interpretation 2. Conversational: as opposed to the ways in which jokes or riddles are announced or set off in special ways from basic conversation, proverbs are utilized as part of the conversation 3. Traditional: use over a period of time 4. Spoken: as opposed to sung, chanted, etc. 5. Fixed form: this can either be attributed based on the fossilized linguistic form of the proverb, or on the fact that the traditional nature of the proverb thereby gives it a fixed quality. 6. Didactic: or at least potentially didactic 7. General: impersonal and general in nature 8. Figurative: using an analogy or some other non-literal imagery 8

Named for the character of Sam Weller from Charles Dickens‘ Pickwick Papers (1837), a Wellerism consists of a triadic structure: a proverbial statement, the identification of a speaker, and an unexpected setting: ―Everyone to his own taste,‖ as the farmer said when he kissed the cow (Mieder 2004: 15).

As the Call, so the Echo

5

9. Prosodic: this is optional, but does serve to separate proverbs from jokes, etc. 10. Not entertaining: proverbs entertain only incidentally, it is not their primary purpose. 11. Not humorous: again, a proverb can be humorous; but humor is not its express purpose. This feature matrix is a useful construction and does serve Norrick‘s purpose in distinguishing the proverb from other genres linguistically; yet, it is purely structural and, as anthropologists well know, any model developed to account for human behavior can only rely so much on structuralist explanations. Qualities that are notably absent from his feature matrix are the culture-specific quality of the proverb—in content, employment and imagery— and the socio-cultural motivation for their use, extending beyond their base didactic nature. First, it is important to highlight what is right about Norrick‘s feature matrix definition. His points regarding the potential free conversational turn and the conversational quality of proverbs are paramount in an accurate definition. Past definitions have often required that proverbs be complete sentences, but this clearly becomes erroneous when a proverb such as ―like father, like son‖ is considered. Thus, Norrick‘s point—that a proverb need not be a complete sentence, but rather a complete and independent thought (or message) in and of itself—is an important one. Likewise, it is useful to be mindful of the conversational nature of the proverb, as this element will become more significant in a discussion of their use. Further, the mention of prosody in Norrick‘s matrix follows previous scholarship, which cites rhyme, rhythm, and repetition as frequent, but not necessary, features of proverbs.. Finally, the absence of humor or entertainment as the primary purpose or focus of a proverb needs no further discussion. To confirm traditionality, Norrick relies on a proverb‘s appearance in the Oxford Dictionary of English Proverbs; and, for purposes of his feature matrix, this seems appropriate, or at least acceptable. But, in a larger framework of definition, traditionality is a slippery feature. Certainly, a proverb whose lineage can be recounted over generations or centuries deserves the badge of traditionality, but where is the line drawn? Generally born of communities and widely popularized by literature and human interaction, even the proverbs we categorize as traditional were young once. The modern proverb ―different strokes for different folks‖ was popularized by Sly and the Family Stone in 1969. Believed to have been an expression common within the Black community since the 1950s, today it is in widespread use as an American proverb. Likewise, John Gray‘s book title Men are from Mars, Women are from Venus has become an oft quoted proverb in modern American culture, though it was only published in 1992. In short, the protean nature of language necessarily requires that we side here with the folk definition of a proverb—which emphasizes widespread use within a community of speakers—rather than focusing on traditionality as a defining feature. One of the most fascinating qualities of the proverb can be best appreciated by simultaneously considering two seemingly-contradictory features noted by Norrick: its spoken nature, and its fixedness of form. Thanks to the written word, the majority of our most well-known proverbs are an inheritance from cultures far removed from us in time and space; yet, the fundamental nature of proverbs is conversational. As a spoken form, we might expect proverbs to be passed from generation to generation with as much consistency as the messages sent in a children‘s game of telephone; and, yet, many proverbs –like ―where there is smoke there is fire‖—have been documented word for word in dozens of European

6

Kristyl Kepley

languages.9 In point of fact, proverbs are both fixed and variable. As part of our own individual processes of enculturation, we acquire the fixed forms of expressions like proverbs much like we acquire the words of our language.10 Once we master these expressions, we can modify them for emphasis, humor, or other effect. Norrick notes that for well known proverbs, the entire proverb can be summoned, simply with the mention of one crucial recognizable component or ‗kernel.‘11 Thus a singular word or phrase can invoke an entire realm of traditional cultural experience for commentary: a quality which allows for limitless experimentation and expression within the language. In order to be recognizable, a proverb must be frozen to some extent, but, to be profitable and adaptable to the reality of modern human experience, proverbs must live in a constant state of variability. The remaining three features of proverbs, as mentioned by Norrick—didactic, general, and figurative—dovetail with the culture-specific qualities of the proverb—in content, employment, and imagery—and the socio-cultural motivation for their use. The most defining characteristic of the proverb is its didactic nature. No other form of speech—save, perhaps, for fables and fairy tales—possesses advice or instruction as its foremost purpose. For this reason, the cultural element of proverbs must be considered amongst its foremost features. What makes a proverb a proverb is not simply its instructional nature, but the culture-specific content of that instruction. A proverb‘s purpose is to provide individuals with socially constituted moral and ethical advice that will not only assist individuals, but will benefit the community as a whole. In order to be a member of a community, an individual must master an enormous amount of cultural knowledge—what amounts to the whole of their community‘s cumulative experience. Over time, wise, creative and intelligent behavior within a community is noted and appreciated; and, while individuals may not pass on their genes, the memory of their thoughts, actions, works, and words will be preserved in some form (Csikszentmihalyi 1995:127). Every discovery, every profitable behavior, every innovation is a part of the vast store of information that must be available to present and future members of the community to allow for continued success, development and cohesion of that community. This preservation of knowledge cannot be taken lightly; but, how can a community ensure that such a dynamic, diverse and enormous system of knowledge is retained? The answer is Culture. In fact, what we call Culture is mostly just information stored in human brains (Richerson and Boyd 2005: 61); it is information stored in such a way as to become an implicit, pre-supposed, taken-for-granted model of the world for all of those that share that culture. In 1957, Ward Goodenough articulated this cognitive view of culture, saying: …a society‘s culture consists of whatever it is one has to know or believe in order to operate in a manner acceptable to its members…Culture, being what people have to learn as distinct from their biological heritage, must consist of the end product of learning: knowledge…We should note that culture is not a material phenomenon; it does not consist of things, people, behavior, or emotions. It is rather an organization of things. It is the forms of things that

9

Mieder (2004: 11) citing work conducted by Gyula Paczolay. Ray Jackendoff, in fact, estimates that there are as many fixed expressions in American English as there are words; thus, people have approximately 160,000 items memorized and available for use (Glucksberg 2001: 68, citing Jackendoff 1995). 11 For example, ―early bird‖ brings forth ―the early bird gets the worm‖ and ―silver lining‖ summons ―every cloud has a silver lining.‖ 10

As the Call, so the Echo

7

people have in mind, their models for perceiving, relating, and otherwise interpreting them (Duranti 1997: 27).

But constructing models that would allow for the spread of knowledge in this way is no small task; and, Searle (1997: 105) emphasizes the enormity of the prospect: ―the spread of ideas through the repetition of cultural knowledge and values requires the whole apparatus of human consciousness and intentionality.‖ One of the tools for this spread of ideas through repetition is the proverb—which functions as part of a community‘s store of knowledge and, can be considered among the memes12 of a culture. In this capacity, proverbs represent the commonsensical of everyday life—in their own culture-specific way, of course—and are primarily concerned with encouraging the culturally preferred course of action as it applies to morality, or to evaluating and shaping the perception of events; thus, proverbs are used primarily not to describe a situation so much as part of argumentation, direction and over-all advice giving as evaluative directives that reflect a larger cultural schema aimed at influencing other people in the most economical way possible (White 1987: 151). As a result, there are proverbs addressing nearly every possible field of human experience, often—and predictably—with contradictory advice. Mieder (2004 :1) cites the proverb pairs of ―Absence makes the heart grow fonder‖ and ―Out of sight, out of mind‖ as evidence that our most cited—and seemingly accepted—proverbs ―do not represent a logical, philosophical system.‖ Instead, these contradictions illustrate a finer point about the didactic content of proverbs: namely, that they are oligosemic—having their meanings embedded in the culture of which they are a part. Proverbial contradictions like these are widespread throughout the cultures of the world because Culture is not teleological, but the product of generations of individuals with varying experiences. This not only accounts for the fact that proverbs within one cultural system may be contradictory, but also for the fact that particular themes and specific pieces of advice are often present within cultures that have had no contact. We should expect a certain degree of uniformitarianism: that people living in all times and places will have distinctly human experiences which others learn from and that people living in different times and places will often approach human experiences from different perspectives specifically because they are working from different cultural models. Thus, it is not enough merely to say that proverbs are didactic. We must also acknowledge that they possess culturally-specific content. Since proverbs do not exist outside of the cultures that use them, the general character of proverbs must be analyzed through their employment within a culture. Though proverbs are oligosemic, they all do their didactic work in terms of a general, non-committal exposition. The indirect quality of proverbs is often produced by using verbs which are not marked for number or tense, permitting the proverbs to attain a quality of immemoriality which is immune to variations of time, place, or circumstance. In addition, by using qualifiers like ‗all‘ 12

The concept of a meme was proposed by Richard Dawkins as a cultural analogue to the biological gene. The analogy, though, is not an air-tight one. Though we can view the spread of cultural information through memes as similar to the way that biological information is spread via genes, memes are certainly not transmitted faithfully like genes are. Of course, biological genes are subject to mutation and may not always transmit faithfully, either. The literature on whether or not memes are a useful analogy is bountiful, and I will not address it further, except to say that the analogy is obviously not a perfect one, but the basic concept of it can be useful for discussion; and since this is a paper on figurative language, I feel quite comfortable invoking a metaphorical concept as part of my analysis.

8

Kristyl Kepley

or ‗every‘ proverbs do not allow for hedges or exceptions based on particular circumstances; in this way, proverbs are able to maintain their validity within the community as universal principles (White 1987: 153). The cultural importance of proverbs is evidenced in that they are both didactic and generalized. This feature of generalization also permits proverbs to be used as part of a depersonalizing—and therefore, mitigating—strategy, allowing what is said to be said with indirectness and, thus, allow the speaker to save face, if necessary. Norrick (1985: 29) acknowledges that the didactic character of proverbs may, at first, appear to be incompatible with the feature of generality, but this is not the case at all. How can uttering a proverb carry authoritative directive force and get the speaker off the hook at the same time?...By invoking tradition and the community as a whole, the speaker not only disappears as an individual directive agent, he also imposes the weight of social sanctions.

It is not the speaker‘s job to implore and convince. Instead, the speaker is merely acting as a mouthpiece for the general community and, thereby, can avert any caustic or protesting response. Norrick contends that a speaker shows—by employing a proverb—an assumed right to council the hearer because of a relationship between equals or one in which the speaker is ―one-up‖ (ibid). However, employing a proverb may carry with it an assumed right to council the hearer specifically because the speaker—in their capacity as a proverb speaker—is no longer speaking from a place of personal knowledge and authority, but as a representative of the community. The issue of authority and who has the power to speak a proverb is, then, a cultural issue at the core of this feature of generality. Since cultures and communities are composed of complex interpersonal relationships with varying degrees of intimacy and power, neither authority figures nor hierarchies of power are universal systems. Societal structures and the power relationships that exist between people of different ages, genders, social classes, etc., are generally unique to a culture. As a result, proverbs are an important part of the spread of cultural memes because they provide an opportunity for authoritative speech in the mouths of individuals who, for reasons of social construction, might be barred from participation. Thus, it is important to consider the culture-specific conditions of employment when analyzing the general character of proverbs. Finally, and perhaps most evidently, it is necessary to consider the culture-specific imagery apparent in the figurative nature of proverbs. Proverbs are not simply a matter of speaking, but a specific abstract linguistic construction which performs a unique function in language and inhabits a unique place in a cultural matrix. Proverbs fall under the category of what Quinn (1997: 155) calls selective use metaphors: wherein speakers intentionally select specific metaphors to communicate a point, as a part of their inherent communicative task of clarifying just what it is they mean to say. Therefore, it is the use of figurative and metaphoric language to achieve didactic ends in proverbs that necessitates their consideration within a cultural context. Abstract linguistic constructions are a particularly interesting facet of human communication because they depend upon the learning of culturally conventional linguistic structures and on the cognitive skills of categorization and schema formation; this interaction creates powerful possibilities for derivational, analogical and metaphorical construals of objects, actions and events (Tomasello 1999: 157). Proverbs function effectively as communicative devices because they set the listener up to draw practical and specific

As the Call, so the Echo

9

inferences by expressing key propositions embedded in a cultural model; in turn, the interpretation of proverbs is an interactive construction in which the speaker (1) perceives and evaluates a social situation in terms of an abstract cultural model, (2) articulates that point of view in a proverb expressing one or more interlinked propositions, which is then, (3) interpreted by the listener, who expands on those propositions by locating them in the relevant cultural model and drawing appropriate inferences (White 1987:155).

Proverbs, then, not only in the imagery they invoke but also in terms of their cultural schemas—of which they are both product and producer—are distinctly wed to their culture in a way that makes their interpretation impossible outside of its consideration.

Why Use Metaphorical Wisdom? ―For rhetoric as such is not rooted in any past condition of human society. It is rooted in an essential function of language itself, a function that is wholly realistic, and is continually born anew; the use of language as a symbolic means of inducing cooperation in beings that by nature respond to symbols.‖ –Kenneth Burke (1989:188).

Over time, human collaboration has allowed for an incredible array of categories, concepts, abstractions and analogies—which together comprise socially constituted knowledge. Evolutionary epistemologists have suggested that the purpose of knowledge is to ―provide a living species with enough predictable information so as to allow it to make adaptive responses to the environment‖ (Csikszentmihalyi 1995: 123). Socially-constituted knowledge does just this in a way that is more refined and focused because of its culturespecific nature. For the system to work—for knowledge to be effectively and reliably propagated over time—it is helpful if the novices within a community can easily access the information which will permit them to function properly within a community, thereby gaining experience so that they may become future conveyers of cultural information. The purpose, at the cultural level, is to convey knowledge in a way that can be accurately and reliably accessed, interpreted and executed by different people with different levels of cultural (and life) experience. Further, cultural knowledge must be socially distributed, meaning that the individual is not always the end point of the acquisition process and not everyone within a community has access to the same information (Duranti 1997: 30). To believe that culture is what is in the brains of individuals, means that a community‘s theory of the world must be communicated in order to be lived (Duranti 1997: 33). Thus, cultures use metaphorical wisdom with a goal to provide access to this shared cultural knowledge. Framing cultural knowledge in the form of short, pithy proverbial sayings enhances the listener‘s (and the speaker‘s) ability to recall cultural knowledge quickly and easily. In the same way that children are taught the alphabet to a tune, or verses are used to remind us of how many days there are in September, proverbs are an encapsulated, easilyretrievable means for disseminating the cultural knowledge of a community. However, as we have already seen, prosody does not a proverb make; and, if prosody were, in fact, all that was needed, why would proverbs possess a distinctly figurative, metaphorical quality? According to modern research, humans may employ proverbial wisdom because of peculiarities of brain structure and function. Lakoff and Johnson (1980: 3) have insisted that

10

Kristyl Kepley

―our ordinary conceptual system, in terms of which we both think and act, is fundamentally metaphorical in nature.‖ It is interesting, then, that Aristotle categorized ―metaphor‖ as an abnormal part of language, and that the classical assumptions of language, which served as the underpinnings for our own modern conceptions of language, only propagated this view. Lakoff summarized these assumptions thusly: all conventional language is literal; all subject matter can be comprehended literally, without the use of metaphor; and only literal language can be true or false (Katz 1998:20). In short, figurative language was not considered fundamental to language, but simply an ornamental way of describing things. However, recent research has shown that figurative language is compact, vivid and efficient for conceptualization; in fact, ―metaphor might be intrinsically related to the human ability to invent new—and meaningful—concepts that might not be explicable by recourse to some more basic literal description‖ (Katz 1998:21).13 As evidence, Lakoff and Johnson (1980: 115) note how ―so many of the concepts which are important to us are either abstract or not clearly delineated in our experience (emotions, ideas, time, etc.)‖ therefore, in order to truly comprehend those concepts, we seek to understand them through the use of concepts which are concrete and clear to us.14 Michael Tomasello, also, has said that humans create analogies like proverbs when their linguistic inventories are insufficient to meet their communicative needs. That is: It is difficult to image that human beings would conceptualize actions as objects or objects as actions—or engage in anything beyond the most rudimentary forms of metaphorical thinking—if it were not for the functional demands placed on them as they adapt conventional means of linguistic communication for particular communicative exigencies (Tomasello 1999:157).

Metaphor then is a device for expressing abstract concepts in a way that is more closely grounded in physical experience. Therefore we can see how a complementary relationship exists between the conceptual role of the metaphor and the pragmatic role of the proverb. By employing metaphorical imagery from shared-cultural experience, the propositions of a proverb are taken to be self-evident, because of the undeniable truth of the concrete, realworld experience whence the metaphor comes.

Why Is It Important to Study Proverbs? ―What is all wisdom save a collection of platitudes? Take fifty of our current proverbial sayings - they are so trite, so threadbare, that we can hardly bring our lips to utter them. Nonetheless they embody the concentrated experience of the race.‖ –Norman Douglas.

13 14

E.g. Black holes, ―string‖ theory Despite its valuable pragmatic influence, the strict conceptual metaphor view has not fared well empirically. Glucksberg (2001: 105), for example, claims that if our knowledge of abstract concepts is—as Lakoff contends—literally subsumed by our knowledge of concrete concepts, then this system would be incapable of differentiating the literal from the metaphorical. As such, I employ Lakoff‘s metaphoric representations as heuristic, but not categorical.

As the Call, so the Echo

11

Proverbs are important because they represent generalized knowledge and, as such, tell us something about cultural schemas. Studying the ways that speakers interpret proverbs may also provide cognitive scientists with new perspectives on underlying conceptual processes. The ease with which we interpret statements and construct meaning…is absolutely misleading: we feel as if we are doing no work at all. It is like listening to a speaker of English utter scores of syllables a minute: We use complicated unconscious knowledge to understand the speech but feel as if we are passive, as if we merely listen while the understanding happens by magic. With parables and proverbs, just as with language itself, we must see past our apparent ease of understanding if we are to locate the intricate unconscious work involved in arriving at these interpretations (Turner 1996: 6).15

In order to understand a proverb, a listener cannot be a passive bystander; he must translate both the metaphorical imagery and the cultural theory while applying the metaphorical wisdom to his present, real-life, circumstances. In order to accurately interpret the meaning of a proverb, an individual needs to know the particular cultural schema in which the proverb operates. Additionally, proverbs can tell us something about the process of language change as it relates to the communicative needs of speakers and how the interpretation and reanalysis of proverbs over time may lead to the death of some forms and the creation of others.

Cultural Schemas Some theory of the organization of knowledge is necessary if we are to speculate on the nature of higher order comprehension and application of knowledge. In the cognitive sciences, the essence of schema theory is that information processing is, in large part, mediated by learned or innate mental structures that organize related pieces of knowledge (Strauss and Quinn 1997: 49). This theory involves a representation of the structure of the knowledge being used, the new information structures that are being constructed and the entire course of action involved in this process; thus, schema is a process as well as a structure (Cashen 1980: 51). The cognitive study of metaphors and proverbs as cultural schema is associated with the idea that we understand the world—including language—in terms of prototypes which are simplified, generalized views of folk theories of experience (Duranti 1997:38). Cultural schemas work by organizing domains of experience from simple concepts of objects or events to elaborate knowledge systems; they connect and organize interrelated sets of elements and serve as working models for entire domains of cultural activity (Quinn 1997: 140). By sheer necessity, when humans speak to one another, the vast majority of communicated information goes unsaid; yet, the communication is still effective because cultural schemas, which are intersubjectively shared by a community, are constantly being exploited by the interlocutors—albeit subconsciously. D‘Andrade (1987:113) explains that we can identify a schema that is ―intersubjectively shared when everybody in the group 15

This point is most obvious in metonymy, where a part is used to symbolize the whole. So, in the story of the donkey and the ox, the sifted straw is metonymic for luxury and the plough is metonymic of labor and suffering; but, we do not mistake the straw as metonymic of yellow things or the plough as metonymic of man-made things: how we do this is a mystery (Turner 1996: 10).

12

Kristyl Kepley

knows the schema, and everybody knows that everyone else knows the schema, and everybody knows that everyone knows the schema.‖ The two significant consequences of this intersubjective nature of cultural models is that all members treat the intersubjective knowledge as though it were obvious fact, and that most information related to the model need not be explicitly stated (ibid). In other words, these intersubjectively shared schemas fill in the gaps of missing or ambiguous information—all that can be left unsaid because the schemas of speaker and listener are effectively equivalent. In many ways, these schemas are the foundation for interpreting language and human experience within a culture, and they function to determine the meanings that are projected onto present experience by providing the frame through which expectations for the future and memories of past events are constructed.

Cultural Schemas and the Folk Model of the Mind The structures of cultural schema form a world view unique to a particular culture, providing an understanding of the world as it is thought to be; however, these cultural schemas rest upon a folk model of the mind which acts as a standard for determining normality. Roy D‘Andrade (1987: 116) has presented the folk model of the mind as composed of a variety of mental processes and states: perceptions; beliefs / knowledge; feelings / emotions; desires / wishes; intentions; and resolution, will or self-control. Derived from American and European cultural models, his folk model of the mind emphasizes the treatment of the individual mind of a person as both a container—in various states and conditions possessing many possibilities and potentialities simultaneously—and a processor, which is actively engaged in carrying out specific operations, and thus limited to a small number of concurrent actions (ibid). This culture-specific folk model of the mind asserts that feelings and emotions are understood as reactions to the world, mediated by one‘s understanding of the world, and that desire and intention are states of being; further, this model connotes that while feelings cannot be controlled, thoughts can.16 In the same way that cultural models influence individuals‘ behavior, a culture‘s underlying folk model of the mind has serious implications for individuals‘ understanding and interpretation of the world.17 Of course, most people have no organized view of their culture‘s model of the mind, or are even aware that one exists, in large part because a culture‘s folk model of the mind is so fundamental to the cognitive process of thought and organization of knowledge as part of cultural schemas, that people consider these modelbased assumptions to be obvious fact. Proverbs act as schemas because they are simply a concise way of explaining or reinforcing the complex model of the mind that a culture possesses but generally can not articulate.

16

17

This Western belief is not shared by all cultures. Eastern religion and philosophy, for example, stress that individuals can control their feelings. A recent study illustrates this point: researchers found that when trying to determine the emotional state of an individual, North Americans only assess the facial expression of the target individual, while East Asians assess the facial expressions of all surrounding individuals to determine the target individuals‘ emotional state. Researchers conclude that this is the case because, while Western culture‘s focus is on the individual, Eastern cultures are more focused on the collective. (University of Alberta: 2008)

As the Call, so the Echo

13

Processing Cultural Schemas in Figurative Terms Paul Smolensky (1995: 36) suggests that ―we can view the top-level conscious processor of individual people as a virtual machine‖ that runs a program of cultural knowledge. In other words, the cultural rules are formulated in natural language as effective formalizations of underlying cultural knowledge; but, these rules are formulated in terms of concepts consciously used to describe the task domain, and are formulated at the conceptual level (Smolensky 1995: 37). Proverbs—as linguistic formulations of knowledge—are a part of this process. Yet, one of the hallmarks of proverbs is their use of figurative language, so to truly understand the process, we have to understand how the metaphoric and figurative language used in proverbs is processed by speakers.18 Several studies have sought to determine the processes at work in the interpretation of proverbs and other figurative language. In 1976, Brewer, Harris and Brewer conducted a study in which unfamiliar proverbs were paired with either literal or figurative paraphrases (Cashen 1980: 67). When the proverb was presented first, followed by either the figurative or literal paraphrase, there was no difference in the time needed to initiate a similarity judgment. However, when the paraphrase was presented first, followed by the proverb, longer times were needed to make similarity judgments for the proverbs that were paired with figurative paraphrases. The investigators concluded that this extra time was required for the extra step of processing, first, the literal meaning and then the figurative meaning of the unfamiliar proverb (Cashen 1980: 67-68). In an experiment undertaken in 1997, Katz and Turner also found that, even when placed in an elaborate context, a novel proverb took longer to comprehend than either a literal paraphrase of the proverb or a context in which the proverb was used literally (ibid). Subsequent cued recall memory tasks indicated that participants were able to recall the proverbial phrase when given a hint associated either with the literal or figurative sense, but only when the unfamiliar proverb was placed in a context that stressed its figurative meaning; on the other hand, the only effective cue for literal contexts was the hint associated with the literal meaning (Katz et al 1998:166). These results suggest that a literal meaning is activated for unfamiliar proverbs, and the figurative sense is generated only when there is sufficient contextual support for that sense. Subsequent studies dealing with metaphor have not always shown a sufficient difference in the time needed to process the literal and figurative senses of the metaphor, thereby suggesting that a two-stage process is not at work in these cases. However, research on the comprehension of idiomatic phrases has indicated that people interpret them literally until they encounter a ―key point‖ that signals the phrase should be interpreted figuratively (Katz et al. 1998: 165). The difference in interpretation is, in large part, directly related to the polysemic nature of language. Polysemy is the diversity of meaning in words, and is a major obstacle in the interpretation of metaphoric speech, like proverbs. By its very existence, polysemy means that sentence context must control the selection of the appropriate meaning of a word or phrase. For example, psycholinguists have suggested that at the earliest stages of language processing, multiple analyses are rapidly performed, but most of the potential interpretations 18

An incredible volume of research exists on this subject and most of it falls well outside the focus of this paper. See especially Richard P. Honeck‘s A Proverb in Mind: The Cognitive Science of Proverbial Wit and Wisdom (1997).

14

Kristyl Kepley

are quickly abandoned, with the receiver settling on a preferred interpretation long before the utterance is completed. In the case of metaphors, this explanation builds on the polysemous nature of language, as an individual‘s familiarity with oft-used metaphor may likely account for the speed at which the metaphoric interpretation is arrived. In a sense, the processor can skip the two-step process of literal to figurative interpretation because the constituents of a metaphorical phrase have become polysemous through frequent use in the metaphorical context (Glucksberg 2001: 78). Thus, it would seem that when an individual encounters an unfamiliar proverb—most specifically when a proverb is presented out of context—they attempt to understand the expression based, compositionally, on the usual meaning of the words and their syntax. When the proverbs are presented in a context that highlights their figurative sense, however, the conceptual metaphor is engaged and eases comprehension. In short, if we expect that the sentence alone provides sufficient context, then we are left with a result of total unpredictability—and, ultimately failure—in interpreting metaphors because all the traditional paths of interpretation should, theoretically, be eliminated since the metaphor does not have any relevant, literal association with the other words in the sentence. Thus, to interpret figurative language like proverbs, an individual must rely heavily on their store of cultural knowledge to provide context and, thereby, relevant meaning. And, since most figurative language is based upon learned convention, familiarity with a cultural model does not necessarily lead to accurate proverb interpretation, though we would expect that an individual familiar with the cultural model underlying a novel proverb would be more accurate in its interpretation than an individual who is familiar with a different cultural model.

Language Change When I was young, old people around here used to make [words] up all the time. Only a few people did it and they were the best talkers of all…One of them would make a new [word]and right away other people would start to use it…Only the good talkers can make them up like that. They are the ones who really speak Apache. They are the ones who make up ‗wise words‘ and don‘t have to use someone else‘s.‖ –A Western Apache Consultant speaking to the Anthropologist Keith Basso

When an individual is born into a community, they undergo an enculturation process just like all those that have come before them—both in terms of the language they speak and the cultural models they use to interpret their experiences in the world. However, this is not a one-way process in which each individual receives whole and intact specific cultural practice or linguistic items. Instead, individuals face choices from among the various models they are exposed to; these choices are not necessarily conscious and often are the result of who the speaker wishes to be associated with (Mufwene 2002). On the individual level, these competing choices are easy to imagine. The challenge has been in identifying the relationship between individual selections—made by particular speakers—and the group selections that determine the evolutionary trajectory of language (Mufwene 2002). In their landmark paper on a theory of language change, Weinreich, Labov and Herzog (1968: 188) list as their third postulate ―Not all variability and heterogeneity in language structure involves change; but all change involves variability and heterogeneity.‖ As a point of further explanation:

As the Call, so the Echo

15

Where change is involved, a certain variant will occur in the speech of children though it is absent in the speech of their parents, or, more typically, a variant in the parents‘ speech will occur in the speech of their children with greater frequency, and in the speech of their grandchildren with even greater frequency. In the community at large, successive generations will show incremental frequencies in the use of the innovative variation. The logical conclusion, as time goes by, will be the categorical use of that new variant and the elimination of older variants (Chambers 2003: 203).

Thus, most historical linguists hold that speakers change, not languages; and, all language change can be viewed as social change, rather than language change, per se, because it is speakers that both innovate and adopt new linguistic forms. A linguistic variable represents two or more ‗ways of saying the ―same thing‖‘ (Weinreich, et al. 1968:162). Three important theoretical claims are implicit in the sociolinguistic concept of a linguistic variable: first, language change does not involve an abrupt shift, but instead involves a period of variation wherein variants of a single linguistic variable exist; second, individual speakers typically use more than one variant; and third, the pattern of distribution of the variants of the linguistic variable are dependent upon social factors like socio-economic class, gender, and age (Croft 2000:54). Individual speakers, then, exercise competing variants differently for social reasons. However, this does not explain what would trigger the appearance of linguistic variation in the first place. One response (Lightfoot 1991) has been to consider language change as the result of a situation in which each speaker has a single variant of a form in their grammar, but the mix of actual speakers is in a constant state of flux, thus leading to change over generations. While this theory does explain the gradual nature of language change, again, it does not explain how specific variations arise in the first place. One further theory that argues for the action of individual speakers as the cause of language change is the invisible hand theory, put forth by Keller (1994). Keller argues that while attempting to communicate with interlocutors, speakers sometimes innovate in pursuit of their communicative goals. If the innovation leads to a change in language—or even a variation—then that change is an unintended causal effect of an intended action (Croft 2000: 59). If only one speaker innovates in this way, then the innovation will be lost. However, by Keller‘s model, any speaker in the same ecological situation—presumably one composed of the same basic communicative goal, access to a similar cultural model, and a speech act occurring under similar sociolinguistic conditions of age, gender, etc.—will behave in essentially the same way. Unlike the sociohistorical theory, in which speakers choose from already existing variants as a result of social factors in their environment, in Keller‘s model, the only process at work is the creation of the novel variant; the differential replication of that variant is simply a result of subsequent speakers producing that same innovation under the same ecological conditions. Not surprisingly, a great deal of the literature on language change emphasizes variation in phonemes or in grammatical constructions in large part because these are the most befuddling areas of change if we are to accept that an adult‘s grammar is fixed once it is acquired.19 Explaining how variation occurs in the lexicon is a far easier process, since we know that this area is the most flexible in the linguistic system of individuals,20 especially adults. In short, an 19 20

Again, a vast array of research exists in this area, but falls outside the focus of this paper. See Aitchison 2000 for evidence that a speaker‘s lexicon changes throughout adulthood. See Kerswill 1996, regarding speaker‘s ability to continue acquiring additional social and stylistic registers through adulthood.

16

Kristyl Kepley

individual‘s linguistic ‗system‘ is sensitive to exposure to the language use in the surrounding community—especially when it comes to items as easily acquirable as a word, a phrase or a proverb. Roman Jakobson‘s (1960) functional account of language behavior listed three important functions of language use, which linguists have invoked to explain language change: Referential function—communication of information Poetic Function—creativity Phatic Function—solidarity and conformity with social norms.

In regards to lexical selection, the referential function is related to normal replication. The purpose of language is to convey information; so, an underlying principle of communication is ―talk in a way in which you believe the other would talk if he or she were in your place‖ (Keller 1994:99). Thus, to follow this mechanism, an individual must conform to the linguistic convention of the community in which the speech takes place. Conformity to convention, then, is the engine of normal replication and accounts for the vast majority of our speech acts. The poetic function of language allows for increased expressivenss and, thereby, is a source for altered replication, or the creation of new variants. All human languages possess the capacity for linguistic creativity, and we actively engage in this practice on a regular basis. However, the majority of these inventions and innovations make ‗one time only‘ appearances, or become esoteric references amongst nuclear families or tightly-knit social circles. The phatic function explains the acceptance of a convention or innovation. In other words, the engine of variant selection is talking like others talk—generally, ‗others‘ that you want to be like or be accepted by. This hypothesis has been analyzed in many different forms: social psychology‘s theory of accommodation, Le Page‘s (1985) theory that talking involves acts of identity, and Labov‘s proposals regarding prestige and covert prestige to explain approaches to variation selection are but a few of these. However, none of these sufficiently account for how individual selections become group selections that determine the course of a language (nor do any of them contend to). Individuals sample their language from across a fairly large number of people in their peer group and from the generations above them. All other things being equal, learning algorithms point to the adoption of the most common variant within the system. Now, when a new variation first arises within a speech community, it is—by definition—very rare; and, thus, most speakers will not be exposed to it or will consider it an aberration and ignore it. Nettle (1999) calls this the threshold problem, because variations can only become fixed if they can pass a threshold of frequency use. However, there must be more to this equation because the logic inherent in a learner adopting whatever norm is most common around him, always leads after a few generations to ―homogeneity if the starting point is heterogeneous and stasis if the starting point is homogenous‖ (Keller 1994: 99). We know this is not the case with language since language is anything but static. In fact, in real language change situations, rare variants spread through communities in a sharply rising s-shaped curve. Nettle (1999: 99) hypothesizes that the reason for this is that learners weigh all the sources which they are exposed to equally, but use a biased learning strategy which may sometimes favor variants and, thereby, allow them to overcome the threshold of rarity.

As the Call, so the Echo

17

In order to test his hypothesis, Nettle (1999) conducted a computer simulation designed to replicate language change. After running the simulation through 200 generations with no functional biases and no status differentials between individuals, variations in language arose but could never take hold and spread because of the overwhelming majority.21 Additionally, variations that possessed a functional bias—making them either easier to acquire or utilize— were not able to take hold and spread in the simulated speech community. When social status between individuals was differentiated, however, a contrasting result emerged. These simulations showed that when a hyper-influential individual possessed a rare variant, that variation quickly spread through the immediate neighborhood. If this diffusion happened quickly enough, the rare variant attained a critical mass, overcoming the threshold problem, and replacing the previously dominant form in a sharply rising curve similar to that identified in real linguistic change (Nettle 1999: 111). We can view the engine of language change as the result of competition and selection occurring within the communicative system available to speakers. As with the evolution of species, the evolution of language is shaped by fitness. So McMahon‘s (1994: 248) ―real actuation question‖—that being: ―why some of these [linguistic] innovations die out and others catch on, spreading through the community, or why certain instances of variation become changes and others don‘t‖— can be answered in the same way that Darwin answered his propositions about biological variation: fitness. Dealing specifically with individual words, Worden (2002: 92-93) lists six major selection pressures that can be analyzed to determine fitness; however, these characteristics can be applied to determine the fitness of any linguistic structure—here, I have adapted them from Worden‘s work to apply specifically to proverbs: Useful Meaning: A proverb will tend to be used frequently if it expresses a meaning which people find useful and wish to express often. Productivity: The fitness of a linguistic species depends upon how well it cooperates with other species, so the fitness of a proverb depends not only on its meaning, but on the productivity of the constructs in which it operates. Economy: Building on the work of Zipf and others, Keller (1994: 101) explains the economic principle of language use thus: ―Talk in such a way that you do not expend superfluous energy.‖ So, if the same kernel of cultural wisdom can be expressed using various proverbs, and one is richer and more economical than the others, then that proverb tends to be used more frequently and, thereby, learned by more members of the community, and is, therefore, more likely to be continued into the next generation of speakers. Minimizing Ambiguity: Though proverbs are by their figurative and general nature somewhat ambiguous, a proverb that does not require prodigious feats of ambiguity resolution will tend to be selected for. Ease of learning: studies have shown that in order for a child to learn a word, they must collect six examples of its use in unambiguous sentences where all the other words in the sentence are familiar and the child can, from these, infer the intended meaning nonlinguistically (Worden 2002:92). Though I know of no study that mirrors this acquisition process in children as it relates to proverbs, studies do indicate that increased familiarity with a proverb leads to increased interpretability among children.22 Thus, the meaning of a proverb 21

Under extreme conditions—such as high parochialism and mutation rates—variants can achieve local persistence, but the result is an odd heterogeneous equilibria unlike what we see in real world language change (Nettle 1999: 110). 22 A 1996 study at the University of Oregon, Eugene examined concreteness and familiarity of expressions as important factors in the development of proverb comprehension in children and adolescents. As grade level increased, performance in interpretation of proverbs improved; additionally, concrete proverbs were more

18

Kristyl Kepley which is used frequently in less ambiguous constructs—or where the figurative sense of the proverb maps more seamlessly onto the literal meaning of the proverb—can be more readily inferred, and thus, we can expect that this proverb is more fit and more likely to survive.23 Social Identification: just as word evolution is continually shaped by the social aspirations of individuals, so the use of proverbs is concomitant with the social groups that individuals wish to be a part of, or those in which they find themselves. As a result, the fitness of a proverb depends upon the favor it finds within a social network of speakers. In a very real way, this favor is based upon a further application of the previously mentioned five features and how the proverb in question meets the communicative needs of this sub-group of speakers.

Since language change is a dynamic process, at any one moment in time the variants within a community will differ. As a result, apprehending a change requires numerous observations of the feature that is believed to be undergoing change, and in order that researchers can make some sense out of the appearance of these variants, it is necessary to analyze how these variants are distributed in the speech community and what they signify socially at any one moment (Chambers 2003: 205). Since proverbs—under the best of circumstances—are not necessarily even used on a daily basis by most American speakers, attaining and analyzing a corpus of actual proverb use across different social, age, and gender groups is nearly an impossible task. Thus, the pilot study undertaken in this paper attempts to analyze the potential for change in this category by attempting to predict the fitness of some traditional and modern proverbs in terms of how they are currently understood—if not necessarily used—within various groups.

Research Design In 1982, Geoffrey White, interested in how models of analogical reasoning and mappings between cognitive domains were constructed, undertook a study of proverbs which gave advice on managing a personal dilemma. He sought to determine how people go about paraphrasing the proverbs and drawing out of them the common and contrasting themes of cultural knowledge. In his study, 11 proverbs were presented to 17 participants who were then asked to explicate them. Individuals were to read the proverbs, paraphrase them and then group them according to meaning—with a further explanation regarding the groupings. The proverbs chosen said something about the relations between a person and a problem or goal— carrying an explicit or implied recommendation about the appropriate response to a problematic situation.24 White (1987:159) concluded that his data indicate that proverbial reasoning involves an inferential process that moves from the (1) assertion about some aspect of the person or easily interpreted than abstract proverbs. The results concerning familiarity support a view that comprehension develops through meaningful exposure to proverbs (Nippold and Haq 1996). 23 Further speculation about the effects of prosody may be relevant here. It is worth noting that studies have found: that a proverb which names a single object before mentioning a plurality of objects—e.g. ―One foot is better than two crutches‖—those that name the present before the future—e.g. ―Here today, gone tomorrow‖—and those that name something close before something far—e.g. ―A bird in the hand is worth two in the bush‖— are favored by speakers and, thus, have increased fitness and a better chance of survival (Norrick 1985:48). 24 The eleven proverbs he used were: 1. Every cloud has a silver lining. 2. God helps those who help themselves. 3. The grass is always greener on the other side of the fence. 4. There‘s no use crying over spilt milk. 5. Where there‘s a will, there‘s a way. 6. Necessity is the mother of invention. 7. Rome wasn‘t built in a day. 8. The squeaky wheel gets the grease. 9. You can‘t have your cake and eat it too. 10. Don‘t make a mountain out of a mole hill. 11. Time heals all wounds (White 1987:158).

As the Call, so the Echo

19

problem, (2) to an expansion of its psychological implications based on a cultural model held by that person, (3) to inferences about an appropriate response to the problem. Undoubtedly, in order for his results to be reliable, each participant would have to be accurately interpreting the proverb with the culturally accepted meaning. White noted that all the participants reported that the proverbs were ―well known‖—except for three people who said they did not know ―The squeaky wheel gets the grease‖ (ibid). For me, however, this was the most interesting part of the study: how did these 17 native English speakers acquire these proverbs in the first place? How did young people interpret the proverbs of their parents and grandparents? Were their interpretations successful? Would there be a difference in the success rate of their interpretations if they concerned proverbs introduced more recently? These questions necessarily demanded that a similar analysis be conducted on older generations‘ interpretations of proverbs. Did older speakers have a greater rate of success in interpreting traditional proverbs than younger speakers did? And, was there a difference in the interpretation success rate of older speakers when they interpreted ‗traditional‘ vs. ‗modern‘ proverbs? These questions seemed, further, to demand that a middle group of speakers also be consulted. Would their rates of interpretative success be better overall? Or would they more clearly mirror one group over another? Thus, the current research is an apparent-time study designed to determine the accessibility/understandability and, thus, the fitness level of six traditional and six modern proverbs amongst three different age groups. Like all apparent-time studies, the inferences of this study depend upon the validity of the hypothesis that the language use of speakers in these age groups will remain essentially the same over the course of their lives. I followed the same procedure used by White in his study, with only a few modifications. There were 39 participants, 17 men and 22 women, ranging in age from 18-79. All participants were native English speakers, or possessed native-like competency. All participants had had at least some college education, and were middle or upper middle class. For purposes of analysis, these speakers were divided into three age groups. The younger group consisted of 15 individuals, aged 18-24, with a mean age of 18.33 years. The middle group consisted of 15 individuals, aged 30-45, with a mean age of 30.33 years. The older group consisted of 9 individuals aged 50-79, with a mean age of 61.78 years. I supplied participants with a list of 12 proverbs; six of the proverbs were traditional and in common use, six others were new proverbs brought into usage within the last 15 years, or so. The proverbs still said something about the relations between a person and a problem or goal, and carried an implied recommendation about the appropriate course of action. The six traditional proverbs were: 1. Don‘t count your chickens before they‘ve hatched.25 2. When in Rome, do as the Romans do.26 3. Beggars can‘t be choosers.27 25

This proverb was first recorded in 1579, but made famous by Aesop‘s fables—first published in 1666. It means: don‘t assume that you will get things you want, until you have them (Bartleby.com). 26 This proverb comes from a letter to St. Ambrose, the 4th century bishop of Milan, from St. Augustine in which he explained that when it comes to what days to fast, etc., different customs prevail in different cites and it is best to do what others do while in Milan, and to do what the Romans do while in Rome (Knowles 2000). It means: When you are in an unfamiliar situation, follow the lead of those who are familiar with a situation (Bartleby.com).

20

Kristyl Kepley 4. Let sleeping dogs lie.28 5. Fight fire with fire. 29 6. There‘s more than one way to skin a cat.30

Of these, Mieder (2004) lists the first four as being used with high frequency in the United States; but I contend the remaining two are in common use as well.31 When new proverbs catch on, it is because they capture features of experience that have already become important to many—or at least to most individuals within a specific community—and because they draw on metaphors that many have already come to acknowledge as good exemplars of those features (Quinn 1997:151). In searching for modern proverbs, I found that most of them were vulgar or, at least, indelicate in nature—perhaps even greater evidence of the way that proverbs reflect the cultural schemas of their formulators. Thus, many of the proverbs used in this portion of the study were also vulgar or indelicate. Additionally, many of these modern proverbs surfaced, themselves, with variants.32 This is to be expected, as even older literary sources illustrate very clearly that today‘s well-known proverbs existed in many variants until one dominant wording finally became the standard that we recognize today (Mieder 2004). The six modern proverbs used in this study were: 7. 8. 9. 10. 27

Life is like a box of chocolates.33 Keep it Real.34 It‘s not the size of the ship, it‘s the motion of the ocean.35 Don‘t get high on your own supply.36

First found in John Heywood‘s 1546 collection of proverbs. It means: people who depend on the generosity of others are in no position to dictate what others give them (ibid). 28 This proverb is from Chaucer‘s Troilus and Criseyde, published in 1380, in which he states ―It is nought good a slepyng hound to wake.‖ It means: do not stir up a problem that has lain quiet for some time (ibid). 29 Plato and Erasmus cautioned that one should not ad fire to fire. The warning was first incorporated into idiom in Shakespeare‘s Coriolanus. It means: To combat one evil or one set of negative circumstances by reacting in kind (freedictionaary.com). 30 This proverb was first published in John Ray‘s collection of English proverbs in 1678. It was also used in Mark Twain‘s A Connecticut Yankee in King Arthur‘s Court. It means: many tasks can be accomplished in several ways (Bartleby.com). 31 A quick Google search, in March 2008, produced the following number of hits: ―Don‘t count your chickens before they‘ve hatched‖—13,000. ―When in Rome, do as the Romans do‖—103,000. ―Beggars can‘t be choosers‖—154,000. ―Let sleeping dogs lie‖—204,000. ―Fight fire with fire‖—504,000. ―There‘s more than one way to skin a cat‖—34,200. 32 A quick Google search, in March 2008, produced the following number of hits for the ‗modern‘ proverbs and their variants: ―Life is like a box of chocolates‖—312,000. ―Keep it Real‖—2,810,000. ―It‘s not the size of the ship, it‘s the motion of the ocean‖—65,000. ―Don‘t get high on your own supply—18,000. ―Don‘t let your mouth write checks your ass can‘t cash‖—300,000. ―Don‘t hate the playa, hate the game‖—118,000. An interesting side note, here, is that even without including closely related variants in the search, the number of hits for ‗modern‘ proverbs is, in many cases, equal or greater than that for ‗traditional‘ proverbs. This may be because the majority of computer and internet users and posters are from the younger generations. 33 Made famous by screenwriter Eric Roth in Forrest Gump, the line was adapted from the novel by Winston Groom, where it was originally ‗being an idiot is no box of chocolates‘. This expression roughly means that life is unpredictable. Its proverbial status has been explored by Winick (1998). 34 Briefly meaning to stay true to yourself and your circumstance and not attempt to portray yourself as something you are not, the origin of this expression is claimed by many rappers of the early 1990s—most notably Milkbone and T-Iowe. Urbandictionary.com defines it as: ―To not be fake. Be yourself.‖ 35 A proverb used most frequently in reference to sexuality to indicate that one cannot make accurate judgments about an item‘s usefulness based on its size.

As the Call, so the Echo

21

11. Don‘t let your mouth write checks your ass can‘t cash.37 12. Don‘t hate the playa, hate the game.38 Participants were asked to read the proverbs and then briefly write, in plain language, what they believed the proverb to mean. Participants were also instructed to respond to each proverb, even if they were unsure of its meaning or had never heard it before. The paraphrases were then scored for accuracy. For an interpretation to be considered ―successful‖ or ―accurate‖ it had to, at least loosely, conform to the culturally accepted interpretations of the proverb;39 and, in general, there were no significant problems in making determinations regarding the accuracy of interpretations.

Results In order to mitigate the impact of any extreme scores, the mean accuracy in the interpretation of all the traditional proverbs was calculated along with the mean accuracy in the interpretation of all the modern proverbs for each respondent. These results were then compared with the mean scores of individuals in each of the three age categories, again in order to mitigate the effect of extreme scores. An accurate interpretation received a score of 1, and inaccurate interpretation received a score of 3; thus, the higher a proverb‘s mean score, the more often it was interpreted incorrectly. Figures 1 and 2 illustrate the mean interpretation accuracy score for each individual proverb. In general, the traditional proverbs were interpreted slightly better than the modern proverbs. The mean accuracy score for all the proverbs was 1.413; while the mean accuracy score for the traditional proverbs was 1.311. Figure 1 illustrates that the two proverbs people had the most trouble interpreting were ―Let sleeping dogs lie‖ and ―Fight fire with fire.‖ The modern proverbs had a mean accuracy score of 1.515, with ―Don‘t get high on your own supply‖ and ―Don‘t hate the playa, hate the game‖ being misinterpreted far more often than the other proverbs.

36

Originally used literally by screenwriter Oliver Stone in Scarface (1983), this proverb has been extended to apply to situations wherein it is unwise for an individual to hoard all their resources. A proverb used to express that it is unwise to make promises you are unable to fulfill; it is ascribed to the band Stetasonic, and was used on their 1991 album Blood, Sweat and No Tears. 38 Attributed to rapper Ice-T as a track title on his 1999 album Seventh Deadly Sin. This proverb roughly means that we should not find fault with the actor, when it is the circumstance the actor finds himself in which is to blame. 39 These definitions were provided by Bartleby.com, freedicitonary.com, The New Dictionary of Cultural Literacy, and the Urbandictionary.com. 37

Mean AccuracyKristyl Rating Kepley of Individual Traditional Proverbs

22

2.5

Mean

2.0

1.5

1.0 Chickens

Romans

Beggars

Dogs

Fire

Ship

High

Checks

Cat

Mean Accuracy Rating of Individual Figure 1. Mean Accuracy Rating of Individual Traditional Proverbs. Modern Proverbs 2.5

Mean

2

1.5

1 Chocolate

Real

Player

Figure 2. Mean Accuracy Rating of Individual Modern Proverbs.

The results showed that age played a statistically significant role in one‘s ability to successfully interpret ‗traditional‘ and ‗modern‘ proverbs. Overall, the 18-25 age group had a more difficult time than the other groups with accurate interpretation of the traditional proverbs. The scatter-plot in figure 3 shows a marked increase in the interpretation accuracy score (representing a greater number of inaccurate interpretations) amongst the 18-25 age group. On the other hand, the great majority of speakers aged 30-45 and 50+, scored an average of between 1 and 1.33—representing, respectively, zero or one interpretation errors.

As theby Call, so the Echo Mean Accuracy Score Individual Respondent for Traditional Proverbs

23

2.75

2.50

Mean Score

2.25

2.00

1.75

1.50

1.25

1.00 15

30

45

60

75

Age

Figure 3. Mean Accuracy Score by Individual Respondent for Traditional Proverbs.

The scatter-plot in figure 4 represents the mean accuracy score by individual in interpreting modern proverbs. Here, the 50+ group has extremely varied scores, with seven of the nine participants scoring 1.6, 2.0 or 2.3—representing 2, 3 and 4 errors, respectively, out of 6 possibilities. A one-way ANOVA indicated that for traditional proverbs, there was a significant effect (F (2, 36) = 17.905, p < 0.000). The mean difference in accuracy of interpretation between individuals in the 18-25 age group, and those in the 30-45 age group was .537 (p < 0.000). The results were similar when the 18-25 age group was compared with the 50+ group. Here, the mean difference was even greater, though, at .574 (p < 0.000). However, when the average accuracy scores for the mid-range age group of 30-45 year olds was compared with the accuracy scores of the older group of 50+, no significant difference was observable (p = 0.756) indicating that any difference between the accuracy of interpretations across these two groups was most certainly due to a reason other than age. A Mean Accuracy Score by Respondent for Pearson correlation test showed significance atIndividual the .01 level. Modern Proverbs 2.75

2.50

Mean Score

2.25

2.00

1.75

1.50

1.25

1.00 15

30

45

60

75

Age

Figure 4. Mean Accuracy Score by Individual Respondent for Modern Proverbs.

24

*

Kristyl Kepley

The mean difference is significant at the 0.5 level.

Figure 5. LSD Post Hoc Tests comparing means of Traditional and Modern proverb accuracy scores by age group.

A one-way ANOVA indicated that for modern proverbs, there was not a significant effect across age groups in accuracy of interpretation (F (2, 36) = 2.186, p = 0.127). Individuals in the 18-25 age group, and those in the 30-45 age group showed a mean difference of .084, (p = 0.558) indicating that any difference between the accuracy of interpretations across these two Mean Accuracy of Interpreting Traditional groups was likely due to a reason other than age.

Proverbs by Age Group

1.7

Mean of Traditional Proverbs

1.6 1.5 1.4 1.3 1.2 1.1 1.0 18-24

30-45

Age Group Figure 6. Mean Accuracy of Interpreting Traditional Proverbs by Age Group.

50+

Mean Accuracy ofCall, Interpreting Modern As the so the Echo Proverbs by Age Group

25

1.8

Mean of Modern Proverbs

1.7 1.6 1.5 1.4 1.3 1.2 1.1 18-24

30-45

50+

Age Group Figure 7. Mean Accuracy of Interpreting Modern Proverbs by Age Group.

Paired Samples Correlations age_group 18-24

30-45

50+

N

Correlation

Sig.

Pair 1

Average_Traditional_ Proverbs & Average_ Modern_Proverbs

15

.057

.840

Pair 1

Average_Traditional_ Proverbs & Average_ Modern_Proverbs

15

.135

.631

Pair 1

Average_Traditional_ Proverbs & Average_ Modern_Proverbs

9

.367

.331

Figure 8. Correlational coefficients for mean accuracy scores on traditional and modern proverbs by age group.

When the 30-45 age group was compared with the 50+ group, the mean difference showed significance (p = 0.046); however, the difference between the accuracy rate for the youngest and oldest groups was not significant (p = 0.129). When comparing the overall performance of 18-25 year olds, the difference in means between their accuracy scores on the traditional proverbs as a whole and the modern proverbs as a whole was not significant (t(14) = 1.132, p = 0.277), with them doing only slightly better overall on interpreting modern proverbs. However, 30-45 year olds did show a significant, though slight, difference (t(14) = -

26

Kristyl Kepley

2.982, p = 0.010), performing better on traditional proverbs. The 50+ age group showed a significant difference in the accuracy of their mean scores (t(8) = -5.367, p = 0.001), performing better on the traditional proverbs. We should expect that if what was being measured with this experiment was an overall ability to interpret proverbs, then there would be a high degree of correlation between a group‘s average accuracy score for traditional proverbs and for modern proverbs. This is clearly not the case, however. Instead, correlations between accuracy scores for the two different groups of proverbs increase for the groups with age. There is virtually no relationship between the accuracy scores for modern and traditional proverbs among the 1825 group; older speakers do show some relationship, though it is a weak one. Beyond the simple accurate/inaccurate scoring system, note was taken of how participants went about interpreting these proverbs. Most of the interpretations were ―straight‖ brief explanations of a proverbs meaning in plain language; but, there were some exceptions. Older participants were more likely to use other proverbs or metaphorical language as interpretations of the prompt. For example, one 50+ participant responded to the proverb ―Don‘t hate the playa, hate the game‖ with the interpretation ―This can apply to many things. Hate war, not the solider. Hate a law, not the enforcer.‖ Another 50+ participant interpreted ―Don‘t let your mouth write checks your ass can‘t cash‖ as ―Don‘t talk the talk unless you can walk the walk.‖ One 31 year old participant responded to ―Fight fire with fire‖ as ―Don‘t bring a knife to a gun fight.‖ Only one participant from the 18-25 age group employed metaphorical speech in their interpretation of a proverb; for ―Keep it real‖ this participant wrote, ―Don‘t beat around the bush. Just be yourself.‖ When asked to paraphrase the 12 proverbs, participants in the 18-25 age group occasionally employed literal translations to explain the meaning of proverbs in their paraphrases. This was a strategy almost never used by participants in the other groups. Sometimes these literal translations were correct, and sometimes they were not. For example, one participant in the 18-25 age group wrote the paraphrase of ―When in Rome…‖ as ―When visiting a different country, just do what they do‖—while true, this paraphrase does not activate the full spectrum of cultural meaning embedded in this proverb. Sometimes literal interpretations were employed which seemed to indicate that the participant understood the proverb‘s meaning, but couldn‘t find another way to express it. For example, when responding to ―Beggars can‘t be choosers,‖ a participant from the 18-25 age group wrote, ―This means that beggars can‘t choose what they want, they should settle for anything.‖ Other times, literal translations from this group seemed to be forced attempts at making sense of a confusing proverb. This was especially the case with the proverb ―Let sleeping dogs lie,‖ which one participant interpreted as ―dogs who are sleeping tell lies‖ while another wrote ―don‘t wake a mean animal.‖ Figure 9 illustrates this trend: no participants from the two older groups employed a literal translation to interpret this proverb, while more participants in the 18-25 age group interpreted it literally than accurately.

As the Call, so the Echo

Literal Interpretations of "Let Sleeping Dogs Lie" by Age Group

27

14 12

Count

10 8 6 4 2 0

18-24

Ag e

Accurate

30-45

Gr

ou p

50+

Literal

n

tatio

rpre Inte

Figure 9. Literal Interpretations of "Let sleeping dogs lie" by Age Group.

Of the six traditional proverbs, ―Let sleeping dogs lie‖ and ―Fight fire with fire‖ were the two interpreted with the least amount of accuracy. For the modern proverbs, ―Don‘t hate the playa, hate the game‖ and ―Don‘t get high on your own supply‖ were misinterpreted most often. The 18-25 and 30-45 age groups accurately interpreted ―Don‘t hate the playa, hate the game‖ the same number of times, and misinterpreted it as ―Don‘t take things personally‖ the same number of times. This meaning was provided by the 50+ group as often as the accurate interpretation. Interestingly, in the 18-25 age group and the 30-45 age group, there were respectively 3 and 4 interpretations of the proverb as meaning ―Don‘t be jealous.‖ This was not found at all in the responses of the 50+ group, but it accounted for 21.5% and 26.6%, respectively, of the two younger groups‘ responses to this proverb. See figure 10, below:

Figure 10. Interpretations of "Don't hate the playa, hate the game" by age group.

An even more interesting result, featured in figure 11, was found for the responses to ―Don‘t get high on your own supply.‖ This proverb had an interpretation accuracy score of 2.26—by far the worst of all the proverbs.

28

Kristyl Kepley

Only five respondents gave interpretations that were straight, brief paraphrases in plain language reflecting the meaning of ―Don‘t hoard your own resources‖—the definition traditionally ascribed to the proverb. Ten other respondents provided interpretations which showed they interpreted the overall meaning of the proverb in the same way, but they either made a direct reference to the real-life situation that this proverb is figuratively invoking— that of a drug dealer who uses his own merchandise to get high—or gave a literal businessadvice interpretation; ―Do not use the drugs you should be selling,‖ and ―It is uneconomical to sidestep the market and consume your own product‖ are representative of these responses. However, the most interesting finding here was that 16 respondents reported the proverb to mean ―Don‘t be cocky.‖ This was more common among the 50+ group, accounting for nearly 78% of the responses to this proverb, but it also made up 36% of the 18-25 age group‘s responses, and 26% of the 30-45 age group‘s responses.40

Approaches to Interpreting "It is not the size of the ship..." by Age Group Figure 11. Interpretations of "Don't get high on your own supply" by Age Group.

10

Occurences

8

6

4

2

0

Straightforward

18-24 30-45

Age Gro

up

50+

Sexual Connotation

of od tion th a Me pret er Int

Figure 12. Approaches to Interpreting "It's not the size of the ship..." by Age Group. 40

Interpretation responses could be both literal and accurate.

As the Call, so the Echo

29

One final element of interest was the responses of the participants to the proverb ―It‘s not the size of the ship, but the motion of the ocean.‖ This proverb was viewed by most participants as a reference to sexuality and they seemed unable to separate the general proverbial meaning from the context in which it is most frequently used. Thus, even when the interpretations indicated that a broader meaning was understood by the participant, the interpretations were usually couched in references to sex: for example, ―good sexual technique prevails over physical prowess‖ and ―sexual satisfaction is more than the equipment.‖ Some respondents did not even try to develop a more general meaning: one male participant wrote: ―A man‘s sexual aptitude has no correlation with his penis size. (Usually spoken by men with small penises.)‖ One female participant simply wrote: ―I disagree!! I like a nice size ship.‖ Overall, the 30-45 age group seemed most comfortable invoking the purely sexual meaning of the proverb, while the 18-25 age group tried to avoid any direct references to sex, though it was clearly on their minds: ―Size doesn‘t matter, it is how the man uses it.‖

Conclusion ―For time changes everything. There is no reason why languages should be exempt from this universal law.‖ –Ferdinand de Saussure

White (1987: 152) has argued that in order to understand the use of proverbs it is more useful to ask ―what is the speaker trying to do with proverbs?‖ than ―what is the speaker trying to say?‖ In invoking a proverb, the speaker is aligning themselves, to some degree, with the traditions, beliefs and prejudices of the community. In most cultures, the speaker of a proverb is also presenting themselves as an authority or a credible source. One consequence of this is that younger speakers use fewer proverbs than older ones—presumably either because they do not see themselves as authority figures, they do not wish to align themselves with the ‗old-fashioned‘ wisdom of the community, or perhaps because they are not familiar enough with the proverbs in the first place. Like coining new slang, the production of proverbs can be used to create community among those ―who understand it and can figure out the point of the expression‖ (Cacciari 1998: 141). This kind of linguistic creativity can reify a sub-group and further confirm their exclusivity. If a newly coined statement contains elements of cultural wisdom and employs familiar proverbial traits, than we might expect it to catch on, first within the area or subgroup in which it was coined; and, with the reach of mass media, it may spread very quickly through music, film, television or print media.41 The findings, here, that younger speakers were able to interpret most of the traditional proverbs without great difficulty would seem to be consistent with what we know about parent and peer influence on younger speakers. For example, in normal cases of transmission, 41

A fascinating example of how quickly a modern proverb could spread through media networks was evidenced in March 2008 when Danny Noriega, a contestant on the wildly popular American Idol television program used the cell phone texting terminology ―tmth‖—meaning ―too much too handle‖—on a broadcast watched by nearly 30 million people. Previously used almost exclusively by speakers under 25, this researcher was surprised to overhear a 40-something woman using it the next day. A quick Google search found 7,000 ―tmth‖ hits for the three week period following that broadcast, while in the entire eleven months previous to that, only 5,500 hits for ―tmth‖ were generated. While this is not evidence of a proverb spreading, it does illustrate how rapidly linguistic creations can spread via media.

30

Kristyl Kepley

children probably carry many markers of their elders‘ speech—especially in situations in which ―their most intense and only sustained contacts are with those elders‖ (Chambers 2003: 185). Caregivers, concerned with providing cultural and ―life‖ instruction, seem likely to invoke proverbs in their interactions with children;42 and a European study (cf. Hain 1951) noted that older speakers tended to cite proverbs when speaking authoritatively to and for the benefit of younger hearers. Norrick (1985: 28) hypothesizes, further, that it is just for this reason that younger speakers may make humorous use of proverbs and why they don‘t employ them very much amongst themselves. As with other forms of speech, speakers use proverbs under the assumption that they are familiar to the receiver, and will be understood. While literal use of language is accessible to virtually anyone in the speech community, a figurative use may be accessible only to those who share specific information about the knowledge, beliefs, and cultural models of a subgroup. This may explain why some proverbs are not familiar to all speakers. As younger speakers age and begin to be more influenced by the language, cultural models and popular imagery of their peers, we can assume that any proverbs that have not become a common part of the idiolects of most of these speakers will not be used with much frequency within their youth sub-groups. Likewise, modern or newly coined proverbs may rarely be invoked in communication with individuals outside of the sub-group—either because the use of the proverb is a part of group membership and covert prestige, or because the speaker does not expect that they will be understood. This may account for the reasons that older speakers are not as good at interpreting ‗modern‘ proverbs: they simply have not heard them used in context as often. The concept of ―playing the talk market‖ may account for why the speakers in the middle-age range are successful at interpretations of both the traditional and modern proverbs (Chambers 2003: 199). We expect that rebellious adolescents will use language to distinguish themselves from other groups; but, today, more than ever we must expect that young people—and here I refer to individuals who would fall into the 18-25 and 30-45 age groups of this study—are ―exposed to a greater inventory of linguistic variants because they are exposed to a wider circle of acquaintances‖ (Chambers 2003:189). The increase in physical mobility, and the luxury of immediate and pervasive communication via cell phones and the internet, has lead to increased exposure to linguistic variants and often to preferences not witnessed in the speech of middle-aged and older adults. Chambers (2003: 199) notes that no developmental studies have been conducted documenting the stages at which young speakers begin to accommodate their adolescent accents to the pressures and the needs of the workplace; however, he hypothesizes that in subjects in whom the market pressure evokes a linguistic response, this reaction might be measurable either as an increase in the range of variable use or as a general increase in standard variants, depending upon whether the individual is restricting their accommodative behavior to the workplace and other ―adult‖ contexts, or if they have extended this behavior to all settings. This age-graded-type

42

I know of no studies that have observed and documented the rate of proverb use between parents/caregivers and children in America. However, studies in other cultural settings have shown this to be the case (cf. Penfield and Duru 1988 for discussion of proverb use among the Igbo). While we can not make the leap that proverb use for children in one society equals proverb use for children in another society, we can at least hypothesize that it is likely that younger children learn most of their cultures‘ traditional proverbs from parents and caregivers, rather than from their peers.

As the Call, so the Echo

31

hypothesis serves as a good explanation for why individuals in the 30-45 age range are comfortable and familiar with the modern proverbs and the traditional proverbs. After middle age and the formative linguistic periods, speakers reach a point wherein, at least subconsciously, they view their inventory and range of socially significant speech styles and variants to be sufficient for all the situations they are likely to find themselves in. It is reasonable that the oldest generation in the study would not be quick to incorporate the newly coined modern proverbs since, as Chambers (2003: 203) reports: even when linguistic changes take root in the speech of younger people in the same community, the older people usually remain impervious to it, or nearly so. That is perhaps a linguistic reflex of the conservatism that often accompanies aging, but is also a function of the slowing of the language-learning capability beyond the critical period.

I would also argue that two other reasons probably buttress this explanation, firstly that the sub-culture-specific imagery that is often invoked in the ―modern‖ proverbs is not part of the cultural models familiar to older speakers, and thus do not work to produce a strong link between the metaphoric imagery and the didactic goal of the proverb;43 and, older speakers may not find the didactic content of specific ‗modern‘ proverbs to be useful or necessary to fulfilling their communicative needs as speakers. This final point should not be particularly surprising if we consider that proverbs are born out of a communicative need within a cultural group, and if the cultural group we are considering here is one composed of young or adolescent speakers, then it only stands to reason that the communicative needs and memes of sub-group information that must be passed on are specific to that group, and would not necessarily be applicable to the life experience of people outside of that sub-group. In conclusion, age was a clear and significant factor in the accuracy of interpretation of proverbs but, it seems, for different reasons. In support of Katz and Turner‘s 1997 study (Katz et al. 1998), participants from the 18-25 age group had consistent difficulties with accurate interpretations, and they were more likely to employ ineffectual, literal interpretations as a strategy for deciphering an unknown proverb. The 50+ group did significantly better with the interpretation of traditional proverbs and the 30-45 group did very well on both types of proverbs, outperforming the other two age-groups on modern proverbs, and rating only slightly less accurately than the 50+ group on traditional proverbs. The lack of a correlation between performance on interpreting modern proverbs and that of interpreting traditional proverbs indicates that interpreting proverbs accurately is not a skill; instead, accuracy depends upon exposure to the proverb in contexts. As we would expect, older speakers did better than younger speakers on traditional proverbs, since they have likely had more contact with these proverbs in context. The fact that the 18-25 age group did not outperform the 50+ speakers on modern proverbs may be explained by the fact that the overall proverb use within their age group is comparatively very low, and they have not yet been exposed to even these proverbs sufficiently in context to interpret them accurately. Additionally, the younger participants may be less adept, overall, at the task of composing a 43

Here, specifically, I am making reference to ―Don‘t hate the playa, hate the game‖. While games and players are familiar to older speakers, the function of invoking the ―playa‖ as an urban figure struggling to survive in a hostile ―playing‖ environment is not fully realized in the cultural perceptions of most older speakers who are likely unfamiliar with the urban or gang-related subculture that is invoked with this proverb—or at least they are likely to be more unfamiliar with it than their children and grandchildren who have had more contact with this sub-culture through media sources.

32

Kristyl Kepley

literal paraphrase for a figurative proverb. Also, the 50+ age group may simply be better at guessing the meanings of unfamiliar proverbs because they have had more practice at applying literal interpretations to figurative language over their life spans. The 30-45 age group performs well on both proverb types because they straddle both sub-cultures: being not so far-removed from youth culture, they are likely to have had multiple exposures to the modern proverbs; additionally, they have simply been interacting with the larger, traditional culture longer, and have had more opportunity to encounter the traditional proverbs than have the youngest group of speakers. Is the language changing? If these results represent an age-graded phenomena wherein individuals begin to use more proverbs as they get older and earn authority within the community, then the answer is no. The variance in interpretation accuracy is simply the result of age. However, it is possible that these results represent a possible trend. While, there is not sufficient evidence to suggest that the modern proverbs studied here are going to become the traditional proverbs of tomorrow, there is evidence that some traditional proverbs may be losing their fitness. For example, although the proverb ―Let sleeping dogs lie‖ was on Mieder‘s (2004) list of 75 most used American proverbs, it was widely misinterpreted by the 18-25 age group. The core of a proverb is its didactic content; if that meaning is lost, then the proverb becomes effectively useless. Speakers do not retain elements of their language that do not perform a communicative function. So, if this group of young people is representative of other speakers of the same age, then it stands to reason that this proverb may fall out of the lexicon sooner than later. Harkening back to theories of language change, in the majority of cases—like those of phonemic realization, grammatical constructions, etc.—the invisible hand theory does not seem plausible, since once ecological conditions were established for a community of speakers, the innovative form would, according to the theory, spread through the community quite rapidly; yet, we know that language change takes place over a long period of time. It is possible, however, that in the case of lexical adoption and, specifically here, proverb proliferation, that the invisible hand theory could be at work specifically because novel proverbs are coined and then, presumably, propagated by individuals who exist in the same ecological condition: having need of the same basic communicative goal, access to a similar cultural model, and speaking under similar sociolinguistic conditions of age, etc. This process fits the characteristics of the invisible hand theory with random, unintended consequences resulting from in-the-moment linguistic creativity by influential members of the speech community. While it is purely speculation, this may explain why the proverbs ―Don‘t hate the playa, hate the game‖ and ―Don‘t get high on your own supply‖ had more than one widely-held interpretation. Perhaps there is no need in the community for the warning ―do not hoard your resources‖ and thus, ―Don‘t get high on your own supply‖ is starting to mean ―Don‘t be cocky.‖ Alternatively, ―Don‘t hate the playa, hate the game‖ may simply be in a state of instability, with two possible interpretations competing for dominance and, eventually, standardized meaning. Without further study, it is impossible to determine the meaning of these results. They could indicate change in progress, or they could simply show that this expression carries different meanings in different contexts. It is interesting to note, however, that both of these expressions are the product of popular media forces: the film Scarface and the rapper Ice-T. Nettle‘s discussion of a high-influence individual causing a rapidly spreading trend in a community of speakers may be applicable to these cases. The popularity

As the Call, so the Echo

33

of these two media forces may account for their familiarity and widespread use, but would not necessarily require that these proverbs were reproduced with a consistent meaning. This was only a pilot study, and its limitations have proven to be infinite. As stated earlier, the true test of fitness for all the proverbs studied here could be more satisfactorily investigated, and their frequency, method, and motivating factor of use more clearly established in a corpus of observed use. Presently, assembling such a corpus of natural speech for speakers of different ages seems an overwhelming and impossible task. Until technology makes this a more accessible path to pursue, a study that compared the interpretations of traditional and modern proverbs which sought to convey the same didactic message would be particularly informative. If pairs of proverbs—one traditional and one novel—conveying the same basic lesson were interpreted by participants, the results might tell us if the existence of the new proverbs indicated a move toward changing lexemes or if the dominant cultural images—from which proverbial metaphors are drawn—are actually changing. It would also be helpful to conduct a study which sought to determine if modern proverbs conveyed meanings consistent with traditional proverbs, or if a change in underlying cultural knowledge could be observed.

References Aitchinson, Jean. (2000). Psycholinguistic perspectives on language change. Handbook of historical linguistics, ed. Richard D. Janda and Brian D. Joseph. Oxford: Blackwell. Bonvillain, Nancy. (2000). Language, culture, and communication: The meaning of messages. (3rd ed.) Upper Saddle River, New Jersey: Prentice Hall. Bornstein, Valerie. (1991). A case study and analysis of family proverb use. Proverbium,8,19-28. Burke, Kenneth. (1989). On symbols and society. Ed. Joseph R. Gusfeld. Chicago: University of Chicago Press. Cashen, Gail L. (1980). The identification, interpretation and judgment of figurative language by high school and college students. Diss. Hofstra University. Chambers, J.K. (2003). Sociolinguistic theory. (2nd ed.). Oxford: Blackwell. Croft, William. (2000). Explaining language change: An evolutionary approach. Harlow, England: Longman Linguistics Library. Csikszentmihalyi, Mihaly. (1995). Evolutionary hermeneutics. Rethinking knowledge: Reflections across the disciplines. Ed. Robert F. Goodman and Walter R. Fisher. Albany: State University of New York Press. 123-143. D‘Andrade, Roy. (1987). A folk model of the mind. Cultural models in language and thought. Ed. Dorothy Holland and Naomi Quinn. Cambridge: Cambridge University Press. 112-149. Duranti, Alessandro. (1997). Linguistic anthropology. Cambridge: Cambridge University Press. Glucksberg, Sam. (2001). Understanding figurative language: From metaphors to idioms. Oxford: Oxford University Press. Hain, Mathilde. (1951). Sprichwort und Volkssprache. Eine volkskundlich-soziologische Dorfuntersuchung. Gießen: Wilhelm Schmitz. Harvey, David. (1990). The condition of postmodernity. Malden, MA: Blackwell Publishers.

34

Kristyl Kepley

Holland, Dorothy and Naomi Quinn. (1987). Cultural models in language and thought. Cambridge, Cambridge University Press. Jakobson, Roman. (1960). Linguistics and poetics. In Style and language. Ed. Thomas A. Sebeok. Cambridge, MA: MIT Press. Katz, Albert N. (1998). Figurative language and figurative thought: A review. In Figurative language and thought. (3-43). New York: Oxford University Press. Katz, Albert N., Mark Turner, Raymond W. Gibbs, Jr., and Christina Cacciari. (1998). Counterpoint commentary. In Figurative language and thought. (158-192). New York: Oxford University Press. Keller, Rudi. (1994). On language change: The invisible hand in language. London: Routledge. Kerswill, Paul. (1996). Children, adolescents and language change. Language variation and change, 8, 177-202. Knowles, Elizabeth (ed.) (2000). Oxford dictionary of phrase and fable. Oxford: Oxford University Press. Lakoff, George and Mark Johnson. (1980). Metaphors we live by. Chicago: University of Chicago Press. LePage, Robert B. and Andree Tabouret-Keller. (1985). Acts of identity. Cambridge: Cambridge University Press. Lightfoot, David W. (1991). How to set parameters: Arguments from language change. Cambridge, MA: MIT Press. McMahon, April. (1994). Understanding language change. Cambridge: Cambridge University Press. Mieder, Wolfgang. (2004). Proverbs –A handbook. Westport, CT; Greenwood Press. Mufwene, S. (2002). Competition and selection in language evolution. Selection, 3, 45-56. Nettle, D. (1999). Using social impact theory to simulate language change. Lingua, 108, 95117. Nippold, Marilyn A. and Faridah Serajul Haq. (1996). Proverb comprehension in youth: The role of concreteness and familiarity. Journal of speech and hearing research, 39, 166176. Norrick, Neal R. (1985). How proverbs mean: Semantic studies in English proverbs. Amsterdam: Mouton. Penfield, Joyce and Mary Duru. Proverbs: Metaphors that teach. Anthropological quarterly, 61.3, 119-128. Quinn, Naomi. (1997). Research on shared task solutions. In A cognitive theory of cultural meaning. Ed. Claudia Strauss and Naomi Quinn. Cambridge: Cambridge University Press. 137-188. Richerson, Peter J. and Robert Boyd. (2005). Not by genes alone: how culture transformed human evolution. Chicago: University of Chicago Press. Saville-Troike, Muriel. (1989). Ethnography of communication: An introduction. (2nd Ed.) Oxford: Basil Blackwell. Searle, John R. (1997). The mystery of consciousness. New York: The New York Review of Books. Smolensky, Paul. (1995). On the proper treatment of connectionism. In Cynthia MacDonald and Graham MacDonald (Eds.), Connectionism: Debates on psychological explanation. Oxford: Blackwell.

As the Call, so the Echo

35

Strauss, Claudia and Naomi Quinn. (1997). A cognitive theory of cultural meaning. Cambridge: Cambridge University Press. Tomasello, Michael. (1999). The cultural origins of human cognition. Cambridge: Harvard University Press. Tuner, Mark. (1996). The literary mind: The origins of thought and language. Oxford: Oxford University Press. University of Alberta (2008, March 7). When It Comes To Emotions, Eastern And Western Cultures See Things Very Differently. ScienceDaily. Retrieved March 16, 2008. Weinreich, Uriel, William Labov, and Marvin I. Herzog. (1968). Empirical foundations for a theory of language change. Directions for historical linguistics: A symposium, ed. Winfred P. Lehman and Yakov Malkiel. Austin: University of Texas Press, pp. 95-188. White, Geoffrey M. (1987). Proverbs and cultural models. Cultural models in language and thought. Ed. Dorothy Holland and Naomi Quinn. Cambridge, Cambridge University Press. 151-172. Winick, Stephen D. (1998). The proverb process: intertextuality and proverbial innovation in popular culture. Diss. University of Pennsylvania. Worden, Robert. (2002). Linguistic structures and the evolution of words. In Linguistic evolution through language acquisition. Ed. Ted Briscoe. Cambridge: Cambridge University Press.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 2

SPECIFYING COORDINATION AN INVESTIGATION INTO THE SYNTAX OF DISLOCATION, EXTRAPOSITION AND PARENTHESIS Mark de Vries University of Groningen, Groningen, The Netherlands

Abstract Coordination is a syntactic construction with a varying semantics. I argue that besides additive, disjunctive and more specialized kinds of coordination, there is another main type, namely one that can be characterized as specification. Grammatical configurations involving specifying coordination can be found at diverse levels of sentence structure. This provides an new perspective on a range of syntactic phenomena, such as apposition, parenthesis, left- and right-dislocation, and extraposition. Certainly, all these constructions are different, and they have been studied before in isolation – but they also have properties in common. I propose a basic typology in terms of four independent factors: restrictiveness, anchoring, backgrounding, and place of attachment. A common problem for the analysis of non-restrictive material is that it is a priori unclear how to incorporate such phrases and clauses in the host structure, since the syntactic relationship between the two cannot be expressed in terms of regular subordination, which is an automatic result of syntactic merger according to the standard view. Furthermore, several types of phrases or clauses can show up either in a dislocated position or as an intervening parenthesis, which is also unexpected from a traditional perspective. These problems can be solved by means of the particular mode of construal discussed in this chapter, specifying coordination, which is then divided into two kinds: i) restrictive specifying coordination, which can be used to explain extraposition, ii) non-restrictive (parenthetical) specifying coordination, which triggers a secondary message (a proposition if the parenthesis is anchored). Thus, it is shown that superficially very different syntactic constructions make use of the same abstract mechanisms.

38

Mark de Vries

1. Introduction Typical examples of coordination involve the combination of comparable constituents by means of the coordinative conjunctions and or or. Evidently, the meaning of these two central coordinators is different, which already suggests that coordination is a syntactic construction with a varying semantics. This idea will be confirmed in section 2, where a typology of coordinators is presented. With this as a background, I will argue in this chapter that there is another main type of coordination, namely specifying coordination. As we will see in sections 3-6, this opens up a completely new perspective on phenomena as diverse as extraposition, appositive relative clauses, intervening parentheses, and dislocation constructions, all of which can be shown to involve coordination-like characteristics. The most transparent form of specifying coordination is found in the (non-restrictive) appositional construction. Examples are my best friend, John or my neighbor, a nice guy. Here, two comparable phrases are combined; the second can be understood as a specification or explication of the first. As far as I know, Kraak and Klooster (1968) were the first to subsume appositions under coordination. The similarities between standard coordination and apposition were also stressed in Quirk et al. (1985) and Sturm (1986). Later, Koster (2000) and Rijkhoek (1998) used the idea of specifying coordination as a construction type to explain certain properties of extraposition. The present chapter, which expands on ideas I expressed in earlier work, tries to strengthen these views not only by accumulating more evidence and providing more precise analyses, but also by going beyond previous proposals in that it seeks to generalize specifying coordination to parenthetical construal. In doing so, it provides a possible approach to what is in fact a long-standing problem in generative linguistics, namely the question how to incorporate non-restrictive construction types into the grammar. This chapter is organized as follows. Section 2 contains a typology of coordinators and related linkers in English and Dutch, the two languages to which the empirical domain of the discussion will mainly be limited. Section 3 introduces the concept of specifying coordination, and shows that appositions have many characteristics of standard coordination. Furthermore, it is argued in some detail why extraposition is to be analyzed in these terms as well. Section 4 contains a discussion of non-restrictive construal. It is claimed that this always brings about a secondary message, which is then syntactically analyzed in terms of a nonrestrictive specifying coordination phrase. It is argued that there is an implicit predicative secondary proposition in appositional constructions. A structural proposal is suggested, which explains the similarities with appositive relative clauses and the existence of intermediary types. Section 5 is about anchoring. It is argued that many dislocation constructions and parentheses are anchored, like appositions. In each case, the anchor is interpreted as the subject of an implicit predication. Furthermore, backgrounding constructions are shown to be syntactically similar to afterthoughts, but semantically and pragmatically different in that they do not provide new information. Section 6 expands the idea of anchoring such that a discourse category like ‗sentence topic‘ may constitute an anchor, and it contains a discussion of left-dislocation constructions. Section 7 provides an overview of different types of parentheses, which result from a number of interacting parameters, including the possibility of anchoring, the information status, and the place of attachment of a particular parenthetical phrase or clause. Finally, section 8 is the conclusion.

Specifying Coordination An Investigation into the Syntax of Dislocation …

39

2. A Typology of Coordinators in English and Dutch In constructing a sentence, phrases can be combined by means of coordination. This syntactic operation is extraordinarily flexible:   

Coordination can be applied at any level of syntactic structure. Coordination can be applied both iteratively and recursively. There are various coordinators, which trigger different semantic relationships between the conjuncts.1

The first property is illustrated in (1a-c), where nouns, prepositional phrases and clauses are coordinated, respectively. The relevant constituents are put between brackets; coordinators are italicized. (1) a. Yesterday, I bought some [spoons and forks]. b. I am afraid we have mice [[behind the cupboard] or [in the attic]]. c. [[John is going to the cinema] and [I would like to accompany him]]. The difference between iteration and layered recursion is exemplified in (2a/b). In (2a), the subject simply consists of a list of participants in the event. In (2b), however, different instances of coordination are nested, which leads to subgroups, as is indicated by the brackets.2 (2) a. [John, Paul, Mary, and Susan] are going to the cinema tonight. b. [either [John and Paul], or [Mary and Susan]] are allowed to leave the house. Thus, coordination is a syntactic construction. The associated semantics depends on the particular coordinator that is used (regardless of whether it is overt or implicit). The central coordinators are and and or, which correspond to an additive and disjunctive relationship, respectively. Their meanings correlate with the homonymous Boolean operators. It is likely that all languages have the means to express such relationships. Zwart (2005) reports on noun phrase coordination, based on a survey of 162 languages from different families. Although there is some variation regarding the way coordination is instantiated syntactically (for example, one may distinguish the comitative strategy and the summary strategy from standard coordination; see also Dik 1968, Mithun 1988, Stassen 2000/2003, Haspelmath 2007), apparently every single one of these languages is able to relate phrases in the intended way. 1

The term conjunct is somewhat confusing. It refers to one of the coordinated phrases, whether the coordination as a whole constitutes conjunctive coordination, disjunctive coordination, or something else (see table 1). 2 Here, I am presupposing that distributive constructions are structurally similar to additive constructions. See De Vries (2005) and the references there for a more sophisticated discussion.

40

Mark de Vries

Let us consider the semantic types of coordination in English and Dutch in some more detail.3 Apart from standard coordination with and and standard disjunction with or, there are more specialized coordinators that have a more limited distribution. In Dutch, the contrasting maar ‗but‘, the explicative want ‗for‘, and the concluding dus ‗so‘ are also coordinators. For conjoined main clauses this can be shown unequivocally by the verb second test. In (3a), the second clause has obligatory V2; therefore, it is a main clause, which contrasts with the otherwise comparable subordinate clause in (3b). Moreover, there is no inversion in (3a), so the relevant linkers cannot be analyzed as (conjunctive) adverbs.4 (3) a.

Joop gaat naar huis en/of/maar/want/du Jaap gaat naar de film. Joop goes to home and/or/but/for/so Jaap goes to the movies ‗Joop is going home, and/or/but/for/so Jaap is going to the movies.‘ b. Joop gaat naar huis omdat/terwijl Jaap naar de film gaat. Joop goes to home because/while Jaap to the movies goes ‗Joop goes home because/while Jaap goes to the movies.‘ The linker noch ‗nor‘ does trigger inversion, but, unlike conjunctive adverbs such as daarom ‗therefore‘ and toch ‗still‘, it has characteristics typical of a coordinator: it can link categories other than clauses (as in leuk noch aardig ‗nice nor friendly‘ or jongens noch meisjes ‗boys nor girls‘), it cannot be preceded by a central coordinator (*en noch), and it is obligatorily left-peripheral. Let us call it a semi-coordinator. The same can be said about alsmede and alsook ‗and also‘. In English, the equivalents of the above-mentioned coordinators in Dutch can also be considered coordinators (coordinating conjunctions; Dutch, ‗nevenschikkende voegwoorden‘). A general test is that clauses introduced by a coordinator cannot be preposed (4a), unlike subordinated clauses (4b): (4) a.

(* And/or/but/for/so John is staying home,) Mary is going to the movies (, and/or/but/for/so John is staying home). b. (Because/while/since/although John is staying home,) Mary is going to the movies (, because/while/since/although John is staying home). A complete overview of coordinators in English and Dutch is provided in table 1.

3

4

In this chapter the discussion is restricted to English and Dutch, mainly. For a discussion of semantic types of coordinators, especially in Slavic languages, see also Malchukov 2004. Perhaps confusingly, the word dus ‗so‘ can function as an adverb as well. Compare (3a) to i) and ii), which are also main clauses:i) … (en) Jaap gaat dus naar de film. ii) … (en) dus gaat Jaap naar de film. [here, topicalization of dus leads to inversion]. In both cases, the clause can be introduced by the coordinator en ‗and‘, which is impossible in (3a): …(*en) dus Jaap gaat naar de film. This follows if dus in (3a) is analyzed as the coordinator itself.

Specifying Coordination An Investigation into the Syntax of Dislocation …

41

Table 1. Coordinators (central, specialized, and semi-) in English and Dutch type of coordination additive (conjunctive, copulative) negative additive disjunctive (alternative)

adversative (contrasting) causal (explicative) illative (concluding) † ††

^ ^^ * **

# ## ###

coordinator (English) and

coordinator (Dutch)

but**, (yet**)

en, (alsmede†^), (alsook†^), annex### (noch*^) of, ofwel††^^, oftewel†,††^^, danwel†^^ slash## [/], schuine streep## c.q. [casu quo] (respectievelijk**) maar**, doch†

for# (so**), ergo†

want## dus**, ergo†

(nor*) or slash## [/] c.q. [casu quo] (respectively**)

Formal register. Emphatic and exclusive. Often used with an initial coordinator (see table 2). Triggers inversion in a main clausal conjunct. Optionally triggers inversion in a main clausal conjunct. Renders both conjuncts negative. To be distinguished from its use as an adverb. Only if it is initial and not preceded by a central coordinator. To be distinguished from its use as a preposition or specialized complementizer. To be distinguished from its use as a noun. To be distinguished from its use as an adjective.

It is worthwile noting that some coordinative constructions can be combined with a socalled initial coordinator, which is in fact a focus adverb (Hendriks 2004, De Vries 2005, Johannessen 2005). An example is given in (5). (5) Zowel Joop als Anna is vorige week getrouwd. Both Joop and Anna is last week married ‗Both Joop and Anna got maried last week.‘ Such correlative constructions (also called ‗reeksvormers‘ in Dutch) are always semantically distributive. In (5), Joop and Anna cannot be understood to be married to each other: there must have been two wedding events. The agreement on the auxiliary is obligatorily singular. An overview of such coordinative combinations is given in table 2. They can be additive, disjunctive or adversative, but, as far as I am aware, there are no combinations with a causal or illative meaning.5

5

The correlative resultative construction so...that (Dutch, zo...dat) is subordinative, which is evident from the verb final word order of the that-clause in Dutch.

42

Mark de Vries

Often, coordinated phrases or clauses are semantically symmetrical (balanced), that is, they have a comparable status with respect to the context and with respect to each other. This, however, is by no means necessary. Table 2. Correlative coordinators in English and Dutch type of relation additive (conjunctive, copulative) negative additive disjunctive (alternative) adversative (contrasting) * #

correlative (English)

correlative (Dutch) en...en, zowel...als

both...and neither...nor* either...or, whether…or not only...but(...)also

noch...noch* of/ofwel/hetzij/danwel...of /ofwel/hetzij/danwel# niet alleen...maar(...)ook

Renders both conjuncts negative. Also triggers inversion in main clausal conjuncts. Not everybody likes every combination, but all sixteen combinations have been attested. Generally, there is a preference for equal parts.

Example (6) illustrates temporal consecution, (7a/b) involve implications, and (8) is the famous balansschikking (lit. ‗balance-arrangement‘) in Dutch (see Welschen 1999 for extensive discussion): (6) Let us plant these seeds now, and harvest the crop this summer. (7) a. Give him the money, or he will shoot you. b. Say that again, and I will leave the house. (8)

De kat was nog niet van huis of de muizen begonnen te dansen. The cat was yet not of house or the mice started to dance ‗As soon as the cat had left the house, the mice started to dance.‘ (fig.)

Especially the last two suggest that constructions can be syntactically coordinative, but semantically subordinative (cf. Culicover and Jackendoff 1997). Of course, causal and illative coordinations also belong to this type. Interestingly, the reverse pattern is also possible. A number of prepositions and prepositional phrases semantically act like coordinators. These are called insubordinators in Van der Heijden (1999). An example is given in (9): (9) Iedereen behalve Joop droeg een jas. Everyone except Joop wore a coat ‗Everyone except Joop was wearing coats.‘

Specifying Coordination An Investigation into the Syntax of Dislocation …

43

Here, iedereen ‗everyone‘ and Joop are noun phrases that can be used as the subject of the predicate droeg een jas ‗wore a coat‘. The relation between the two is determined by the insubordinator behalve ‗except‘. Semantically, the construction expresses subtractive coordination (‗but not‘). Despite that, a number of tests show that insubordinators are syntactically subordinative. First, the conjoined phrase can be preposed, which is impossible with coordinated phrases; compare (10a) to (10b), where the underscore indicates the normal position: (10) a.

[Behalve Joop] droeg iedereen eenjas. Except Joop wore:SG everyone a coat ‗Except Joop, everyone was wearing coats.‘ b. * [En Joop] droeg(-en) Anna _een jas. And Joop wore(-PL) Anna a coat ‗And Joop, Anna _ were wearing coats.‘ Also note that the verb obligatorily shows singular agreement with the subject in (10a). Furthermore, insubordinative clauses display verb final. Since clausal complements of behalve ‗except‘ start with the standard complementizer dat ‗that‘, for which V-final is expected anyway, I will illustrate this with the comparative insubordinator zoals ‗like‘: (11) Ik zou graag boeken willen kunnen schrijven zoals Rembrandt schilderijen I would gladly books want can write like Rembrandt paintings Kon maken. Could make ‗I would like to be able to write books like Rembrandt could draw paintings.‘ Here, the clause starting with zoals has the word order of a subordinate clause, which differs from the word order in clauses coordinated to a main clause; compare (3) above, for instance. Table 3 contains an overview of insubordinators.6 Several of these are semantically comparative (which is close to additive), or subtractive (which is close to adversative). Notice that many insubordinators can take a clausal complement starting with the subordinator dat ‗that‘ in Dutch. Ellipsis phenomena, in particular gapping, are possible in any clausal construction that is semantically coordinative. The examples in (12) show this for semantically balanced coordination and insubordination;7,8 (13) shows that regular subordination and semantically unbalanced coordination do not license gapping.

6

Here is an alphabetical list of regular subordinators (subordinating conjunctions) in English. They are closely related to (or even indistinguishable from) prepositions: after, although, if, because, before, lest, once, since, that, though, till, unless, until, whether, while. 7 Interestingly, the equivalents of (12b/c) without gapping necessarily involve subordinate clauses: ... behalve dat Joop een aap tekende ‗except that Joop drew a monkey‘; ... als Joop apen kan tekenen ‗as Joop can draw monkeys‘. 8 According to the reviewer, the equivalent of (12b) in Spanish or Catalan is uncacceptable.

44

Mark de Vries (12) a.

Joop kocht appels en Anna _ bananen. Joop bought apples and Anna bananas b. Niemand tekende een dier behalve Joop eenaap _. Nobody drew an animal except Joop a monkey c. Niemand kan zo goed dieren tekenen als Joop apen _. Nobody can so good animals draw as Joop monkeys ‗Nobody can draw animals so well as Joop monkeys.‘ (13) a. * Joop brought the cake because Anna _ the flowers. b. * Joop brought the cake, for Anna _ the flowers. Table 3. Insubordinators (including insubordinative phrases) in English and Dutch type of relation comparative

insubordinator (English)

negative comparative additive

as little as*, no more than* let alone* as well as with**, together with next to** except (for), besides†, apart from, excluded# without**

subtractive

negative substitutive adversative * **

† †† # ##

^

as, like (as...)as (more...)than

instead of contrary to

insubordinator (Dutch) zoals, evenals, net als, gelijk†† (even/zo...)als (meer...)dan evenmin als*, zomin als* laat staan* behalve^ met**, samen met naast** behalve^, met uitzondering van, uitgezonderd## zonder** in plaats van in tegenstelling tot

Renders both ‗conjuncts‘ negative. Only in the (positive or negative) comitative sense. To be distinguished from its use as a conjunctive adverb. Flemish. Follows the second ‗conjunct‘. Optionally follows the second ‗conjunct‘. Behalve can be used both subtractively and additively, depending on the context and the use of negation and adverbs like ook ‗also‘.

Thus, it seems that the possibility of gapping is semantically constrained, rather than syntactically. (See Van der Heijden 1999 for more discussion and Dutch data.) On a general note, it is thinkable that insubordinative phrases lacking a verb always involve ellipsis, even in apparently simple cases such as (9). If so, it needs to be established whether comparative

Specifying Coordination An Investigation into the Syntax of Dislocation …

45

deletion or subdeletion (see Bresnan 1973, Chomsky 1977, den Besten 1978, Corver 2006, among others) can be generalized to these constructions, or if different types of ellipsis must be distinguished. These issues are beyond the scope of this chapter. Table 4. Conjunctive adverbs (and some comparable transitional phrases) in English and Dutch Type of relation Contrasting

Substitutive Substitutive-concluding Concluding

Additive

Negative additive Temporal-additive

Temporal†

Comparison Explication, illustration

Conjunctive adverb (English) However, though* Conversely, (by contrast), (on the other hand) Nevertheless, nonetheless, (despite that), (then again) Still, yet## Anyway, anyhow, (even so), (all the same), (at any rate) Otherwise** Contrarily, rather**, (on the contrary) Instead (in brief), (to sum up) Therefore, thus So##, hence, henceforth Consequently, accordingly, (as a result) Furthermore, moreover, additionally, (in addition) Also, ...too, plus Incidentally, besides, (by the way) Neither Finally, lastly Next#, then** (at the same time) First(ly), second(ly), … (in the first/second/… place) Meanwhile, (in the meantime) Again, (once more) Hitherto Likewise, similarly, analogously Namely, (that is), (after all) (in other words) Specifically (for example), (for instance) (to illustrate) (in fact)

Conjunctive adverb (Dutch) Echter Daarentegen, (aan de andere kant) Niettemin, nochtans, evenwel, desondanks, desalniettemin Toch Althans, (hoe dan ook) Anders Integendeel (in plaats daarvan) Kortom, samengevat Daarom Dus## Dientengevolge, bijgevolg, derhalve, zodoende, deswege Verder, bovendien, daarenboven, daarnaast, buitendien, hierenboven Ook, tevens Trouwens, overigens Evenmin Ten slotte Dan* Tegelijk, tegelijkertijd (ten eerste), (ten tweede), … (in de eerste/tweede/… plaats) Intussen, ondertussen, onderwijl Opnieuw, andermaal, wederom (tot dusver), (tot zover), (tot nu toe) Evenzo, evenzeer, (net zo) Namelijk, immers (met andere woorden) (in het bijzonder) Bijvoorbeeld (om een voorbeeld te geven) (in feite)

*

To be distinguished from its use as a subordinating conjunction or preposition. To be distinguished from its use as a regular adverb. # To be distinguished from its use as an adjective. ## To be distinguished from its use as a coordinator or regular adverb. † To be distinguished from temporal adverbs relating to events (instead of the discourse). **

Next, let us have a look at conjunctive adverbs and similar phrases (Dutch, ‗voegwoordelijke bijwoorden‘). These are adverbs that relate a proposition to the discourse (context), instead of modifying the proposition itself. Crucially, they do not syntactically conjoin clauses (although this may apparently be the case if they figure in juxtaposed sentences). Common examples are however, furthermore, and finally. An overview is

46

Mark de Vries

presented in table 4. Importantly, conjunctive adverbs must be semantically distinguished from modal sentence adverbs such as certainly or probably, which involve an evaluation of the truth or probability of the relevant proposition by the speaker. Both, of course, involve some kind of meta-information. In English, conjunctive adverbs and phrases are often left-dislocated, but in principle they can also be parenthetically constructed in other positions (whether intervening or rightperipheral), even though not every single one of them has all of these possibilities. An illustration is (14): (14)

(However,) Joop (,however,) did not like this proposal (, however).

In Dutch, conjunctive adverbs are often constructed as regular adverbs. If they are fronted to the first position, they trigger inversion, as is shown in (15): (15) a.

Joop heft daarentegen voor het voorstel gestemd. Joop has conversely for the proposal voted ‗Conversely, Joop has voted for the proposal.‘ b. Daarentegen heeft Joop voor het voorstel gestemd. However, they can also be treated as parentheses: (16) (?Daarentegen,) Joop (,daarentegen,) heeft voor het voorstel gestemd (?, daarentegen). Some others must be left-dislocated: (17) a.

Integendeel, Joop heft voor het voorstel gestemd. on.the.contrary Joop has for the proposal voted b. * Integendeel heeft Joop voor het voorstel gestemd. c. * Joop, integendeel, heeft voor het voorstel gestemd. d. * Joop heeft voor het voorstel gestemd, integendeel. At present it is unclear to me how to predict these differences from the lexical semantics of the relevant conjunctive adverbs, so I will leave the issue aside. Finally, to be complete, there are some frequent correlative combinations that are semantically coordinative, but do not involve a coordinator; see table 5. To summarize, there are several true coordinators and other elements that behave semantically comparable. There is a clear overlap in the various semantic relationships that are brought about by such constructions. Proper coordination itself must be defined as a syntactic construction, which, depending on the particular coordinator, can express a variety of meanings. This, then, will serve as a background for the next sections, in which I will propose that there is another semantic main type of coordination, which will shed light on

Specifying Coordination An Investigation into the Syntax of Dislocation …

47

appositional constructions, extraposition, and various types of parentheses. As a result, the meaning overlap between conjunctive adverbs and true coordinating constructions will be expanded even further. Table 5. Correlative combinations in English and Dutch Type of relation Contrasting

Comparative Correlation

Combination (english) For one thing...for another thing, on one hand...on the other hand – Like...so The [+comparative]...the [+comparative]

Combination (dutch) Enerzijds...anderzijds aan de ene kant...aan de andere kant Nu eens...dan weer (‗at one time...then again‘) Zo(als)...zo Hoe [+comparative]...hoe/des te [+comparative], des te [+comparative]...des te [+comparative]

3. Specifying Coordination at Various Levels 3.1. Appositions As Conjuncts Appositional constructions consist of an anchor and the actual apposition, a nonrestrictive modifier, which provides a specification or alternative description of the anchor.9 A typical example is given in (18), where Joop is the anchor and my roommate the apposition: (18) Have you met Joop, my roommate? The intonation suggests that the apposition is a kind of parenthesis (more evidence for this is presented in section 4). Even so, there are clear indications that an apposition is structurally attached to its anchor, and hence forms a constituent with it. In (18), the anchor is the direct object. In (19a) it is the subject, in (19b) the indirect object, in (19c) the complement of an adverbial prepositional phrase. In all cases, the apposition appears adjacent to the anchor. (19) a. Joop, my roommate, gave me a birthday present. b. I gave Joop, my roommate, a birthday present. c. I was given a present by Joop, my roommate, yesterday. Possible hypotheses are that appositions are right-adjoined to the anchor, or that they are parenthetical orphans forced into the desired position by some adjacency constraint. 9

Just to be clear, this chapter does not concern restrictive modifiers (as in my cousin John), whose semantic and structural relationship with respect to the modifyee is quite different – and therefore should not be included under the definition of apposition, in my view.

48

Mark de Vries

However, I will argue that there is a more explanatory analysis, which involves structural coordination of an apposition to its anchor. Several facts support this view (these will be illustrated directly below):       

Like conjuncts, appositions can be attached to phrases of any syntactic category. Normally, an apposition is functionally equivalent to the anchor with respect to the context. This is similar to the constraint concerning coordination of likes. Like conjuncts, appositions can be iterated. As in complex coordinate structures, appositions can be arranged recursively. Like conjuncts, nominal appositions usually receive the same Case as the anchor in many languages. Appositions behave like conjuncts with respect to constraints on movement. Coordinators may be used as linkers in appositional constructions.

Canonical appositional constructions involve noun phrases. But this is certainly not the only possibility. In (20a-d) we find anchor-apposition pairs that are prepositional phrases, adjectival phrases, clauses, and verbs, respectively: (20) a. b. c. d.

Let us meet over there, in the pub. She is intelligent, even bright. The bullet went astray, (that is,) it missed the target. Bill moved, (or rather) ran down the hill.

In (20c/d), that is and or rather are linkers commonly used in appositional constructions. As was illustrated in section 2, regular coordination can also be applied at any level of syntactic structure; compare (1), for instance. Despite this remarkable flexibility, not anything goes. The examples in (21) and (22) show that both parts must be functionally equivalent. (21) a. Joop was ill-humoured and in a state of distress yesterday. b. b. # Joop was ill-humoured and in the courtroom yesterday. (22) a. Joop was ill-humoured, (that is,) in a bad mood, yesterday. b. b. # Joop was ill-humoured, (that is,) in the courtroom, yesterday. Notably, there is no strong syntactic requirement that all conjuncts must be of equal category (Williams‘s 1981 Law of Coordination of Likes), but there must be a functionalsemantic correspondence (see Dik 1968, Sag et al. 1985, and Hendriks 2001 for discussion, among others). This is the case for coordination and apposition in a similar fashion.

Specifying Coordination An Investigation into the Syntax of Dislocation …

49

The phrases in (23) show that appositions can be iterated. What is important is that this is possible in principle, regardless of the issue whether such combinations are frequent or not. (23) a. Joop, my roommate, a nice guy, also a linguist, ... [gave me this present.] b. [He is] smart, really intelligent, bright, even. c. [They still have not found] dictator X, our enemy, our worst nightmare, ... Iteration is common for conjunctions and disjunctions, but impossible with the more specialized coordinators; for instance, we have X or Y or Z…versus X but Y (*but Z…), or X for Y (*for Z…). Even a layered recursive arrangement is possible. This is related to the fact that there are several semantic subtypes of appositions, as will be explained in more detail section 4. Appositions can be identificational (as in my roommate, Joop), predicational (as in Joop, a nice guy), or inclusive (as in some guys, Joop for instance). A complex example showing recursive groupings of the pattern [[A, B], [C, D]] is provided in (24), which can be compared to recursive coordination in (2b). (24) [[Insects, those six-legged animals], and in particular [the mosquito, the most annoying of all]], swarm around in big numbers during the summer. Here, and in particular is a linker used for inclusive appositions. Next, we predict the existence of a pattern in which a nominal apposition receives the same Case as its anchors. Since this cannot be empirically established in English or Dutch (because these languages do not have morphological Case markings on full noun phrases), consider the following equivalent examples from Icelandic (25) and Russian (26).10 Here, the relevant appositions are italicized. (25) a.

Árið 1973 tók Skylab tvö dýr, köngulærnar Arabellu og year:DEF 1973 took skylab two:ACC animals:ACC spiders:ACC Arabellu:ACC and Anítu út í geiminn. Anítu:ACC out in space:DEF ‗In 1973, Skylab took two animals, the spiders Arabella and Anita, into space.‘ b. Þær urðu að búa til mikilvægasta verfæri sitt, vef, í lausu lofti. They had to make to most.important tool:ACC their:ACC web:ACC in free air ‗They had to conduct their major device, a web, in free fall.‘ c. Festingin milli líkamshlutanna, mittið, er mjög mjó. connection :DEF.NOM between body.parts:DEF waist:DEF.NOM is very narrow ‗The connection between the segments, the waist, is very thin.‘ d. Þeir gáfu Anítu, hinni köngulónni, þetta líka. 10

Many thanks to Herman Heringa, and indirectly Gunnar Hrafn Hrafnbjargarson and Evguenia Markovskaya, for providing me with these data.

50

Mark de Vries They gave Anita:DAT other:DAT spider:DAT this too ‗They gave it to Anita, the other spider, as well.‘ (26) a. V 1973 Skulab vzjal dvuh životnyh, aukov Arabellu I Anitu, in 1973 Skulab took two:ACC animals:GEN spiders:ACC Arabella:ACC and Anita:ACC v kosmos in space. ‗In 1973, Skylab took two animals, the spiders Arabella and Anita, into space.‘ b. Oni dolžny byli sozdatj svoe osnovnoe prisposoblenie, pautinu, v svobodnom padenii. They must were create their major device:ACC web:ACC in free fall ‗They had to conduct their major device, a web, in free fall.‘ c. Svjazj meždu otdelami, talija, očenj tonkaya. connection:NOM between segments waist:NOM very thin The connection between the segments, the waist, is very thin.‘ d. Oni dali ih Anite, drugomu pauku, tože. They gave them Anita:DAT other spider:DAT too ‗They gave it to Anita, the other spider, as well.‘ a.

In each instance, the Case of the apposition is adapted to that of the respective anchor. In (25a/b) and (26a/b) this is accusative; in (25c) and (26c) it is nominative; in (25d) and (26d) it is dative. If appositions were only parenthetical adjuncts, these data would require an independent explanation. However, if appositions fall under the umbrella of coordination, they simply conform to a well-known pattern.11 From this perspective, it is also expected that appositions comply to restrictions on displacement known for conjuncts (in particular, the Coordinate Structure Constraint, which goes back to Ross 1967). Compare the examples in (27) through (31), which show the impossibility of preposing, stranding in the middle field – by moving the first conjunct or anchor –, and extraction (in this order). In each case, an underscore indicates the unmarked position of the moved constituent, which is italicized. The examples in (27) – (29) are in Dutch. First, (27) shows that topicalization of the complete coordinate or appositional construction is perfectly fine; then, (28) and (29) show that topicalization of either the second conjunct/apposition or first conjunct/anchor is impossible. (27) a.

11

Joop en Anna ben ik _ vandaag niet tegengekomen. Joop and Anna am I today not encountered

Case distribution is not the only pattern known. It seems that in some instances a default Case shows up on the apposition, depending on the particular language/dialect and the syntactic funtion of the anchor. Interestingly, similar phenomena have been reported for regular coordination; see especially Johannessen 1998. In appositional constructions, we found examples involving nominative and oblique Case. Currently, Herman Heringa is running a cross-linguistic investigation concerning this issue. Thanks to Henk van Riemsdijk and Alexander Grosu (p.c.) for bringing this to our attention.

Specifying Coordination An Investigation into the Syntax of Dislocation …

51

‗Joop and Anna, I haven‘t encountered _ today.‘ b. Joop, mijn buurman, ben ik _ vandaag niet tegengekomen. Joop my neighbor am I today not encountered ‗Joop, my neighbor, I haven‘t encountered _ today.‘ (28) a.

* En Anna ben ik Joop _ vandaag niet tegengekomen. And Anna am I Joop today not encountered ‗And Anna, I haven‘t encountered Joop _ today.‘ b. * Mijn buurman, ben ik Joop _ vandaag niet tegengekomen. My neighbor am I Joop today not encountered ‗My neighbor, I haven‘t encountered Joop _ today.‘ (29) a.

* Joop ben ik _ en Anna vandaag niet tegengekomen. Joop am I and Anna today not encountered ‗Joop, I haven‘t encountered _ and Anna today.‘ b. * Joop ben ik _, mijn buurman, vandaag niet tegengekomen. Joop am I my neighbor today not encountered ‗Joop, I haven‘t encountered _ my neighbor today.‘ The examples in (30) and (31) show that wh-extraction with P-stranding out of a complement PP within an indefinite noun phrase is possible in English, but not if this noun phrase is a second conjunct or apposition. (30) a. Which topic did you study books about _? b. Which construction did you discover an interesting aspect of _? (31) a. * Which topic did you study articles on coordination and books about _? b. * Which construction did you discover wh-movement, an interesting aspect of _? Thus, restrictions on displacement work out similarly for regular conjuncts and appositions. Finally, (32) shows that the central coordinators or and and, and even but in some cases, can be used as linkers in appositional constructions.12

12

For attributive appositions, asyndetic construal (ø) is the most common. An example is Joop, a nice guy.

52

Mark de Vries (32) a. Joop lives in The Netherlands, or Holland, as it is often called. b. Joop loves to visit capital cities, and/but especially Amsterdam. Comparable examples in Dutch are the following: (33) a.

Joop is een taalkundige, en wel een syntacticus. Joop is a linguist and indeed a syntactician ‗Joop is a linguist, namely a syntactician.‘ b. Joop woont in de ‗metropool van het Noorden‘, ofwel Groningen. Joop lives in the metropole of the Northor.indeed Groningen ‗Joop lives in the ‗metropole of the North‘, that is, Groningen.‘ c. Joop houdt van mensapen, maar in het bijzonder van chimpansees. Joop loves of apes but in the particular of chimpanzees ‗Joop loves apes, but in particular chimpanzees.‘ Here, the specifying linkers en wel and ofwel contain the coordinators en and of. Similar patterns have been reported for other languages. In Czech, neboli ‗or/that is‘ can be used as an appositional linker, and it contains the central coordinator nebo ‗or‘ (Heringa 2007). All of this strongly suggests that appositional constructions involve structural coordination (see also Kraak and Klooster 1968, Quirk et al. 1985, Sturm 1986, Koster 2000). Of course, this type of coordination is semantically different from the types listed in table 1. Thus, we are led to the conclusion that there is yet another semantic type, which can be called specification. Thus, specification constitutes another main type of coordination. Compare the examples in (34): (34) a. b. c. d.

The Netherlands and Belgium (additive coordination) The Netherlands or Belgium (disjunctive coordination) not The Netherlands, but Belgium (adversative coordination) The Netherlands, Holland (specifying coordination)

In (34a-c) the two conjuncts have a different referent. This is not the case in (34d). An apposition provides an alternative description of the anchor, or, depending on the particular subtype, a property or subset of the anchor. In section 4, I will show that the appositional construction involves more than just this: appositional constructions, by definition, are non-restrictive; furthermore, the apposition can be understood as a predicate of the anchor. However, it will become clear that this must be attributed to other factors. For now, it is sufficient to recognize that apposition involves structural coordination, and, consequently, that there is a type of coordination, specification, that is semantically different from the generally acknowledged types such as conjunction and

Specifying Coordination An Investigation into the Syntax of Dislocation …

53

disjunction. This enforces the idea that coordination is a syntactic construction with a varying semantics. What does syntactic coordination look like? I will assume the basic structure in (35). For the present purposes, the functional layer needed to account for distributivity effects (see De Vries 2005) will be ignored. (35)

In this representation, the coordinator is a functional head projecting into a coordination phrase, in which the conjuncts are the specifier and the complement, hence the arguments. See Munn (1993) and Johannessen (1998), among others, for discussion. The CoP creates a structural position for the conjunction. The coordinative head can be instantiated by an additive conjunction (and), disjunction (or) or some other linker (possibly lexically zero).13 Therefore, the semantics of the construction is determined by the lexical choice of the head, the coordinator. Needless to say, this approach neatly corresponds to the insights underlying X-bar theory. Moreover, it accounts for the fact that the coordinator forms a constituent with the second conjunct, which was already convincingly shown in Ross (1967). In section 3.2 I will argue that apposition is not the only phenomenon that involves specifying coordination. Extraposition is to be analyzed in these terms as well. In section 4, I will return to appositional constructions.

3.2. Extraposition in Terms of Coordination 3.2.1. Introduction A typical case of extraposition is illustrated in Dutch in (36a/b). The original position of the relative clause die een rode hoed droeg ‗who wore a red hat‘ is next to its antecedent een man ‗a man‘ (36a); in the extraposed order, it surfaces to the right of the participle gezien ‗seen‘ (36b). This shift to the right is optional.14 The same can be done with a prepositional phrase such as met een rode hoed ‗with a red hat‘. (36) a. 13 14

Ik heb een man die een rode hoed droeg gezien. I have a man who a red hat wore seen

A discussion of phrasal linkers is beyond the scope of the present chapter. The obligatory position of complement clauses to the right of the verb can be explained by a VO basis. Nominal phrases are moved into the middle field. See Zwart (1994) and subsequent work for discussion. In this chapter, I will only discuss true extraposition, which is optional.

54

Mark de Vries ‗I saw a man who wore a red hat‘ b. Ik heb een man gezien die een rode hoed droeg.

Building on ideas by Koster (2000), I argue in this section that extraposition involves coordination. More specifically, the analysis entails that an extraposed constituent is basegenerated within a second conjunct attached to the relevant clause at the required structural level, ranging from VP to CP, depending on the position of the anchor. The type of coordination involved is specifying coordination. Differently from Koster‘s original proposal, it is claimed that the coordination is syntactically balanced, and that there is phonological deletion inside the second conjunct. In what follows, I will use examples from Dutch mainly. This is convenient, since Dutch has an easily detectable ‗right sentence bracket‘: the final verb, verb cluster or verbal particle. (Recall that only the finite verb in main clauses is moved to the second position.) The discussion is limited to restrictive relative clauses and prepositional phrases at first, but section 3.2.5 broadens the empirical scope.

3.2.2. Why Extraposition Is Not Rightward Movement, Stranding, or BaseAdjunction Several potential analyses of extraposition have been proposed. Possibly, it involves rightward movement (Reinhart 1980, Baltin 1983, Büring and Hartmann 1995, among others). Alternatively, the extraposed material could be stranded in a right-peripheral position, while the antecedent moves leftward (Kayne 1994).15 These ideas are roughly sketched in (37). A third approach is that the extraposed material is base-generated right-peripherally, to which I will return shortly. (37) a. Ik heb [een man ti] gezien [die een rode hoed droeg]i. (rightward movement) b. Ik heb [een man]i gezien [ti die een rode hoed droeg]. (stranding) There are strong arguments against these analyses. The most obvious objection to stranding is that extraposition from embedded positions is possible. The reason is that what needs to be moved leftward is not a constituent in this analysis. Even the sentence in (38) would be a problem, let alone examples such as (39): (38) Ik heb [de man] gezien die een rode hoed droeg. I have the man seen who a red hat wore ‗I saw the man who wore a red hat‘ (39) De politieagent heeft [de papieren van de man] gecontroleerd die een rode hoed droeg. The police.officer has the papers of the man checked who a red hat wore ‗The police officer checked the papers of the man who wore a red hat.‘ 15

A related variant is Wilder‘s (1995) ‗copy plus scattered deletion‘.

Specifying Coordination An Investigation into the Syntax of Dislocation …

55

In (38), there is extraposition from a simple definite noun phrase, which is very common. This would require movement of the determiner plus the noun (stranding the modifier), but these do not form a constituent. The problem is amplified in (39); see the schematic illustration in (40): (40) … V [D [N [P [D [N modifier]]]]] Here, what has to be moved to the left of the verb is the underlined part. Clearly, this is highly undesirable, given general assumptions about syntactic processes. Other strong arguments against the stranding theory are the fact that there can be extraposition from subjects and other constituents (see section 3.2.3 below), and the resulting mirror pattern in the case of multiple extraposition; these patterns cannot be explained. Furthermore, stranding in the middle field is impossible; compare (41a)-(41d), for instance: (41) a.

Heb je de mannen gezien die een rode hoed droegen? Have you the men seen who a red hat wore ‗Have you seen the men who wore red hats?‘ b. Heb je de mannen die een rode hoed droegen gezien? c. * Hoeveel mannen heb je die een rode hoed droegen gezien? how.many men have you who a red hat wore seen d. Hoeveel mannen heb je gezien die een rode hoed droegen? If indeed (41a, b and d) can be derived, there is no way to exclude (41c) in the stranding approach. Let us consider the rightward movement theory instead. Unfortunately, the facts concerning extraposition from embedded positions constitute stark evidence against this approach as well. Consider the difference between the sentences in (42): (42) a.

Ik heb [een man uit Denemarken] ontmoet. I have a man from Denmark met ‗I met a man from Denmark.‘ b. Ik heb [een man _ ] ontmoet uit Denemarken. (extraposition) c. *Waaruit heb je [een man _] ontmoet? (wh-movement) where.from have you a man met ‗From where did you meet a man?‘ Since wh-extraction of an adjunct PP modifier from NP is unacceptable, as is shown in (42c), it is remarkable that extraposition of a PP from such an island position is fine (42b). This problem is aggravated by the fact that even extraposition from deeply embedded positions is possible in Dutch, as was already shown in (39). Compare also the sentences in (43):

56

Mark de Vries (43) a.

De politieagent heeft de papieren van een man uit Italia gecontroleerd. The police.officer has the papers of a man from Italy checked ‗The police officer checked the papers of a man from Italy‘ b. De politieagent heeft de papieren van een man gecontroleerd uit Italië. c. * Uit welk land heft de politieman de papierenvan een man gecontroleerd? From which country has the policeman the papers of a man checked ‗From which country did the police officer check the papers of a man?‘ The PP modifier uit Italië ‗from Italy‘ is embedded in the complex phrase [de [papieren [van [een [man [uit Italië]]]]]] ‗the papers of a man from Italy‘ in (43a). Thus, the (supposed) original position of the PP is inside a strong island. As (43c) shows, wh-extraction from an embedded noun phrase is completely impossible. Therefore, extraposition out of embedded positions is expected to be equally ungrammatical, contrary to fact. Unless, of course, one is prepared to accept that movement to the left behaves completely differently from movement to the right, which goes against the idea that syntactic relationships are calculated in terms of hierarchy. In the case of movement, what should be relevant is that the landing site c-commands the original position (in a Minimalist framework, this is the consequence of the strict extensional character of a Merge-based derivation). There is another asymmetry between wh-movement and extraposition: the first can be unbounded, whereas extraposition is clause-bound. The latter is known as the Right Roof Constraint (Ross‘s 1967 Upward Boundedness Constraint); compare the illustrations in (44) and (45): (44) a.

Wat zei Jan [dat Anna dacht [dat Joop gedaan had]]? What said Jan that Anna thought that Joop done had ‗What did Jan say that Anna thought that Joop had done?‘ b. Wat is nu wel duidelijk [dat Joop gedaan heeft]? What is now indeed clear that Joop done has ‗What is quite clear by now that Joop has done?‘ (45) a.

[Dat Joop een misdaad gepleegd heft die hem zal berouwen] is nu that Joop a crime committed has which him wil regret is now wel duidelijk. Indeed clear ‗That Joop has committed a crime that he will regret is quite clear by now.‘ b. * [Dat Joop een misdaad gepleegd heeft] is nu wel duidelijk die hem zal berouwen. ‗That Joop has committed a crime is quite clear by now that he will regret.‘

Specifying Coordination An Investigation into the Syntax of Dislocation …

57

In (45a), there is extraposition of the relative clause within the higher subject clause; in (45b), it is extraposed to the right of the matrix, which is sharply unacceptable. Thus, the movement approach to extraposition makes the wrong predictions. Therefore, let us turn to the possibility of base-generation. This, too, can be done in a number of ways. The most straightforward idea is right-adjunction (see Culicover and Rochement 1990, for instance).16 This assumption, however, is problematic. First, note that Dutch allows for topicalization of a remnant VP after object scrambling out of it. In (46), the indirect object has been scrambled out of the verb phrase, which itself is preposed, dragging along the direct object: (46) [ti een boek gegeven]j heb ik hemi niet tj. A book given have I hem not lit. ‗Given a book, I have not to him.‘ Now consider the following examples: (47) a.

Ik heb de man gezien die een rode hoed droeg. I have the man seen who a red hat wore ‗I saw the man who wore a red hat‘ b. [De man gezien die een rode hoed droeg] heb ik niet. c. * [Gezien die een rode hoed droeg] heb ik (niet) de man (niet). In (47a), there is extraposition of a relative clause, which, supposedly, is right-adjoined to the verb phrase. In (47b), the complete phrase containing the antecedent, the participle, and the extraposed clause is topicalized; this is some extended projection of the VP (say, AgrOP). However, (47c) shows that topicalization of the VP without the antecedent is unacceptable. This is entirely unexpected, since remnant VPs may be topicalized, as was shown above. Importantly, the pattern in (47) can be replicated with modifiers other than relative clauses, such as prepositional phrases, result clauses, and complement clauses of nouns and adjectives. Therefore, a construction-specific rule such as ―a relative clause must be preceded by its antecedent‖ is of no use.17 The general pattern is that remnant VPs containing extraposed material are inert (see Kaan 1993). Neither of the theories mentioned so far is able to explain this pattern. However, rather than elaborating on the disadvantages of analyses that were already shown to be problematic,18 I would like to present a promising alternative.

16

An alternative is proposed by Haider (1997), who assumes that the extraposed constituent is generated deeply embedded within the VP. 17 Preposing of a relative clause alone is also unacceptable: * [Die een rode hoed] droeg heb ik de man _ gezien ‗who a red hat wore have I the man seen‘. This can be attributed to other factors, such as the island status of (complex) noun phrases. 18 See De Vries (2002: chapter 7) for an extensive overview.

58

Mark de Vries

3.2.3. Extraposition as Specifying Coordination Koster (2000, and earlier lecture notes) proposed that extraposed constituents are basegenerated right-peripherally, but in a way quite different from the adjunction analysis. They can be analyzed as second conjuncts in a so-called colon phrase. These are attached at the relevant level of the main projection line – to be precise, the minimal part of the matrix clause containing the unmarked position of the extraposed material functions as the first conjunct. An example is (48): (48) a. Ik heb [:P [een man gezien] [:‘ : [die een rode hoed droeg]]]. b.

Here, een man gezien ‗a man seen‘ is an extended projection of the verb such as vP or AgrOP (I will remain agnostic as to the analysis of the middle field in Dutch). The object een man is the antecedent of the extraposed relative clause die een rode hoed droeg ‗who a red hat wore‘. The colon is a functional head that functions as a particular kind of coordinator; its meaning is described roughly as ‗namely‘. Clearly then, we can identify the colon phrase as an instance of specifying coordination. In the next section, I will propose a modification of Koster‘s analysis, but first, let us see why it is a good idea to subsume extraposition under coordination (see also Rijkhoek 1998, who focuses on result clauses). First, extraposition from embedded positions is no longer a problem, since there is no movement relationship between the two positions; therefore, no barrier needs to be crossed; see (49), for instance, which corresponds to (43b): (49) De politieagent heeft [:P [[de papieren van een man] gecontroleerd] [ : [uit Italië]]]. Second, we can now explain Kaan‘s generalization, which says that (remnant) VPs containing extraposed material are inert, as was shown in (47). Consider (50):

Specifying Coordination An Investigation into the Syntax of Dislocation …

59

(50) a. Ik heb [:P [de man gezien] [ : [die een rode hoed droeg]]]. b. [De man gezien die een rode hoed droeg] heb ik niet. The extraposed order is repeated in (50a). If we topicalize the complete colon phrase, we get (50b), which is fine. What cannot be derived, and correctly so, is (51): (51) * [Gezien die een rode hoed droeg] heb ik (niet) de man (niet). This follows from the fact that the participle plus the relative clause (gezien die een rode hoed droeg) is not a constituent in (50a). Also, de man cannot be scrambled out of the (extended) verb phrase in (50a) because that would be a violation of the Coordinate Structure Constraint, so an alternative derivation in terms of remant movement is not available. Third, it is correctly predicted that extraposition can take place from any constituent of the clause. This is illustrated in (52): (52) Joop heft [:P [iemand het cadeau gegeven] [ : [die ‘t echt verdiende]]]. Joop has someone the present given who it really deserved ‗Joop gave someone the present who really deserved it.‘ b. Er liep [:P [een man op straat] [ : [die ik niet kende]]]. There walked a man on street who I not knew ‗There was a man walking in the street who I didn‘t know.‘ c. [:P [Iemand heft me een brief geschreven] [ : [die ik niet ken]]]. Someone has me a letter written who I not know ‗Someone wrote me a letter who I don‘t know.‘ d. Ik ben [:P [ergens geweest] [ : [waar ze gratis koekjes uitdelen]]]. I am somewhere been where they free cookies distribute ‗I have been somewhere where they hand out cookies for free.‘ e. [:P [Alleen die boeken heb ik weggegeven] [ : [die ik niet meer wil hebben]]]. Only those books have I away.given which I not anymore want have ‗Only those books I have given away that I don‘t want anymore.‘ f. [:P [Hoeveel boeken heb je weggegeven] [ : [die je al hebt gelezen]]]? How.many books have you away.given which you already have read ‗How many books did you give away which you have already read?‘ a.

In (52a), the antecedent of the relative clause is the indirect object, in (52b) it is the (thematic) subject, in (52c) the subject of a ditransitive clause, in (52d) a locational adverbial phrase, and in (52e) and (52f) even a topicalized phrase and a wh-phrase (the object). In each case, the first conjunct of the colon phrase is as big a part of the matrix as necessary; this varies from VP to CP. Fourth, multiple extraposition creates mirror effects. If there is extraposition from both XP1 and XP2, and XP1 precedes XP2, the order of the extraposed constituents after the right sentence bracket is mirrored, namely EX2 > EX1. Some examples are provided in (53) and

60

Mark de Vries

(54), where the dependencies are marked with italics and underlining. In (53), there is a comparative phrase related to the subject and a prepositional modifier related to the object; in (54), there are two relative clauses. (53) a.

Meer jongens hebben de man gezien met de rode hoed dan meisjes. More boys have the man seen with the red hat than girls ‗More boys saw the man with the red hat than girls.‘ b. * Meer jongens hebben de man gezien dan meisjes met de rode hoed. (54) a.

Uitgerekend die man heft het boek geschreven dat de Pulitzer-prijs heeft precisely that man has the book written that the Pulitzer Prize has gewonnen die laatst nog overwoog zijn schrijverscarrière te beëindigen. Won who lately still considered his writer‘s.career to end ‗That man, of all people, wrote the book that won the Pulitzer Prize who only recently considered giving up his writer‘s career.‘ b. * Uitgerekend die man heeft het boek geschreven die laatst nog overwoog zijn schrijverscarrière te beëindigen dat de Pulitzer-prijs heeft gewonnen. The obligatory nested configuration is expected in the specifying coordination approach. A schematic analysis is given in (55a); a concrete example is (55b), which corresponds to (53a): (55) a. (…) [:P1 [ XP1 … [:P2 [ XP2…] [ :2 EX2]]] [ :1 EX1]] b. [:P1 [ Meer jongens hebben [:P2 [de man gezien] [ :2 [met de rode hoed]]] ] [ :1 [dan meisjes]]]. There is no way of deriving the incorrect word order in (53b) or (54b). Fifth, since specifying coordination is asymmetric (the second conjunct specifies the first), it is predicted to be impossible that extraposable material shows up to the left of the associated position. This is illustrated in (56): (56) a.

Ik heb een man gezien die een rode hoed droeg. I have a man seen who a red hat wore ‗I saw a man who wore a red hat.‘ b. * Ik heb die een rode hoed draagt een man gezien. Notice that this is a problem for theories making use of adjunction (whether basegenerated or after movement), since there would be no hierarchical distinction between (56a), where the relative clause would be right-adjoined to the VP (schematically, ...[VP [VP DO V]

Specifying Coordination An Investigation into the Syntax of Dislocation …

61

RC]), and (56b), where it could be left-adjoined to the VP (schematically, ...[VP RC [VP DO V]]). Sixth, extraposed material is known to be an island for extraction, even if it is not an island in its unmarked position. This is the so-called freezing effect; see the illustrations in (57) and (58): (57) a.

Waar heb je een artikel over _ geschreven? Where have you an article about written ‗What did you write an article about?‘ b. *Waar heb je een artikel geschreven over _? (58) a.

Waar is hij altijd dol op _ geweest? Where is he always fond on been ‗What has he always been fond of?‘ b. * Waar is hij altijd dol geweest op _? This contrast follows from the assumption that extraposed phrases are the second conjunct of a specifying coordination phrase, given the Coordinate Structure Constraint. Thus, the ungrammaticality of (57b) and (58b) can be compared to (59): (59)* What did you write a book about linguistics and an article on _? Here, the wh-phrase is moved from the second conjunct only, which is unacceptable. I conclude that the idea that extraposition involves structural coordination is strongly empirically supported. However, in the next subsection I will argue that this is not the end of the story.

3.2.4. Why Extraposition Involves Deletion In section 2 we saw that conjuncts are not necessarily of the same category, but it is certainly not the case that anything goes: conjuncts must be functionally equivalent. From this perspective, consider again the schematic representation of an extraposed relative clause or prepositional modifier in (60): (60) Ik heb [:P [een man gezien] [ : [{die een rode hoed droeg} / {met een rode hoed}]]]. Here, the first conjunct is a verb phrase, but the second a clause or a prepositional phrase. Clearly, these are not functionally equivalent. This problem can be resolved. Let us hypothesize that the second conjunct involves more than what can be overtly detected. In (61), the specifying second conjunct repeats the first conjunct, but it also includes the overtly extraposed material in its canonical position; the repeated material is phonologically deleted. In order to distinguish this theory from Koster‘s

62

Mark de Vries

original one, I will change the notation of the specifying coordinative head to and: instead of just a colon. (61) Ik heb [and:P [een man gezien] [ and: [een man die een rode hoed droeg gezien]]]. Both conjuncts are now (extended) verb phrases, hence functionally (and categorially) equivalent. The second conjunct contains more information than the first, so it can be interpreted as a specification. The relative clause is generated adjacent to its antecedent within the second conjunct. The deletion involved can be compared to gapping; this will be discussed below. The representation in (61) has a number of additional advantages over (60), and in fact over any other analysis of extraposition in terms of base-generation. The basic reason is that (61) incorporates some traits of the rightward movement analysis without taking over its drawbacks. First, the extraposed modifier is generated in its canonical position next to the modifyee (here, man). Therefore, the required (restrictive) interpretation is directly represented in the structure. No additional mechanism that links the extraposed constituent to its associated position in the matrix is necessary. Furthermore, it is now possible to maintain the raising analysis of relative clauses in extraposition contexts. Evidence for this theory is accumulated in Vergnaud (1974), Kayne (1994), Bianchi (1999), De Vries (2002), Bhatt (2002), and others. The basic idea is that the head noun of a (restrictive) relative clause finds its origin within the relative clause; it is then raised to a left-peripheral position. This is shown in (62): (62) (the) [meni that I saw ti yesterday]. See section 4.3 for more structural details. A full discussion of this theory is beyond the scope of this chapter, however. What is relevant at this point is that it is incompatible with the original coordination approach to extraposition. In (60), the antecedent and the relative clause are generated separately, so there cannot be raising of any kind. In (61), on the other hand, the head noun man is present within the second conjunct, so raising can take place as usual. Another advantage of the deletion approach is the following. Consider the example in (63a) and the analyses in (63b/c), slightly more detailed than hitherto. Here, (63b) would be Koster‘s original analysis, and (63c) is the present view: (63) a.

Ik heb alleen die man beleefd gegroet die een rode hoed droeg. I have only that man politely greeted who a red hat wore ‗I politely greeted only that man that wore a red hat.‘ b. [CP IkS heb [:P [XP [DP alleen die man]DO (X) [vP beleefd [vP tS gegroet tDO]]] [ : die een…droeg]]] c. [CP IkS heb [and:P [XP1 [DP alleen die man]DO [vP beleefd [vP tS gegroet tDO]]] [ and: [XP2 [DP alleen die man die een rode hoed droeg]DO (X) [vP beleefd [vP tS gegroet tDO]]] ]]]

Specifying Coordination An Investigation into the Syntax of Dislocation …

63

Again, there is a relative clause related to the object of the matrix clause. This object has been scrambled to some middle field projection XP across an adverb, which I take to be adjoined to the vP for ease of representation. Crucially, the vP contains a trace of the subject as well. In the :P analysis (63b), the subject must be moved from the vP inside the first conjunct to the first sentence position. However, this would constitute a clear violation of the Coordinate Structure Constraint. In the and:P analysis, this problem does not arise, since there is a representation of the thematic subject position inside the second conjunct as well. Therefore, the subject is moved in an across-the-board fashion. As is well-known, ATB movement is not subject to the Coordinate Structure Constraint (Ross 1967), witness (64), for instance: (64) Whati did Mary buy ti and John sell ti? Finally, I would like to mention the fact that the scope of an extraposed constituent is determined by the associated position in the matrix. This is stressed in particular by Büring and Hartmann (1995), and confirmed with a systematic overview of Dutch data by De Vries (2002). The hierarchical difference between subjects and objects is clear and uncontested, so let us focus on the more intricate double object construction. Consider the example in (65a), which shows binding of a pronoun inside a relative clause related to the direct object by a quantified indirect object. In (65b) the reverse pattern is sought for: a quantified direct object binds a pronoun inside a relative clause related to the indirect object; this, however, is unacceptable. Thus, (65) confirms the idea that IO asymmetrically c-commands DO. The inverse scope reading in (65b), although imaginable, is unavailable to me. (65) a.

Ik heb iedereeni het verhaal dat hiji wilde horen verteld. I have everyone the story that he wanted hear told ‗I told everyone the story he wanted to hear.‘ b. # Ik heb de persoon die heti wilde horen elk afzonderlijk verhaali verteld. I have the person who it wanted hear every single story told ‗I told the person who wanted to hear it each story.‘ Interestingly, extraposition of the relative clauses in (65) does not alter the judgments: (66) a. Ik heb iedereeni het verhaal verteld dat hiji wilde horen. b. # Ik heb die persoon elk afzonderlijk verhaali verteld die heti wilde horen. The pattern can be confirmed by a Condition C test; see (67) and (68). In (67a) the direct object Joop zelf ‗Joop himself‘ does not c-command the coreferent Joop embedded inside the indirect object; therefore, there is no Condition C violation. In (67b) the R-expression de headhunter inside the direct object is c-commanded by the (pronominal) indirect object, which does lead to a violation (on a coreferential interpretation). Again, extraposition in (68) does not alter the judgments.

64

Mark de Vries (67) a.

Ik heb de headhunter die zijni/Joopsi college natrok Joopi (zelf) I have the headhunter who his/Joop‘s colleague investigated Joop himself aanbevolen. recommended ‗I recommended the headhunter who investigated Joop‘s colleague Joop himself.‘ b. # Ik heb hemi een man die de headhunteri niet kende aanbevolen. I have him a man who the headhunter not knew recommended ‗I recommended him a man who the headhunter did not know.‘ (68) a. Ik heb die headhunter Joopi (zelf) aanbevolen die zijni/Joopsi collega natrok. b. # Ik heb hemi een man aanbevolen die de headhunteri niet kende. If extraposition from an object involves right-adjunction to the verb phrase, these facts would be problematic. Under a coordination approach, this is not the case. An indirect object is attached at least one Merge step higher than the direct object; therefore, a specifying coordination phrase containing material related to IO is attached higher than one relating to DO; thus the required hierarchical distinction reflected in c-command relations remains intact. There is, however, a potential problem for the original :P analysis, which bears on the fact that A-bar movements are reconstructed. If an object is wh-moved or topicalized, it is still under the scope of the subject, witness (69) for instance: (69) a. # Which mani did Mary say hei invited _? b. Himselfi , Johni likes _ very much. In (69a), he c-commands the trace of which man, which leads to a Principle C violation. In (69b), the anaphor is bound by the subject John despite the fact that it has been topicalized. In short, A-bar movements do not alter scope relations. If we combine this with the insight that extraposition does neither, we can put up an interesting test: an extraposed relative clause relating to a topicalized object should still be under the scope of the subject. That this is correct is shown by means of quantifier binding in (70): (70) Die ene, unieke gelegenheid zal niemandi zich laten ontgaan waarbij hiji de koningin That one unique opportunity will nobody SE let escape whereby he the queen persoonlijk te spreken krijgt. Personally to talk gets ‗That once-in-a-lifetime opportunity, nobody will let escape, whereby he will get to speak to the Queen in person.‘ The schematic :P analysis is shown in (71): (71) :P [TOPDO Aux S V tDO] [ : RC]]

Specifying Coordination An Investigation into the Syntax of Dislocation …

65

Topicalization finds place within the first conjunct; therefore, the pronoun within the relative clause is outside the scope of the subject quantifier, and cannot be bound, contrary to fact. By contrast, the and:P analysis has no such problems, as is clear from the representation in (72): (72)

[and::P [TOPDO Aux S V tDO] [ and: [[TOP+RC]DO Aux S V tDO]]]

Here, the second conjunct contains all relevant material. The relative clause attached to the topicalized object can reconstructed as usual (compare (69)). To sum up, there is not only clear evidence for coordination in extraposition constructions, but also for silent repeated material. At this point, it is time for a short digression on the nature of the deletion involved. The deletion needed in the and:P analysis has three important characteristics. First, the deletion may involve nonconstituents and discontinuous material. Second, deletion of all material that is repeated is obligatory. Third, the deletion is directed forward; moreover, it concerns a second conjunct. All of these characteristics are familiar from the phenomenon known as gapping. Gapping is forward deletion of material in a second conjunct of a coordination phrase. The examples in (73) show that deletion may involve discontinuous strings. Here, stressed constituents are capitalized. (73) a.

a. JOOP heft MIEKE EEN BOEK gegeven en JAAP heft MONIQUE EEN CD gegeven. Joop has Mieke a book given and Jaap has Monique a CD given ‗Joop gave Mieke a book, and Jaap Monique a CD.‘ b. JOOP heft HEM een boek gegeven en JAAP heft HAAR een boek gegeven. Joop has him a book given and Jaap has her a book given ‗Joop gave him a book, and Jaap her.‘ c. HIJ heft Mieke EEN BOEK gegeven en ZIJ heft Mieke EEN CD gegeven. He has Mieke a book given and she has Mieke a CD given ‗He gave Mieke a book, and she a CD.‘ d. JOOP heft een boek over SYNTAXIS gelezen en JAAP heft een boek over Joop has a book about syntax read and Jaap has a book about FONOLOGIE gelezen. Phonology read ‗Joop read a book about syntax, and Jaap about phonology.‘ Furthermore, it is clear that remnants must provide new information:19 (74) Joop heft Mieke vanochtend een boek gegeven en Jaap/*Joop heft Monique/*Mieke Joop has Mieke this.morning a book given and Jaap/Joop has Monique/Mieke 19

Notice that the verb (or verbs) represents old information, and it is deleted. If this were not the case, all its arguments must be projected according to the Head Condition (Fiengo, 1974; see also G. de Vries 1992 for discussion and references); as a consequence, there would be no deletion at all, which is not what we are looking for. Therefore, it is not possible to introduce a new verb in the second conjunct while deleting arguments that represent old information.

66

Mark de Vries gisteren/*vanochtend een CD/*boek gegeven. yesterday/this.morning a CD/book given ‗Joop gave Mieke a book this morning, and Jaap/*Joop Monique/*Mieke a CD/*book yesterday/*this morning.‘

Thus, the deletion involved in the and:P does not lead to additional stipulations; it simply conforms to the pattern already known to be available for regular gapping constructions. We now have an additional reason why extraposition conforms to the Right Roof Constraint. Recall that this concerns examples such as (75): (75)* [That Joop committed a crime _ last year] is quite clear by now that he will regret. A schematic and:P analysis is the following: (76)* [and:P [CP1 [CP2 C [IP2 α N β]] IP1 ] [ and: [CP1 [CP2 C [IP2 α N RC β] IP1 ]] ]] In (76), there is deletion across a clause boundary. This is impossible in gapping constructions, as well; see (77), for instance: (77)* Joop said that Mieke bought a book, and Jaap said that Monique bought a CD. Therefore, from the present perspective the Right Roof Condition on extraposition reduces to a more general condition on deletion in gapping configurations. Of course, it needs to be clarified in turn why forward deletion is sensitive to clause boundaries. This general issue is beyond the scope of this chapter. It is worth mentioning that an analysis in terms of recoverability was proposed by G. de Vries (1992).

3.2.5. Extrapose a? The and:P analysis, and base-generation analyses more generally, predict that optional extraposition is only possible if the remainder is syntactically well-formed, and hence can be generated to begin with. We do not expect full arguments to be extraposable, since that would leave an illegitimate gap in the matrix. Possible exceptions to this rule of thumb are cases in which the argument position can be filled by an empty category pro. Extraposed phrases, then, are usually modifiers or complements. At least eight different construction types fit this pattern:        

relative clauses prepositional modifiers of N (adjuncts and complements) result clauses comparative clauses second conjuncts clausal complements of N prepositional modifiers of A clausal complements of A

Specifying Coordination An Investigation into the Syntax of Dislocation …

67

The previous subsections contain several examples of the first two. Illustrations of the last six are provided in (78): (78) a.

Joop heft zo hard gewerkt dat hij er zowaar moe van was. Joop has so hard worked that he there actually tired of was ‗Joop worked so hard that he actually got tired.‘ b. Joop heft meer artikelen gelezen dan boeken. Joop has more articles read than books ‗Joop has read more articles than books.‘ c. Ik heb Joop gezien en Jaap. I have Joop seen and Jaap ‗I saw Joop and Jaap.‘ d. d. Joop heft het gerucht gehoord dat Mieke zwanger is. Joop has the rumour heard that Mieke pregnant is ‗Joop heard the rumour that Mieke is pregnant.‘ e. Joop is altijd dol geweest op appels. Joop is always fond been on apples ‗Joop has always been fond of apples.‘ f. Joop is altijd bang geweest dat zijn vrouw hem zou verlaten. Joop is always afraid been that his wife him would leave ‗Joop has always been afraid that his wife would leave him.‘ In each case, the and:P analysis schematically looks as follows: (79) ... [and:P [XP1 α YP] [ and: [XP2 [α EX] YP]]] At the relevant point in the matrix a specifying coordination phrase and:P is inserted. The remainder of the matrix, XP1 (consisting of α YP), is put in the first conjunct. The second conjunct is a more specific copy of XP1, namely some XP2 that contains an additional phrase or clause EX related to α. Repeated material is phonologically deleted. One more concrete example is (80), which corresponds to (78e): (80) Joop is altijd [and:P [VP1 [AP dol] geweest] [ and: [VP2 [AP dol [PP op appels]] geweest]]] It is always the case that the second conjunct specifies the first. Intuitively this makes sense: in a discourse, one can add information only to something that has already been mentioned. Also the fact that there is forward deletion is pragmatically expected: it makes sense to not pronounce again what has already been pronounced (and has the same function), but not the other way around. There are also construction types without a ‗first part‘ that may or must involve a rightperipheral positioning, namely adverbial phrases and clauses, secondary predicates, heavy NPs (including free relatives), and prepositional and clausal complements of V. In De Vries (2002), following suggestions by Koster (1999), I assumed that all of these involve extraposition such that EX in the second conjunct replaces a pro or weak pronoun in the first.

68

Mark de Vries

In the present chapter, however, I will show that we must distinguish between several different types (depending, for instance, on whether there is backgrounding or not), and that neither constitutes extraposition proper (as described in this section), although indeed many of them do involve some kind of specifying coordination; see section 5 below.

3.3. Intermediate Conclusion This section introduced the concept of specifying coordination. The abstract coordinator and: projects as a regular X-bar category. Its meaning can be described as a relation of specification, such that the second conjunct provides an alternative description of the first conjunct, or a more specific description by adding information. If specifying coordination is used at the constituent level, it can be used to represent an appositional construction. If and:P is inserted in the spine of the clausal structure, it can be used to represent extraposition, an idea which turned out to have substantial explanatory power. Thus, we obtain a nice generalization over two seemingly unrelated constructions. Needless to say, there is also a major difference between the two: extraposed constituents are interpreted restrictively, whereas appositions are non-restrictive. In the next section, I will argue that non-restrictive construal requires a second type of specifying coordinator.

4. Non-Restrictive Construal and Secondary Propositions 4.1. More on Appositions In this section we will have a closer look at appositional constructions. It will become clear that appositions involve coordination on two levels at the same time: constituent coordination, as was argued in section 2, and also implicit coordination of a secondary proposition to the primary proposition, which is semantically at the sentence level. This secondary proposition is a copular clause in which the apposition is a predicate of the anchor. More generally, a secondary message reflects parenthetical (non-restrictive) construal. Syntactically, this is brought about by means of a particular type of specifying coordinator called Par, whose properties will be discussed in more detail in the next subsection. Section 4.3 contains a structural proposal for appositions and appositive relative clauses. Section 5 continues with anchored parentheses in general, stretching the ideas to right-dislocation constructions and what will be called interthoughts. Consider the following two sentences. In (81a), Joop is an apposition; in (81b) Joop is a regular (disjunctive) conjunct. (81) a. Anna invited her neighbor, Joop, for dinner. b. Anna invited Mary or Joop for dinner. Both involve structural coordination, but there are essential differences. The apposition Joop in (81a) constitutes an alternative description of the anchor (her neighbor); in (81b) it is

Specifying Coordination An Investigation into the Syntax of Dislocation …

69

an alternative to the anchor (Mary). In the former Joop is constructed non-restrictively; in the latter it is restrictive information. The proposition in (81b) is true if either Mary is invited for dinner or Joop is invited for dinner by the subject (or both). Sentence (81a) is true if Anna invited her neighbor for dinner. That her neighbor can be identified as Joop is another matter. Suppose that Anna invited her neighbor for dinner, but that Joop is not Anna‘s neighbor. Then (81a) is not simply false, it is an incoherent message, comparable to a presupposition violation. A possible response could be ―Well, that‘s true, but note that her neighbor is not Joop.‖ Thus, it can be concluded that (81b) contains one complex proposition, but that (81a) contains in fact two propositions (see also Berckmans 1994, Dever 2001, Corazza 2005, and Potts 2005 for discussion). This difference is reflected in the intonation. Regular conjuncts are included under the intonation contour of the matrix. If there is multiple coordination, as in John, Mary, Anna, and Bill, the main stress shifts to the last conjunct. By contrast, an apposition normally receives an independent pitch accent. It does not influence the pronunciation of the anchor or other elements in the matrix; it only temporarily interrupts the intonation contour of the matrix. There are other tests that confirm the non-restrictive character of appositions. For instance, appositions are not implied in VP ellipsis contexts; see (82), for instance: (82) Anna admires Joop, her best friend, and Monique does, too. Here, the second, elliptical clause means that Monique also admires Joop. There is no implication concerning a possible friendship between Monique and Joop. Also, notice that an appositional construction such as (83a) is equivalent to constructions with a relative clause (83b) or a parenthetical (83c): (83) a. Joop loves Anna, his girlfriend. b. Joop loves Anna, who is his girlfriend. c. Joop loves Anna – she is his girlfriend. The clausal character of appositional constructions is confirmed by the fact that sentential adverbs can be used (see also Quirk et al 1985:1314 for English): (84) Joop, toen nog een tiener, componeerde zijn eerste symfonie in 1970. Joop then still a teenager composed his first symphony in 1970 ‗Joop, still a teenager at the time, composed his first symphony in 1970.‘ Thus, there is clear evidence that appositions provide additional information that does not belong to the primary proposition. There must be an implicit secondary proposition, with an independent truth value. Let me informally define this term: (85) Definition A secondary proposition is a proposition that is presented or perceived as relevant sideinformation to a primary proposition.

70

Mark de Vries

The intuition that the additional proposition invoked by an apposition is secondary can be reinforced as follows. Consider the sentence in (86), which contains the propositions (i) and (ii): (86) Anna invited Joop, her best friend, for dinner. (i) Anna invited Joop for dinner. (ii) Joop is her best friend. Scenario A is that (i) is true but (ii) is false. As I stated before, a possible response is ―Well, that‘s true, but note that Joop is not her best friend.‖ It would be infelicitous to say: ―That‘s not true. Joop is not her best friend. (But you are right that she invited him for dinner.)‖ Reversely, scenario B is that (i) is false and (ii) is true. In that case, an appropriate response is ―No, that is incorrect. She did not invite him for dinner. (But indeed, they are best friends.)‖ By contrast, it would be infelicitous to say ―That is correct, they are best friends. (However, note that she did not invite him for dinner.)‖ An appositional construction is predicational (see especially Doron 1994). More precisely, the secondary proposition involved constitutes an implicit copular clause in which the anchor is the subject, and the apposition the predicate. This intuition is strengthened by the fact that there are striking similarities between regular predicates and appositions. For instance, a nominal predicate allows for a particular i-within-i configuration that is unacceptable in an argument position: (87) a. Joop is [hisi own cook]i. b. Joop, (still) [hisi own cook]i, produced a delicious meal last night. c. # [Hisi/Joopi‘s own cook]i produced a delicious meal last night. More importantly, the two main types of apposition, namely attribution and identification, correspond to the two main types of predication. This is illustrated in (88) and (89). (88) attributive predication a. Joop is a would-be pianist. (copular clause) b. … Joop, a would-be pianist, … (appositional construction) (89) identificational/specificational predication a. Joop‘s piano teacher is Mr. Brendel. (copular clause) b. … Joop‘s piano teacher, Mr. Brendel, … (appositional construction) See Heringa 2007 for more discussion on this particular point, and see Heringa and De Vries 2008 for a detailed semantic classification of appositional constructions in Dutch. If the anchor of an appositional construction is indefinite, a slight complication arises. Consider the following example:

Specifying Coordination An Investigation into the Syntax of Dislocation …

71

(90) A handsome guy, Joop, entered the room. It is important to see that there is no predicate inversion here. The secondary proposition is not that Joop is a handsome guy, which would be an instance of attributive predication, but that a particular handsome guy is identified as Joop, which is a specificational predication. If we want to spell out this proposition overtly, we have to take the scope of the quantifier a into account, which yields the handsome guy that entered the room was Joop. To sum up, appositional constructions involve a secondary proposition that expresses the predicational relationship between the anchor and the apposition. Thus, there is not only coordination on the constituent level, as was argued in section 2, there is also coordination of propositions, which is a more general property of parenthetic material. In the next subsection, it is argued that this is brought about in syntax by a second type of specifying coordinator.

4.2. Parentheses and Non-Restrictive Construal Appositions, and parentheses more generally, are linearly integrated with the host sentence, but they are non-restrictive, that is, they do not restrictively participate in the primary proposition, and they are not categorially selected. There is much discussion in the literature concerning the relation between parentheses and the host sentence (see Espinal 1991, Burton-Roberts 1999, and especially the introductory chapter in Dehé and Kavalova 2007 for an excellent overview and further references), but the bottom line is that parentheses are used to express side-information. Often, this is phonologically marked, but it can be done in various ways, and sometimes there is not even a clear intonational separation between the parenthesis and the host structure (Dehé 2008). Therefore, a general positive characterization of parentheses in phonological or phonetic terms, which is sometimes referred to as the ‗comma reading‘, is untenable. (Despite that, it is clear that change of pitch, speech rate, potential pauses, and so on often play a role; see Altmann 1981, Bolinger 1989, and Pittner 1995, for instance.) Furthermore, since parenthesis encompasses a host of different syntactic constructions (see further below), it is difficult to provide an a priori definition in syntactic terms, as well. Thus, the following semantic definition seems more adequate: (91) Definition Parenthesis is a grammatical construction type that involves a message that is presented or perceived as secondary with respect to the host. Here, the notion message is used as a cover term for propositions, modal propositions, questions, meta-linguistic comments, and so on. Furthermore, the term grammatical construction type indicates that a parenthesis is related to the host sentence by the grammar, from which it follows that they cannot be linearly separate. In other words, parenthetic material is linearly integrated with the primary proposition by either intervention or adjacency (modulo potential other parentheses related to the same host, obviously). Secondary messages, therefore, relate to sentences directly, with obvious effects on the discourse – but it will be clear that parenthesis is not only a pragmatic phenomenon. A concrete example is provided in (92). Here, the parenthesis is printed in italics; the primary and secondary message are explicated below the example.

72

Mark de Vries (92) Waarschijnlijk heft Joop – dat is mijn buurman – de deur groen geverfd. Probably has Joop DEM is my neighbor the door green painted ‗Probably, Joop – he is my neighbor – has painted the door green.‘ Primary (modal) proposition: Probably, Joop has painted the door green. Secondary proposition: He is my neighbor.

The intuition that the parenthesis in (92) constitutes a secondary proposition is corroborated by the fact that it is not included in the scope of the sentence adverb waarschijnlijk ‗probably‘. What is probable is that Joop has painted the door green; the parenthetical is simply not part of this. Parentheses are subject to a variety of semantic and syntactic invisibility effects. For instance, they are not included in cases of VP-ellipsis or pronominal reference to the host sentence. The former was shown for appositions in (82) above; the latter is illustrated in (93), which is a possible response to the utterance in (92): (93) Oh, maar dat vind ik niet leuk. Oh but DEM find I not nice ‗Oh, but I don‘t like that.‘ Here, dat ‗that‘ refers only to the primary proposition in (92), to the exclusion of the secondary one. Furthermore, finite clausal parentheses normally are main clauses (with the exception of appositive relative clauses; see the next subsection), which have an independent illocutionary force. A relevant example is (94). Recall that main clauses in Dutch have V2, but subordinate clauses an exclusively verb final word order. (94) Heeft Joop – ik1 bedoel2 jouw buurman – de deur groen geverfd? Has Joop I mean your neighbor the door green painted ‗Has Joop – I mean your neighbor – painted the door green?‘ b. * Heeft Joop – (dat) ik jouw buurman bedoel – de deur groen geverfd? a.

As (94) shows, V2 in the parenthetical is obligatory. Similarly, a main clause order can be observed in reporting and comment clauses: (95) ―De aarde,‖ zei1 Joop2 gisteren, ―is niet helemaal rond.‖ The earth said Joop yesterday is not entirely round ‗ ―The earth,‖ Joop said yesterday, ―is not entirely round.‖ ‘ Notice that there is subject-verb inversion in this kind of parentheticals (see De Vries 2006b for discussion and references.) Finally, parenthetical construal is different from subordination in that it blocks syntactic and semantic relationships between elements in the host and elements inside the parenthesis.

Specifying Coordination An Investigation into the Syntax of Dislocation …

73

For instance, there cannot be wh-movement from a parenthesis into the host. See (96), which is completely unacceptable: (96)* What did the police – the commissioner suspected Hank stole _ – search his house? Generally, elements inside a parenthesis cannot be licensed (or ruled out) by elements from the host. (The reverse is also true, for the simple reason that elements inside the parenthetical are embedded, hence not in a c-commanding position.) This is investigated in detail in De Vries (2007a), and I will not repeat it here. To sum up, parentheticals are linearly integrated with the host sentence, but they show a variety of invisibility effects. They are semantically non-restrictive, and involve a secondary message. Syntactically, they are semi-independent (main) clauses. How can these properties be represented in the syntactic structure? I should note that there is no agreement in the literature concerning the question whether parentheses are to be represented in syntax in the first place. According to the ‗orphanage‘ view, this is not the case (Haegeman 1991, Burton-Roberts 1999, Peterson 1999); according to the ‗integrated‘ view, it is (Espinal 1991, Ackema and Neeleman 2004, Fortmann 2005). I am convinced that the latter approach is right. Generally, I believe that syntax should reflect semantics as closely as possible. But there is also a practical and very straightforward reason to reject orphanage. It is a truism that parentheses, like any other linguistic material, are both interpreted and pronounced. Therefore, they must be present at the LF interface and the PF interface. According to standard assumptions concerning the organization of grammar, there is only one way to get at these interfaces, namely via the overt syntax. If a parenthesis were to be added after spell-out, say, at or after the LF interface, there would be no way to pronounce it, or the reverse. Therefore, it can be concluded that there is such a thing as parenthetical, non-restrictive construal in syntax.20 This amounts to acknowledging the fundamental distinction between hypotaxis and parataxis as observed by traditional grammarians. I will assume that parenthetical construal is triggered by a functional head that can be called Par, which will be identified as a particular kind of specifying coordinator below. First, let us turn to the structural embedding of parentheticals in the host. Unlike appositions, but like adverbial phrases and clauses, parentheticals do not have an anchor. Furthermore, they are not categorially selected, and their position within the host is relatively free. An example is the comment clause in (97). See also Schelfhout (2006) and Stoltenburg (2003) for a corpus-based investigation of possible positions of parentheses in Dutch and German, respectively. (97) Joop (X) vertrekt (X) morgen (X) naar Amerika (X). X = dacht Anna Joop leaves tomorrow to America thought Anna Undoubtedly, discourse considerations play a role in the decision where to put a parenthesis, but what is relevant for our purposes is that there are no clear syntactic restrictions. Thus, it is likely that parentheses are structurally adjoined to an arbitrary projection of the host. 20

See De Vries (2007a) for an attempt to translate this into a more formal proposal in terms of the Minimalist Program (Chomsky 2005, and subsequent work). See also Fortmann (2005) for a comparable attempt to derive the effects of non-restrictive construal in an LFG framework.

74

Mark de Vries

In order to distinguish parentheses from restrictive adjuncts, I propose that they are embedded in a ‗parenthetical phrase‘. See the representation in (98), in which the extended parenthetical projection ParP of XPpar is adjoined to some projection ZP of the matrix: (98)

Here, the head Par triggers the non-restrictive interpretation of XPpar with respect to the host. Also, ParP screens off XPpar from potential licensing/c-command relationships by elements in the host (Y for instance). As there are many types of parenthetical phrases, the internal composition of XPpar in (98) varies, with different properties associated with each. Nevertheless, there is a common basis. The representation in (98) reflects what all parentheses have in common, namely their non-restrictive construal with respect to the host, and the correlated invisibility effects. The head Par can be seen as a non-restrictive specifying coordinator. If there is no anchor, a parenthesis is simply a specification of the discourse. In this respect, it is noteworthy that some parentheticals can be introduced by an overt coordinator, witness (99) for instance. See Blakemore (2005) and Kavalova (2007) for more discussion on and-parentheticals. (99) He asserted – and I wonder what you think about it – that the prisoners should be released. b. This man stole my bicycle, or at least so I think. a.

Recall that parentheses entail a secondary message. On some level of abstraction, this proposition is coordinated to the discourse. It seems that in (99), the overt coordinator is a spelled-out Par head. The next section contains a structural proposal of appositions and nonrestrictive relative clauses, which are concretely anchored parentheses.

4.3. Appositions and Non-Restrictive Relative Clauses: A Tentative Structural Proposal In section 3.1 it was established that appositions involve coordination at the constituent level. Furthermore, appositions are parentheses, hence non-restrictively constructed. In the

Specifying Coordination An Investigation into the Syntax of Dislocation …

75

previous subsection it was argued that non-restrictive construal is triggered by the head Par. Thus, we arrive at (100), which reinforces the idea that Par is a specifying coordinator: (100) [ParP anchor [ Par apposition]] Notice that we now have two specifying coordinators, namely and:, which is used for extraposition, and its non-restrictive counterpart Par, which is used for parenthesis. However, (100) cannot be the complete analysis. Section 4.1 showed that appositional constructions do not only involve (non-restrictive) constituent coordination, they also involve a predicative secondary proposition. Thus, there must be an implicit copular clause. In order to find out how this can be structurally represented, I will make a comparison with nonrestrictive relative clauses. In De Vries (2006), I argued that appositive relative constructions involve specifying coordination, such that the antecedent is in the first conjunct, and the relative clause is embedded inside the second conjunct. The complete second conjunct is a full noun phrase. Usually, it is a kind of semi-free relative, and hence contains a restrictive relative. Thus, the construction as a whole looks like an appositional construction. An example such as Joop, who is my neighbor is analyzed roughly as Joop, i.e., (the/some)one who is my neighbor. Strong evidence for this view comes from the fact that an additional internal head is possible, although this belongs to a very formal register. See the illustrations in (101): (101) ―De avonden‖, welk boek van Reve veel gelezen wordt, is herdrukt. ―De avonden‖, which book of Reve much read is, has.been reprinted ‗ ―De avonden‖ [the evenings], which book by Reve is read by many people, has been reprinted.‘ b. Hond en kat zijn als water en vuur, welk feit reeds lang bekend is. Dog and cat are like water and fire, which fact already long known is ‗Dogs and cats are like water and fire, which fact has been well-known for ages.‘ a.

Examples in English are provided in (102), taken from Smits (1988:287/8). See also Cardoso (2008) for illustrations in Old Portuguese, and De Vries (2004) for a discussion of Dutch data. Notice that in (101b) and (102b), the antecedent is the entire first clause. (102) a. My dog, which faithful animal has guarded me for years, died last week. b. It might be a bit crowded at the airport, in which case we had better be early. Crucially, this is impossible in restrictive relative constructions, which is an argument for the raising analysis of relative clauses mentioned before in section 3.2. A more detailed structural representation of the man who/that is my neighbor is given in (103): (103) [DP D [CP [DP-rel NPi [ Drel ti]]j C [IP tj .........................]]] The man (who) (that) is my neighbor

76

Mark de Vries

The complete structure in (103) is a DP, an extended noun phrase, and D (the) is the external determiner, which takes scope over the noun phrase and the restrictive relative clause. The latter is the complement of D. Within the relative CP, the head noun is generated as an argument of the embedded verb. This argument is DPrel, the head of which is the relative pronoun, which functions as an operator. If the complementizer C is spelled out, Drel is lexically empty, and the other way around (in English). Next to raising (that is, operator movement dragging along the NP embedded in DPrel), the internal NP is moved to the left periphery (whether this is DPrelinternally or to a higher layer within CP does not concern us, here). In this position it can be associated with the external determiner. Clearly, there is no room for an additional noun phrase. In an appositive construction such as Joop, who is my neighbor, the situation is different. A structure similar to the one in (103) is coordinated as a second conjunct DP2 to the antecedent DP1; schematically, this is shown in (104a), and in some more detail in (104b): (104) a. [ParP DP1 [Par DP2]] b. [ParP DP1 [Par [DP2 D2 [CP [DP-rel [NP N]i [ Drel ti]]j C [IP tj ..........................]]]]] Joop who is my neighbor D2 and N are related; usually, they are abstract: a silent equivalent to a pronoun. In that case, what we encounter is an ordinary appositive relative construction. Interestingly, there are no less than seven variants with essentially the same meaning, as is shown in (105). Notice that I made up these minimally different examples for ease of exposition, but all types can be attested cross-linguistically (and diachronically), and I find all seven acceptable in Dutch, with some differences in markedness.21 (105)English Dutch a. b. c. d. e. f. g.

Joop, who is my neighbor Joop, die mijn buurman is Joop, he who is my neighbor Joop, hij die mijn buurman is Joop, the one who is my neighbor Joop, degene die mijn buurman is Joop, the man who is my neighbor Joop, de man die mijn buurman is Joop, the which man is my neighbor Joop, dewelke man mijn buurman is Joop, which man is my neighbor Joop, welke man mijn buurman is Joop, my neighbor Joop, mijn buurman

In languages that can use a complementizer instead of a relative pronoun in appositive relatives, we also find the patterns [DP, C …], [DP, pron C …], and [DP, (D) N C …] as alternatives for (105a-d). Rather than analyzing appositive relative clauses as appositions or the other way around, I would like to suggest that all the variants just mentioned are different realizations (spell21

Especially (105e) is a bit archaic. Notice that appositive relatives containing an internal head are more readily accepted if they are extraposed; also the presence of a pied piped preposition – as in (102) – seems to be of influence.

Specifying Coordination An Investigation into the Syntax of Dislocation …

77

outs) of one and the same underlying schema, which can be characterized as a complete restrictive relative construction (a complex noun phrase) that is added as a non-restrictive specifying conjunct to another noun phrase. First consider (106): (106) [ParP DP1 [Par [DP2 D2 [CP [DP-rel [NP N]i [ Drel ti]]j C [IP tj ............................]]]]] Joop (i.e.) (the) (one/man) (who) (that) is my neighbor If the NP is overt in (106), it acts as an external head within the second conjunct, which is a restrictive relative construction. In the case of an internal head, as in (105e/f), the relative pronoun is a dependent pronoun, so the original order of Drel and NP stays intact; see (107): (107) [ParP DP1 [Par [DP2 D2 [CP [DP-rel [ Drel [NP N]]]j C [IP tj ..........................]]]]] Joop (i.e.) (the) (which) (man) (that) is my neighbor It is worth noticing that it can now be explained why an appositive relative clause is a subordinate clause (without V2 in Dutch), unlike regular parentheticals, and despite the fact that it constitutes a secondary proposition. The reason is straightforward: it is not the relative clause itself that is parenthetically constructed; rather, it is embedded in a DP that is nonrestrictively attached to the anchor. Finally, consider the appositional construction. In this case, none of the elements in the CP domain is pronounced. Furthermore, the interpretation is always predicative. This means that the verb inside IP must be an abstract copula. (By contrast, the embedded predicate in a regular appositive relative clause can be anything.) This is not surprising: a silent verb can only be interpreted as a copula. In many languages, finite predicative clauses do not (necessarily) contain an overt copula (for an overview, see Stassen 2008). Even in English and Dutch, predication can be expressed covertly, namely if it is secondary. Examples are given in (108): (108) a. Mary considers John a fool. b. John came home drunk. What is implied here is that John is a fool (according to the subject) and that John was drunk (when he came home). Clearly then, we can make use of this in appositional constructions, which, after all, also contain a secondary predication. The resulting analysis is sketched in (109):22 (109) [ParP [DP1 anchor] [Par [DP2 D [CP N Drel COP apposition ]]]] Here, COP is the copula be (possibly, with the required inflectional features). I should stress that D+N is a pronoun whose interpretation depends on the particular type of apposition. It can be animate or non-animate, for instance, and its quantificational status may 22

Within the second conjunct, there is a similarity with Kayne‘s (1994) analysis of reduced relatives and adjectives, which are also predicative. Kayne suggests that the yellow book may be derived from a structure that can be paraphrased as ‗the book that is yellow‘.

78

Mark de Vries

vary (specific, non-specific or universal). Furthermore, notice that a predicate noun normally takes over the Case of its (small clause) subject. Thus, in (109), the apposition acquires its Case via the subject, which in turn takes over the Case of the first conjunct. To be sure, the Case distribution in appositive relative clauses is different. If certain elements of the clausal structure are overt, internal Case becomes available for a potential relative pronoun and an internal head (if present). Thus, the activation of some relevant parts of the embedded clause will block the kind of Case attraction found in appositions. Possibly, this difference is related to a difference in finiteness. However interesting, these issues are beyond the scope of the present chapter, and I will leave a further exploration of the internal structure of the embedded clause in (109) and its semantics for future research. Before closing this section, I would like to point out another similarity between appositive relative clauses and appositions, namely the possibility of a non-nominal antecedent. Some examples are given in (110) and (111). (110) a. John is over there, in the park. b. The bullet went astray, (that is,) it missed the target. (111) a. The meeting lasted from 9 till 12, which is a long time. b. John is leaving tomorrow, which is too bad. It seems that the first conjunct in such constructions can be any XP. Still, the second conjunct must be a complex DP in each case, because it constitutes a restrictive relative clause construction. This gives [ParP XP [ Par DP], which is syntactically unbalanced. The reason that this is possible is that the head of such a DP is a pronoun. And pronouns may refer to concepts, places, times, events, facts, and entities, implying that they can refer to any syntactic category (see also Jackendoff 1977:175). Thus, as long as a certain functionalsemantic equivalence is maintained, constructions such as (110) and (111) are acceptable.23 To summarize, appositions and appositive relative clauses can be analyzed as nonrestrictive secondary predicates. They are embedded in a parenthetical phrase, of which the first conjunct is the anchor/antecedent, and the second conjunct a complex DP that is structurally equivalent to a restrictive relative construction. We paid some attention to the internal make-up of this relative construction, and to the possibility of spelling out various combinations of all the structural ingredients. In the next section, we will shift our attention to the anchor, and to the structural position of the parenthetical phrase within the matrix.

23

Lehmann (1984:277) notes that relative clauses appositive to non-nominal antecedents are less common than those appositive to nominal antecedents. Presumably, the same can be said about appositions. This is in line with the present analysis, since syntactically unbalanced coordination is generally more marked than balanced coordination.

Specifying Coordination An Investigation into the Syntax of Dislocation …

79

5. Anchored Parentheses 5.1. Introduction We defined parenthesis as a grammatical construction type that involves a secondary message. Let us now explicitly distinguish between those parentheses that are related to an anchor within the matrix clause, and those that are not: (112) Definition An anchored parenthesis is a parenthesis (i) that is coordinated to some anchor within the host sentence, and/or (ii) that involves an (implicit) predication, such that some anchor within the host sentence is interpreted as the subject, and the overt parenthetical material as the predicate. In the case of an apposition or appositive relative clause, there is a concrete anchor to which the parenthesis is attached. For regular parentheticals this is not the case; these are simply adjoined at some position within the matrix (see also section 4.2). Here are some further examples of the latter: (113) Joop heeft – heb ik het niet gezegd? – de deur inderdaad groen geverfd. Joop has have I it not said the door indeed green painted ‗Joop – I told you so – has indeed painted the door green.‘ b. Joop heeft – en ik vertrouwde hem al niet – het geld gestolen. Joop has and I trusted him yet not the money stolen ‗Joop – and I didn‘t trust him anyway – has stolen the money.‘ c. Joop had, zo zei hij, het geld alleen even geleend. Joop had so said he the money only a.while borrowed ‗Joop, so he said, had just borrowed the money for a while.‘ a.

Parentheses often contain a pronoun coreferring with the primary proposition as a whole or a constituent thereof. For instance, hem ‗him‘ in (113b is coreferent with Joop in the matrix, and zo ‗so‘ in (113c) refers to the entire matrix. This regular linking device between propositions is also highly frequent in subsequent sentences, and it should not be confused with anchoring, which means that the entire parenthetic phrase is interpreted as a secondary predicate. What is special about appositional constructions is that the anchor of predication (the subject of the implicit copular clause) coincides with the anchor of coordination (the first conjunct). In the following subsections we will see that this is not a necessary condition for parenthetical anchoring. Right-dislocated phrases such as afterthoughts also involve an implicit predication relation between the anchor and the overt parenthetical material; still, the overt anchor and the afterthought are not coordinated at the constituent level. Furthermore, it will become clear that backgrounding constructions involve anchoring as well. The difference

80

Mark de Vries

between them is simply that backgrounded information is not new. Thus, we will develop a generalized perspective on parenthetic secondary predication.24

5.2. Afterthoughts Afterthoughts are right-dislocated phrases of a certain kind. They constitute new information; furthermore, they receive an independent pitch accent, which makes them clearly distinct from extraposed phrases. Arguably, they have a lot in common with appositions. Consider the following examples: (114) Ik heb Joop gezien – een leuke jongen. I have Joop seen a nice guy ‗I saw Joop – a nice guy.‘ b. Ik heb Joop gezien – gekleed in een cowboypak! I have Joop seen dressed in a cowboy.suit ‗I saw Joop – dressed in a cowboy suit!‘ a.

In both cases, there is an implicit secondary proposition in which the italicized phrase is understood as a predicate of a constituent in the host clause (here, Joop). In (114a), this proposition is equivalent to Joop is a nice guy; in (114b), which is ambiguous, it is either Joop was dressed in a cowboy suit, or I was dressed in a cowboy suit. Accordingly, afterthoughts are also anchored parentheses according to the definition in (112). However, differently from the situation in appositional constructions, the anchor of an afterthought is structurally separated from it. In (114), the subject Joop figures in the middle field, whereas the afterthought surfaces right-peripherally. It seems almost as if afterthoughts are extraposed appositions. If so, the analysis would be roughly as in (115): (115) Ik heb [and:P [Joop gezien] [and: [[ParP [Joop] [ Par [een leuke jongen]]] gezien]]] Although there are no syntactic arguments to exclude (115) to my knowledge, I prefer a somewhat simpler analysis, which involves right-adjunction of ParP to the matrix clause, as is shown in (116). In this case, the first conjunct in the parenthetical phrase must be a silent pronoun that is coreferent with the overt anchor: (116) [CP [CP Ik heb Joopi gezien] [ParP PROi [ Par [een leuke jongen]]]] Here, ParP is adjoined, as is the case for non-anchored parentheses; at the same time, it acts as a bivalent specifying coordinator, as in regular appositional constructions. From a performance perspective, there is reason to prefer (116) over (115). Namely, in (116), the afterthought is added after the the derivation of the main sentence (the primary proposition)

24

See also Van der Wouden (2000) for a discussion of right-peripheral phrases (‗appendices‘) in Dutch.

Specifying Coordination An Investigation into the Syntax of Dislocation …

81

has been completed, contrary to the situation in (115). Intuitively, this is right: extraposition must be planned, but an afterthought need not be – hence the name. At this point, let us consider what may constitute an anchor. As was shown around (90) for appositions, the subject of the implied predicative clause includes the scope of the anchor. The same is true for afterthoughts. Thus, in (117), the secondary proposition is equivalent to the handsome boy that I have seen is Joop: (117) Ik heb een knappe jongen gezien – Joop. I have a handsome boy seen Joop ‗I saw a handsome boy – Joop.‘ Interestingly, it is also possible that the anchor is an (abstract) representation of the event expressed by the host. This is illustrated in (118): (118) Ik heb Joop gezien – gisteren. I have Joop seen yesterday ‗I saw Joop – yesterday.‘ b. Ik heb de paus gezien – op het Sint-Pietersplein. I have the Pope seen on the Saint Peter‘s Square ‗I saw the Pope – on Saint Peter‘s Square.‘ c. Ik heb de paus gezien – gisteren! Op het Sint-Pietersplein! I have the Pope seen yesterday on the Saint Peter‘s Square ‗I saw the Pope – yesterday, on Saint Peter‘s Square.‘ a.

In these examples, the afterthought is a phrase that relates to the event denoted by the primary proposition, or more precisely, the temporal frame and the locational frame of the event, respectively. Accordingly, the implicit predications are roughly this event was yesterday in (118a), and this event was on St. Peter‘s Square in (118b). The sentence in (18c) shows that these possibilities can also be combined. The type of predication is clearly attributive (as in (114) but unlike (117)), so the afterthought is interpreted as a property of the relevant event. On a final note, it is worth commemorating that the end result of uttering, say, (118a) is that the hearer is informed that the speaker saw the Joop yesterday, which equals the situation after the message ―I‘ve seen Joop yesterday‖, in which all information is constructed restrictively. However, the discourse in (118a) is put together differently by means of a different syntax. The message that the particular event of seeing Joop was yesterday is presented as secondary information surfacing at the right periphery. This may lead to different pragmatic entailments; for instance, an element of emphasis or surprise might play a role.

5.3. Interthoughts The examples in the previous subsection showed that afterthoughts can be of any lexical category. They are all phrasal, but still, they constitute a full secondary proposition because of

82

Mark de Vries

the implicit predication. Regular (non-anchored) parentheses also invoke a secondary message. However, since there is no such implicit relationship, these are often clausal (with the exception of interjections and some other types; see sections 6 and 7). It is important to see that it is not the right-peripheral position of afterthoughts that triggers anchoring, witness the following examples containing right-peripheral parentheticals:25 (119) Joop weigerde mee te doen – (en) gelijk heeft-ie. Joop refused with to do and right has-he ‗Joop refused to participate – (and) right he is.‘ b. ―Mieke had het niet moeten doen,‖ beweerde Joop. Mieke had it not should do claimed Joop ‗ ―Mieke shouldn‘t have done it,‖ Joop claimed.‘ a.

Therefore, anchoring is a factor independent from surface position. Non-anchored parentheses may show up at various positions intervening in the host (see (97) and (113), for instance), and also on the edges of the host, depending on the particular type and the pragmatic import. This raises the question whether it is possible to construct anchored parentheses of the intervening type – that is, non-restrictive secondary predicates that are neither right-dislocated, nor directly coordinated to the anchor. This is indeed the case; see the examples in (120): (120) Joop opende – nog steeds dronken – de deur. Joop opened yet still drunk the door ‗Joop opened – still drunk – the door.‘ b. Joop heeft – vorige maand, nota bene – zijn paper al ingezonden. Joop has last month nota bene his paper already submitted ‗Joop already submitted his paper – last month, would you believe?‘ a.

In (120a), the secondary predicate relating to Joop is presented as an aside; hence, there is a secondary proposition Joop was still drunk. In (120b), the time parenthetical is anchored to the time frame of the event expressed by the host sentence (compare (118a)), yielding a secondary proposition this event was last month, nota bene. Thus, there is a construction type that we may call ‗interthought‘.26

25

Notice that the parenthetical reporting clause in (119b) contains an empty object that is coindexed with the quote (see De Vries 2006b discussion and references). This has nothing to do with anchoring; see the text below (113). 26 An interthought may happen to surface adjacent to the anchor, in which case it is hard to distinguish from an apposition. A relevant example, to be compared to (120a) is Joop – nog steeds dronken – opende de deur ‗Joop, still drunk, opened the door.‘

Specifying Coordination An Investigation into the Syntax of Dislocation …

83

As was shown in (98), a parenthetical phrase can be adjoined to any projection of the matrix. The examples in (120) suggest that this is also the case for anchored ParPs. The structural analysis of (120a) is sketched in (121): (121) [CP Joops opendev [XP [ParP PROs [Par [AP nog steeds dronken]]] [XP [de deur]o [vP ts tv to]]]] Here, XP is some projection in the middle field that contains the definite object. The parenthetical phrase is left-adjoined to XP; the implicit subject PRO is coreferent with Joop. The matrix subject and verb are in the first and second position of the sentence, which precede the intervening ParP. To sum up, we have seen that anchored parentheses can be constructed at the constituent level (appositions), right-peripherally (afterthoughts), and in intervening positions (interthoughts). In the next subsection, it will be shown that the information-structural status of such parentheses can be varied.

5.4. Backgrounding All parentheses discussed so far involve new information, and all carry a pitch accent. We will now see that each of the three types of anchored parentheses can be backgrounded. Let us start with the right-peripheral type. If an afterthought is backgrounded, the result is the familiar right-dislocation construction; see (122): (122) Ik heb ‘m gisteren nog gezien, Joop. I have him yesterday still seen Joop ‗I saw him only yesterday, Joop.‘ Here, Joop must be pronounced with a low, flat intonation, and is intended as a specification of ‘m ‗him‘. Although ‘m is a pronoun, whose referent is therefore known in the discourse, the speaker feels that he or she needs to remind the hearer of the intended referent. The secondary proposition implied in this construction is the specificational predication He is Joop. This means that the pronoun is identified as Joop (and not that the name of a particular person is Joop, which would constitute an attributive predication). Furthermore, notice that the pronoun ‘m is still anaphoric, not cataphoric, since it is coreferent with a preceding referential expression in the discourse; only this link is facilitated by repeating the R-expression, which is then related to the pronoun by means of a specificational predication. Interestingly, the dramatic difference between afterthoughts and backgrounding rightdislocation is the result of changing just one parameter, namely the information-structural status of the parenthesis in terms of [+/- new]. If the parenthesis is old information, it automatically follows that the anchor is pronominal, since only in that case the parenthesis can be a specification. Namely, if the anchor itself were already a (familiar) R-expression, there is no way the parenthesis could be a meaningful addition/reminder. It also follows that backgrounding must be anchored: how else could it be a reminder of something? Like afterthoughts, backgrounded parentheses can be of any syntactic category. This is illustrated in (123), where the pronominal anchor is underlined:

84

Mark de Vries (123) a.

Joop is er vorig jaar ook geweest, in Parijs. Joop is there last year also been in Paris ‗Joop was also there last year, in Paris.‘ b. Ik heb het je toch gezegd, dat het niet zou lukken? I have it you after.all said that it not would succeed ‗I told you so, that it wouldn‘t work.‘ In (123a), the relevant constituent is a prepositional phrase, in (123b) a subordinate clause. Furthermore, the anchor can be the time frame of the event; see (124), which can be compared to (118a): (124) Ik heb de paus ook op tv gezien, gisteren. I have the Pope also on TV seen yesterday ‗I saw the Pope on TV, too, yesterday.‘ The regular restrictive position of time adverbials is in the middle field. In (124), however, gisteren ‗yesterday‘ is backgrounded as a reminder of when the event expressed in the host took place. Thus, there is a secondary proposition stating that the temporal value of this event equals yesterday, even though this information is not new in the discourse. Backgrounding in the right periphery is the most common, and the most natural. Nevertheless, it is also possible to construct backgrounded appositions and backgrounded intervening parentheses, witness (125) and (126): (125) Ik heb hem – Joop (dus) – vorige week nog gezien. I have him Joop so last week yet seen ‗I saw him – Joop, (that is) – only last week.‘ (126) Ik heb hem gisteren – Joop (dus) – nog gevraagd om het rapport in te leveren. I have him yesterday Joop so yet asked for thereport in to hand ‗I asked him only yesterday – Joop (, that is) – to hand in the report.‘ In both cases the hearer is reminded that hem ‗him‘ is to be understood as Joop. However awkward it may seem, similar constructions can be found with other categories. An further example is (127): (127) Ik heb Joop toen (– gisteren –) gevraagd (– gisteren –) of hij me wilde helpen. I have Joop then yesterday asked yesterday if he me wanted help ‗I then asked Joop – yesterday, that is – if he wanted to help me.‘ Here, too, gisteren ‗yesterday‘ is a backgrounded specification of time, as in (124). The anchor is overtly present by means of a non-specific time adverbial toen ‗then‘, to which the backgrounded phrase may be coordinated. I conclude that backgrounded phrases are anchored parentheses that do not provide new information. The structural analysis is exactly the same as that for non-backgrounded phrases.

Specifying Coordination An Investigation into the Syntax of Dislocation …

85

They can be attached to the host as an apposition (hence coordinated at the constituent level to an overt anchor), as an interthought (an adjoined intervening parenthesis containing a PRO substitute of the anchor), or as an afterthought (a right-peripherally adjoined parenthesis containing a PRO substitute of the anchor).

5.5. Intermediate Conclusion Let us take stock. All parentheses involve a secondary message. Phrasal parentheses can be anchored, which means that they are understood as the predicate of some anchor in the host. Thus, parenthetical anchoring is analyzed as non-restrictive secondary predication. Like regular parentheses, anchored parentheses may surface in various positions. In particular, these positions are i) right-peripheral, which yields and afterthought; ii) intervening, which yields an interthought; and iii) coordinated to the anchor, which yields an apposition (or appositive relative clause). In the first two cases, the anchor can also be an abstract representation of the event expressed by the host. A more detailed proposal for appositions was suggested in section 4.3, which also took into account the similarities between appositions and apositive relative clauses. In both cases, the anchor is the first conjunct of a non-restrictive specifying coordination phrase ParP, and the apposition or relative clause is embedded inside the second conjunct, which itself is a complete relative construction, hence a complex DP. In appositional constructions, the anchor of predication coincides with the anchor of coordination. The implicit embedded clause is necessarily copular, and the overt parenthetic material is the internal predicate. This analysis can be straightforwardly transferred to afterthoughts and interthoughts. The only difference is that such parentheses are not coordinated at the constituent level to the anchor. Instead, they are adjoined at the relevant projection of the matrix, and the overt anchor is represented by means of a PRO substitute as the first conjunct of ParP. So far, I have argued for three parameters concerning parentheses. The first two concern the possibility of anchoring and the structural position. The third concerns the information status of parentheses: they may involve information that is either new or old in the discourse. I argued that this has consequences for the interpretation and the intonation, but not for the syntactic analysis. Parentheses that constitute old information are backgrounded. These are obligatorily anchored to a pronoun (or some abstract aspect of the event expressed by the matrix). I showed that there are backgrounded appositions, backgrounded interthoughts and backgrounded afterthoughts. The last construction is the most common; it is known as (backgrounding) right-dislocation.

6. Left-Dislocation and Discourse-Anchoring Dislocation is not only possible on the right, but also on the left. Therefore, let us have a closer look at left-dislocation and some related phenomena in this section. In various ways, these involve the combination of meta-information concerning the discourse with the primary proposition.

86

Mark de Vries

6.1. Hanging Topics An interesting example of parenthesis to the left of the sentence is the construction known as hanging topic left-dislocation (HTLD). Two examples of hanging topics in Dutch are the following. In (129), the context is put between brackets. The difference with contrastive topics is discussed in section 6.3. (128) Die jongen, heb je die al eens eerder gezien? That boy have you DEM already once before seen ‗That boy, have you seen him before?‘ (129) [A: Heb je nog gezeild in de vakantie? B: Nee. A: Waarom niet?] Have you still sailed in the vacation no why not [B: Och...] zeilen, ik ben er niet zo dol op. Well sailing I am there not so fond of ‗[A: Have you been sailing during the vacation? B: No. A: Why not? B: Ah well,] sailing, I am not so fond of it.‘ A hanging topic in Dutch is interpreted as a sentence topic (to be precise, an ‗aboutness topic‘ in the sense of Reinhart 1982). In (128), this is die jongen ‗that boy‘; in (129) it is zeilen ‗sailing‘. A hanging topic can be preceded by the phrase wat betreft ‗as for‘ (lit. ‗what concerns‘), which confirms the interpretation as a topic. Notice, however, that the discourse function of hanging topics is subject to cross-linguistic variation; see Prince (1998), Frey (2005), and De Vries (to appear) for a discussion of English, German, and Dutch, respectively. There are several indications that hanging topics are dislocated, hence parenthetical. First, there is an in situ pronominal representative in the sentence; these are underlined in (128) and (129). Second, they precede a complete main clause with a spelled-out first constituent (recall that Dutch has V2). Third, they are intonationally separated from the main clause. At first sight, HTLD seems to be the mirror pattern of backgrounding right-dislocation (BRD). This, however, is not completely true. As we have seen, BRD may involve all kinds of syntactic categories; by contrast, HTLD is restricted to nominal phrases. Compare (130a) to (130b), for instance: (130) a.

Joop heft er vanochtend gezeten, in de tuin. (BRD) Joop has there this.morning sat in the garden. ‗Joop has been sitting in the garden this morning.‘ b. * In de tuin, Joop heeft er vanmorgen gezeten. (HTLD) I believe that this is related to the fact that aboutness topics are always nominal phrases (see also Frascarelli and Hinterhölzl 2007). Thus, backgrounded phrases, unlike hanging topics in Dutch, are not necessarily sentence topics. What is of interest for the present purposes is the question how hanging topics are related to the host structure. As they are not restrictively constructed with respect to the main clause, they must involve a parenthetical phrase. Their meaning can be paraphrased as follows:

Specifying Coordination An Investigation into the Syntax of Dislocation …

87

concerning the sentence to follow, the aboutness topic is X. In other words, there is an (abstract) discourse category S-TOPIC, and for this sententence, it is specified as X. I think the resemblance with anchored parentheses discussed in the previous section is too close to be accidental. We have already seen that anchors can be abstract aspects of events. Let us now go one step further, and hypothesize that certain discourse categories relevant for the interpretation of a sentence may also serve as an anchor. A hanging topic, then, involves a non-restrictive specifying coordination phrase [ParP S-TOPIC [Par DP]], in which DP is the overt topical phrase specifying the sentence topic. Furthermore, this ParP must be left-adjoined to the sentence; see the structure in (131):27 (131)

In a way, HTLD has the opposite effect of an afterthought. We do not add information to a completed sentence, we start out saying explicitly what the sentence is about. Let us call the particular kind of anchoring in HTLD discourse-anchoring. It is a method of combining meta-information with the primary proposition. In the remainder of this subsection and in the next, I will show that there are more constructions of this type. Van Riemsdijk (1997) distinguishes between HTLD and ‗loose aboutness leftdislocation‘, as attested in French and Chinese, for instance (for the latter, see Chafe 1976:50), but not in Germanic. In this construction, the dislocated constituent is not pronominally represented in the clause, unlike the situation in HTLD. An example is (132), taken over from Van Riemsdijk (1997:4) [my translation]: (132) Oh, tu sais, moi, la bicyclette, je n‘aime pas me fatiguer. Oh you know me the bicycle I NEG-like not me tire ‗Oh, you know, as for me, concerning bicycles, I don‘t like to tire myself.‘ Here, the relevant constituent is la bicyclette ‗the bicycle‘. (Interestingly, the preceding moi, ‗me‘ seems to be a hanging sentence topic (HTLD); but this is another matter.) I think the hanging constituent in a loose aboutness construction must be analyzed as an explicitly spelled-out discourse topic. This kind of topic, which must be distinguished from sentence topics, is pragmatically frame-setting, but not necessarily represented in a particular sentence. It may be active during a series of sentences. Let us refer to this discourse category as D-TOPIC. 27

I should note that I proposed a different (simpler) structure in De Vries (2007b), namely one in which the sentence itself is a specification of the topical phrase. I now think that this is not precise enough.

88

Mark de Vries

If the analysis of HTLD in (131) is correct, we can now easily explain the similarities between hanging topic LD and loose aboutness LD. Analogously to the first, the latter may be structurally represented as [CP [ParP D-TOPIC [Par DP]] [CP…]]. The example in (132) shows that other sentence-initial parentheses may precede it. Since the relevant parentheses are analyzed as adjoined phrases, this is not unexpected: a priori, there is no reason to exlude multiple adjuncts.

6.2. Non-Restrictive Epithets and Vocatives There are also constructions that involve discourse-anchoring that are not inherently leftperipheral. Consider the following examples, which contain a parenthetical epithet: (133) a.

Sukkel, ik zei toch dat dat niet kan. Dope I said indeed that DEM not can ‗Idiot, I told you that that was impossible.‘ b. Ga weg, engerd! Go away creep ‗Go away, creep!‘ In each case, the italicized phrase is an (emotional) qualification of the hearer. What the speaker means is that the hearer is an idiot or a creep in (133a/b), respectively. This message is not the primary proposition, but a secondary one, attached to the main clause. As is the case for hanging topics, there is no concrete anchor in the sentence. Clearly, however, HEARER is an accessible discourse category. Thus, the parentheses in (133) can be straightforwardly analyzed as (non-restrictive) attributive specifying phrases coordinated to a discourse anchor: [ParP HEARER [Par epithet]]. In turn, the parenthetical phrase can be adjoined anywhere in the sentence. In this case, there is no general pragmatic preference for an initial position, which is why epithets of this kind can be found in left-dislocated, right-dislocated, and also intervening positions. The above paves the way for an analysis of vocatives, as well. Consider the sentences in (134), where Joop is a vocative phrase: (134) a.

Joop, heb jij een auto? Joop have you a car ‗Joop, do you have a car?‘ b. Ik denk, Joop, dat dit geen geode beslissing is. I think Joop that this no good decision is ‗I think, Joop, that this is not a good decision.‘ c. Wat zullen we doen, Joop? What shall we do Joop ‗What shall we do, Joop?‘

Specifying Coordination An Investigation into the Syntax of Dislocation …

89

As we know from appositions, secondary predications can be identificational (specificational). Thus, it seems possible to transfer the proposal for epithets to vocatives, which yield the following parenthetical phrase: [ParP HEARER [Par vocative]]. At first sight, it seems silly to state that the hearer can be identified as, in this case, Joop. After all, the message is addressed to Joop, and, obviously, Joop knows that he is Joop. But it is not selfevident that Joop is paying attention and/or knows for sure that the message is directed towards him. Therefore, the pragmatic import of a vocative can roughly be paraphrased as ‗I am talking to you‘. It gives the hearer a clue that the message is directed at him or her, or it simply confirms that this is the case, thereby strengthening the social aspects of the communicative situation.

6.3. A Note on Contrastive Left-Dislocation and Topicalization In section 6.1, hanging topics were analyzed as left-adjoined, discourse-anchored phrases. This construction must be distinguished from contrastive left-dislocation (and from clitic leftdislocation in Romance languages and Greek). CLD, it can be argued, is closely related to topicalization.28 Some miminal pairs are provided in (135): (135) Joop, ik ken ‘m niet. (HTLD) Joop I know him not ‗Joop, I don‘t know him.‘ b. Joop, die ken ik niet. (CLD) Joop DEM know I not ‗Joop, I don‘t know.‘ c. Joop ken ik niet. (topicalization) Joop know I not ‗Joop, I don‘t know.‘ a.

In the HTLD construction (135a), there is an in situ pronominal correlate of the hanging topic (here, ‘m). In the CLD construction (135b), there is an accompanying pronoun that is obligatorily a distal demonstrative, and obligatorily adjacent to the left-peripheral phrase. In the regular topicalization construction, there is no pronominal correlate at all: the topicalized phrase is clearly a constituent of the sentence. CLD, like topicalization, but unlike HTLD, may involve any syntactic category; see (136), for instance (compare also (130) above): (136) In de tuin (daar) heft Joop vanmorgen gezeten. In the garden there has Joop this.morning sat ‗In the garden is where Joop has been sitting this morning.‘

28

Dutch and German CLD is also comparable to English topicalization, but not to English HTLD (cf. Frey 2005, contra Zwart 1998).

90

Mark de Vries

The interpretation of such preposed phrases is variable (see De Vries, to appear, for discussion). In (136), the prepositional phrase is frame-setting, and possibly strongly contrastive, but it is not an aboutness topic. As is well-documented in the literature, CLD is also different from HTLD in that it shows connectivity effects. This concerns a host of different phenomena: Case matching between the dislocated phrase and the demonstrative (which yields internal Case), reconstruction for binding purposes and idiomatic interpretations, and locality effects that are indicative of Abar movement. See, among others, Jansen (1981), Van Haaften, Smits and Vat (1983), Van Riemsdijk (1997), Grohmann (2003), Shaer and Frey (2004), Alexiadou (2006), De Vries (2004, 2007b). I will not repeat all the arguments here. What is crucial is that CLD behaves on a par with regular topicalization: in every respect, the leftmost phrase acts as an integrated constituent of the sentence. This suggests that it is not dislocated at all, despite appearances. In De Vries (2007b), I proposed that the left-dislocated constituent is generated inside the clause, and subsequently moved to the left periphery (see also Van Haaften, Smits and Vat 1983). This solves all connectivity effects at once. Of course, it raises the question where the demonstrative comes from. The surprising anwser is that it is this pronoun that is ‗dislocated‘, not the full left-peripheral phrase XP. Namely, it can be attached as an apposition (on the constituent level) to XP. See the schematic representation in (137): (137) [CP [ParP XP [Par DEM]]i V2 ... ti ...] Here, ParP containing XP is A-bar moved like any topicalized phrase, which explains all reconstruction effects, the internal Case, and locality. Like regular appositions, the demonstrative is adjacent to its anchor, and shows a matching Case between XP and the demonstrative. The V3 order is only apparent, since ParP is one constituent. How meaningful is a pronominal apposition? Is this not a vacuous specification? Normally, it would be, but it is no coincidence, I believe, that the pronoun in a CLD construction is a demonstrative. A demonstrative does add information, namely deixis and contrast. This is exactly what is needed in CLD. A paraphrase of the idea is ‗XP, namely that person/thing/... (and nothing else)‘. Notice that there are more ways of creating a meaningful demonstrative apposition. For instance, there are ‗summary appositions‘, as in your pencils, paperclips, rubber bands, staples: all that. Furthermore, appositions can be disambiguating. In a particular context, say a family reunion, I could say Joops nicht, DIE ‗Joop‘s niece, that one‘ in Dutch, presupposing that he has more than one niece. Interestingly, this complex phrase in turn can be used in a CLD construction, yielding Joops nicht, DIE, die heeft wel een hond ‗Joop‘s niece, that one, she does have a dog.‘ In this case, I interpret the first die as disambiguating, and the second as contrastive with respect to the predicate have a dog. Thus, there is support for the idea that it can be meaningful to use a (demonstrative) pronoun as an apposition. This lends credibility to the analysis of CLD in (137), which is a bit counterintuitive at first sight, but which has a strong explanatory power.

7. Parentheses: An Overview In this chapter, I defined parenthesis as a grammatical construction type that entails a secondary message (often a proposition), which is a message that is presented or perceived as

Specifying Coordination An Investigation into the Syntax of Dislocation …

91

relevant side-information to a primary message. There is a major distinction between parentheses that are anchored and those that are not. The latter include interjections, conjunctive adverbs and other hedges, and various kinds of parentheticals, such as reporting and comment clauses, secondary declaratives, and wh-parentheticals, each of course with its own characteristics, depending on the internal make-up. Anchored parentheses include appositions, afterthoughts, and backgrounded phrases. These are interpreted as non-restrictive secondary predicates. Any constituent of the host sentence may serve as an anchor. This is also the case for the host itself, an abstract representation of an event expressed by the host (or the temporal or locational frame thereof), or even a discourse category such as TOPIC or HEARER, which leads to a new perspective on hanging topics and certain parenthetical epithets. The possibility of anchoring can be viewed as a parameter for classifying grammatical construction types. A second parameter is the possibility of backgrounding, which essentially means that the relevant parenthesis does not constitute new information. A third parameter is the structural position of attachment. We may distinguish between four possibilities: a leftperipheral position, a right-peripheral position, an intervening position, or a coordinated position (at the constituent level). Table 6 shows how these parameter values can be systematically combined, leading to a variety of construction types. As is clear from the table, there are four main classes: regular parentheses, anchored parentheses, backgrounded parentheses, and discourse-anchored parentheses. As was explained in section 5.4, only anchored phrases can be backgrounded. Furthermore, it turns out that anchored parentheses cannot be left-dislocated, unless they are discourse-anchored. This makes sense, since a constituent cannot be interpreted as a predicate of something that has not yet been introduced. Finally, it seems that discourse-anchored parentheses attached at the constituent level do not exist, which is a consequence of the fact that discourse categories relate to full sentences. Left- and right-dislocated parentheses must be distinguished from other construction types that target the left or right periphery. Although it is often possible to detect the difference between restrictive and non-restrictive construal by listening to the intonation, there are also syntactic ways to do so, namely by investigating connectivity effects. A schematic overview is provided in table 7, based on sections 3.2, 5.2, 5.4, 6.1, 6.3, and the references there: Clearly then, extraposition is not parenthesis on the right, and topicalization and contrastive left-dislocation are not parenthesis on the left.29 Finally, let us see if we can combine various parentheses in the right and left periphery. The example in (138) shows that a backgrounded phrase and an afterthought may cooccur: (138) Ja, ik heb ‘m gezien, Joop – op, school. Yes I have him seen Joop on school ‗Yes, I saw him, Joop – at school.‘

29

Concerning the demonstrative in CLD, see section 6.3. See also Zwart (2001) on backgrounding right-dislocation, and the comments in De Vries (2007b).

92

Mark de Vries Table 6. Parentheses: some parameters and terminology Anchoring

Information New

Anchored Old

Discourseanchored

Not Anchored

Old

New

Old

Position Right-dislocated Intervening Constituent level Left-dislocated Right-dislocated Intervening Constituent level Left-dislocated Right-dislocated Intervening Left-dislocated Constituent level Right-dislocated Intervening Constituent level Left-dislocated

Name of construction Afterthought Interthought Apposition, appositive relative

Backgrounding right-dislocation Backgrounded parenthesis Backgrounded apposition

…

Anchored parenthesis

Backgrounding parenthesis

Discourse-anchored parenthesis

Parenthesis (non-anchored, non-backgrounding)

Table 7. Some distinguishing properties of left- and right-peripheral phrases Position Ld

Rd

Construction Hanging topic left-dislocation Contrastive left-dislocation Topicalization Afterthought Backgrounding right-dislocation Extraposition

Connectivity effects No Yes Yes No No Yes

Categories Dp Xp Xp Xp Xp Xp

Here, Joop is backgrounded, and specifies ‘m ‗him‘. The phrase op school ‗at school‘ is stressed and functions as an afterthought. The reverse order (afterthought – backgrounded phrase) sounds unnatural to me. Probably, this is the result of the general informationstructural tendency to push new information to the right edge. Other combinations are also possible. The example in (139) contains (in this order) an interjection, a vocative, and two afterthoughts: (139)Ik heb de paus in het echt gezien, hoor, Joop – op het Sint-Pietersplein, met zijn I have the Pope in the real seen hear Joop on the St-Peter‘s.Square with his Kalotje op! Calotte on ‗I saw the Pope in the flesh, really, John – on Saint Peter‘s Square, wearing his calotte!‘ In the left periphery, too, several items can be combined. A worst case scenario is (140), which combines an interjection, a wh-parenthetical, a vocative, and a hanging topic (in this order):

Specifying Coordination An Investigation into the Syntax of Dislocation …

93

(140) Tja, wat denk je, Joop, het tentamen syntaxis, heb je dat nu gehaald of niet? Well what think you Joop the examination syntax have you that now passed or not ‗Well, what do you think, Joop, the syntax examination, have you passed it or not?‘ We can also combine HTLD with CLD; see (141). Here, gisteren ‗yesterday‘ must be stressed. (141) (Wat betreft) die jongen, gisteren, toen heb ik ‘m nog gezien. As concerns that boy yesterday then have I him still seen ‗(As for) that boy, yesterday I still saw him.‘ The order between gisteren and (wat betreft) die jongen cannot be reversed. This is consistent with the view expressed here that CLD is not true dislocation.

Conclusion Coordination is a general syntactic configuration that can be applied at any structural level. The semantics and the phonological shape depend on the particular coordinative head. I provided a near-complete inventory of coordinators, insubordinators, coordinative combinations, and conjunctive adverbs in English and Dutch, along with some tests how to distinguish between these different elements. Crucially, I argued that there is a further main type of coordination, called specifying coordination, with several subtypes. In each case, the second conjunct provides an alternative description, a more specific description, an example, or a property of the first conjunct. Roughly, then, the asymmetrical semantic relationship between the two conjuncts can be described as specification, hence the name. Interestingly, the acknowledgment of this type of coordination extends the overlap in possible meanings with conjunctive adverbs. Explicative adverbs such as namely can even be used as linkers in some specifying coordinative constructions. The appositional construction is the canonical type of specifying coordination. I have shown that this construction behaves on a par with standard constituent coordination in numerous respects. A second construction that shows characteristics of coordination is extraposition. In this case, the extraposed part is attached as (a part of) a specifying conjunct to the matrix clause at the level of VP up to CP, depending on the correlated structural position in the matrix. I argued for a particular implementation of this theory, which involves structural repetition of the relevant part of the matrix clause in combination with phonological deletion of what is by then old information, similar to the situation in ordinary gapping constructions. This turns out to have many empirical and theoretical advantages, such as the prevention of CSC violations, the possibility of maintaining the independently motivated raising analysis of relative clauses, and a more direct relationship between the syntactic structure and the semantic interpretation. A central hypothesis in this chapter is that non-restrictive construal can be analyzed as a particular kind of specifying coordination. I argued that every parenthesis involves a secondary message (often a regular proposition, but it can also be a modal proposition, a question, or a meta-linguistic comment). Thus, parenthesis implies coordination of propositions (etc.) at the semantic level. Structurally, a parenthesis is attached to the host as a

94

Mark de Vries

Parenthetical Phrase, whose head is a non-restrictive specifying coordinator. If the secondary proposition is semi-independent, as in parentheticals, for instance, there is no anchor, and the parenthesis is adjoined anywhere in the host. If there is an anchor, the situation is different. Again, the appositional construction seems to be the most straightforward type. Here, the apposition is coordinated to the anchor. It is also interpreted as a specifying or attributive predicate of the anchor. Therefore, the secondary proposition in an appositive construction is an implicit predicative clause. I pursued the idea that this clause is syntactically represented, and offered a detailed structural proposal that takes into account the similarities with nonrestrictive relative clauses. This approach also sheds new light on dislocation constructions and certain interrupting phrases that I dubbed interthoughts. I illustrated the similarities between these and appositions, and concluded that afterthoughts and interthoughts are adjoined anchored parenthetical phrases of which the first conjunct is a silent pronominal representation of the anchor in the matrix. In general, anchored parentheses are to be analyzed as non-restrictive secondary predicates. Interestingly, what constitutes the anchor is not necessarily an overt constituent of the matrix; it can also be a more abstract aspect of the event, or, in the case of hanging topics and non-restrictive epithets, a certain discourse category. Clearly then, anchoring is a factor independent of the position and way of structural attachment. Thus, we can set up a typology of parenthetical constructions. In this respect, a third parameter of interest is the information status of parentheses. I argued that backgrounding constructions are syntactically similar to other anchored parentheses, but semantically different in that they constitute old information, and hence function as reminders; this has phonological consequences as well. The perspective put forward in this paper stresses the similarities between several kinds of parentheses, which are difficult to explain in a construction-specific approach. This is not to deny that there are distinctions, but I believe it should be possible to attribute such differences to variation in the internal make-up of parentheses. What is of interest here, is that core properties of parenthetical construal can be attributed to the way parentheses are related to the host structure. An important key to this end is the concept of specifying coordination. Needless to say, the grammatical apparatus is preferably not to be enriched for constructionspecific purposes. Therefore, it is important that the particular proposal based on a detailed analysis of appositions can be generalized to other constructions. I argued that this is indeed the case.

Acknowledgments I thank Adriana Cardoso, Jordi Fortuny, Herman Heringa, Marlies Kluck, Jan Koster, Radek Šimìk, and Jan-Wouter Zwart for their comments and questions. This research was financially supported by the Netherlands Organisation for Scientific Research (NWO).

References Ackema, Peter and Neeleman, Ad. 2004. Beyond Morphology: Interface Conditions on Word Formation. Oxford: Oxford University Press.

Specifying Coordination An Investigation into the Syntax of Dislocation …

95

Alexiadou, Artemis. 2006. Left Dislocation (Including CLLD). The Blackwell Companion to Syntax, ed. by Martin Everaert, Henk van Riemsdijk, Rob Goedemans and Bart Hollebrandse, chapter 37 in Volume II. Altmann, Hans. 1981. Formen der ―Herausstellung‖ im Deutschen: Rechtsversetzung, Linksversetzung, Freies Thema und verwandte Konstruktionen. Tübingen: Niemeyer. Baltin, Mark. 1983. Extraposition: Bounding versus Government-Binding. Linguistic Inquiry 14, 155-162. Berckmans, Paul. 1994. Demonstration, Apposition, and Direct Reference. Communication and Cognition 26, 499-512. den Besten, Hans. 1978. On the Presence and Absence of Wh-Elements in Dutch Comparatives. Linguistic Inquiry 9, 641-671. Bhatt, Rajesh. 2002. The raising analysis of relative clauses: Evidence from adjectival modification. Natural Language Semantics 10: 43-90. Bianchi, Valentina. 1999. Consequences of Antisymmetry: Headed Relative Clauses. Berlin: Mouton de Gruyter. Blakemore, Diane. 2005. and-parentheticals. Journal of Pragmatics 37, 1165-1181. Bolinger, Dwight. 1989. Intonation and its Uses: Melody in Grammar and Discourse. Stanford, Calif.: Stanford University Press. Bresnan, Joan. 1973. Syntax of the Comparative Clause Construction in English. Linguistic Inquiry 4, 275-343. Büring, Daniel and Katharina Hartmann. 1995. All Right! On Extraction and Extraposition in German, ed. by Uli Lutz and Jürgen Pafel, 179-211. Amsterdam: John Benjamins. Burton-Roberts, Noël. 1999. Language, linear precedence and parentheticals. The clause in English: In honour of Rodney Huddleston, ed. by Peter Collins and David Lee, 33–52. Amsterdam: John Benjamins. Cardoso, Adriana. 2008. Internal and external heads in appositive relative clauses: more evidence for the ‗specifying coordination‘ analysis. Ms. University of Lisbon. Chafe, Wallace L. 1976. Givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View. Subject and Topic, ed. by Charles N. Li, 25-55. New York: Academic Press. Chomsky, Noam. 1977. On Wh-Movement. Formal Syntax, ed. by Peter Culicover, Thomas Wasow and Adrian Akmajian, 71-132. New York: Academic Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Corazza, Eros. 2005. On Epithets qua Attributive Anaphors. Journal of Linguistics 41, 1-32. Corver, Norbert. 2006. Comparative Deletion and Subdeletion. The Blackwell Companion to Syntax (Volume I, Chapter 15), ed. by Martin Everaert and Henk van Riemsdijk, 582637. Malden, MA: Blackwell. Culicover, Peter and Ray Jackendoff. 1997. Semantic subordination despite syntactic coordination. Linguistic Inquiry 28, 195-217. Culicover, Peter and Michael Rochemont. 1990. Extraposition and the Complement Principle. Linguistic Inquiry 21, 23-47. Dehé, Nicole. 2008. Clausal parentheticals, intonational phrasing, and prosodic theory. Ms., Freie Universität Berlin. Dehé, Nicole and Yordanka Kavalova, eds. 2007. Parentheticals. Amsterdam: John Benjamins.

96

Mark de Vries

Dik, Simon. 1968. Coordination: its implications for the theory of general linguistics. Amsterdam: North Holland. Doron, Edit. 1994. The discourse function of appositives. Proceedings of IATL 9, 53-62. Hebrew University, Jerusalem. Espinal, M. Teresa. 1991. The Representation of Disjunct Constituents. Language 67, 726– 762. Fiengo, Robert. 1974. Semantic Conditions on Surface Structure. PhD dissertation, MIT. Fortmann, Christian. 2005. On parentheticals (in German). Proceedings of the LFG05 Conference [Bergen], ed. by Miriam Butta and Tracy Holloway King, 166-185. Frascarelli, Mara and Roland Hinterhölzl. 2007. Types of Topics in German and Italian. On Information Structure, Meaning and Form, ed. by Susanne Winkler and Kerstin Schwabe, 87-116. Amsterdam: John Benjamins. Frey, Werner. 2005. Pragmatic properties of certain German and English left peripheral constructions. Linguistics 43, 89-129. Grohmann, Kleanthes. 2003. Prolific Domains. Amsterdam: John Benjamins. van Haaften, Ton, Rik Smits and Jan Vat. 1983. Left dislocation, connectedness, and reconstruction. On the formal syntax of the Westgermania ed. by Werner Abraham, 133154. Amsterdam: John Benjamins. Haegeman, Liliane. 1991. Parenthetical adverbials: The radical orphanage approach. Aspects of Modern English Linguistics: Papers Presented to Masatomo Ukaji on His 60th Birthday, ed. by Shuji Chiba et al., 232-254. Tokyo: Kaitakusha. Haider, Hubert. 1997. Extraposition. Rightward Movement, ed. by Dorothee Beerman, David Leblanc and Henk van Riemsdijk, 115-151. Amsterdam: John Benjamins. Haspelmath, Martin. 2007. Coordination. In: Shopen, Timothy (ed.) Language typology and syntactic description, vol. II: Complex constructions. 2nd ed. Cambridge: Cambridge University Press, 1-51. Heijden, Emmeke van der. 1999. Tussen nevenschikking en onderschikking. Utrecht: LOT. Hendriks, Petra. 2004. Either, both and neither in coordinate structures. The Composition of Meaning: From Lexeme to Discourse, ed. by Alice ter Meulen and Werner Abraham, 115-138. Amsterdam: John Benjamins. Heringa, Herman. 2007. Appositional constructions: coordination and predication. Proceedings of the Fifth Semantics in the Netherlands Day, ed. by Marlies Kluck and Erik-Jan Smits, 67–82. Den Haag: RS Staten Kopie. Heringa, Herman and Mark de Vries. 2008. Een semantische classificatie van apposities. Nederlandse Taalkunde 13. Jackendoff, Ray. 1977. X‘-syntax: A study of phrase structure. Cambridge, Mass.: MIT Press. Jansen, Frank. 1981. Syntactische konstrukties in gesproken taal. Amsterdam: Huis aan de drie Grachten. Johannessen, Janne Bondi. 1998. Coordination. Oxford: Oxford University Press. Johannessen, Janne Bondi. 2005. The syntax of correlative adverbs. Lingua 115, 419-433. Kaan, Edith. 1993. Extraposition from NP in Dutch: consequences of Minimalism. Groninger Arbeiten zur germanistischen Linguistik 36, 144-151. Kavalova, Yordanka. 2007. And-parenthetical clauses. Parentheticals, ed. by Nicole Dehé and Yordanka Kavalova, 145–172. Amsterdam: John Benjamins. Kayne, Richard. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press. Koster, Jan. 1999. Empty objects in Dutch. Ms, University of Groningen.

Specifying Coordination An Investigation into the Syntax of Dislocation …

97

Koster, Jan. 2000. Extraposition as parallel construal. Ms, University of Groningen. Kraak, Remmert and Wim Klooster. 1968. Syntaxis. Culemborg: Stam-Kemperman. Malchukov, Andrej. 2004. Towards a Semantic Typology of Adversative and Contrast Marking. Journal of Semantics 21, 177-198. Mithun, Marianne. 1988. The Grammaticalization of Coordination. In: John Haiman and Sandra Thompson (eds.) Clause Combining in Grammar and Discourse, 331-359 Amsterdam: John Benjamins. Munn, Alan. 1993. Topics in the Syntax and Semantics of Coordinate Structures. PhD Dissertation, University of Maryland, College Park. Peterson, Peter. 1999. On the boundaries of syntax. The Clause in English: In Honour of Rodney Huddleston, ed. by Peter Collins and David Lee, 229-250. Amsterdam: John Benjamins. Pittner, Karin. 1995. Zur Syntax von Parenthesen. Linguistische Berichte 156, 85-108. Potts, Christopher. 2005. The Logic of Conventional Implicatures. Oxford: Oxford University Press. Prince, Ellen. 1998. On the limits of syntax, with reference to left-dislocation and topicalization. The Limits of Syntax, ed. by Peter Culicover and Louise McNally, 281– 302. New York: Academic Press. Quirk, Randolph et al. 1985. A comprehensive grammar of the English language. London: Longman. Reinhart, Tanya. 1980. On the Position of Extraposed Clauses. Linguistic Inquiry 11, 621624. Reinhart, Tanya. 1982. Pragmatics and linguistics: An analysis of sentence topics. Bloomington, Indiana: Indiana University Linguistics Club. van Riemsdijk, Henk. 1997. Left Dislocation. Materials on Left Dislocation, ed. by Elena Anagnostopoulou, Henk van Riemsdijk and Frans Zwart, 1-10. Amsterdam: John Benjamins. Rijkhoek, Paulien. 1998. On Degree Phrases and Result Clauses. PhD dissertation, University of Groningen. Ross, John R. 1967. Constraints on Variables in Syntax. PhD dissertation, MIT. Reprinted as Infinite Syntax! 1986. Norwood, New Jersey: ABLEX. Sag, Ivan, Gerald Gazdar, Thomas Wasow and Steven Weisler. 1985. Coordination and How to Distinguish Categories. Natural Language and Linguistic Theory 3, 117-171. Schelfhout, Carla. 2006. Intercalations in Dutch. PhD dissertation, Radboud University Nijmegen. Shaer, Benjamin and Werner Frey. 2004: ‗Integrated‘ and ‗Non-Integrated‘ Left-peripheral Elements in German and English‘. Proceedings of the Dislocated Elements Workshop, ZAS Berlin, November 2003 (ZAS Papers in Linguistics 35) ed. by Benjamin Shaer, Werner Frey and Claudia Maienborn, 465-502. Berlin: ZAS. Smits, Rik. 1988. The relative and cleft constructions of the Germanic and Romance languages. Dordrecht: Foris. Stoltenburg, Benjamin. 2003. Parenthesen im gesprochenen Deutsch. InLiSt – Interaction and Linguistic Structures, No. 34. Stassen, Leon. 2000. AND-languages and WITH-languages. Linguistic Typology 4, 1-55.

98

Mark de Vries

Stassen, Leon. 2003. Some universal characteristics of noun phrase conjunction. In: Frans Plank (ed.) Noun phrase structure in the languagues of Europe, 761-817. Berlin: Mouton de Gruyter. Stassen, Leon. 2008. Zero Copula for Predicate Nominals. The World Atlas of Language Structures Online, ed. by Martin Haspelmath, Matthew Dryer, David Gil and Bernard Comrie. Munich: Max Planck Digital Library, chapter 120. URL: . Sturm, Arie. 1986. Primaire syntactische structuren in het Nederlands. Leiden: Martinus Nijhoff. Vergnaud, Jean-Roger. 1974. French Relative Clauses. PhD dissertation, MIT. de Vries, Gertrud. 1992. On Coordination and Ellipsis. PhD dissertation, Tilburg University. de Vries, Mark. 2002. The Syntax of Relativization. Utrecht: LOT. de Vries, Mark. 2004. Hoofd-interne relatiefzinnen in het Nederlands. Nederlandse Taalkunde 9, 193-230. de Vries, Mark. 2005. Coordination and Syntactic Hierarchy. Studia Linguistica 59, 83-105. de Vries, Mark 2006a. The Syntax of Appositive Relativization: On Specifying Coordination, False Free Relatives and Promotion. Linguistic Inquiry 37, 229-270. de Vries, Mark. 2006b. Reported Direct Speech in Dutch. Linguistics in the Netherlands 23, 212-223. de Vries, Mark. 2007a. Invisible constituents? Parentheses as B-Merged adverbial phrases. Parentheticals, ed. by Nicole Dehé and Yordanka Kavalova, 203–234. Amsterdam: John Benjamins. de Vries, Mark. 2007b. Dislocation and backgrounding. Linguistics in the Netherlands 24, 235-247. de Vries, Mark. To appear. The left and right periphery in Dutch. The Linguistic Review. Welschen, Ad. 1999. Duale syntaxis en polaire contractie: negatief gebonden of-constructies in het Nederlands. PhD thesis, VU University Amsterdam. Wilder, Chris. 1995. Rightward movement as leftward deletion. On Extraction and Extraposition in German, ed. by Uli Lutz and Jürgen Pafel, 273-309. Amsterdam: John Benjamins. Williams, Edwin. 1981. Transformationless Grammar. Linguistic Inquiry 12, 645–654. van der Wouden, Ton. 2000. Focus on appendices in Dutch. Linguistics in the Netherlands 17, 233-244. Zwart, Jan-Wouter. 1994. Dutch is Head Initial. The Linguistic Review 11, 377-406. Zwart, Jan-Wouter. 2001. Backgrounding (‗right dislocation‘) in Dutch. Manuscript, University of Groningen. Zwart, Jan-Wouter. Where is Syntax? Syntactic Aspects of Left Dislocation in Dutch and English. The Limits of Syntax, ed. by Peter Culicover and Louise McNally, 365-393. New York: Academic Press.

Reviewed by Jordi Fortuny

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 3

LEARNER PERCEPTION AND STRATEGIES FOR PRAGMATIC ACQUISITION: A GLIMPSE INTO ONLINE LEARNING MATERIALS Julie M. Sykes1 and Andrew D. Cohen2 1

University of New Mexico, Department of Spanish and Portuguese, MSC03-2100 1 University of New Mexico Albuquerque, New Mexico 87131-0001 2 University of Minnesota, Program in Second Language Studies, 214 NCCE, 315 Pillsbury Dr. SE, Minneapolis, MN 55455 USA

Abstract Research on second language (L2) pragmatics indicates that explicit instruction helps learners improve their pragmatic performance and that interlanguage pragmatic (ILP) development is an important component of communicative competence. A pedagogical model for pragmatics needs to address the true complexity of pragmatic development. Moreover, an informed approach to L2 pragmatic instruction calls for the inclusion of multiple knowledge areas and a strategy-based understanding. CALL technologies play an increasingly important role in ensuring that pragmatics instruction is comprehensive in nature. Likewise, an inclusive model would incorporate strategies for the learning and performance of pragmatics. The intention is to assist learners in gaining the necessary skills to tackle L2 pragmatic complexities in a variety of contexts. Moreover, innovative technologies have the potential not only to aid in the creation of effective online materials for learning pragmatics, but also to provide a valuable environment for interaction. At this point, little research has empirically analyzed the use of an online, strategy-based model for learning L2 pragmatics. Even less has specifically targeted learners‘ perceptions of this space. In order to arrive at a greater understanding of these issues, this paper compares learner perception data from two research projects which have utilized CALL materials to enhance learners‘ strategies for L2 pragmatics. The first study was an in-depth, qualitative research project addressing the use and effectiveness of the first-ofits-kind learner self-access website in L2 Spanish pragmatics with an extensive strategy 

Phone: (505) 277-5616, Fax: (505) 277-3885, Email: [email protected]

100

Julie M. Sykes and Andrew D. Cohen overlay. The second was an extensive study which analyzed the use of the first synthetic immersive environment (a 3-dimensional gaming space designed with a specific educational outcome) used as an instructional space for learning Spanish pragmatics, specifically requests and apologies. This chapter compares learners‘ perceptions of their own strategy development within these two unique, online contexts, and utilizes patterns found in qualitative interview data to provide insights into how learners view both types of CALL environments for pragmatic development. The overarching goal is to consider potential roles that each of these mediated contexts could play in a strategies-based approach to L2 pragmatics. Future application of the findings to materials design and creation, research, and pedagogy will also be presented.

Introduction Research on second language (L2)1 pragmatics suggests that explicit instruction helps learners improve their pragmatic performance (e.g., Bardovi-Harlig, 2001; Kasper and Rose, 2002) and that interlanguage pragmatic development is an important component of communicative competence in an L2. Various models have explicitly noted pragmatics as an essential component of communicative competence. For example, in Bachman‘s (1997) model, pragmatics is included as both the necessity for illocutionary competence (i.e., the ability to carry out certain functions in a language) and sociolinguistic competence (i.e., knowing when to use which forms and functions). Both branches suggest pragmatic competence is essential to communicative success. Thorne (2005) further advocates the importance of pragmatics in communicative competence by suggesting a re-orientation from a focus on L2 communicative competence to a focus on intercultural competence. This reorientation emphasizes the critical connection between language and social practice as related to the negotiation of interactional patterns in intercultural communication. In addition, it places L2 pragmatics at the core of language proficiency, a contrast from its current peripheral or equal position. As suggested in each of these models, pragmatic competence plays a significant role and should be considered an essential component of L2 curricula. A pedagogical model for pragmatics needs to address the true complexity of pragmatic development. Moreover, an informed approach to L2 pragmatic instruction calls for the inclusion of multiple knowledge areas and a strategy-based understanding (Judd, 1999; Cohen, 2005). Despite the established theoretical importance of pragmatics, there is still a large gap between theory and the practical application of pragmatic instruction (e.g., Bardovi-Harlig, 2001; Felix-Brasdefer, 2002; Rose, 2005; LoCastro, 2003). One aim of this paper is to start to bridge the gap between theory and practice through the exploration of learners‘ perceptions of two online environments for interlanguage pragmatic (ILP) development.2 Computer Assisted Language Learning (CALL) technologies play an increasingly important role in ensuring that pragmatics instruction is comprehensive in nature. Innovative technologies have the potential not only to aid in the creation of effective online materials for 1

For the purposes of this chapter, L2 will serve as a generic label, including both the context where the language is spoken widely and the context where it is not. In principle, pragmatic development in an L2 will be faster in the former context than in the latter, but it depends largely on how the learner makes use of the available resources. 2 Defined by Kasper and Schmidt (1996) as ―the development and use of strategies for linguistic action by nonnative speakers‖ (p. 150); ILP development specifically refers to pragmatics in an L2.

Learner Perception and Strategies for Pragmatic Acquisition

101

learning pragmatics (e.g., CLEAR, 2005; Cohen and Ishihara, 2005; Cohen and Sykes, 2006), but also to provide a valuable environment for interaction (Furstenberg and Levet, 2001; Belz, 2002, 2003; Biesenbach-Lucas, 2005; Sykes, 2005; Sykes, Oskoz, and Thorne, 2008). In addition, an online environment offers a ready opportunity to provide a series of strategies that learners could use to enhance their learning and performance of pragmatics. The intention of each of the mediated environments explored in this study is to assist learners in gaining the necessary skills to tackle L2 pragmatic complexities in a variety of contexts. At this point, little research has empirically analyzed the use of an online, strategy-based model for learning L2 pragmatics. Even less has specifically targeted learner perception of this space. Learner perception is critical due to the impact it can have on, for example, motivation, intake, and willingness to practice in the online environment.3 In order to arrive at a greater understanding of these issues, this paper compares learner perception data from two research projects which have utilized CALL materials to enhance learners‘ strategies for L2 pragmatics. The first study (S1) was an in-depth, qualitative research project addressing the use and effectiveness of the first-of-its-kind learner website in L2 Spanish pragmatics with a comprehensive strategy overlay (Cohen and Sykes, 2007; Sykes and Cohen, 2007). The second (S2) was an extensive study which analyzed the use of the first synthetic immersive environment (a 3-dimensional gaming space designed with a specific educational outcome) used as an instructional space for learning Spanish pragmatics, specifically requests and apologies (Sykes, 2008). This chapter compares learners‘ perceptions of their own strategy development within these two unique, online contexts, and draws some conclusions about how learners view CALL environments dealing with pragmatics. We end the chapter by posing questions for future research. The overarching goal of this chapter is to consider the potential roles that each of these mediated contexts could play in a strategies-based approach to L2 pragmatics.

Relevant Background Information Prior to a description of the two studies themselves, it is necessary to examine issues associated with this line of investigation. This section starts by describing relevant terminology. We then briefly deal with approaches to the learning of pragmatics and explore the potential role of mediated contexts (i.e., an instructional website and online virtual space) in pragmatic instruction.

A Definition of Pragmatics Due to the wide-variety of contexts in which it is used, it is especially important to delineate what is meant by the term pragmatics. For our purposes, pragmatic ability will be explicitly defined as, the various manners (i.e., linguistic and nonlinguistic) in which meaning is communicated and interpreted in interaction, as well as the sociocultural factors (individual

3

A complete review of the role of motivation and perception as related to second language acquisition (SLA) is beyond the scope of this work. See Gass and Selinker (2001, chapter 12) for an overview of these factors as related to SLA.

102

Julie M. Sykes and Andrew D. Cohen

and collective) which influence the communicated and interpreted messages (Sykes, 2008, p. 5). 4 This definition is a combination of three canonical definitions of pragmatics and offers a comprehensive view of the important elements of each. We briefly explore each definition of pragmatics contributing to the current conceptualization. First, Yule (1996) defines pragmatics as the ability to deal with ―meaning as communicated by a speaker and interpreted by a listener…and [to be able to interpret] people‘s intended meanings, their assumptions, their purposes or goals…‖ (Yule, 1996, p. 3-4). Yule‘s definition is important because it establishes the position meaning carries as a central component of interaction. Furthermore, it introduces the essential roles both expression and interpretation play in the transfer of meaning. Crystal (1997) explicitly states the importance of sociocultural considerations in his definition of pragmatics—―the study of language from the point of view of the users, especially the choices they make, the constraints they encounter in using language in social interaction, and the effects their use of language has on the other participants in the act of communication‖ (p. 301). Finally, LoCastro (2003) expands Crystal‘s definition to include non-linguistic factors within interaction. She notes, ―pragmatics is the study of speaker and hearer meaning created in their joint actions that include both linguistic and nonlinguistic signals in the context of socioculturally organized activities‖ (p. 15). Inclusion of both linguistic and non-linguistic elements within interaction is an important contribution of LoCastro‘s definition. From our perspective each element presented in the aforementioned definitions is critical to the development of a comprehensive pragmatic system. As one can imagine, acquiring the necessary skills encompassed in a comprehensive pragmatic system is quite complex, especially in a second language. Speakers must know what to say, how and when to say it, in relation to whom they are speaking. That is, interlocutors must have a sense of both the pragmalinguistic and sociopragmatic norms of the 5 speech community in order to successfully perform any language function. Pragmalinguistic norms encompass the linguistic forms that are used to carry out language functions. For example, when apologizing, a speaker needs to select from a set of linguistic structures forms that are appropriate for the given context (e.g., a speaker-oriented form, Discúlpame., and/or a hearer-oriented form, Lo siento.). In order to be sociopragmatically appropriate, the speaker needs to access knowledge about when, why, and with whom to use the various forms. This involves, for example, considerations of social class, age, gender, role within the interaction, and the cultural assumptions relevant to the interaction. In the case of an apology, for example, this would be the knowledge of when an apology should actually be made. The knowledge and skills needed to successfully perform each of these components can be mind boggling, especially for language learners and their instructors, who are attempting to incorporate pragmatics instruction in the classroom or even during a study abroad stay.

4

It is important to point out that numerous conceptualizations of pragmatics exist. The use of the current definition is not meant to ignore additional definitions, but rather establish common ground for understanding the theoretical conceptualization of pragmatics. For additional exploration of this area see LoCastro (2003) and Crystal (1997). 5 Beebe and Waring (2001) and Márquez-Reiter and Placencia (2005) both note the difficulty in separating pragmalinguistic and sociopragmatic aspects of language. The terms are used here to distinguish the various components under consideration and are not meant to indicate mutually exclusive elements.

Learner Perception and Strategies for Pragmatic Acquisition

103

Approaches to Pragmatic Instruction Given the complexities associated with ILP pragmatic development, it is essential to utilize a multi-faceted approach to instructional practices. We must consider how learning of pragmatics actually takes place. Memorizing a list of specific linguistic forms based on a dialect or region in isolation is not sufficient for developing pragmatic abilities. Instead, it is advisable to focus on the skills and strategies necessary to be ―pragmatically adaptable‖ in a variety of contexts (e.g., region, age, social class, and setting). One step is to enhance the strategies that learners have for performing L2 pragmatics, providing the learners with examples drawing on the target-language pragmatic system. This approach is especially valuable in promoting the acquisition of sociopragmatic skills. In an exploration of contextual knowledge as related to language study, Kramsch and McConnell-Ginet (1992) note the complexity which ―context of interaction‖ entails: Sociocultural contexts cannot be reduced to an inventory to be ―mastered‖…they are not only too rich and various but also in constant flux as people reshape them through speaking and other forms of social interaction. Yet students can come to understand the centrality of context to linguistic communication and can develop some ethnographic skills to help them better understand the relevant contexts for their own uses of the target language. (p. 5)

Kramsch and McConnell-Ginet advocate the instruction of a set of ethnographic skills to deal with the constantly dynamic variable of context, a critical component of ILP development. The advantage of this type of approach is that it prepares learners for a wide variety of encounters. This also brings us closer to Thorne‘s (2005) conceptualization of intercultural competence which ―ideally moves learners from simulated classroom-based contexts toward actual interaction with expert speakers of the language they are studying‖ (p. 3). Furthermore, development of these skills would afford learners the ability to deal with contextual differences when talking with NSs. In the area of pragmatic instruction, two researchers have proposed models for addressing a set of pragmatic skills. First of all, Judd (1999) presents an approach to teaching pragmatics in the classroom which entails teacher analysis of the speech act, heightening of cognitive awareness skills, determining if students have the receptive skills to recognize speech act within different contexts, encouraging controlled production of the speech acts in different types of activities, and free integrated practice. Cognitive awareness raising is ―designed to make learners consciously aware of differences between the native and target language speech acts‖ (p. 154). Learners‘ speech act production is elicited through different types of written and oral activities (e.g., cloze passages, role-plays, simulations) designed to help learners ―actually incorporate the patterns into their own speech production‖ (Judd, 1999, p. 158). The various skills proposed could be used in combination, or separately, as a way to isolate specific pragmalinguistic and sociopragmatic aspects of the speech act under investigation. While Judd‘s approach tends to overemphasize pragmalinguistic skills, the model could also apply to sociopragmatic awareness. Most importantly, it emphasizes the notion of the need to develop a comprehensive set of pragmatic skills. Another recent approach to learning pragmatics central to this study is a strategies-based approach, introduced by co-author Cohen (2005). Distinct from teacher strategies used for instructional purposes as well as learners‘ general website-use strategies, Cohen offers a taxonomy of strategies intended specifically for the learning and performance of L2

104

Julie M. Sykes and Andrew D. Cohen

pragmatics. This taxonomy includes three types of strategies; these serve as the focus of this analysis. Table 1 lists all strategies from the taxonomy.

Table 1. Taxonomy of Strategies for Learning Speech Acts, Adapted from Cohen (2005)

Speech Act Learning Strategies

Communicative Act Use Strategies

Metapragmatic Considerations

L1: Access published material dealing with speech acts. L2: Ask native speakers to model performance of the speech act. L3: Identify the second language speech acts learners want/need to focus on. L4: Gather information (through observation, interviews, written materials, movies, radio) on how the speech acts are performed. L5: Conduct one‘s own "cross-cultural" analysis (e.g., identify norms and semantic formulae, determine similarities and differences). L6: Observe what native speakers do by noting what they say, how they say it, and their non-verbal behavior. U1: Remain true to one‘s own identity and subjectivity while still being aware of appropriate performance of the speech act. U2: Use communication strategies to get the message across (e.g., "I'm not sure how to say this right", repair when necessary, attempt to follow native speaker examples). U3: Ask native speakers for feedback. U4: Practice! (role plays, imaginary situations, conversations with native speakers) U5: Devise and utilize a memory aid for retrieving the speech act material that has already been learned. U6: Learners determine their learning style preferences and try approaches that are consistent with their individual style. M1: Be conscious of the necessity for pre-planning. M2: Decide on a focus. Performance? Comprehension? Both? M3: Monitor various elements of the communicative act (e.g., level of directness, terms of address, timing, organization, sociocultural factors).

The combination of these strategies recognizes learning pragmatics as the development of a skill set necessary for effective communication that moves beyond a specific speech act or appropriate semantic formulae used in realizing a language function.6 By developing the strategies necessary for dealing with various language functions in communication, learners will be prepared to handle a wide variety of contextual situations, as opposed to being restricted just to those they have encountered previously. Empirical research is still needed to confirm if this approach is effective and helpful for ILP development; however, a consideration of this type of approach is warranted. In order to begin investigation into a strategy-based approach to ILP development, the design of the 6

Semantic formulae refer to the speech act-specific strategies which alone or in combination with other strategies serve to constitute the speech act (e.g., the strategies of ―acknowledging responsibility‖ or ―offering a repair‖ in the speech act of apologizing).

Learner Perception and Strategies for Pragmatic Acquisition

105

instructional materials used in the current analysis drew extensively on the taxonomy proposed by Cohen (2005). By looking at how learners interacted with these materials, as well as through surveys tracking their perceptions, it was possible to analyze what they learned, their application of the strategies targeted through instruction, and their attitudes towards this process. This work focuses on the learner perception piece of the puzzle.

The Role of CALL in ILP Development Addressing ILP development in formal instructional settings can present a number of difficulties, many of which likely contribute to the lack of pragmatics instruction in the L2 classroom. Some of these barriers include a lack of theoretical support in course design and competency considerations (Bardovi-Harlig, 2001; Felix-Brasdefer, 2002; Rose, 2005; LoCastro, 2003), a lack of authentic curricular materials and appropriate input (BardoviHarlig, 2001; Félix-Brasdefer, 2002; LoCastro, 2003, a lack of instructor expertise (LoCastro, 2003), a focus on ―micro-level‖ features instead of ―macro-level‖ competence (BardoviHarlig and Dörnyei, 1998), limited time and attention in FL classroom (Kasper and Schmidt, 1996), perceptions and behavior unique to each individual learner (Kasper, 1997, BardoviHarlig, 2001), assessment and feedback challenges (Roever, 2004; Cohen, 2004; Salaberry and Cohen, 2006), and immense variation (dialect, social, individual) (Márquez-Reiter and Placencia, 2005). However, the emergence of innovative technologies allows us to overcome many of these barriers in order to make learning pragmatics a reality. Computer Assisted Language Learning (CALL) technologies play an increasingly important role in ensuring that pragmatics instruction is comprehensive in nature. Innovative technologies have the potential not only to aid in the creation of effective online materials for learning pragmatics (e.g., CLEAR, 2005; Cohen and Ishihara, 2005; Cohen and Sykes, 2006), but also to provide a valuable environment for interaction (Furstenberg and Levet, 2001; Belz, 2002, 2003; Biesenbach-Lucas, 2005; Sykes, 2005). In addition, an online environment offers a ready opportunity to provide a series of strategies that learners could use to enhance their learning and performance of pragmatics. Furthermore, online interactive environments offer features which make them an accessible vehicle for presenting information on pragmatics, and for dealing with issues associated with ILP development such as emotional sensitivity unique to individual learners, feedback and assessment difficulties, and the phenomenon of individual, social, and dialectal variation. Some of these features include: (a) opportunities to focus on different/multiple aspects of the language including discourse functions and syntactic complexity (e.g., Abrams, 2006; Furstenberg, et. al., 2001; Payne and Whitney, 2002; Sotillo, 2000; Sykes, 2005; Vandergriff, 2006), (b) effective, multi-level feedback with minimal instructor interference (e.g., Godwin-Jones, 2004; Linden and Rochon, 2003; Sotillo, 2005), (c) low risk practice situations with high emotional connection (e.g., De Freitas, 2006; Gee, 2003, 2005; Mistral, 2007; Wilcox et. al., 2006), and (d) paced, individualized interaction (e.g., De Freitas, 2006; Gee, 2003a, 2005). While a detailed exploration of each of these features is not possible within the scope of this work,7 their mention is warranted here due to mediated environments analyzed for the current work (to be described in detail in the following section). 7

See Sykes (2008) for a complete explanation of the role of collaborative environments such as SIEs in pragmatic development.

106

Julie M. Sykes and Andrew D. Cohen

The goal of the two studies to be reported on below was to provide learners with two novel online environments to assist them in gaining the necessary skills to tackle L2 pragmatic complexities in different contexts. At this point, little research has empirically analyzed the use of an online, strategy-based model for learning L2 pragmatics. Even less research has specifically targeted learner perceptions about such online environments. In order to arrive at a greater understanding of how learners perceive each of the mediated contexts, this paper compares learner perception data from two research projects which have utilized CALL materials to enhance learners‘ strategies for L2 pragmatics. In doing so, it sought to answer two research questions.

Research Questions 1. How do learners perceive their own strategy development for Spanish pragmatics through the use of two distinct CALL environments, a self-access website and a synthetic immersive environment? 2. What are learners‘ perceptions of learning Spanish pragmatics in each of these two CALL environments?

Methodology Both a self-access website and a synthetic immersive environment were included in this study in order to better understand learner perception about CALL materials. Both contexts were created with the aim of improving learners‘ strategic abilities in pragmatics. The two studies utilized similar research methods in order to analyze learners‘ perceptions. Moreover, both studies addressed the same speech acts and had a specific focus on learner strategies for learning and performing pragmatics, eliminating the effect that the content being taught would have on differences in learner perception. The first data set is taken from an in-depth, qualitative research project (S1) addressing the use and effectiveness of the first-of-its-kind learner website in L2 Spanish pragmatics with an explicit strategy overlay (Cohen and Sykes, 2007; Sykes and Cohen, 2008).8 The second is a subset of data taken from an extensive study (S2) analyzing the use of the first synthetic immersive environment (a 3-dimensional gaming space designed with a specific educational outcome) used as an instructional space for learning Spanish pragmatics, specifically requests and apologies (Sykes, 2007). The participant groups and methods employed for each of the studies are presented below followed by an explanation of their use for the current analysis.

Participants The participants in Study 1 (S1) consisted of a group of ten volunteers (N=10) who were 9 the first to express interest in the study. The participant group in Study 2 (S2) includes a 8 9

For a detailed report of this study see Sykes and Cohen (2008) Approximately 50 students volunteered for the study, indicating a strong interest in the subject matter as well as the project. The group was limited to 10 due to budget constraints.

Learner Perception and Strategies for Pragmatic Acquisition

107

subset of volunteers (N=25) who agreed to participate in one-on-one interviews and entrance and exit surveys in addition to their work in the virtual space as part of the advanced Spanish course they were taking. This group (S2) represents approximately 50% of the participants in the full-scale study (N=53). The demographic composition of both the S1 and S2 groups is fairly similar and, therefore, allows for comparison of the two data sets. In addition, both represent the typical composition of advanced Spanish classes and the large, mid-western university where both studies took place. Table 2 below presents the demographic compilation of each of the participant groups. Table 2. Demographic Information about the Participant Groups Participant Group 1 (S1)

Participant Group 2 (S2)

Native Language

Female (N=5) Male (N=5) Range 19-37 Average Age: 22.6 English

Female (N=22) Male (N=3) Range 19-25 Average Age: 20.1 English

Dominant Language

English

English

Experience Learning Spanish

High school and university classroom; 40% (N=4) study abroad Length of stay: 1 month-1 year

High school and university classroom; 32% (N=8) study abroad Length of stay: 3 weeks-1 semester

Gender Age

Study Abroad

As can be seen in Table 2, the participants represented learners of Spanish from ages 19 to 37. While the S1 group was balanced in terms of gender, the S2 group was predominantly female. This difference can be attributed to the subject recruitment method. The participants in S1 were recruited outside of the classroom context, allowing for the creation of a balanced group. On the other hand, S2 was made up of student volunteers from a research project studying classroom activities. In the classes used for the study, the majority of the enrolled students were female. Therefore, it is to be expected that the subset of volunteers was also predominantly female. Both groups S1 and S2 came from a variety of language learning contexts. Some began their studies earlier than others (high school vs. university), yet, all were enrolled in upper-division Spanish courses at a large mid-western university. While class placement alone is not a strong indicator of proficiency level, it can be determined that all learners in both participant groups were at least at a minimum level of proficiency. In order to be enrolled in these courses, students were required to pass a university-administered, language proficiency exam (LPE) indicating that they were at or above the intermediate-mid level on the ACTFL proficiency scale.10 Therefore, it was expected that all participants were at least at the Intermediate-High level for Reading and Listening and an Intermediate-Mid

10

While pragmatics tends to be considered primarily part of oral communication (i.e., speaking and listening), it is essential to examine reading and writing as well, since much of the communication takes place in a multimodal environment that requires the use of written text as well.

108

Julie M. Sykes and Andrew D. Cohen

level for Speaking and Writing.11 All participants in both groups were native speakers of English, and had fairly high average GPAs in Spanish (S1 = 3.63; S2 = 3.625). While very few participants (S1: 30%; S2: 16%) in either group spoke languages other than Spanish, some participants did report having studied French (S1: N=1, S2: N=2), Mandarin (S1: N=1), German (S2: N=1), Hindi (S2: N=1), and Portuguese (S1: N=1). It is important to bear in mind that both learner groups accessing the CALL materials for learning Spanish pragmatics were relatively diverse. This allows for a more comprehensive glimpse into how a variety of different types of learners perceive the CALL materials for developing a strategic approach to ILP development.

Data Collection Procedures and Instrumentation Each of the studies employed in this analysis consisted of an entrance survey, pretest, treatment period, post test(s), exit survey, and one-on-one interview with the researcher. The data collection procedures and instrumentation for each study are presented below. Detailed descriptions of the outcome measures (pre- and posttests) are included for informational purposes; however, the results of the outcome data are not reported as a component of the current analysis in this chapter.12

Study 1 (S1) In this study (Cohen and Sykes, 2007; Sykes and Cohen, 2008), all subjects attended a general descriptive session about the project and then completed a paper-and-pencil entrance survey. This survey was designed to obtain important demographic information about each of the participants as well as to collect data concerning the types of learning and performance strategies learners perceived themselves to already be using (based on Cohen, 2005). Information from this initial survey was important in establishing a baseline of perceptive strategy use. Approximately one week after completing the entrance survey, participants completed a pretest designed to evaluate their level of pragmatic knowledge. The pretest took approximately one hour and consisted of two parts—(1) a written multiple-rejoinder discourse completion task (DCT),13 with five situations based on material from the instructional website, Dancing with Words: Strategies for Learning Pragmatics in Spanish, calling for two requests, two apologies, and a service encounter, and (2) a three-part role-play in a synthetic immersive environment assessment (SIEA) area: a request to borrow their host sister's course notes, a service encounter with a street vendor (buying souvenirs), and apologizing to their sister for spilling Coke on the notes in their backpack and ruining them. Given the use of DCTs in this study, it is important to note concerns raised as to their validity for data collection in pragmatics (e.g., Golato, 2003; Roever, 2004; Garces-Conejos Blitvitch, 2006). It has been claimed, for instance, that the use of DCTs produces data that are not representative of language use, but rather reflect speaker intuitions or perceptions. Other commonly asserted limitations of the DCT include the lack of genuine interaction in 11

For a detailed explanation of these proficiency levels as indicated by ACTFL, see www.ACTFL.org. See Sykes and Cohen (2008) and Sykes (2008) for an analysis of the outcome data for each study. 13 A DCT is a paper-and-pencil task in which participants receive a context and then one, or more, turns of the conversation. They are then asked to fill in what they would say in each of the blanks. Different formats (e.g., open-ended vs. close-ended, multiple rejoinder possibilities) have been designed with the intention of eliciting data more like natural speech. 12

Learner Perception and Strategies for Pragmatic Acquisition

109

responding to DCT vignettes, the limited contextual information (making it difficult for participants to put themselves in the situation), the use scenarios that are too complex making them inaccessible to learners (or even NSs), the over-reliance on the intuition of the speaker, and the disconnect between what they say they would do and what they actual do (FélixBrasdefer, 2003; Golato, 2003; Roever, 2004). Nevertheless, these limitations should not be taken as grounds for ruling out the use of the DCT for pragmatics research. In fact, as confirmed by Kasper and Rose (2002), when used correctly, elicited data (e.g., through a DCT) can be quite useful. ―Contrary to the popular perception that ―inauthentic‖ equals ―invalid,‖ interactions arranged for research purposes can be most useful sources of data if used judiciously‖ (Kasper and Rose, 2002, p. 80). DCTs actually present a number of important benefits. In this case, the use of a DCT allowed the researchers to access what the learners know about each of the speech acts, as well as to isolate the specific semantic formulae utilized. In addition, the DCTs were utilized in conjunction with the virtual role-play scenarios in order to triangulate the assessment data. The synthetic immersive environment assessment (SIEA) is an important contribution of this project because it allowed for authentic interaction to occur between a native speaker and the participants in a realistic and three-dimensional interactive space. Utilizing much of the content from the instructional materials, we created a virtual online environment for the purpose of assessing Spanish pragmatics. This new virtual space allows for assessment of both speech act performance as well as students‘ use of the resources within the virtual space for successful interaction. It builds on experiences over the last decade with synchronous computer-mediated communication (SCMC) (Payne and Whitney 2002; Belz 2003, 2005; Payne and Ross 2005; Sykes 2005) as well as current educational gaming research (e.g., Gee, 2003, 2005; DeFreitas, 2006).14 The SIEA tool was an adaptation of one component of an online virtual world named Croquelandia that was developed by Sykes and a team of programmers as part of a larger project at the University of Minnesota. Croquelandia the learning environment used in S2. The graphics in the space were created utilizing photos taken by Sykes in the Spanishspeaking world. The photos were then adapted and redesigned into the space by the graphic design and programming team. In the SIEA, learners were able to move their avatar15 throughout the environment, interact with built-in content and non-player characters (NPCs), and talk with a native speaker. See Figure 1 for images of the SIE assessment space based on Croquelandia. The creation of this SIEA space provided a low-risk, yet authentic, space for interaction and shows promising possibilities for future pragmatics research and instruction.

14

See Sykes (2007) for further discussion of the relationship between SCMC and SIEs as related to learning L2 pragmatics. 15 An avatar is the virtual representation of one‘s character in the online virtual space. One‘s avatar is controlled by the user and can interact with the digital space.

110

Julie M. Sykes and Andrew D. Cohen

Figure 1. Sample Virtual Settings in SIEA Space.

After completing the pretest, participants then participated in a content orientation session which consisted of a one-hour introduction to pragmatics and to the strategies taxonomy (Cohen, 2005). Following the content orientation session, participants then completed online modules from Dancing with Words: Strategies for Learning Pragmatics in Spanish. The Dancing with Words website is a content site (http://www.carla.umn.edu/speechacts; Cohen

Learner Perception and Strategies for Pragmatic Acquisition

111

and Sykes 2006), developed over eleven months that was launched in August of 2006. It consists of an introductory module as well as eight additional modules – (1) Compliments, (2) Gratitude and Leave Taking, (3) Requests, (4) Apologies, (5) Invitations, (6) Service Encounters, (7) Advice, Suggestions, Disagreements, Complaints, and Reprimands, and (8) Considerations for Pragmatic Performance. It includes unscripted video interchanges between natives of various regional varieties of Spanish and utilizes scaffolding for the purpose of addressing the learners‘ varying levels of language/pragmatic ability. Speech acts are dealt with sequentially: first as a core, then in interaction, and then as a naturally occurring sequence. As previously mentioned, one important feature of this site that makes it especially useful for this project is the extensive strategies overlay that goes along with the materials (Cohen, 2005).16 The three modules completed for this project were requests, service encounters, and apologies. It took participants approximately 1-2 hours to complete each of the modules, for a total of between 3-6 hours of self-access content time with the website. All online activity was recorded using Camtasia Screen Capture (http://www.techsmith.com/camtasia.asp) software. The screen recordings allowed for observation and analysis of the participants‘ interaction with the website as well as the responses given to each of the interactive activities. After the completion of their final module, all participants took part in a reflective interview with one of the researchers. Each interview lasted approximately fifteen minutes and was recorded and transcribed for future analysis. No more than 48 hours after completing the last module, participants completed an immediate posttest, similar to the pretest, but with varied situations and contexts in the SIEA. The students had to make a request to borrow money from their host sister, who only had a large croquedo bill which they needed to get changed and return the change. The service encounter was the same, but the apology was for losing the rest of the money. There was also a delayed posttest, which was the same as the pretest. The data set used from this study for the current analysis included the survey results for the entrance and exit surveys as well as the interview data.

Study 2 (S2) In this study (Sykes, 2008), participants completed two modules in a 3-dimensional, immersive environment as part of their course in Spanish. All students enrolled in the course participated in the instructional activities which were also recorded and observed for research purposes. In addition, a subset of volunteers (S2) also participated in additional researchoriented activities outside of their classroom activities. Since the focus of this article were the perceptions as reported by a subset of 25 participants, the following section describes the nature of their participation in the study. First, all the student subjects participated in a general introduction to the project and completed an entrance questionnaire administered online. Much like Study 1, the entrance survey in this study was designed to glean important demographic and experiential information from each of the learners prior to beginning the instructional activities in order to establish a baseline for their perceptions as to strategy use. One section of this survey was parallel to the strategies section used in Study 1. As was previously mentioned, this set of

16

For additional background information sp_pragmatics/for_researchers.html

on

the

website

see

http://www.carla.umn.edu/speechacts/

112

Julie M. Sykes and Andrew D. Cohen

questions was used to collect data concerning the types of learning and performance strategies learners perceived themselves to already be using (based on Cohen, 2005).17 During Week 3 of the semester, learners participated in two class sessions dedicated to pragmatics and the course project. Class Session 1 a theoretical introduction to pragmatics was presented, based on what the students read for homework, and a pretest DCT was included as part of the classroom activities. This class session was taught by the researcher, an experienced instructor, in all five class sections to maintain consistency throughout the groups. Prior to arriving in class, learners were assigned two introductory readings, chapter 1 from LoCastro‘s (2003) textbook, and chapter 7 from Koike and Klee (2003). These readings provided the theoretical basis for the in-class discussions on pragmatics and introduced a number of important concepts in pragmatics with which learners needed to familiarize themselves, including a strategies-based approach. Class Session 1 was a typical class session dedicated to an introduction of pragmatics. The class session consisted of an interactive PowerPoint presentation and discussion of important concepts related to the assigned reading. Between Class Sessions 1 and 2, all participants were encouraged to review the articles and project description. During Class Session 2, participants had the opportunity to practice in the synthetic immersive environment (SIE), watch a number of online tutorials introducing them to the technology, and ask any questions they might have had. The tutorials were available for consultation throughout the course of the project. During Weeks 3-7 of the semester, learners were given the opportunity to work in groups of 4 in the online SIE to improve their skills in Apologies and Requests. Participants worked in the SIE in order to learn everything they could about the designated speech acts and compile a 5-7 minute presentation summarizing what they learned. Part 1 (Weeks 4-5) addressed either Requests or Apologies and Part 2 (Weeks 5-7) addressed the speech act not covered in Part 1. The order and distribution of the environment assignments and speech acts varied by group in order to eliminate any task affect on the results. Consequently, the treatment occurred in two different sequences. Half of the group followed Sequence 1 (i.e., Part 1 working on apologies; Part 2 working on requests) and the other half followed Sequence 2 (i.e., Part 1 working on requests; Part 2 working on apologies). All interaction occurring in the online collaborative environments was recorded and archived for analysis using a database built-in to the SIE. In between Parts 1 and 2, all participants took part in a mid-point, one-on-one interview with the researcher. This interview focused on learners‘ initial perceptions of the space as well as suggestions for what they would like to see in the future. In addition, the questions focused specifically on learners‘ experience with the module they had recently finished, either requests or apologies. Each interview lasted between 15-30 minutes. All responses were recorded and transcirbed for future analysis. Before continuing with a description of the study itself, it is important to further describe the SIE environment. The SIE used in this study was an online virtual world named Croquelandia that was designed by Sykes and a team of system architects at the University of Minnesota as part of a larger project, specifically for learning and practicing Spanish pragmatics. The Spanish model is the first environment of its kind designed for language 17

The items in each survey for S1 and S2 were quite similar. However, some minor differences did occur. These included one additional item on the survey in S1. Therefore, the pre-post test design was critical in making the results comparable.

Learner Perception and Strategies for Pragmatic Acquisition

113

learners and is designed to serve as a model for similar SIEs in other languages. The space was developed utilizing five guidelines for the creation and use of CALL materials to aid in interlanguage pragmatic development. These are:18 Design 1. Objectives should be co-operative, dynamic, and socially-constructed in order to allow for authentic and realistic pragmatic practice. 2. Educational outcomes of the activities should be explicitly defined and supported with accurate content. Task Creation and Administration 3. Tasks should reflect the determined objectives and outcomes through multiple involvement opportunities, realistic contexts, and varied participant roles. 4. The platform should be user-friendly and supported by initial training and ongoing support. Feedback and Assessment 5. Instructors should play a passive, yet interested, role during tasks. Feedback should be reserved for follow-up and processing. It is important to note that the SIEA space used in Study 1, represents one component of the larger Croquelandia project. As was mentioned previously, all of the graphical content in the space was created utilizing photos taken by Sykes in the Spanish-speaking world. Sound recordings of street noises were also made using an iPod and iMic. The photos and sounds were then adapted and redesigned into the space by the graphic design and programming team. In the space itself, learners can collaborate and interact in three primary areas – their host family‘s house, a central plaza and market place, and the university. Learners could move seamlessly between the three spaces using an interactive map in which they could click on the area to which they wanted to travel (see Figure 2). In addition, the participants were able collaborate with their group members using voice or written chat and could interact with the environment by clicking on different items, walking around the space, and ‗talking‘ with non-player characters (computer-generated avatars present in each of the spaces). Figure 3 shows the main square and marketplace in Croquelandia. All NPC interactions were created from roleplay conversations with a variety of native speakers. See Figure 4 for an example of this type of interaction. All native speakers were given a context and then asked to react to the other person in the conversation (the researcher). Three or four conversations were recorded for each conversation with different semantic formulae carrying different pragmatic meanings. These conversations were then parsed and used to create a ―choose your own adventure‖ conversation tree for the learner to

18

For a detailed description of each of these guidelines see Sykes (2008).

114

Julie M. Sykes and Andrew D. Cohen

choose from. The NPC in the virtual space reacts based on what the native speaker actually did in the ―real world‖ conversation.

Figure 2. Navigation Map in Croquelandia.

Figure 3.Sample of Market in Croquelandia.

Figure 4. Sample NPC Interaction.

Learner Perception and Strategies for Pragmatic Acquisition

115

Figure 5. Detailed Description of User Interface.

In Croquelandia, learners were also able to create their own content and leave messages for their group members using an asynchronous bulletin board feature. See Figure 5 for a detailed description of the interactive features of the user interface. After completing all activities in the SIE, the participants worked as a group to present their findings about Apologies and Requests in Spanish. The class assignment required participants to create a two-part presentation in Spanish and was worth 10% of the students‘ grade in the course. Following the presentations an in-class discussion was used to summarize the content learned. The group presentations and the discussion were videotaped for future analysis. It should be noted that the post test DCT occurred prior to the class presentations in order to ensure that any changes could be attributed to work done in the SIE. Finally, after completing the group presentations and all other activities, the participants completed an exit survey similar to the entrance survey as well as a final one-on-one interview. Each interview lasted between 10-15 minutes and addressed overall perceptions of the project and experience in the SIE. All responses were recorded and transcribed for analysis.

Current Project As can be seen from the detailed descriptions above, each of the studies was complex. However, the similarities between components of the two data sets allows for logical comparisons to be made using the measures each has in common.

116

Julie M. Sykes and Andrew D. Cohen

Figure 6. Data Collection Procedures for Studies 1 and 2

Therefore, the goal of this research project was not to compare all data from both studies, but rather to utilize data from three specific measures to better understand the learners‘ perceptions regarding strategic development in each of the two environments. This chapter compares the entrance and exit survey data specifically addressing learners‘ perceptions of the strategies used for learning and performing pragmatics. It also considers the one-on-one interview data from both studies. Figure 6 below provides a graphical comparison of each study. The data being utilized for the current analysis are noted with an asterisk (*).

Learner Perception and Strategies for Pragmatic Acquisition

117

Procedures for Data Analysis Survey and interview data from each of the aforementioned studies were used in order to answer the proposed research questions. First, entrance and exit survey data from both studies were analyzed to examine learner perception of strategy use before and after completing the online activities (Research Question 1). In each of the surveys, the participants reported their use of each strategy from Cohen‘s (2005) taxonomy prior to the treatment activities and after completion of the instructional materials. In the case of strategy U6 (learning style), the group in Study 1 (S1) was queried using one survey item whereas, in Study 2 (S2), two items were used to examine individual style (i.e., cognitive style and personality). In this case, the two items in S2 were averaged for the purposes of this comparison. The survey responses indicated perceived strategy use in each of the three areas reported by Cohen as being important for strategic development in pragmatics-(1) Learning Strategies, (2) Use Strategies, and (3) Metapragmatic Strategies. For further explanation of each of these areas of strategic development see the Relevant Background Information section of this chapter. In each survey, the respondents evaluated their own strategy use on an ordinal scale with five categories and a sixth ―Don‘t know‖ option. For analysis purposes, the mean of all responses was calculated by assigning a numerical value to each response and then calculating the average response value and standard deviation. The numerical values corresponding to each qualitative response are as follows: Seldom/Never=1, Somewhat Likely/Sometimes=2, Likely/Often=3, Very Likely/Very Often=4, Almost Always=5, and Don‘t Know=0. Based on this scale of numerical assignment, a higher mean points to a higher perception of that strategy being used by the learners themselves. In order to compare the two data sets, the change between the entrance and exit means was calculated for each strategy. An analysis of the change rate based on the entrance and exit responses for each group helps eliminate overall differences in perceived strategy use between the two groups and maintain a consistent value for comparison. The change rate of perceived strategy use was calculated by subtracting the entrance score from the exit score. Thus, a positive change rate indicates increased perceived usage whereas a negative number shows a decrease in perceived usage on the part of the learners. For the purposes of this analysis, a change rate of less than .5 was considered as no change. A change rate between .5 and 1 in either direction was considered noteworthy and indicative of a minor increase or decrease. A change rate greater than 1 in either direction was considered as indicative of a moderate change. After calculating the values, rates of change for each type of pragmatic strategy (learning, use, and metapragmatic) were compared for both studies. Due to small sample size and different group of learners in each study, the results could not be verified through calculations of statistical significance. Nevertheless, some change patterns worthy of discussion did emerge. These will be presented in the following section. For research question 2, patterns in the interview data were found through qualitative categorization of each set of interviews. The themes found common to each and relevant for the comparison of learner perception of each of the spaces are reported in the Findings section.

118

Julie M. Sykes and Andrew D. Cohen

Findings The survey and interview data from both Studies 1 and 2 provided material for responding, at least in a preliminary fashion, to each of the research questions addressed in this study. The goal of this analysis is to explore findings that can then be used as a starting point for additional empirical research in this area. This section presents results and interpretations for each research question, as well as the limitations of the present study.

Research Question 1 How did learners perceive their own strategy development for Spanish pragmatics through the use of two distinct CALL environments, a self-access website and a synthetic immersive environment?

Results In order to answer the first research question, learners‘ perceptions of strategy use were measured using the entrance and exit survey data from both studies. In this section, we report the results of the comparison of the survey data for each of the three categories of pragmatic strategies. First, in the learning strategies category, we see one strategy (L5=.1) in S1 and four strategies (L1=.3; L2=-.045; L3=.295; L5=.375) in S2 that we consider as no change. As can be seen in Table 3, the strategies which exhibited minor changes in perceived use by the learners are strategies L4 (S1=.65; S2=.53) and L6 (S1=.7; S2=.5). Each exhibited fairly equivalent rates of change in both studies. In addition, we also see a minor change in strategy L1 (.6) in S1. Finally, in S1, both strategies L2 (1.1) and L3 (1.63) showed moderate changes. Both of the strategies that had a moderate change rate in S1 exhibited no change in S2. Overall, results for the learning strategies category suggest an increased perceived use of learning strategies after having participated in the online website activities (S1). However, in the SIE (S2), there was some evidence of an increase in perceived strategy use in 2 of the 6 strategies and no change in the others. The larger increase in perceived strategy use is confirmed in that the average rate of change for this category in S1 was .79 (minor change) and for S2 was .32 (no change). Figure 7 summarizes the results of this category in graphical format.

Table 3. Reported Average of Learning Strategies in S1 and S2 Study 1 (S1) S1 Entrance Mean SD

S1 Exit Mean SD

Change

Study 2 (S2) S2 Entrance Mean SD

S2 Exit Mean SD

Change

L1: I will refer to published material (e.g., articles, websites) dealing with communicative acts.

2

.94

2.6

.84

0.6

1.45

.71

1.75

.98

0.3

L2: I will ask native speakers to model how they perform the communicative act.

2.4

1.17

3.5

.87

1.1

1.625

.71

1.58

.92

-0.045

L3: I will identify the communicative acts (i.e., requests, apologies, compliments) that I want/need to focus on. L4: I will gather information (through observation, interviews, written materials, movies, radio) on how the communicative acts are performed.

2.2

1.54

3.83

1.06

1.63

2.58

1.43

2.875

1.07

0.295

2.9

1.28

3.55

1.30

0.65

2.625

1.36

3.16

1.34

0.535

L5: I will conduct my own cross-cultural analysis (e.g., identify norms and strategies specific to a given communicative act like ―requesting,‖ determine the similarities and differences between my first language and Spanish).

3.2

.91

3.3

1.05

0.1

2.625

1.51

3

1.25

0.375

L6: I will pay attention to what native speakers do by noting what they say, how they say it, and their nonverbal behavior.

3.4

1.26

4.1

.87

0.7

3.125

1.65

3.625

1.01

0.5

Strategy

120

Julie M. Sykes and Andrew D. Cohen

Reported Average of Learning Strategies in S1 & S2 L1-Published Materials L2-NS Model S1 Entrance

L3-Identification

S1 Exit

L4-Gather Information

S2 Entrance L5-Cross-cultural Analysis

S2 Exit

L6-Attention to NS 0

1

2

3

4

5

Figure 7. Reported Average of Learning Strategies in S1 and S2.

In terms of the use strategies category, a similar picture emerges. Once again, we see an increase in perceived use for more strategies in S1 (4/6) than in S2 (2/6). Table 4 reports the means, standard deviation, and change for each individual strategy in the entrance and exit surveys for both studies. Utilizing the comparison of entrance and exit survey data for both studies, we see that for strategy U2 there was very little change in either study (S1=.2; S2=0). Furthermore, no change is found for strategy U5 (.45) in S1 and strategies U3 (.04), U4 (.035), and U6 (0.105) in S2. In terms of minor change, Strategy U1 (S1=.06; S2=.62) exhibits equivalent rates of change in both studies and, in S1, strategies U4 (.85) and U6 (.5) also present evidence of a minor increase in perceived use. In the use category, the only strategy which exhibits a moderate change is U3 (1.05) in S1. As is the case with the learning strategies, there are no instances of moderate change in S2. Overall, results for the use strategies category indicated a tendency for increased perceived deployment of use strategies after having participated in the online website activities (S1) and little change in perceived strategy deployment upon participation in the SIE (S2). In general, participation in the website resultd in a minor change rate in the use category in S1 (.60) and no overall change in S2 (.18). Figure 8 summarizes these results of this category in graphical format. The final category of strategies to be analyzed is that of metapragmatic strategies. In this category, a slightly different pattern emerges than that found in the learning and use categories.

Table 4. Reported Average of Use Strategies in S1 and S2 S1 Entrance

S1 Exit

S2 Entrance

S2 Exit

Strategy

Mean

SD

Mean

SD

Change

Mean

SD

Mean

SD

Change

U1: I will remain true to my own cultural identity and personal values while still being aware of the cultural expectations of native speakers. U2: I use communication strategies to get the message across (e.g., "I'm not sure how to say this right," repair when necessary, attempt to follow native speaker examples). U3: I ask native speakers for feedback on my pragmatic abilities.

3.5

.97

4.1

.87

0.6

3.29

1.60

3.91

1.10

0.62

3.6

.84

3.8

.91

0.2

3.16

1.20

3.16

1.23

0

2.75

.92

3.8

1.03

1.05

1.54

1.11

1.58

.82

0.04

U4: I practice (e.g., role-plays, imaginary situations, conversations with native speakers) in order to improve my pragmatic skills. U5: I will devise and utilize memory strategies for retrieving the communicative act materials that has already been learned.

2.75

1.31

3.6

1.17

0.85

1.875

1.30

1.91

.97

0.035

2.6

.69

3.05

1.30

0.45

2.29

1.28

2.79

1.25

0.5

U6: I will determine my style preference as a learner and try approaches that are consistent with my individual style.*

3.1

.99

3.6

1.07

0.5

3.395

1.38

3.29

1.55

-0.105

122

Julie M. Sykes and Andrew D. Cohen

Figure 8.Reported Average of Use Strategies in S1 and S2.

Table 5 reports the mean, standard deviation, and change for the metapragmatic strategies category. In this case, 2 of the 3 strategies in S1 exhibit no change (M2=-0.2; M3=.45) and only one strategy in S2 shows no change (M3=.46). For both M1, a minor perceived rate of change was found in both studies (S1=.5; S2=.79). Strategy M2 is interesting in that it is the only strategy in which we see a minor change in S2 (.66) and no change in S1. There are no moderate changes in this category. Overall, results suggest that participation in the synthetic immersive environment (S2) most likely had a higher impact on learners‘ perceived use of metapragmatic strategies than the website (S1), albeit the change was minor in S2. This was the only category in which the S2 group indicated a higher overall rate of change. The overall average strategies for Studies 1 and 2 in this category were .25 (no change) and .63 (minor change), respectively. Figure 9 is a graphical representation of the data from the metapragmatic strategies category.

Figure 9-Reported Average of Metapragmatic Strategies in S1 and S2.

Learner Perception and Strategies for Pragmatic Acquisition

123

Table 5. Reported Average of Metapragmatic Strategies in S1 and S2 S1 Entrance

S1 Exit

S2 Entrance

S2 Exit

Strategy

Mean

SD

Mean

SD

Change

Mean

SD

Mean

SD

Change

M1: I will be conscious of the necessity for preplanning.

3

.81

3.5

1.35

0.5

2.91

1.22

3.7

1.08

0.79

M2: I will decide what my focus is. Performance? Comprehension? Both? M3: I will monitor my performance of communicative acts (e.g., level of directness, terms of address, timing, organization, sociocultural factors).

3.3

1.15

3.1

1.19

-0.2

2.25

1.40

2.91

1.24

0.66

2.6

1.07

3.05

1.21

0.45

2.33

1.41

2.79

1.25

0.46

In general, these data suggest that the website had a minor impact on S1 learners‘ perceived use of learning and use strategies. In the case of strategies L2, L3, and U3 we see a moderate increase in perceived use. S1 demonstrates no effect on learners‘ metapragmatic strategies. The opposite appears to be true for S2. In S2, the data exhibits no change in the learning and use categories overall (with a few exceptions indicating minor changes at the strategy level) and a minor increase in perceived use of metapragmatic strategies. In all cases (S1 and S2) where there is evidence of minor or moderate change, there is always an increase in perceived use, not a decrease. The following section addresses some implications of these results.

Discussion The previous comparison of perceived strategy use data are a sign that participation in different types of mediated contexts (i.e., a self-access website and an immersive space) may have had a minor impact on perceived strategy use. In looking at all three categories, it can be preliminarily concluded, based on the scale used in this study, that the users of the self-access website in S1 reported a moderate increase in the use of learning and use strategies for pragmatic development with no change in their use of metapragmatic strategies. However, the users of the synthetic immersive environment in S2 indicated a minor increase in the perceived use of metapragmatic strategies for pragmatic development and no change in the learning and use categories. In analyzing possible explanations for the preliminary differences found between S1 and S2, it is important to first examine the emerging differences at the category level. We will then identify unique findings for some of the specific strategies themselves. First of all, the category-level difference between S1 and S2 should be understood as emerging patterns worthy of future attention, and not as conclusive results. The minor increase in learning and use strategies in S1 and no evidence of change in S2 could be attributed to the type of strategies-based instruction each group of learners received through the website and SIE. In

124

Julie M. Sykes and Andrew D. Cohen

the website, explicit identification and exploration of each of the strategies was included as part of the instructional activities. In the SIE, the strategies-based approach entailed experiential learning. That is, instead of explaining to the learners how they might use a specific strategy to improve their pragmatic abilities (the case of the website), the SIE quests and activities required that learners implement each of the strategies through practice and use. Both received introductions to a strategic approach to pragmatics; however, the distinct delivery method of instruction may explain the differences found in the learning and use strategies categories. At the same time, delivery method does not explain why there is no evidence of change in S1 and a minor change in S2 in the metapragmatic category. If context of instruction were the only factor, it would be expected that learners in S1 would also exhibit an increase in perceived use of metapragmatic strategies since they received explicit instruction similar to that of the learning and use categories. Therefore, other possible explanations should be explored. First of all, we do not deny that the differences between S1 and S2 could be explained by inherent distinctions in each group of participants since they represent two different groups of learners. In other words, it could be that the participants in S1 were naturally drawn to learning and use strategies while those in S2 leaned towards metapragmatic differences. However, since both groups exhibit similar characteristics including age, proficiency level, learning context, and study abroad experience, we are assuming, for the purposes of this analysis, that they are comparable groups. Taking this into consideration, there are more than likely additional interpretations of the data. One plausible explanation for the difference is that metapragmatic strategies were more often employed and practiced by the participants in S2 as they worked in the immersive space. For example, in order to complete a quest successfully, learners had to select who they were going to talk to (M1, M2) and decide what they were going to work on before beginning (M1, M2). In addition, monitoring one‘s own pragmatic behavior (M3) was a central component of the SIE experience and was built into the immersive space itself. These findings are congruent with those from current educational gaming research demonstrating the positive impact that engagement1 (i.e., the complex, immersive, emotional experience of participating in an activity) can have on cognitive processing of certain skills (Gee, 2003; Johnson, 2005; Prensky, 2001; Taylor, 2006). Therefore, it could be the case that the SIE is especially useful for developing metapragamatic strategies and that learners were more aware of their importance based on their experience in the virtual world. In addition to category-level comparisons, some differences warranting consideration also emerge at the strategy level. First of all, in the learning and use strategies categories, strategies L2 (i.e. Ask native speakers to model how they perform the communicative act.), L3 (i.e., Identify the second language speech acts learners want/need to focus on.), and U3 (i.e., Ask native speakers for feedback on your pragmatic abilities.) are especially noteworthy because we find a moderate increase in perceived use from the participants in S1 and no change in S2. One feasible explanation for the distinction may be the emphasis on the use of a native speaker (NS) as a resource for pragmatic learning in the website itself. In the activities on the website, learners utilize models and examples in improving their abilities. However, for additional information they would need to talk with NSs in the real world and, as a result, 1

In discussing digital game-based learning, Prensky (2001) defines twelve reasons why games are especially engaging. These include fun, play, rules, goals, interactivity, outcomes and feedback, adaptive capabilities, winning, conflict and challenge, problem solving, interaction, and representation and story (p. 106).

Learner Perception and Strategies for Pragmatic Acquisition

125

are instructed to do so. In the case of the SIE, learners have the ability to ―talk‖ with virtual NSs throughout their experience and repeat their own observation of behavior as many times as they wish. Therefore, participants in S2 may not perceive explicitly asking for help of NSs as a necessary resource for pragmatic development, we do not intend to imply that learners should always try to emulate NS behavior.2 Strategy U6 (i.e., Determine your style preference as a learner and try approaches that are consistent with that individual style.) also warrants special attention. Once again, in S1 there is an increase in perceived usage (.05) and no change (-0.105) in S2. As mentioned in the Procedures for Data Analysis section, this discrepancy could be due to the measure itself in that the queries used were distinct in each survey. Overall, a comparison of perceived strategy use show a minor increase in perceived use in the S1 group for learning and use strategies and no change for metapragmatic categories. The opposite was true in S2 where there was no overall change in the learning and use categories and a minor change in the metapragmatic category. This provides preliminary indications that online environments may be effective for increased awareness in strategy use and that the type of mediated context may have an impact on the type of perceived strategy use (i.e., whether strategies for learning pragmatics, for use or material already learned, or for metapragmatic evaluation of performance). In considering these results, it is critical to remember that the findings are based on learners‘ perceptions, and not on actual use. Learners reported which strategies they would use for improving their pragmatic abilities based on their experience in each of the environments. With preliminary empirical indication of perceived change rates (i.e., none, minor, and moderate), the next step, to be discussed in detail shortly, is to critically validate, first, that the use of these strategies does indeed enhance pragmatic abilities and, second, that learners are actually using the strategies they say they use.

Research Question 2 What were learners‘ perceptions of learning Spanish pragmatics in each of these two CALL environments?

Results A qualitative analysis of the interview data from both studies reveal notable patterns that should be considered for future research and development in this area. This section highlights some of the most salient patterns from the one-on-one interviews in each study and compares some of the differences found between participants who accessed the learning material in the website and those who utilized the SIE for learning. The three most salient findings from the interviews were:

2

Although not appropriate for discussion in this work, the notion of the ‗native speaker‘ can be problematic, especially when dealing with pragmatic development. See, for example, Kramsch and McConnell-Ginet (1992), Kramsch(1997), and Thorne (2003), among others.

126

Julie M. Sykes and Andrew D. Cohen (1) There was strong motivation and interest in learning pragmatic content and developing ILP competency. This motivation appears to be enhanced through participation in the SIE. (2) In most cases, more was better. Learners wanted a variety of activities, contexts, and practice spaces. For the best experience for all learners, online pragmatics materials should be delivered in a variety of ways. (3) Learners were high-demand consumers of technology; however, some leniency was given for new and innovative initiatives.

First of all, in both studies, participants reported a strong interest in the content material and motivation to continue their own ILP development. It can be concluded that interaction with the strategies-based materials produced positive results and was well-perceived by the participants in each of the studies. Extracts (1) and (2) both provide notable indications of this positive experience. Extract (1) is from S1 and Extract (2) is from S2. (1) …what you are getting at with the program is really, really positive because I really don‘t think there is enough emphasis on real world application.…what I am always super, super frustrated with is you always end up with a class full of people who can write A+ papers and perfect grammar, and they can‘t speak it to save their lives…the fact that you‘re emphasizing a lot more on real world situations than on grammar is something that the Spanish curriculum desperately needs. – Paco, S1 (2) The best [aspect of the project] is we‘ve never talked about pragmatics at all before and I‘ve always taken Spanish and never [learned about pragmatics]. The only thing I really knew is because one of my best friends is from Spain. She lived with my family for a little bit. She never said sorry for anything. That was how I knew that. So... that was actually neat learning about it in a classroom setting. It‘s a really cool idea. I liked the whole program being able to do that. –Roxy, Interview 2, S2

In each of these studies we see positive motivation and interest in the content itself. This suggests that both mediated contexts presented a useful and motivating approach for learners to improve their pragmatic abilities. Furthermore, a strategies-based approach appeared to be interesting to the learners and something they were willing to address as part of their pragmatic repertoire. Of all participants interviewed for both studies, none indicated a lack of interest in the pragmatic content. An additional aspect of learner perception of the SIE activities in S2 was the ―Wow factor‖ associated with the innovative use of a virtual immersive space for learning. While work in this area is growing, Croquelandia is the first 3-dimensional immersive space for learning pragmatics. The participant in Extract (3) provides a noteworthy and common, reflection noted by participants of S2. (3) I just think it is so cool that I‘ve probably talked to tons of friends, I‘m playing a video game for my Spanish class right now, and like no one understands it‘s not like a real video game but I just think it‘s like such a neat idea. I can see it working for you know,

Learner Perception and Strategies for Pragmatic Acquisition

127

becoming a part of a curriculum of a class, especially a class, like I took a colloquial class, and integrating something like this into that could be very useful and it is a way to get people excited rather than reading like this is what you say for this and this is what you say for this...last Thursday, I think, I came home and my friends were going out and I said I‘m going home and play the video game for awhile, I want to see what happens, like if I can find Ana, so I think it‘s really the idea of it I have never even considered that there was this was such an option ... – Lisa, Interview 1, S2

When asked what their favorite aspect of the project was, 17 of the 25 participants in S2 pointed to a feature unique to the SIE itself, in addition to the content being learned. Features of the SIE that were noted by learners included the immersive aspects, the quest and quest completion, interaction with simulated characters within the immersive space, and the ability to ―get involved‖ in what was happening. The second common pattern found in both sets of interviews was an indication that more content and activities with a variety of choices was better than limited, structured content. Learners expressed a desire for a variety of activities, contexts, and practice spaces. Moreover, for the best experience for all learners, this material should be delivered in a variety of means. Among the participants in both studies, it was evident that different types of materials are needed, and essential, to an effective self-access learning module. Learners have varied reactions to the same activities and find different mediated contexts useful for different types of activities. As can be seen in Extract (4), the participants in S1 had very different responses to the same activity. (4) [The website] kind of puts into written order what you kind of hear on your own, but you don‘t really know how to order it…it helps to have it all written down and put together. –Susana, S1 I‘m kind of a fan of interactive things like if they were drop-down boxes. –Henry, S1 I don‘t need like fancy stuff to help me learn, I guess. –Abril, S1

This varied approach supports the inclusion of a variety of types of feedback and activities. Similar evidence for the inclusion of varied approaches can be found in the interviews from S2 as well. In Extract (5), from S2, we also see very different reactions to the same activities. (5) I mean it‘s just a fun way to do it. I think the whole game thing. Just because we are so absorbed in technology. I think it is a good way to do it and... I don‘t know, I‘m kind of a nerd and I like to learn things like that, so I don‘t know that was just fun...kind of fun for me because I feel like I learned a lot from it. –McKenna, Interview 2, S2 Personally, I would just prefer to do talking with people in class like actually being able to talk not having it as kind of a pseudo-conversation. I don‘t know. I prefer not to do it like online because I really don‘t like using computers that much. –Roxy, Interview 2, S2

128

Julie M. Sykes and Andrew D. Cohen

McKenna really enjoyed the type of learning experience that was possible in the SIE as well as the in-class group presentation; whereas Roxy preferred not to use the online virtual environment. This is worth noting because the same participant expressed strong motivation for learning pragmatics (See Extract 2). This would suggest that although there was a preference for a different type of learning, the use of the virtual space did not inhibit her interest in the content. The addition of varied content and activities in mediated spaces affords learners the opportunity to choose what they would like to complete, as well as the features with which they would like to interact. The final pattern emerging from the interviews to be discussed here is that learners are high-demand consumers of technology and expect online activities to work in accordance with their expectations. At the same time, some leniency was afforded new and innovative content or activity types. In both studies, learners expressed some frustration with the technology. For example, in Study 1, learners were frustrated with the sometimes choppy video clips and the appearance of too many pop-up windows; nevertheless, all indicated an interest in and an enjoyment of the topic being studied. Excerpt 6 is an example of the type of comments made. (6) the audio was really clunky especially in the…apologies section like it was hard to keep paying attention ‗cause you‘re like (ugh) the audio is bothering me... But other than that it was pretty good. –Javier, S1

The distinction between technological expectations and innovative learning spaces and content was especially evident in S2, where learners were the first group of students to use the SIE in the classroom. Of the 25 participants interviewed, all reported some frustration with the space ―crashing‖ or not being able to access the SIE; yet, the majority were quick to clarify that they understood technical difficulties were part of a new process and were willing to be patient because the activity was especially engaging and interesting. Many also made explicit distinctions between the technological component and the pragmatic content. Excerpt (7) is an example of this type of comment. (7) About the whole experience the best thing was having fun with something new and learning something new ... I mean it was a fun, it was a fun experience. Having it, having a new experience on a whole was really good...The two worst things [about the experience], probably, and this isn‘t anything that... I mean, we all understood that it was a new program and stuff, but just technical errors. I mean that was sort of a bummer that it froze up my computer all the time but I can‘t really think of anything else. –John, Interview 2, S2

Interestingly, despite a number of problems logging on and using the SIE at the beginning of the project, this participant was especially enthusiastic about the project. She was willing to overlook the technical issues. This was the case for the majority of participants in S2. When asked if the project should be repeated as part of the course, 18 of 25 reported that it should be repeated at the same level (a 6th semester Spanish course). Of the seven who said it should not be repeated, 4 of those said that it should be used again, but at a lower level.

Learner Perception and Strategies for Pragmatic Acquisition

129

Only three of the 25 were hesitant in using it again and all three did note that if the technology were more consistent, the activity should be repeated.

Discussion The patterns found in the interview data suggest directions for future research, as well as directions for pedagogy. These include a strong motivation and interest in learning pragmatics, a desire for variety and complexity, and a flexible acceptance of technology. In this section we briefly discuss these common themes. First of all, as noted above, the interviews in S1 and S2 revealed strong motivation and interest in learning pragmatic content and developing ILP competency. Future research and curriculum development should harness this motivation and interest in the topic whenever possible. Research has shown that pragmatics is teachable (e.g., Bardovi-Harlig, 2001; Judd, 1999; Kasper and Rose, 2002) and that CALL environments are a feasible way to make instructional materials accessible to learners (Furstenberg and Levet, 2001; Belz, 2002, 2003; Biesenbach-Lucas, 2005; Sykes, 2005; Sykes, Oskoz, and Thorne, 2008). In this case, both the interactive website and synthetic immersive environment show promising possibilities. In addition, the motivation for pragmatic development appears to be enhanced through participation in the SIE. These results are congruent with educational gaming research indicating a higher level of engagement with different types of immersive spaces (e.g., DeFreitas, 2005; Dror, 2007; Gee, 2003; Prensky, 2001). Learners‘ positive perception of the SIE used in S2 serves as a strong indicator for continued research and development in the use of massively multiplayer online games (MMOGs), virtual social spaces (e.g., Second Life, Active Worlds), and SIEs (e.g., Croquelandia) for the development of pragmatic skills.3 Furthermore, these results suggest that, in most cases, more is better. Learners want a variety of activities, contexts, and practice spaces. For the best experience for all learners, online pragmatics materials should be delivered in a variety of ways. This could be in the form of multiple approaches (Judd, 1999) and strategies (Cohen, 2005) as well as varied types of activities within the mediated environments themselves. Based on the preliminary results of research question 1, one might even suggest that various types of mediated spaces be utilized simultaneously. Finally, in each of these studies we see that learners are high-demand consumers of technology; however, some leniency is given for new and innovative initiatives. This pattern indicates that researchers and practitioners should not be afraid to take educated risks with innovative educational technologies. We are not advocating uninformed implementation and use, but rather emphasizing the need to try new approaches without fear of learner repercussion. If the materials are research-based, tested, and well-explained to the learners, it is likely that the learners will appreciate the effort and availability of the pragmatics material, and will be more patient if technological problems arise.

3

See Sykes, Oskoz, and Thorne (2008) for a continued discussion of the use of emerging digital technologies in L2 development, including pragmatics.

130

Julie M. Sykes and Andrew D. Cohen

Limitations As is often the case with studies that are exploratory in nature, this study presents a number of limitations. In this section we present the limitations unique to each study as well as considerations relevant to the current analysis. First, in S1, the small sample size makes statistical comparisons impractical and inhibits the feasibility of generalizing the results on a larger scale. While the ten participants were most likely representative of learners in advanced courses, a larger sample size would be needed to confirm the findings of this study. Participants in this study were also paid for their participation; consequently, there may have been an above-normal desire to do well and participate. At the same time, since participants were participating during their personal time, there may have been a tendency to rush through the measures or mark responses at random in order to finish more quickly. Likewise, the subjects in S2 represented volunteers who agreed to participate in the survey and interview activities outside the framework of their normal coursework. Thus, they may have represented a highly-motivated subset of the students studying Spanish at the institution. While the sample size was larger, it was unbalanced in terms of gender and still may not have been entirely indicative of the perceptions of students on a more general scale. In this case, the survey was administered online. While it is not expected this caused any difficulties for the participants, there is no way of confirming any discrepancies in responses. It is also important to address some of the limitations of the analysis itself, which were true of both studies. First of all, it is critical to note that in strategies research, self-report data is not always accurate and not necessarily indicative of how the strategies are actually used (Cohen, 2005). In other words, intuition and action do not always correspond and selfreporting is not always the most accurate means of data collection. In this case, learners‘ perceptions of strategies in the learning and performance of pragmatics were targeted. However, future research should also utilize other measures for confirming actual strategy use to complement the self-reported perception data used in this analysis, which would suggest an increased awareness of learning, use, and metapragmatic strategies for pragmatics justifying the value of future research. An additional consideration was the use of two distinct sets of students to compare the perceived strategy use of each of the mediated contexts. Future studies should consider perceived strategy development in each of the mediated contexts by the same learners. Finally, this chapter does not offer an in-depth analysis of all data from both studies nor a complete analysis of the attributes inherent to each of the online environments. While this is beyond the scope of the chapter, additional considerations of the nature of online spaces should also be considered prior to generalizing these results to all CALL materials (See, for example, Thorne and Payne, 2005, Sykes, Oskoz, and Thorne, 2008).

Suggestions for Future Research While the results presented here can only be suggestive of what might happen, many of the issues raised in this study signal important catalysts for large-scale empirical research in the area of ILP development. Furthermore, they would suggest that a strategies-based approach to learning pragmatics be explored further in future work. Research in this area would likely find a wealth of information in further examining the use of strategy instruction

Learner Perception and Strategies for Pragmatic Acquisition

131

for learning pragmatics as well as content delivery in an online, self-access format as well as in SIEs. Some questions to consider include, but are not limited to, the following: 







How do learner perceptions and actual strategy use compare when considering pragmatic development? To what extent do learners use different strategies in mediated and non-mediated contexts? To what extent is there an impact of strategy development on actual pragmatic abilities? If so, in what ways is ILP development impacted? Where should emphasis on strategy development be placed? At what levels? What is the impact of various digital environments on the instruction and use of learning, use, and metapragmatic strategies? If deemed relevant, how can the power of various mediated contexts be used to enhance and encourage strategy development? How do various mediated contexts (i.e., websites, SIEs, MMOGs, wikis) impact strategy use and perception? What is the impact of these spaces on ILP abilities?

Future projects should examine perception, effectiveness, and use of a strategy-based approach to ILP development and explore further use of CALL technologies for pragmatic development.

Conclusion Despite the aforementioned limitations, results from these two studies carry preliminary implications for further exploration of the development and use of CALL materials to aid in learning pragmatics. The results of this chapter are not meant to suggest the only way of learning these pragmatic skills or that all CALL materials are necessarily superior just because of their existence online. Rather, we suggest they are a viable option and a practical means for introducing pragmatics instruction in L2 classrooms, especially in cases where face-to-face time may be limited. In a general discussion of Digital Game-Based Learning, Prensky (2001) states, ―Learning is a big job. No one method works alone or by itself for everything. Digital Game-Based Learning is great in that it motivates and teaches in ways that other methods seldom do. But it is neither the unique solution to all training [learning] problems nor a pancea‖ (p. 7). In the same vein, CALL materials are clearly not the only answer; nevertheless, mediated contexts are a possibility for the future of pragmatics instruction, especially from the point of view of the learners. They show an initial impact on learners‘ perceived strategy use. In terms of pedagogy and content development, results from both studies show that there was motivation to improve ILP abilities on the part of the learners who participated in this study. While it remains to be seen how these findings are applicable to all learners, there are preliminary indications of interest and strategy development. The inclusion of pragmatics materials in curricula and learning materials is highly beneficial, and alternative means for learning should be considered further. In addition, in cases where there is not sufficient time for a complete exploration of all information in the classroom setting, self-access materials, such as the Dancing with Words site or online collaborative spaces such as Croquelandia will also likely prove to be beneficial. It is our recommendation, based on this preliminary work,

132

Julie M. Sykes and Andrew D. Cohen

that learners be encouraged to explore materials dealing with L2 pragmatics on their own and be given resources (CALL-related or others) to do so. Empirical research efforts, informed risk-taking, and practical implementation are critical to making this happen.

References Abrams, Z. (2006). From theory to practice: Intracultural CMC in the L2 classroom. In Ducate, L. and Arnold, N. (Eds.), Calling to CALL: From theory and research to new directions in foreign language teaching (pp. 181-209). San Marcos, TX: Computer Assisted Language Instruction Consortium. Bardovi-Harlig, K. (2001). Evaluating the empirical evidence: Grounds for instruction in pragmatics? In Rose, K. and Kasper, G. (Eds.), Pragmatics in Language Teaching (pp. 13-32). Cambridge: Cambridge University Press. Bardovi-Harlig, K. and Dörnyei, Z. (1998). Do language learners recognize pragmatic violations? Pragmatic vs. grammatical awareness in instructed L2 learning. TESOL Quarterly, 32, 233-59. Beebe, L. and Waring, H. Z. (2001). Sociopragmatic vs. pragmalinguisitc failure: How useful is the distinction? NYSTESOL Applied Linguistics Winter Conference SUNY Graduate Center. Belz, J. A. (2002). Social dimensions of telecollaborative foreign language study. Language Learning and Technology, 6, 60-81. Belz, J. (2003). Linguistic perspectives on the development of intercultural competence in telecollaboration. Language Learning and Technology, 7, 68-99. Biesenbach-Lucas, S. (2005). Communication topics and strategies in e-mail consultation: Comparison between American and international university students. Language Learning and Technology, 9, 24-46. CLEAR (2007). Multimedia interactive modules for education and assessment (MIMEA). East Lansing, MI: Center for Language Education and Research, Michigan State University . Cohen, A. D. (2004). Assessing speech acts in a second language. In Boxer, B. and Cohen, A. D. (Eds.), Studying speaking to inform second language learning (pp. 302-27). Clevedon, England: Multilingual Matters. Cohen, A. D. (2005). Strategies for learning and performing L2 speech acts. Intercultural Pragmatics, 2, 275-301. Cohen, A. D. and N. Ishihara (2005). A web-based approach to strategic learning of speech acts. Minneapolis, MN: Center for Advanced Research on Language Acquisition (CARLA), University of Minnesota, 57 pp. . Cohen, A. D. and Sykes, J. M. (2006). The development and evaluation of a self-access website for learning Spanish speech acts. Paper presented at the Paper presented at the Annual Joint AAAL-ACLA/CAAL Conference, Montreal, CN, June 17, 2006. Cohen, A. D. and Sykes, J. M. (2007). Strategies, CMC, and Learning Pragmatics. 17th International Conference on Pragmatics and Language Learning Honolulu, HI. Crystal, D. (Ed.) (1997). The Cambridge Encyclopedia of Language. New York: Cambridge University Press.

Learner Perception and Strategies for Pragmatic Acquisition

133

deFreitas, S. (2006). Learning in immersive worlds: A review of game-based learning. JISC E-Learning Programme. http://www.jisc.ac.uk/media/documents/programmes/ elearning_innovation/gaming%20report_v3.3.pdf Dror, I. (Ed.). (2007). Cognitive Technologies and the Pragmatics of Cognition. Philadelphia, PA: John Benjamins Publishing. Félix-Bradesfer, C. (2002). Refusals in Spanish and English: A cross-cultural study of politeness strategies among speakers of Mexican Spanish, American English, and American Learners of Spanish as a Foreign Language. Unpublished Doctoral Dissertation, University of Minnesota, Minnesota. Félix-Bradesfer, C. (2003). Validity in data collection methods in pragmatics research. In Kempchinsky, P. and Piñeros, C.-E. (Eds.), Theory, practice, and acquisition (pp. 23957). Somerville, MA: Cascadilla Press. Furstenberg, G., Levelt, S., English, K. and Maillet, K. (2001). Giving a virtual voice to the silent language of culture: The CULTURA Project. Language Learning and Technology, 5, 55-102. Garces-Conejos Blitvich, P. (2006). Interlanguage pragmatics: A response to Andrew Cohen‘s ―Strategies for learning and performing L2 speech acts‖ published in Vol.2, No.3, of Intercultural Pragmatics, Intercultural Pragmatics, 3, 213-223. Gass, S. and Selinker, L. (2001). Second Language Acquisition: An Introductory Course. Mahwah, NJ: Lawrence Earlbaum Associates. Gee, J. P. (2003). What video games have to teach us about learning and literacy. New York, NY: Palgrave Macmillan. Gee, J. P. (2005). Why video games are good for your soul. Sydney, Australia: Common Ground. Godwin-Jones, R. (2005). Messaging, Gaming, Peer-to-peer Sharing: Language Learning Strategies and Tools for the Millennial Generation. Language Learning and Technology, 9, 17-22. Golato, A. (2003). Studying compliment responses: A comparison of DCTs and recordings of naturally occurring talk. Applied Linguistics, 24, 90-121. Johnson, S. (2005). Everything Bad is Good For You. New York: Riverhead Books. Judd, E. L. (1999). Some issues in the teaching of pragmatic competence. In Hinkel, E. (Ed.), Culture in second language teaching and learning (pp. 152-66). Cambridge: Cambridge University Press. Kasper, G. (1997). Can pragmatic competence be taught? Second Language Teaching Curriculum Center, www.lll.hawaii.edu/nflrc/NetWorks/NW6. Kasper, G. and Rose, K. (2002). Pragmatic development in a second language. Oxford, UK/Malden/MA: Blackwell. Kasper, G. and Schmidt, R. (1996). Developmental issues in interlanguage pragmatics. Studies in Second Language Acquisition, 18, 149-69. Kramsch, C. and McConnell-Ginet, S. (1992). Text and Context: Cross-Disciplinary Perspectives on Language Study. Canada: Heath and Company. Linder, U. and Rochon, R. (2003). Using chat to support collaborative learning: Quality assurance strategies to promote success. EMI, 40, 75-89. LoCastro, V. (2003). An introduction to pragmatics: Social action for language teachers. Ann Arbor, MI: The University of Michigan Press.

134

Julie M. Sykes and Andrew D. Cohen

Márquez Reiter, R. and Placencia, M. E. (2005). Spanish Pragmatics. New York, NY: Palgrave Macmillan. Mistral, P. (2007). Second Life ballet fills the SIM-Linden suggests selling tickets. SecondLife Herald www.secondlifeherald.com. Payne, J. S. and Ross, B. (2005). Working memory, synchronous CMC, and L2 oral proficiency development. Language Learning and Technology, 9, 35-54. Payne, S. and Whitney, P. J. (2002). Developing L2 oral proficiency through synchronous CMC: Output, working memory, and interlanguage development. CALICO, 20, 7-32. Prensky, M. (2001). Digital Game-Based Learning. St, Paul, MN: Paragon House. Roever, C. (2004). Difficulty and practicality in tests of interlanguage pragmatics. In Boxer, B. and Cohen, A. D. (Eds.), Studying speaking to inform second language learning (pp. 283-301). Clevedon, England: Multilingual Matters. Rose, K. (2005). On the effects of instruction in second language pragmatics. System, 33, 385-99. Salaberry, R. and Cohen, A. D. (2006). Testing spanish. In Salaberry, R. and Laffard, B. (Eds.), The art of teaching spanish: Second language acquisition from research to praxis Washington, D.C.: Georgetown University Press. Sotillo, S. (2000). Discourse functions and syntactic complexity in synchronous and asynchronous communication. Language Learning and Technology, 4, 82-119. Sotillo, S. (2005). Corrective feedback via instant messenger learning activities in NS-NNS and NNS-NNS dyads. CALICO, 22, 467-96. Sykes, J. (2005). Synchronous CMC and pragmatic development: Effects of oral and written chat. CALICO, 22, 399-432. Sykes, J. (2008). A dynamic approach to social interaction: Synthetic immersive environments and Spanish pragmatics. Unpublished Doctoral Dissertation, Department of Spanish and Portuguese, University of Minnesota, Minneapolis, MN. Sykes, J. and Cohen, A.D. (2008). Observed learner behavior, reported use, and evaluation of a website for learning Spanish pragmatics. In M. Bowles, R. Foote, & S. Perpiñán (Eds.), Second language acquisition and research: Focus on form and function. Selected proceedings of the 2007 Second Language Research Forum. Summerville, MA: Cascadilla Press. Sykes, J., Oskoz, A, and Thorne, S. (2008). Web 2.0, Synthetic Immersive Environments, and the Future of Language Education. CALICO Journal, 25 (3), 528-46. Taylor, T.L. (2006). Play Between Worlds. Cambridge, Massachusetts: The MIT Press. Thorne, S. (2005). Pedagogical and praxiological lessons from internet-mediated intercultural foreign language education research. In Belz, J. and Thorne, S. (Eds.), Internet-mediated Intercultural Foreign Language Education Boston, MA: Thomson and Heinle. Vandergriff, I. (2006). Negotiating common ground in computer-mediated versus face-to-face discussions. Language Learning and Technology, 10, 110-38. Wilcox, L., Allison, R., Elfassy, S. and Grelik, C. (2006). Personal space in virtual reality. ACM Transactions on Applied Perceptions, 3, 412-28. Yule, G. (1996). Speech acts and events. Pragmatics (pp. 47-58). Oxford: Oxford University Press.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 4

"BEST SUCCESS THROUGH LANGUAGE LOSS?" AN INCIPIENT AUSTRIAN SOCIOLINGUISTIC STUDY ON OPEN QUESTIONS IN EDUCATION AND MIGRATION RESEARCH1 Katharina Brizic Austrian Academy of Sciences, DINAMLEX Wohllebengasse 12-14, 1040 Vienna, Austria, University of Vienna, Austria Institute of Linguistics

Abstract Which languages are actually transmitted in immigrant families? This is a crucial point in the present fierce European debate on family-language use and immigrant pupils‘ linguistic/educational success. Quantitative sociology often arrives at the conclusion that maintaining immigrant-family languages is counterproductive to children‘s educational success. Qualitative linguistics, in contrast, mainly reveals a positive maintenance effect on children‘s school (language) proficiency. The incipient project deals with the question of which of these approaches is justified and how such inconsistencies could emerge. It furthermore investigates why some groups, such as the Turkish immigrants, regularly score so low in regard to educational as well as linguistic success, whether this has anything to do with their language use, as often assumed, and what new answers to these issues would mean for the present fierce educational policy debates in many European and other immigration societies. From a preceding explorative study and a new explanatory model the following hypotheses have emerged: a. The interdisciplinary inconsistencies could be caused by the continuing wide gap between qualitative and quantitative data-collection methods, which will thus in the present project be overcome as far as possible by interdisciplinary methods as well as theory. b. The methods in turn have to be highly sensitive when actual familial languageuse patterns (and thus also stigmatised languages or even language loss) are at stake. c. 1

The project to be presented here was accepted after peer review in October 2007 and started on March 1st 2008 (conceived for three years). It is funded by FWF (Austria's main body for research funding) and located at the Austrian Academy of Sciences, Vienna.  Email: [email protected], Tel.: 0043 - 664 - 32 68 008

136

Katharina Brizic However, when language proficiency, use, transmission and loss appear as collective phenomena, they can only result from high societal pressure on a whole group, and thus from social inequality – here: language-political inequality. The sending as well as the receiving societies‘ education and language policies have paramount influence on how people construct (or deconstruct), show (or hide) and transmit (or give up) their languages and identities. d. If language transmission/ shift can be uncovered in its many facets, it should be possible to examine the overall hypothesis stating: family-language loss does not facilitate school language acquisition, just because language acquisition is a holistic process and highly depends on the quality of parental input. If parents, in contrast, transmit the language(s) they feel "at home" with, children score higher in both family and school languages. The incorporation of several disciplines concerned (sociology, socio-/psycholinguistics, research on language shift etc.) is intended to make the present approach innovative. Thus, for methodology and theory, but also for educational policy practice in sending and receiving societies ample output can be expected.

1. Introduction Immigrants who give up their languages achieve educational success. Given practised multilingualism in modern societies (Krumm 2002: 34) this finding is contentious, but nevertheless it is gaining ground. Large-scale European educational surveys such as PISA (Entorf and Minoiu 2004; Bacher 2005) and international meta-studies like Esser's (2006) clearly document the particularly weak position of immigrant pupils whose families stick to their languages. Language loss, in contrast, opens up the possibility of shifting to the apparently much more promising familial communication media of the country of immigration's school language. It is not until closer inspection that these results arouse scepticism: the more accurately familial language use is investigated (depending on family members, proficiencies and sites of language use), the more it turns out to be the reverse. Familial shift to a societally dominant language is especially found together with a particularly weak proficiency in the dominant, "new" language as well as in the "old" family language (Afshar 1998: 257-261; Wodak and Rindler-Schjerve 1985: 15). But a close inspection is hard to implement, especially in large-scale surveys, and has hence rarely been done. This is why migration research has up to now had to be content with two highly controversial results as well as with the unconnected coexistence of their sources of origin: the rather quantitatively oriented sociological versus the rather qualitatively oriented linguistic paradigms.2 The question is still open as to how such contradictory results could emerge at all and if they could be affiliated with each other. Incidentally, this is not the only open question in the area of immigrant languages and educational success. There is a second one intrinsically tied to the first: Why do specific groups of origin regularly score so low in regard to both linguistic and educational success, with the Turkish group being especially relevant for Europe (Olechowski et al. 2002; Weiss et al. 2006)? In spite of wide interdisciplinary consensus in this respect, the demand for the reasons is still unfulfilled (cf. e.g. Esser 2006: 547). 2

There exist some connecting acitivities now in regard to language and educational success (e.g. Esser 2006); however, they are only seemingly interdisciplinary given that they ignore a good portion of linguistic research on language acquisition (considered as "methodically inappropriate", cf. Esser 2006: 94ff.).

"Best Success Through Language Loss?"

137

Meanwhile, the unsolved questions are fiercely debated not only in the European scientific world but have reached the public domain as well. The first and apparently negative coherence, namely between immigrant-language use and educational success, has again brought into the crossfire immigrant multilingualism as well as the teaching of home languages; and although home-language teaching has at all times been a controversial matter, it is at present growing even more acute due to views such as Esser's (2006: 514) "Tower-ofBabel dilemma" approach. The second coherence, namely between origin and school success, has even brought "origin-specific culture" into the discussion as being the cause of origin-specific failure (e.g. Korkisch 2004). Consequently and inavoidably, particularly "visible" immigrant groups are hit by this development (Kohlbacher and Reeger 2003: 356). The process is in turn potentiated by or potentiating an atmosphere within parts of the Central European public which has become "anti-Turkish" in other fields also (cf. ESI 2007). Interdisciplinary contradictions and origin-specific failure – both unexplained phenomena but of high topicality and societal explosiveness: how can these challenges be best addressed by migration research in terms of a constructive way out?

2. State of The Art – With Special Consideration of the Austrian Situation 2.1. Insufficient Data Basis It is still basically difficult to systematically investigate the characteristics of specific Austrian immigrant groups' educational success at all. This is due to the inadequateness of currently available official data3: in official records, immigrants can only be identified if they hold or have held foreign citizenship (Herzog-Punzenberger 2003b: 1126). To gain insight into the second and third generations' size and educational situation, one has to rely on estimations; the group with immigrant background from Turkey should thus comprise altogether about 200.000 or more persons (Herzog-Punzenberger 2003b: 1127). This also applies for linguistic affiliations, which are largely unrecorded in official sources. To give an example: while Austrian ethnologists assume a Kurdish background for about 30% of Turkish families (Six-Hohenbalken 2001; Busch 2006: 17), the "Viennese Schulmatrik"4 indicates a Kurdish affiliation for barely 2% (Knapp 2006). This clearly reveals the effort which would be required to carry out any large-scale investigation of coherences between educational success, origin and multilingual every-day reality (cf. Busch 2006) in Austria.

3 4

As to the quality and comparability of various European data bases cf. http://www.compstat.org. Based on the information given by the parents at school enrolment (but under rather unfavourable conditions such as time pressure, language barriers etc.).

138

Katharina Brizic

2.2. Origin-Specific Failure However, for Central Europe in general it has clearly been stated that educational success is closely tied to origin and social-group membership (Kronig et al. 42). This also becomes manifest in the phenomenon of origin-specific failure which affects mainly three immigrant groups: Moroccan and Turkish pupils (cf. Fase 1994; Müller 1997; Rüesch 1998; Stanat 2003; Tilmatine 1997; Maas and Mehlem 2002; Andriessen and Phalet 2002; Crul and Doomernik 2003) as well as pupils with a Bengali background (Fase 1994; Haque 2000). Of these groups, in Austria only the Turkish group is represented to a larger extent. Similarly to international findings (cf. Söhn 2005: 11) in Austria the Turkish group's failure can most easily be observed in the form of its generally problematic school and labour-market careers (Herzog-Punzenberger 2003a; 2003b; Weiss et al. 2006: 11). This leads to another finding concerning a field plainly central for any school career: language. In practically every Austrian study on language acquisition pupils of Turkish origin score lower than e.g. those from former Yugoslavia (who are investigated comparatively often), in German as well as in the tested first language (Olechowski et al. 2002; Peltzer-Karpf et al. 2006). As has already been stated, be it in Austria or internationally, the search for the reasons has so far remained ineffective (cf. Esser 2006: 547; Stanat 2003; Söhn and Özcan 2006).

2.3. Interdisciplinary Inconsistencies The delicate and challenging search for the reasons for the origin-specific failure has in the meantime permeated various disciplines and branches of migration research. The same is true for the fierce debate on immigrant languages and multilingualism in general and on their significance for educational success (cf. section 1), maybe unintentionally paralleling the research on conditions and consequences of individual and collective language shift or language "death". The highly fragmented research arena may be the reason for a corresponding fragmentation of findings, as these findings are tied to very different grades of a qualitative-quantitative continuum comprising a wide range of conflicting methods and approaches. To start with research on language shift and ―death‖: this discipline has provided an unmatched clear insight into language loss and shift to a new, societally dominant "first language" and into its individual and collective consequences (Dorian 1982; Fishman 1999; Kouritzin 1999; Wodak and Rindler-Schjerve 1985). The search for the reasons is located mainly on the socio-political macro-level given that only high societal pressure can cause such a substantial affair like the loss of a community or family language (Sasse 1992: 10 f.; Fishman 1999: 451). It is this assumption that resulted in one of the discipline's most elaborated theoretical models: Sasse's Gaelic-Arvanitika Model (Sasse 1992: 19; cf. also Dressler and de Cillia 2006). A certain deficit might only be the fact that research on language shift tends to pay attention almost exclusively to the old, "dying" languages (with only a few exceptions, cf. e.g. Wodak and Rindler-Schjerve 1985). Proficiencies in the newly acquired dominant (school) languages are barely examined, although they play such a decisive role in modern and highly education-related societies. Psycho- and Sociolinguistics in turn provide insight into both family (i.e. first) and school (i.e.societally dominant) language proficiencies with numerous findings in regard to

"Best Success Through Language Loss?"

139

immigration (cf. for Austria Peltzer-Karpf et al. 2006) and to the significance of parental input for any language acquisition (cf. Jeuk 2000; Boeschoten 1990: 139). However, a stronger connection with ethnological research, more precisely: with any research concerning the coutries of origin, is desirable, and this is true not only for linguistics but also for sociological and educational research: for all too often, samples with Turkish or formerYugoslavian background in European receiving societies are perceived as purely Turkish or Bosnian/Croatian/Serbian-speaking, without considering the existence of numerous smaller or stigmatized first languages in the societies of origin at all (cf. e.g. Olechowski et al. 2002: 32). It is not surprising that data thus obtained often show puzzling "first-language-test" results in Turkish or Bosnian/Croatian/Serbian (cf. Olechowski et al.: 48). Moreover, if stigmatized or smaller languages are already hard to uncover in the frame of individual linguistic tests, this can be expected to be even more so the case in large-scale educational, sociological or economic surveys (cf. Weiss et al. 2006; Esser 2006). But it is exactly this "remaining hidden" of languages and of language-use patterns which consequently impedes any verifying – or disproving – of their benefit. Hence, in terms of innovative approaches it is worth taking a look at those qualitative studies that were intent on detecting the large number of actually spoken majority and minority languages in various immigrant groups (Uçar 1996; Thomé 1987; Preibusch 1992; Mehlem 1998; Maas and Mehlem 2002; Jungblut 1990; de Ruiter 1997; Extra and Yağmur 2004). These studies are innovative also because of their macro perspective, which has included the countries of origins' macro-level (e.g. language planning, language and minority policies) in the investigation of family migration histories and backgrounds. The awareness, in turn, that the effects of the macro-level's factors are long-lasting (Nauck and Schönpflug 1997) should be one of sociological migration research's perhaps most important contributions to any qualitative research on language acquisition, since this language acquisition research (with the above-mentioned exceptions) suffers from an almost entire lack of consideration of the macro-level's effects on social groups as well as from a general lack of relations to sociology (Cameron 1990: 84). So far, language acquisition research has thus been able to explain individual failures very well (cf. Gardner and Tremblay 1998; Dörnyei 2002), but not collective (and with it origin-specific) ones. However, in sociological models (cf. a recent model in Portes and Zhou 2001) the macrolevel of the receiving countries traditionally is well-established. But this again does not (or only nominally) apply for the macro-level of the countries of origin (cf. Esser's 2006: 93 ff. model) nor for educational backgrounds in these countries: both macro-level and educational backgrounds in the sending societies are barely considered (with few exceptions; cf. e.g. Feliciano 2006). It thus too often remains unknown if immigrant groups have been subject to a "downwards mobility" (i.e. to a loss of social status caused by migration) or if they experience a persistent low social standing – and thus a double deprivation – in both the countries of origin and immigration. However, if in fact the receiving societies' educational and language politics are macro-factors with long-lasting effects (cf. OECD 2005; Portes and Zhou 2001), something similar has definitely to be assumed for the sending countries' macro factors (cf. again Maas and Mehlem 2003). Bringing together the various approaches outlined above as well as their highly contradictory results, we can identify the following tendencies concerning immigrant languages and educational success:

140

Katharina Brizic

Explorations and results (of whatever discipline) are less critical in regard to immigrantlanguage loss the more tightly and exclusively they have focused on school-language proficiency and on individual, intra-group or economic factors. Accordingly this is true when they have excluded qualitative methods and findings (cf. the concept of Esser's 2006 metastudy). Explorations and their results, in contrast, are more critical in regard to immigrantlanguage loss (i.e. potential-oriented in regard to language maintenance) the more tightly they have focused on the probands' proficiencies in all their languages, including qualitative investigations of the languages actually spoken in the families; this also holds the more they have incorporated the consideration of the macro-level in the receiving and sending countries5 (in detail see Brizić 2007, ch. 4 and 5.1). The brief outline above – a condensed version of a long-term literature research's results – shows how fruitful a closer collaboration of all involved disciplines could be, given the deep-running fragmentation of the research landscape on immigration, language (loss) and educational success. The various interconnections and demands indicate that it is in fact time for such a collaboration to open up new theoretical and methodological possibilities, above all in this specific field of handling immigration, where research is constantly challenged anew by open educational policy questions, and where answers for every-day school practice are needed more urgently than ever. The project presented here emerged from this basis, including intensive empirical preparatory work, which I will give an overview of in the following section.

3. Preparatory Work for the Incipient Study: An Explorative Investigation Similarly to the international research landscape on immigration, language and school success, Austrian psycholinguistics is also characterized by open questions and some inexplicable results: In the single Austrian long-term study on language acquisition by immigrant children so far (Peltzer-Karpf et al. 2006), the investigated pupils of Turkish origin (n=23) scored considerably lower than the children with a former-Yugoslavian background (n=37), in German as well as in the tested first language, comparable to international findings. The reasons for the origin-specific failure were unknown. An accompanying explorative investigation was thus conducted (Brizić 2006; 2007) which from the outset aimed at meeting the claims outlined above: namely, detecting the actually spoken familiy languages, connecting the individual children's micro-level6 to the socio-political macro-level, and including the country of origin's background factors in the search for the failure's reasons. With time-consuming one-on-one interviews in Turkish or Bosnian/Croatian/Serbian with at least one parent of each child, it was initially possible to fulfill the first claim:

5

6

Moreover, the explorations first mentioned tend to focus on the immigrant languages‘ benefit for school and labour market (a benefit often shown to be missing, cf. Esser 2006), while the latter aim at investigating the cognitive benefit of multilingualism (independent of its societal valuation; cf. linguistic research on multilingualism: Herdina and Jessner 1997). i.e. the children‘s proficiency in the school language (German) and in the officially given family language (Turkish or Bosnian/Croatian/Serbian respectively), tested at the end of the third year of primary school.

"Best Success Through Language Loss?"

141

detecting the actually spoken languages. What had (following the "Viennese School Matrik"7) been considered a linguistically uniform group of (Bosnian/Croatian/Serbian or Turkish) immigrant families emerged as a composition of numerous language majorities and minorities (i.e. the Romanian, Albanian, Bulgarian, Vlah and Roma minorities and the Slovenian, Makedonian and Bosnian/Croatian/Serbian majorities in the Yugoslavian sample; the Kurdish and Greek minorities and the Turkish majority in the Turkish sample), with multifaceted patterns of language use, culminating in the virtual impossibility of naming a "real" first (family) language at all. However, while linguistic diversity applied to both groups of origin, parental answering behaviour shaped up to be "origin-specific": for the parents from the former Yugoslavia bringing up a possible minority affiliation was quite natural; in the Turkish group the same topic obviously touched a taboo. Furthermore, in the Turkish group the proportion of families having undergone or undergoing language shift was considerably higher (e.g. from Kurdish to Turkish, from Greek to Turkish, or from Turkish to German, cf. Brizić 2007, ch. 6.7). Given these origin-specific phenomena, the idea of further investigations on the countries of origins' macro-level – i.e. the educational and language policies – suggested itself. And in fact, a comprehensive review of literature mainly from Turkish studies and political science revealed a wide, primarily linguistic gap between the Turkish-speaking rural population and the educated stratum (cf. Robins 2000: 69; Mehmet 2002) as well as between Turkey's linguistic minorities and its majority (cf. Andrews 1989; 2002). The first-named gap is the result of a far-reaching, fast and ambitious state-language reform intended above all to benefit the rural population; in practice, the reform remained an affair almost exclusively of the educated stratum (Boeschoten 1997: 363), making the acquisition of school-language "capital" particularly difficult for the non-educated (Steuerwald 1963: 9; Boeschoten 1991; 1997: 376). The second gap results from specific conditions, e.g. profound limitations for language use, which Turkey's roughly 40 minority languages have been exposed to (cf. Cummins 2000: 11; Yağmur 2001); these conditions are, among other things, still causing extensive language shift to the societally dominant Turkish, often without Turkish being acquired to a sufficient degree, due to limited access to schooling for parts of the population (cf. Andrews 1989: 629; 2002; Kula 1998: 174). This situation considerably differs from that in the former Yugoslavia where some minorities were similarly stigmatized (cf. e.g. Janjetović 2001; Dragić 2002) but most areas were provided with entire minority-language schooling systems nevertheless (Bachmaier 1982: 40-42). The macro conditions were unexpectedly clearly mirrored both on the parents‘ and on the children's levels: particularly those families whose languages had suffered from a very low prestige in the society of origin (e.g. in Turkey: Kurdish; the rural Turkish dialects; e.g. in former Yugoslavia: the language varieties of the Roma minority) tended towards language shift. Those children, in turn, whose families had undergone or were undergoing partial or total language shift (making up a particularly high percentage in the Turkish sample) showed not only low linguistic self-confidence, low family-language motivation and low familylanguage proficiency but at the same time also low school-language (i.e. German) proficiency, going hand in hand with a particularly high school-language motivation (cf. already Gardner 1985 and Müller 1997 concerning motivation and self-confidence). The best German results, in contrast, were obtained by those children whose parents had maintained 7

Based on the information given by the parents at school enrolment (cf. above).

142

Katharina Brizic

their own first language(s) (this result being consistent with other findings documenting that under such circumstances parental input is particularly complete and authentic; cf. Wodak and Rindler-Schjerve 1985; Afshar 1998; Lasimbang et al. 1992; Kouritzin 1999).8 Thus, it was in fact a wide connecion that resulted from the explorative investigation: immigrant children's proficiency stems, among other things, from parental experiences during their own language acquisition in the country of origin. – The small study however suffered not only from a lack of representativeness but also from a lack of interdisciplinarity: actually it was conducted as a one-person operation, although with extensive extra-linguistic support. However, the study made it possible to develop a new explanatory model which will be presented in the following section, as it already comprises a substantial part of the new project to be presented here: the model was the project idea's starting point and will be its theoretical framework.

4. Incipient Study - Theoretical Framework In the course of the explorative investigation and as a consequence of it a new explanatory model was developed, its centrepoint being parental linguistic capital (i.e. proficiency9) and parental behaviour in terms of (non)transmission of language(s) and linguistic identity (i.e. self-conception, self-perception and construction of linguistic affiliations, cf. Wodak et al. 1998). This language and identity capital together with intergenerational transmission (cf. figure 1, meso-level) is central just because of its intermediatory function: on the one hand, it is highly subjected to socio-political conditions (cf. Figure 1, macro-level), while on the other hand it represents an essential fundament for children's language acquisition (cf. Figure 1, micro-level). It is thus hypothesized that the parental capital can be regarded as an essential mediator between a society's educational and language policies and a particular child's individual starting point.10 Geared to this hypothesis the model is entitled Linguistic Capital Model (LCM). To be able to react to the explorative investigation‘s results (cf. section 3) and – above all – to the questions open across the disciplines (cf. sections 2.2 and 2.3) the new model is geared towards the following concepts: 1. On the macro-level: towards a model from the field of research on language shift and death (Sasse 1992: 19), dealing with the outstanding relevance of the society (of origin)'s macro-level; 2. On the meso-level: towards the sociological concept of linguistic and cultural capital (Bourdieu 1983) and intergenerational transmission of capital (Nauck et al. 1998: 720); and 3. On the micro-level:

8

However, the explorative study comprised many more factors than those named here (e.g. socio-economic status, parental educational background etc., cf. Brizić 2007). 9 Being acquired during all levels of (primary, secondary etc.) socialisation, thus having to be viewed as a dynamic variable, changing over time (Herdina and Jessner 1997). 10 This starting point is known to be particularly relevant in educational systems which are highly socially impermeable, as is the case in Austria: the starting point remains effective during the whole career (cf. HerzogPunzenberger 2003a).

"Best Success Through Language Loss?" a.

b.

143

towards hypotheses from the field of psycholinguistics, dealing with children's language-acquisition success or failure, assuming that the acquisition success is linked to the interdependency between an individual's proficiencies in all available languages (cf. Cummins' updated hypotheses, Cummins 2000); and towards findings from the field of educational research, dealing with immigrant pupils' educational success or failure, assuming that educational success is linked to an individual's relationship to all aspects of his or her own (potentially multiple) identities (Badawia 2002; Schiesser and Theurl 2001).

Thus, the interdisciplinary inconsistencies (cf. 2.3) regarding immigrant languages and educational success are not at all mitigated by the Linguistic Capital Model nor by the antecedent explorative investigation. Rather, the inconsistencies are intensified since language loss is here regarded as substantial capital loss. This means the model's clear affiliation to those parts of migration research which can be considered to be critical in regard to immigrant-language loss. A solution to the problem of the inconsistencies would thus be the prerequisite to being able to test the model at all. As for the origin-specific failure (cf. section 2.2), in contrast, the model does in fact allow new explanatory hypotheses. However, it is a clear deficit that the explorative investigation‘s focus has exclusively been on the countries of origins' macro-levels (due to the fact that they are largely missing in the literature, cf. section 2.3), whereas the country of immigration's macro-level has been completely omitted here (due to already existing and highly elaborated immigration-context models, e.g. Portes and Zhou 2001; Esser 2006).

Figure 1. Linguistic Capital Model (originally published in IJAL, see Brizić 2006, 344).

144

Katharina Brizic

However, the central idea of the new model is that collective language behaviour and proficiency, up to language loss and educational failure, can by no means be "individual", "familial" or "cultural" phenomena (cf. section 1) but definitely can be traced back to social – here: education and language-political – inequality. And it is exactly this inequality which cannot exclusively be a context-of-origin feature. It is just as much an immigration-context problem and in this form even more relevant for our objective, since this is the context where the immigrant pupils' whole educational career takes place. The macro-micro connection can thus be assumed to run much more directly in the immigration than in the emigration society and not only via the parents (cf. in Figure 1 the unbroken arrow connecting the country of immigration's macro-level and children's starting point). The theoretical framework for the proposed project is thus the Linguistic Capital Model presented here, characterized by the consideration of the emigration and immigration contexts' macro-levels and their far-reaching effects on language behaviour, proficiency and educational success of second/third generation immigrant pupils.

5. Incipient Study - Aims of Analysis 5.1. Survey of Main Aims It is this project's essential aim to uncover coherences that have remained hidden so far or have, if investigated, resulted in highly differing findings. Of all the model's levels, the mesolevel will occupy a central position in the investigation: familial language use in all its facets from language transmission to language loss11 will be the main issue (see aim A, sections 5.2.1. to 5.2.3.), given that it is exactly this level which obviously has all too often eluded closer inspection but at the same time is activating the most fierce debates at present. We will thus have to explore which languages are actually used and how. And we will furthermore shed light on the question if language use might in fact be the cause of origin-specific differences (cf. section 2.2) and if it is in fact able to polarize research on immigration, multilingualism and school success by producing interdisciplinary inconsistencies (cf. section 2.3). When these facets have been examined, the overall hypotheses concerning interdependencies between social conditions and individual success will be focused on: the Linguistic Capital Model (see aim B, section 5.3.1.) will be tested, based on and facilitated by the preceding close inspection of language behaviour, language transmission and language loss. This should finally allow the opening up of basically new methodological, theoretical and practical possibilities. There the focus will lie on innovative potentials that will possibly be essential for further research and education-political practice. It seems to us that this "so what" question refering to the overall sense of the project (see aim B, section 5.3.2.) is vital, given the whole subject's societal topicality, its extremely controversial nature and the proposed project's potential to respond to existing demands. 11

"Language loss" is defined here as the abandonment of one or more "old" languages in order to shift to one or more "new" family languages, be it in the (grand)parents‘ generation or between (grand)parents and children (for existing definitions cf. Dressler and de Cillia 2006). In regard to personal, temporal or any other characteristics, language shift can be sited on any conceivable place on a continuum. One of this project‘s most important tasks will be the mapping of the extraordinarily many facets and degrees of language use, shift and loss in migration.

"Best Success Through Language Loss?"

145

5.2. Aim A. Detailed Look at Familial Language Transmission and Loss 5.2.1. Origin-Specific Differences? "Origin" and Type of Migration Here, our main hypothesis states that the ability to acquire language(s) is, similar to any other acquisition process, inherent to all humans and independent of a specific origin (Chomsky 1957; Wode 1995). It is characterized by individual variation; but this variation in turn cannot be said to vary depending on ethnicity – and even if so, this does not provide any explanation in itself. Rather it calls for breaking down the very blanket term of "origin" to analyse the components which it might in the background be consisted of. In our case where "Turkish origin" is the point at issue, this origin will be differentiated in itself, i.e. investigated and compared in terms of three different manifestations (see Figure 2: aim 5.2.1.): firstly, in the form of international migration – here: families from Turkey living in Vienna/ Austria; secondly, in the form of migration within national borders – here: families from Eastern Anatolia in Western-Turkish metropolitan centres; and thirdly, in the form of non-migration – here: families in Western-Turkish metropolitan centres who have never migrated. All three groups geographically are of Turkish origin, and two of them are also migrants. Regardless of such major commonality, they are hypothesized to be characterized by outstanding linguistic diversity and a high number of language use patterns.12 Thus, on closer inspection the apparently unifying feature is expected to fade in favour of high intra-group variance.

Figure 2. aims 5.2.1. to 5.3.2., sample and levels of investigation.

12

However, many more variables geared to recent migration research will be included in the investigation (cf. Söhn 2005).

146

Katharina Brizic

5.2.2. Origin-Specific Differences? "Origin" and Country of Origin According to the same overall hypothesis as before, "Turkish origin" will again be differentiated, this time not in itself but from an external perspective (see Figure 2: aim 5.2.2.): Viennese families of Turkish origin will be compared to families with formerYugoslavian background also living in Vienna/ Austria. The two samples will be compared particularly in regard to their language-use patterns – from language transmission to language loss – across generations. One main hypothesis states that on closer inspection the apparently distinctive features are expected to fade in favour of high inter-group similarities, e.g. in terms of high linguistic diversity. However, based on the explorative study's findings (cf. section 3) a further hypothesis states: the language-use pattern language loss/ shift is expected to be more frequent and at the same time much more sensitive to detect in the Turkish group than in the former-Yugoslavian one, due to language-political factors in the societies of immigration and of origin which possibly favour language shift and/or stigmatisation in the Turkish group. 5.2.3. Interdisciplinary Inconsistencies? "Origin" and Methods of Data Collection Here, the main hypothesis states that interdisciplinary inconsistencies primarily originate from fundamentally differing methods of data collection: the interdisciplinary variation from "language-maintenance-critical" to "language-maintenance-friendly" in regard to the findings (cf. section 2.3.) parallels the methodological variation on a continuum e.g. from "qualitative" to "quantitative", among other aspects, which can encourage or impede the emergence of languages and language-use patterns during data collection. Stigmatized languages, for instance, can be assumed to be detectable only with highly sensitive methods (cf. section 3). The same can be assumed for language shift/ loss which persons concerned tend to hide when the former "first" language has had a rather negative connotation (cf. Bott-Bodenhausen 1996: 52). But it is exactly this "remaining hidden" of languages and of language-use patterns which accordingly impedes any investigation of their consequences (cf. section 2.3). The importance we attach to this fact is so great that in addition to the qualitative data-collection parts we will conduct a quantitative survey (including also the qualitative sample) and access a supplementary official data source (see Figure 2: aim 5.2.3.). We hypothesize that contradictory results can be compared and explained very well by applying all three methods to a single sample of Turkish and former-Yugoslavian families.

5.3. Aim B. Examining the Theoretical Framework and Bringing Together the Different Analyses 5.3.1. Best Success through Language Loss? The Linguistic Capital Model The preceding work packages should have provided a basis for mapping immigrant multilingualism more vividly than has been possible in large-scale investigations up to now. Such a demanding and detailed mapping of actual language-use patterns and of all languages and identities represented on the familial level is regarded as indispensable: without it, no investigation of any "utility" of multilingualism and immigrant-family language maintenance or loss is possible at all (see Figure 2: aim 5.3.1.). The criticisms of the state of current research as previously stated (cf. section 2.3) as well as the criticisms of the explorative investigation presented above (cf. section 3) will also be considered: On the socio-political

"Best Success Through Language Loss?"

147

macro-level, not only the country of origin but also and above all the country of immigration will be included in the examination of the model, given that both societies, and particularly the immigration society, have strongly affected the families' linguistic proficiencies and ways of behaviour. On the parents' and children's levels, in turn, the special consideration of all languages, language-use patterns and aspects of linguistic identities will be crucial; and on the children's level, not only school-success variables but also factors indicating cognitive achievement will be surveyed (cf. footnote 5). The Linguistic Capital Model's overall hypothesis states that familial language transmission actually is highly determined by societal esteem or disrespect: the worse the conditions for transmission, the less family languages will be transmitted to the following generations. A further hypothesis says that families of Turkish origin are particularly affected by macro conditions non-conducive to language transmission; we will have to investigate in detail which factors in the sending and, above all, in the receiving societies could be responsible for a possibly origin-specific transmission behaviour. This behaviour in turn is assumed to be decisive for children's early language acquisition: the more parents transmit those languages which they feel "at home" with, be it emotionally or in terms of proficiency, the more children's language acquisition – including school-language acquisition – will benefit from parental input; the same can be assumed for other cognitive and educational abilities which require linguistic knowledge. In this process, the macro-level of the country of origin is seen to be effective primarily via parental language behaviour. The macro-level of the country of immigration, in contrast, is additionally seen to be effective in an absolutely direct manner, i.e. via school and non-school-related, implicit and explicit educational policy measures. The project thus clearly focuses on the new models' primarily language-related (individual, familial and societal) factors and their consequences for second-generation immigrant children's language acquisition and school career. However, it goes without saying that many more factors than the ones constituting the model must be investigated (e.g. the families' socio-economic status, the parents' educational background etc.), the selection in our project being geared mainly to recent migration research on similar topics. However, in the end it should be possible to clearly isolate the effects of traditionally investigated factors from the effects the new model's factors entail.

5.3.2. So What? Theoretical, Methodological and Educational Policy Consequences The preceding work packages' output can be assumed to be rich. Accordingly, its utilisation should be rich as well: while from the investigation of interdisciplinary inconsistencies (aim 5.2.3.) primarily methodological innovations can be expected, the survey of origin-specific differences (work package 5.2.1., 5.2.2., 5.2.3.) and of the new model's hypotheses altogether (work package 5.3.1.) should result in implications targeted at theoretical as well as educational policy innovations. We attach equal importance to all of them: the issue's topicality and societal explosiveness substantiates this claim.

148

Katharina Brizic

6. Incipient Study – Methods 6.1. Research Locations and Compilation Of Sample The project focuses on immigrant families of Turkish origin in Vienna/ Austria (cf. Figure 2: grey fields). Additionally, in order to carry out specific comparisons and analyses as described above, a small number of immigrant families with former-Yugoslavian background living in Vienna and a small number of non-migrants and migrants within national borders living in Istanbul and Ankara/ Turkey will be included in the investigation, as well as an official Viennese data source and the results of a quantitative survey with families of any origin living in Vienna (cf. Figure 2: white fields). In order to comprehensively map and comprise the families' linguistic backgrounds, three generations will be dealt with: children, parents and grandparents. The children's generation connects the sample's several geographical parts: in either case, the children are pupils in the fourth grade of school, and thus are close to the transition from primary to secondary school. Access to the sample will be provided via the children's schools13 which will in each case (Vienna, Istanbul, Ankara) be located in districts with a high proportion of (international or intra-national) migrants and will be selected randomly. Our Turkish cooperation partners' vitally important support is reliably assured. The number of qualitatively investigated children (cf. Figure 2: micro-level) and their parents (cf. Figure 2: meso-level) will be just small enough to allow time-consuming qualitative methods and just high enough to finally also allow quantitative multivariate analyses. This possibility of walking the borderline between qualitative and quantitative methods is substantial for the proposed project.

6.2. Variables and Levels of Investigation, Methods of Data-Collection and Analysis The choice of variables will be geared above all to the new explanatory model, but also to recent migration research on similar topics in several disciplines (cf. e.g. Söhn 2005). Thus, on the macro-level (cf. Figure 2) the variables will comprise numerous aspects of educational and language policies in the societies of origin and of immigration (here: Turkey and Austria); of special interest will be the question if there exist any aspects of Austrian policies which discriminate against Turkish immigrants in comparison to others. Data collection regarding such macro variables is usually carried out by comparing micro data from quantitative sources (cf. Esser 2006); however, this is not possible in the Austrian case: the quality of existing data does not allow comparisons between particular groups of origin (cf. section 2.1). Information will therefore be collected by intensive literature research, mainly comprising political-science literature, methodologically geared to Portes and Zhou (2001; Portes and Rumbaut 2001). 13

The application for research permittance and access to Viennese schools has been submitted to the Viennese school board and can – on the basis of former cooperations – be regarded as promising. A similar application has been submitted to Turkish authorities in Vienna and has, after an official hearing, been forwarded to Ankara with the Turkish Embassy‘s particularly positive recommendation. However, even without a research permittance, an informal out-of-school access to families in Turkey would be feasible, given our Turkish cooperation partners' contacts and their high interest in our project.

"Best Success Through Language Loss?"

149

The centrepoint of the investigation is located on the meso-level (cf. Figure 2): i.e. familial/ parental language use, parental language proficiencies and transmission behaviour as well as the construction of linguistic identity (cf. Wodak et al. 1998) – these all being factors which make up children's linguistic starting point in the model (see Figure 1). Due to the highly sensitive topics (as e.g. minority languages), data regarding these factors14 will be collected in time-consuming qualitative partly-structured one-on-one interviews with the 260 children's mothers or fathers, to be conducted in the family language(s) as far as possible: there will be at our disposal four Turkish-Kurdish-bilingual (including German: trilingual) interviewers with high expertise in consulting immigrant parents. The choice of the interviewing location will be made together with the parents.15 In addition, interviews with the children's teachers are intended, as well as collecting information about the sample by accessing the official Viennese "School Matrik".16 Furthermore, special attention will be dedicated to conducting G. Extra's quantitative survey "Multilingual Cities" (cf. Extra and Yağmur 2004) aimed at collecting, analysing and comparing data on home-language use amongst primary school pupils in several European capitals. If this survey succeeds in approaching approximately 80% of the Viennese primary schools, as planned17, this would be a considerable advancement regarding the quality of Austrian large-scale data: the survey's highly improved mapping of family languages, combined with other background factors, exceeds the existing quantitative sources by far (cf. section 2.1). On the micro-level (cf. Figure 2), finally, the variables of interest will among others be: children's proficiency in Turkish18 (to be surveyed by language tests developed by Reich and Roth 2003, and following Aksu-Koç et al. 2002; Peltzer-Karpf et al. 2006); a further language-based cognitive proficiency (to be surveyed by a maths test following TIMMS 1997); the variable educational success (to be surveyed in the form of teachers' recommendations regarding their pupils' forthcoming secondary schooling and in the form of children's school grades); several further individual factors making up linguistic identity (such as self-esteem etc., to be surveyed in the interviews with teachers and parents and by means of a questionnaire for children developed during the explorative study); and this investigation's key variable: children's proficiency in the school language of German with the focus on (literary) proficiencies relevant for cognitive and/or educational success (to be surveyed by a test which has to be adapted in the course of this project, following Reich and Roth 2003, Peltzer-Karpf et al. 2006, since satisfactory tests for bilinguals of this age group are rare; cf. Fried 2004; Ehlich 2005; Schroeder and Stölting 2005). The comprehensive data analyses will contain both qualitative and quantitative parts. Both will aim at correlating the variables of the Linguistic Capital Model among themselves as well as in relation to further variables relevant for recent migration research (cf. e.g. Söhn 2005). Initially, the methodological focus will again be on language, identity and its 14

As well as many more factors derived from recent migration research (see above). We were also considering collecting family-language-use data by personal observations in the families. However, such an enterprise turned out to be infeasible, given the large sample and long waiting times for particular use patterns to emerge. 16 Available in the form of "pupils‘ basic data" where children‘s languages are ascertained by means of a rather "rough" data collection method (i.e. under unfavourable conditions, time pressure etc.). 17 In accordance with the Viennese school board and only by mutual agreement with the schools' headmasters. 18 For other family languages such as Kurdish no tests exist so far which could be applied here (see Aksu-Koç 2002). However, we have to expect extensive language shift (cf. Zentrum für Türkeistudien Essen 1998: 190; Skutnabb-Kangas and Phillipson 1995). The families' actual linguistic background will be checked in detail in our study. 15

150

Katharina Brizic

transmission: as for these meso-level variables, the results of a rather "rough" method of data collection (i.e. the Viennese "School Matrik") will be compared to the results of a further quantitative but already highly improved method (i.e. "Multilingual Cities"), both of which are in turn opposed to this project's purely qualitative interviews and their results, with a discourse analysis of selected interviews crowning the analysis. Revealing insights should thus be possible, with a vivid and rounded comprehension of how linguistic identity is constructed (or deconstructed), how multilingual reality is lived (or lost) and how linguistic "origin" is presented (or hidden) in the course of differing methodological approaches. The detailed meso-level results will in turn be checked in terms of their dependency on the sociopolitical macro-level and in terms of their effect on the children's micro-level, including the investigation of direct macro-micro connections. Thus, in the end it should be possible to clearly differentiate between the effects of traditionally investigated factors and the effects the new model's factors entail, and to name those variables which have in our survey turned out to be most significant for immigrant children's school-language acquisition and school success.

Conclusion The proposed project tries to provide a new thematic approach to a rather old issue: it does not go via the home-language-teaching question which is relatively difficult to investigate (cf. Söhn 2005), nor via the cost-benefit calculation regarding language and labour market which easily tends to simplify linguistic matters (cf. Esser 2006). Rather it is launched from the familial input a child gets from the very first day on. As a side effect, concepts such as "origin", "affiliation" and even "first language" are expected to break open in the course of such an investigation: immigrant-family language use is much too multifaceted to stay within these frames. And this is exactly the point where the several groups of origin, in turn, are expected to show much more mutuality than dissimilarity: the "language capital's" character and its conditionality upon the macro-level tie them together. Thus, for the research on other "weak" groups in Central Europe (Moroccans, Bengalis) far-reaching implications may be expected, too. Another innovation is a methodological one: it is the attempt to resolve contradictions by means of different data collection methods in one and the same sample. Moreover, the planned quantitative "Multilingual Cities" survey will considerably improve the state of Austrian large-scale data. But above all, new ways of asking people for their languages should emerge in the end (see also Baur et al. 2004). This would in fact represent a substantial advancement in challenging multilingualism's "benefit" or "disutility": to be able to detect and uncover actual familial multilingualism (as well as its loss) at all. The new model finally represents this project's theoretical innovation, providing possible connections from the micro up to the macro-level, and thus also new explanations. If in fact some of the connections can be evidenced, this would mean a considerable counterbalance making up for the prevailing dichotomy of "immigrant multilingualism versus school success" (cf. Esser 2006: 74 ff.). And it would also mean new recommendations for educational policy and school practice, given that all answers to the open home-languageteaching question and similar issues finally depend on the extensive comprehension of how linguistic capital actually works and operates in individual, familial and societal life.

"Best Success Through Language Loss?"

151

The present project is intended to break new ground in a field where there is high demand in Central European migration research: it is so far the only study, with the single exception of Weiss et al. (2006), that deals with the Turkish group's weak position in Austria. But differently from Weiss et al., it aims at naming causes for the weak performance, as well as potential ways out, and focuses on language, the big issue in regard to educational success in modern societies. However, this huge subject can definitely not be treated completely in the course of only one project – but the project could be a starting point for establishing some new bridgings. In the end, the project touches a delicate topic: language and minority rights, and with it human rights (cf. Clyne 2000; Calvet 1998). This is all the more sensitive since at present the relations between Turkey and the European Union are afflicted exactly with these and related issues. The proposed project, in contrast, represents the special opportunity to face the topic in an entirely different way – not as a point of contention but as something which actually connects the societies of origin and of immigration: like Turkey, for far too long a time (cf. already Lüdi 1990: 119) Europe has had to contend with its own deficiencies in achieving basic linguistic human rights.19

References Afshar, K. (1998). Zweisprachigkeit oder Zweitsprachigkeit? Zur Entwicklung einer schwachen Sprache in der deutsch-persischen Familienkommunikation. Münster: Waxmann. Aksu-Koç, A., Erguvanlı-Taylan, E., and Bekman, S. (2002). Türkiye'de okul öncesi eğitimi: Hizmete duyulan ihtiyaçların belirlenmesi ve çocuğun dil yetisi düzeyinin değerlendirilmesi. Proje raporu (= Preschool education in Turkey: formulating requirements and evaluating children's language proficiency. Project report.) Istanbul: Bosphorus University. Andrews, P. A. (Ed.) (1989). Ethnic groups in the Republic of Turkey. Vol. 1. Wiesbaden: Reichert. Andrews, P. A. (Ed.) (2002). Ethnic groups in the Republic of Turkey. Vol. 2: Supplement and index. Wiesbaden: Reichert. Andriessen, I., and Phalet, K. (2002). Acculturation and school success: a study among minority youth in the Netherlands. Intercultural Education, 13 (1), 21-36. Bacher, J. (2005). Bildungsungleichheit und Bildungsbenachteiligung im weiterführenden Schulsystem Österreichs: Eine Sekundäranalyse der PISA 2000-Erhebung. SWSRundschau, 45 (1), 37-62. Bachmaier, P. (1982). Aspekte der Bildungs- und Wissenschaftspolitik Jugoslawiens 1944– 1979. Vienna: Österreichisches Ost- und Südosteuropainstitut. Badawia, T. (2002). "Der dritte Stuhl": Eine Grounded-theory-Studie zum kreativen Umgang bildungserfolgreicher Immigrantenjugendlicher mit kultureller Differenz. Frankfurt am Main: IKO.

19

I am especially grateful to Barbara Herzog (Political Studies), Reva Akkus (Psychology), Christoph Reinprecht (Sociology), Claudia Römer (Turkish Studies) and Rudolf de Cillia (Sociolinguistics) who have made this interdisciplinary approach possible.

152

Katharina Brizic

Baur, R. S., Chlosta, C., Ostermann, T., and Schroeder, C. (2004). "Was sprecht Ihr vornehmlich zu Hause?": Plädoyer für die Integration von sprachbezogenen Fragen. Essener Unikate, 24, 96-105. Boeschoen, H. (1990). Acquisition of Turkish by immigrant children. A multiple case study of Turkish children in the Netherlands aged 4 to 6. Wiesbaden: Harrassowitz. Boeschoten, H. (1991). The language reform. In H. Boeschoten, and L. Verhoeven (Eds.), Turkish linguistics today (pp. 165-175). Leiden: Brill. Boeschoten, H. (1997). The Turkish Language Reform forced into stagnation. In M. Clyne (Ed.), Undoing and redoing corpus planning (pp. 357-383). Berlin: Mouton de Gruyter. Bott-Bodenhausen, K. (Ed.) (1996). Unterdrückte Sprachen. Sprachverbote und das Recht auf Gebrauch der Minderheitensprachen. Frankfurt am Main: Lang. Bourdieu, P. (1983). Ökonomisches Kapital, kulturelles Kapital, soziales Kapital. In R. Kreckel (Ed.), Soziale Ungleichheiten (pp. 183-198). Göttingen: Schwartz. Brizić, K. (2006): The secret life of languages: Origin-specific differences in L1/L2 acquisition by immigrant children. International Journal of Applied Linguistics, 16 (3), 339-362. Brizić, K. (2007). Das geheime Leben der Sprachen: Gesprochene und verschwiegene Sprachen und ihr Einfluss auf den Spracherwerb in der Migration. Internationale Hochschulschriften, Vol. 465. Münster: Waxmann. Busch, B. (2006). Bosnisch, Kroatisch, Serbisch, Serbokroatisch, Jugoslawisch, Romani oder Vlahisch? Heteroglossie und "muttersprachlicher" Unterricht in Österreich. In P. Cichon (Ed.), Gelebte Mehrsprachigkeit. Akten des Wiener Kolloquiums zur individuellen und sozialen Mehrsprachigkeit, 5./6.11.2005 (pp. 13-27). Vienna: Praesensverlag. Calvet, L.-J. (1998). Language wars and linguistic politics. Oxford: Oxford Univ. Press. Cameron, D. (1990). Demythologizing sociolinguistics: why language does not reflect society. In J. E. Joseph, and T. J. Taylor (Eds.), Ideologies of language (pp. 79-93). London: Routledge. Chomsky, N. (1957). Syntactic structures. Gravenhage: Mouton de Gruyter. Clyne, M. (2000). Promoting multilingualism and linguistic human rights in the era of economic rationalism and globalization. In R. Phillipson, and T. Skutnabb-Kangas (Eds.), Rights to language. Equity, power, and education (pp. 160-163). Mahwah, NJ: Erlbaum. Crul, M., and Doomernik, J. (2003). The Turkish and Moroccan second generation in the Netherlands: divergent trends between and polarization within the two groups. In M. Crul, and H. Vermeulen (Eds.), The future of the second generation. The integration of migrant youth in six European countries (pp. 1039-1064). Staten Island, NY: Center for migration studies of New York. Cummins, J. (2000). Language, power and pedagogy. Bilingual children in the crossfire. Clevedon: Multilingual Matters. de Ruiter, J. J. (1997). L´avenir linguistique de la communauté marocaine aux Pays-Bas. In M. Tilmatine (Ed.), Enseignement des langues d´origine et immigration nord-africaine en Europe: langue maternelle ou langue d´Etat? (pp. 105-120). Paris : INALCO/ CEDREACRB. Dorian, N. C. (1982). Language loss and maintenance in language contact situations. In R. D. Lambert, and B. F. Freed (Eds.), The loss of language skills (pp. 44-59). Rowley, Mass.: Newbury House Publ.

"Best Success Through Language Loss?"

153

Dörnyei, Z. (2002). The motivational basis of language learning tasks. In P. Robinson (Ed.), Individual differences and instructed language learning (pp. 137-158). Amsterdam: Benjamins,. Dragić, D. (2002). Vlasi ili Rumuni istočne Srbije i Vlaško pitanje. Pitanja i odgovori. (=East Serbia's Vlahs or Romans and the "Vlah question". Questions and answers.) Beograd: Helsinški Odbor za Ljudska Prava u Srbiji. Dressler, W. U., and de Cillia, R. (2006). Spracherhaltung, Sprachverfall, Sprachtod. Language Maintenance, Language Decline and Language Death. In U. Ammon, N. Dittmar, K. J. Mattheier, and P. Trudgill (Eds.), Sociolinguistics/ Soziolinguistik (2nd edition, pp. 2258-2271). Berlin/ New York: de Gruyter. Ehlich, K. (Ed.) (2005). Anforderungen an Verfahren der regelmäßigen Sprachstandsfeststellung als Grundlage für die frühe und individuelle Förderung von Kindern mit und ohne Migrationshintergrund. Bonn: Federal Ministry for Education and Science (BMBF). Entorf, H. and Minoiu, N. (2004). What a difference immigration law makes: PISA results, migration background, socioeconomic status and social mobility in Europe and traditional countries of immigration. Darmstadt: Institut für Volkswirtschaftslehre. European Stability Initiative (ESI) (2007). Türkeidebatte in Österreich. Materialsammlung. http://www.esiweb.org/index.php?lang=deandid=127, 26.3.2007. Esser, H. (2006). Sprache und Integration. Die sozialen Bedingungen und Folgen des Spracherwerbs von Migranten. Frankfurt: Campus. Extra, G. and Yağmur, K. (Eds.) (2004). Urban multilingualism in Europe: immigrant minority languages at home and school. Clevedon: Multilingual Matters. Fase, W. (1994). Ethnic divisions in Western European education. Münster: Waxmann. Feliciano, C. (2006). Beyond the family: the influence of premigration group status on the educational expectations of immigrants' children. Sociology of Education, 79, 281-303. Fishman, J. A. (Ed.) (1999). Handbook of language and ethnic identity. New York: Oxford Univ. Press. Fried, L. (2004). Expertise zu Sprachstandserhebungen für Kindergartenkinder und Schulanfänger. Eine kritische Betrachtung (erstellt im Rahmen des Projektes "Schlüsselkompetenz Sprache – Bundesweite Recherche zu Maßnahmen und Aktivitäten im Bereich der sprachlichen Bildung und Sprachförderung in Tageseinrichtungen für Kinder"). München: DJI. Gardner, R. C. (1985). Social psychology and second language learning: The role of attitudes and motivation. London: Arnold. Gardner, R. C., and Tremblay, P. F. (1998). Specifity of affective variables and the trait/ state conceptualization of motivation in second language acquisition. In R. K. Agnihotri, A. L. Khanna, and I. Sachdev (Eds.), Social psychological perspectives on second language learning (pp. 31-52). New Delhi: Sage. Haque, Z. (2000). The ethnic minority "underachieving" group? Investigating the claims of "underachievement" amongst Bangladeshi pupils in British secondary schools. Race, ethnicity and education, 3 (2), 145-168. Herdina, P., and Jessner, U. (1997). Dynamisierung des Fremdspracherwerbs durch Mehrsprachigkeit – theoretische Implikationen. In M. Stegu, and R. de Cillia (Eds.) (1997), Fremdsprachendidaktik und Übersetzungswissenschaft. Beiträge zum 1. VERBAL-Workshop, Dezember 1994 (pp. 45-61). Frankfurt am Main: Lang.

154

Katharina Brizic

Herzog-Punzenberger, B. (2003 a). Die "2. Generation" an zweiter Stelle? Soziale Mobilität und ethnische Segmentation in Österreich – eine Bestandsaufnahme. Vienna: Wiener Integrationsfonds. Herzog-Punzenberger, B. (2003 b). Ethnic segmentation in school and labour market? 40 years legacy of Austrian guestworker policy. In M. Crul, and H. Vermeulen (Eds.), The future of the second generation. The integration of migrant youth in six European countries (pp. 1120-1144). Staten Island, NY: Center for migration studies of New York. Janjetović, Z. (2001). National minorities and non-Slav neighbours in Serbian textbooks. Internationale Schulbuchforschung, 23 (2), 201-214. Jeuk, S. (2000). Psycholinguistische Theorien zum Zweitspracherwerb: ihre Relevanz für die Frühförderung zweisprachiger Migrantenkinder. Deutsch lernen, 25 (3), 195-212. Jungblut, G. (1990). Marokkanische und tunesische Schülerinnen und Schüler in der Bundesrepublik Deutschland. In Landesinstitut für Schule und Weiterbildung (Ed.), Muttersprachlicher Unterricht mit marokkanischen und tunesischen Schülern. Ergebnisse einer Fachtagung. Soest. Knapp, A. (2006). Wiener Schulpflichtmatrik, Stand Juni 2005. Vienna: Stadtschulrat für Wien. Kohlbacher, J., and Reeger, U. (2003). Xenophobie aus der Perspektive der "Anderen" – Erfahrungen und Bewältigungsstrategien betroffener AusländerInnen. In H. Fassmann, and I. Stacher (Eds.), Österreichischer Migrations- und Integrationsbericht. Demographische Entwicklungen, sozioökonomische Strukturen, rechtliche Rahmenbedingungen (pp. 356-366). Klagenfurt/ Celovec: Drava. Korkisch, F. (2004). Terrorismus: Weniger Sicherheit – Die Suche nach der New World Order II. http://www.waldner.priv.at/mirror/korkisch.pdf, 26.3.2007. Kouritzin, S. (1999). Face(t)s of first language loss. Mahwah, NJ: Erlbaum. Kronig, W., Haeberlin, U. and Eckhart, M. (2000). Immigrantenkinder und schulische Selektion. Pädagogische Visionen, theoretische Erklärungen und empirische Untersuchungen zur Wirkung integrierender und separierender Schulformen in den Grundschuljahren. Bern: Haupt. Krumm, H.-J. (2002). "One sprachen konten wir uns nicht ferstandigen. Ferstendigung ist wichtig.": Entwicklung und Tendenzen in der Sprachlehrforschung im Bereich der Migration und Integration. Deutsch als Zweitsprache, 2 (2), 32-39. Kula, O. B. (1998). Durch die interne Migration in der Türkei im Alltagsleben der Familien ausgelöste Veränderungen und ihre Rolle in der Bildung, insbesondere in der Vorschulerziehung. In B. Nauck (Ed.), Erziehung – Sprache – Migration: Gutachten zur Situation türkischer Familien; Analysen; Projekt: Interkulturelle Elternarbeit im Arbeitskreis Neue Erziehung (pp. 145-193). Berlin: Arbeitskreis Neue Erziehung. Lasimbang, R., Miller, C., and Otigil, F. (1992). Language competence and use among coastal Kadazan children. In W. Fase, K. Jaspaert, and S. Kroon (Eds.), Maintenance and Loss of Minority Languages (pp. 333-355). Amsterdam: Benjamins. Lüdi, G. (1990). Les migrants comme minorité linguistique en Europe. In U. Ammon, K. J. Mattheier, and P. H. Nelde (Eds.), Minderheiten und Sprachkontakt. Minorities and language contact. Minorités et contact linguistique. Vol. 4. Sociolinguistica (pp. 113135). Tübingen: Niemeyer.

"Best Success Through Language Loss?"

155

Maas, U., and Mehlem, U. (2002). Schriftkulturelle Probleme der Migration: Kinder marokkanischer Einwanderer in Deutschland. In J. Oltmer (Ed.), Migrationsforschung und Interkulturelle Studien. Zehn Jahre IMIS (pp. 333-366). Osnabrück: Rasch. Maas, U. and Mehlem, U. (2003). Schriftkulturelle Ressourcen und Barrieren bei marokkanischen Kindern in Deutschland: Abschlussbericht des Projekts. Osnabrück: IMIS. Mehlem, U. (1998). Zweisprachigkeit marokkanischer Kinder in Deutschland: Untersuchungen zu Sprachgebrauch, Spracheinstellungen und Sprachkompetenzen marokkanischer Kinder in Deutsch, marokkanischem Arabisch und Berber (Masirisch) in Dortmund. Frankfurt am Main: Lang. Mehmet, Ö. (2002). Fundamentalismus und Nationalstaat. Der Islam und die Moderne. Hamburg: Europ. Verl.-Anst. Müller, R. (1997). Sozialpsychologische Grundlagen des schulischen Zweitspracherwerbs bei MigrantenschülerInnen: Theoretische Grundlagen und empirische Studien bei zweisprachigen und einsprachigen SchülerInnen aus der 6. bis 10. Klasse in der Schweiz. Aarau: Sauerländer. Nauck, B., Diefenbach, H. and Petri, K. (1998). Intergenerationale Transmission von kulturellem Kapital unter Migrationsbedingungen: Zum Bildungserfolg von Kindern und Jugendlichen aus Migrantenfamilien in Deutschland. Zeitschrift für Pädagogik, 44 (5), 701-722. Nauck, Bernhard/ Schönpflug, Ute (Eds.) (1997): Familien in verschiedenen Kulturen. 86 Tabellen. Stuttgart: Enke. OECD (2005). School factors related to Quality and Equity. Results from PISA 2000, http://www.oecd.org/dataoecd/15/20/34668095.pdf, 26.3.2007. Olechowski, R., Hanisch, G., Katschnig, T., Khan-Svik, G, and Persy, E (2002). Bilingualität und Schule – eine empirische Erhebung an Wiener Volksschulen (Endbericht). In W. Weidinger (Ed.), Wissenschaftliche Befunde. Vol. 2. Bilingualität und Schule (pp. 8-63). Vienna: öbv and hpt. Peltzer-Karpf, A., Wurnig, V., Schwab, B., Griessler, M., Akkuş, R., Lederwasch, K., Piwonka, D., Blažević, T. and Brizić, K. (2006). A kući sprecham Deutsch. Sprachstandserhebung in multikulturellen Volksschulklassen: bilingualer Spracherwerb in der Migration. Vienna: Austrian Federal Ministry of Education, Science and Culture (BMBWK). Portes, A. and Rumbaut, R. G. (2001). Legacies: the story of the immigrant second generation. Berkeley: Univ. of California Press. Portes, A., and Zhou, M (2001). The new second generation. Segmented assimilation and its variants. In H. Goulbourne (Ed.), Integration, adaptation and change. Vol. 4. Race and ethnicity (pp. 314-336). London (et al.): Routledge. Preibusch, W. (1992). Die deutsch-türkischen Sprachenbalancen bei türkischen Berliner Grundschülern. Eine clusteranalytische Untersuchung. Frankfurt am Main: Lang. Reich, H. H. and Roth, H.-J. (2003). HAVAS-5. Hamburger Verfahren zur Analyse des Sprachstands bei Fünfjährigen. Landau, Hamburg: unpublished manuscript. Robins, K. (2000). Interrupting identities: Turkey / Europe. In S. Hall, and P. du Gay (Eds.), Questions of cultural identity (5th edition, pp. 61-86). London: Sage Publ. Rüesch, P. (1998). Spielt die Schule eine Rolle? Schulische Bedingungen ungleicher Bildungschancen von Immigrantenkindern – eine Mehrebenenanalyse. Bern: Lang.

156

Katharina Brizic

Sasse, H.-J. (1992). Theory of language death. In M. Brenzinger (Ed.), Language Death. Factual and Theoretical Explorations with Special Reference to East Africa (pp. 7-30). Berlin: Mouton de Gruyter. Schiesser, A. W., and Theurl, P. (2001). Manifeste und schulbezogene Ängste bei Volksschulkindern mit deutscher und nichtdeutscher Muttersprache. In W. Weidinger (Ed.), Ausbildung, wissenschaftliche Perspektiven und empirische Befunde. Vol. 1. Bilingualität und Schule? (pp. 254-275). Vienna: öbv and hpt. Schroeder, C., and Stölting, W. (2005). Mehrsprachig orientierte Sprachstandsfeststellungen für Kinder mit Migrationshintergrund. In I. Gogolin, U. Neumann, and H.-J. Roth (Eds.), Sprachdiagnostik bei Kindern und Jugendlichen mit Migrationshintergrund (pp. 59-74). Münster: Waxmann. Six-Hohenbalken, M: (2001): Migrantenfamilien aus der Türkei in Österreich. Wohnen, Verortung und Heimat; mit einem Exkurs über die Wohnsituation im Aufnahmeland. Vienna: Österreichisches Institut für Familienforschung. Skutnabb-Kangas, T. and Phillipson, R. (Eds.) (1995). Linguistic human rights. Overcoming linguistic discrimination. Berlin: Mouton de Gruyter. Söhn, J. (2005). Zweisprachiger Schulunterricht für Migrantenkinder: Ergebnisse der Evaluationsforschung zu seinen Auswirkungen auf Zweitspracherwerb und Schulerfolg. (= AKI-Forschungsbilanz, No. 2). Berlin: Arbeitsstelle Interkulturelle Konflikte und gesellschaftliche Integration (AKI) and Wissenschaftszentrum Berlin für Sozialforschung (WZB). Söhn, J. and Özcan, V. (2006). The Educational Attainment of Turkish Migrants in Germany. Turkish Studies, 7 (7), 101-124. Stanat, P. (2003). Migration und Sozialschicht. Determinanten der Schulleistungen von Jugendlichen aus zugewanderten Familien. Vortrag, gehalten bei der 64. Tagung der AEPF (Arbeitsgruppe für Empirische Pädagogische Forschung), 29.9. – 2.10. 2003, Universität Hamburg. Hamburg: unpublished manuscript. Steuerwald, K. (1963). Die türkische Sprachpolitik seit 1928. Vol. 1. Untersuchungen zur türkischen Sprache der Gegenwart. Berlin: Langenscheidt. Thomé, G. (1987). Rechtschreibfehler türkischer und deutscher Schüler. Heidelberg: Groos. Tilmatine, M. (1997). L´enseignement des langues maternelles en Europe: la fin d´une mystification? In M. Tilmatine (Ed.), Enseignement des langues d´origine et immigration nord-africaine en Europe: langue maternelle ou langue d´Etat? (pp. 69-101). Paris: INALCO/ CEDREA-CRB. TIMMS (Third International Mathematics and Science Study) (1997). TIMSS mathematics items: released set for population; IEA's Third International Mathematics and Science Study. Third and fourth grades. Chestnut Hill: TIMSS International Study Center, Boston College. Uçar, A. (1996). Benachteiligt: ausländische Kinder in der deutschen Sonderschule. Eine empirische Untersuchung zur Lage der türkischen Kinder in der Sonderschule für Lernbehinderte. Baltmannsweiler: Schneider. Weiss, H., Gapp, P., Strodl, R., Unterwurzacher, A. and Wittmann-Roumi Rassouli, M. (2006). Leben in zwei Welten. Zur sozialen Integration ausländischer Jugendlicher der zweiten Generation. Eine empirische Untersuchung in Österreich. Projektbericht. Vienna: University of Vienna, Institute of Sociology.

"Best Success Through Language Loss?"

157

Wodak, R., de Cillia, R., Reisigl, M., Liebhart, K., Hofstätter, K. and Kargl, M. (1998). Zur diskursiven Konstruktion nationaler Identität. Frankfurt am Main: Suhrkamp. Wodak, R. and Rindler-Schjerve, R. (1985). Funktionen der Mutter beim Sprachwechsel: Konsequenz für die Primärsozialisation und Identitätsentwicklung. Wiesbaden: Vieweg. Wode, H. (1995). Lernen in der Fremdsprache. Grundzüge von Immersion und bilingualem Unterricht. Ismaning: Hueber. Yağmur, K. (2001). Languages in Turkey. In G. Extra, and D. Gorter (Eds.), The other languages of Europe. Demographic, sociolinguistic and educational perspectives (pp. 407-427). Clevedon: Multilingual Matters. Zentrum für Türkeistudien Essen (Ed.) (1998). Das ethnische und religiöse Mosaik der Türkei und seine Reflexionen auf Deutschland. Vol. 1. Wissenschaftliche Schriftenreihe des Zentrums für Türkeistudien. Münster: Lit. Verl.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 5

A CONTRIBUTION TO THE STUDY OF RUSSIAN EUPHEMISTIC YOUTH SLANG Brian Cooper The Publishing Training Centre, UK

Abstract After an introductory discussion of the difficulties of distinguishing between slang and euphemism, an examination is presented of some sexual and scatological aspects of Russian euphemistic youth slang under these headings: Pregnancy and abortion; Contraception; Copulation; Erection; Expletives; Group sex; Homosexuality; Masturbation and menstruation; Oral sex and orgasm; Penis; Prostitute; Virgin and vulva/vagina; Backside and anal sex; and Lavatorial matters and vomiting. The conclusion is reached that slang is mainly a young persons‘ phenomenon and that, though many slang locutions die, a few become established, which justifies the study of slang as one of many sources by which a living language enriches itself and continues to develop. Slang is concerned to an appreciable extent with sex and scatology. Even in the case of euphemistic slang words, the euphemized substitute may itself be contaminated by association with the taboo word and come to be considered unacceptable or not quite decent and therefore to be avoided.

Introduction Although not all usages of a taboo word need necessarily be acquired by its euphemisms, obviously some dysphemistic nuances will almost certainly be acquired, and certain circles of society will be predisposed to using euphemisms of taboo words in this dysphemistic way, perhaps partly motivated by ‗fear of appearing affected or sissy‘ (Steadman: 93), which can be for some a counterbalance to taboo, or by ‗the flavour of forbidden fruit‘ (Steadman: 101). In this area particularly, both euphemism and dysphemism can merge into slang; the borderline between them is not easy to establish: ‗Euphemisms are often hard to distinguish from slang, although slang is usually less transparent in meaning and often serves to reinforce the negative aspects of a topic […] unlike euphemism which attempts to redirect the focus of

160

Brian Cooper

attention‘ (Murphy: 8). Euphemism can strictly speaking affect all varieties of language, from the literary to slang (Kany: vi, reviewed by Widlak: 98; Budagov: 95), but Niceforo argues that the techniques of slang are used to produce the verbal substitutes for forbidden names that are typical of euphemism, including designating a being or object by one of its qualities, metaphor, changing the initial, final or other letters of the name, and borrowing from foreign languages or dialects (Niceforo: 204). He also maintains (217) that the system of conventional names used to avoid taboo words serves a similar function to slang, being aimed at protecting oneself against the foreign. Both might be thought to constitute a ‗parole spéciale‘ (226) or ‗parole de convention‘ (235). Youth slang, of course, follows this same pattern, being meant to hamper the uninitiated from sharing in the intercourse of the young; it ‗results from the favourite game among the young and lively of playing with words and renaming things and actions‘ (Fowler: 316). If both euphemism and slang have many synonyms for sexual matters, it is ultimately for the same reasons, according to Niceforo (264-265); if a thing can be called by its real name, the number of synonyms is usually limited. The present lexical study examines some of the synonyms for sexual and scatological matters among material on recent Russian youth slang assembled by Nikitina.

1. Pregnancy and Abortion One of the euphemisms associated with abortion is zalëčica ‗woman who has had an abortion‘, derived from the slang sense of zalëtnaja ‗pregnant‘, which is based on the standard phrase zalëtnyj gost‘ ‗unexpected visitor‘. Likewise zalët is ‗pregnancy‘ and zaletet‘ ‗fall accidentally pregnant‘: ‗Смотри, чтоб никто не залетел. – Хрен тебе, я с презервативом‘ (Nikitina: 143). A pregnant woman can also be zarjažennaja, literally ‗loaded‘, or ženščina s bemolem (bemol‘ being slang for ‗belly‘ owing to the shape of the sign for a flat in music). Vozdušnyj šar proglotit‘ is ‗to get pregnant‘, while razminirovanie, literally ‗mine sweeping‘, is ‗abortion‘. Ximčistka can be an abortion or the clinic where it is performed: ‗Что, опять в химчистку?‘ (Nikitina: 497).

2. Contraception Euphemistic slang words for a condom include antisemit, because it is anti ‗against‘ semja ‗semen‘; galoša, because it covers and protects like galoshes; parašjut, because it is a safety device; and pokryška, because it covers the penis. The same covering notion is apparent in čexol (‗Дуло-то в чехле?‘, Nikitina: 118; here dulo = penis) and in rubaška, which covers the penis like a nightshirt. Rezin(k)a is the English ‗rubber‘ (‗С резинкой или без?‘, Nikitina: 381), while sest‘ na rezinu is ‗begin to use a condom‘, and francuzskaja šapočka is a ‗French letter‘. Probirka can be a condom, which when extended resembles a test-tube (‗У тебя нет с собой пробирки?‘ – Nikitina: 359), as can šturval because it gives control of the outcome.

A Contribution to the Study of Russian Euphemistic Youth Slang

161

3. Copulation Many euphemisms of this type, especially describing copulation from the male‘s point of view, have a rather violent or aggressive streak, such as drjučit‘, literally ‗to beat with a cudgel‘ < dialectal drjuk ‗cudgel‘, perhaps a hypercorrection of the synonym druk (Vasmer: s.vv. druk, drjučit‘). From this verb comes drjučon ‗prostitute: ‗Дрючон – это чувишучка, которую все подряд трахают‘ (Nikitina: 117). Other verbs of this type include traxat‘ and its compounds, such as ottraxat‘ and potraxivat‘, and, because this verb has become so widespread and dysphemistically tainted, its euphemism traxatet‘. Similarly there are vprysnut‘, literally ‗inject‘; žarit‘, an extension of the demotic sense ‗do something vigorously‘; zagruzit‘ (‗Ну, ты ей загрузил?‘, Nikitina: 137); zapressovat‘; kuvyrkat‘sja, literally ‗turn head over heels‘; mazdat‘, originally ‗hit‘, a back formation from mazdaj ‗must die‘ in the rock opera ‗Jesus Christ Superstar‘ (‗Таких баб маздать‘, Nikitina: 238, whence the noun mazdon ‗copulation‘); mebel‘ dvigat‘; otkukožit‘, from the same root as (s)kukožit‘sja ‗writhe, shiver‘; prodërnut‘, conceivably from the figurative sense ‗tear off a strip, give a rollicking‘; pul‘nut‘, zabit‘ pulju, perhaps from the notion of penetration; pyrnut‘, with the literal meaning ‗jab‘, as likewise in ukolot‘; červjačka zadavit‘; and špoknut‘, literally ‗hit, strike‘: ‗Я бы еë шпокнул‘ (Nikitina: 532). Another common type is that which involves a euphemism for the penis: vitamin X (‗Витамина Х не хватает‘, Nikitina: 61; the Russian letter Х, like its old name xer, is a euphemism based on the first letter of the taboo word xuj, for which see Cooper 1993: 73; 1998: 15, note 9); vzjat‘ na bolt; gradusnik postavit‘; nožik potočit‘; palku kinut‘ (brosit‘, vbit‘, zabit‘, vpravit‘, postavit‘); piston postavit‘ (vstavit‘); ptenčik v gnëzdiško; zavaflit‘ (‗Пришëл чувачок и завафлил их обеих‘, Nikitina: 134 < vaflja ‗penis‘ and ‗fellatio‘, whence vafel‘nica ‗prostitute, mouth‘); otfigačit‘, a euphemism of otxujačit‘ (see Cooper 1998: 17) based on figačit‘ ‗do something energetically‘; and šeršavogo zagnat‘ pod kožu. In some cases the euphemism is for the female or her genitals: obkatat‘ cilindr, obkatka cilindra (cf. English ‗roll‘ – cilindr neobkatannyj = virgin); na gitare igrat‘ – like balalajka, gitara refers to the female genitals; pobatonit‘sja, with baton being an attractive, often loose, young woman; selëdku est‘, where selëdka is another of the many slang words for a girl; and celinu vspaxat‘ (podnjat‘) ‗have sex with a virgin‘, a pun on celka (see section 12). In many cases the euphemism has parallels in English or is actually borrowed from English: zavalit‘ (‗Гога Иру завалил, а я упал на Сильву‘, Nikitina: 134) – cf. valit‘ in the sense ‗tumble, lay‘ (‗Он повалил еë на кровать‘); imet‘ ‗have‘ and its perfectives poimet‘, otymet‘, zaimet‘ (‗Да я не прочь вон ту биксу поиметь‘, Nikitina: 333); pixat‘sja, compounds such as perepixnut‘sja, and the derived nouns perepixon, perepixnin (s povtorinom) ‗(repeat) copulation‘ – cf. English ‗stuff, poke‘ (see Cooper 2003: 7, note 4); pilit‘, its compounds such as zapilit‘, and the derived nouns pila and pilorama (the last implying a prolonged bout of sex) – cf. English ‗saw off a piece‘; poudačit‘sja – cf. English ‗score‘; fakat‘(sja) and fačit‘(sja) – cf. ‗Чем бы дитя ни тешилось, лишь бы не факалось‘, a pun on the saying which normally ends ‗лишь бы не плакало‘ (Nikitina: 471), ‗Я фачился с герлами [= girls]‘ (Nikitina: 474), both derived from the English taboo word fuck and a classic example of borrowing from a foreign language as a method of euphemization, which is further apparent in compounds like otfačit‘ (‗Отфачим по герлушке‘, Nikitina: 303 – see the next paragraph for others) and in the noun fak (‗Снимем вон тех на фак?‘, Nikitina:

162

Brian Cooper

471); čiknut‘sja and počikat‘sja, probably from čikat‘/čiknut‘ ‗snip, cut (noisily), snap off‘ – cf. English slang ‗tear off a piece, take a slice‘ (‗Почикаются и вернутся в школу продолжать образование‘, Nikitina: 346) , but compare also čiksa ‗(flirty) girl, prostitute‘, perhaps from English chick(s); and seksovat‘sja – cf. English ‗have sex‘. In the sense ‗pull‘, i.e. pick up for sex, the main terms are kadrit‘, derived from kadr in the sense ‗member of a cadre‘ – cf. kádrá ‗woman, girl‘, kadriška, kadrica ‗girl‘, kadrëžka ‗pulling, picking up‘, and kleit‘ and its compounds – cf. skleennyj kadr, a man in a longestablished relationship with a girl: ‗Склеить герлу‘, ‗Подклеился вчера к кадришке‘, ‗Лень отклеивать, суетиться‘, ‗Хочешь еë заклеить?‘ (Nikitina: 406, 329, 296, 141). Klej is a man given to ‗pulling birds‘: ‗Он […] известный клей был‘ (Nikitina: 186). Other expressions for ‗pulling‘ include podkatyvat‘ šary (jajca), literally ‗roll the balls up to‘; prifakivat‘sja and podfaknut‘sja, both compounds of fakat‘sja (see above); slomat‘ čuvišku; and snimat‘ (Тебя снимают все‘, Nikitina: 411). Among agent nouns like klej for a man hooked on sex there are babtist, a conflation of babnik ‗womanizer‘ and baptist; traxal‘ščik (‗Он великий трахальщик‘, Nikitina: 449 < traxat‘ – cf. traxalka ‗penis‘ and trax ‗sexual activity‘); mal‘čik-kolokol‘čik – cf. kolokol‘čik ‗penis‘; and a number of loans based on English fuck, such as faker ‗fucker‘, fakir, an extension of the sense of fakir ‗fakir‘ under the influence of faker, and fakmen < fuck + man (= faker).

4. Erection A similar effect of English fuck on an established Russian word (fakel ‗torch‘ < German Fackel, Old High German faccala, Latin facula, a diminutive of fax ‗torch‘) can be seen in a phrase for erection of the penis, fakel razgorelsja (razgoraetsja), in which fakel = penis. An analogous phrase with the same meaning is siška dymitsja (siška = penis). A number of other terms with this implication involve Russian or English words for ‗stand‘: stojačka, stènd(á), dolgostoj (literally ‗long-standing‘, i.e. a prolonged erection) and stolbnjak (stand like a post): ‗У тебя на неë давно стоячка‘, ‗У меня на эту герлу стэнда‘, ‗С бодуна [= when drunk], конечно, фигово, но зато какой столбняк!‘ (Nikitina: 427, 431, 425). The derivative verb stèndovat‘ means ‗have an erection‘. Torčat‘ is commonly used in this sense but, like torčat‘ kak lom and vxodit‘ v stolbnjak, it can denote simply ‗experience pleasure‘: ‗Я с этой музыки торчу‘, Nikitina: 446-447. Another word for an erection is davlenie (‗У него на неë давление‘, Nikitina: 97) and an interesting term for a matutinal erection is šatrovaja bolezn‘, because it raises the bedclothes like a tent.

5. Expletives Yet another way in which English fuck influences Russian is in the use of sexual vocabulary in expletives, whence fakannyj ‗effing‘ (= ëbanyj ‗fucking‘) and fakuški ‗nothing doing, not likely‘ (= xujuški). This type of speech is called in Russian youth slang matil‘da (= mat), rodnaja reč‘ or foršlag(i) (< German Vorschlag ‗proposition‘: ‗Без форшлагов, чуваки, здесь же дамы‘, Nikitina: 483). It makes wide use of English taboo terms in euphemization of Russian ones, whence prikóvo (= xuëvo) and prikóvyj (= xuëvyj), from English prick ‗penis‘: ‗Ну, как тебе тут? – Приково‘, Nikitina: 352. There are, of course,

A Contribution to the Study of Russian Euphemistic Youth Slang

163

several well-established euphemisms that have even found their way into standard Russian dictionaries because they are so widespread: pó figu for pó xuju and na xrená or na figá for na xerá (itself originally a euphemism of na xujá, see Cooper 1993: 73-74). They themselves spawn derivatives, like the augmentative na figišča (‗Чëрт его знает, на фигища ему две любовницы!‘, Nikitina: 275) and the nouns pofigizm ‗couldn‘t care less attitude‘ (= French je m‘en foutisme or its euphemism je m‘en fichisme) and pofigist. Similarly figóvyj = xuëvyj and figóvo = xuëvo (‗Фигово мне тут, домой хочу‘, Nikitina: 477). Fig is the euphemism for xuj on which many derivatives of this type are based – cf. fignja, figovina, figovjak or figacija = xujnja, xuëvina ‗fucking rubbish‘; figet‘ ‗rave‘ and the derived ofigevat‘/ofiget‘ ‗be amazed‘ (= xuet‘, oxuevat‘/oxuet‘), ofigennyj ‗delightful‘ and ofigenno ‗extremely‘ (= oxuennyj, oxuenno); and fiška (= figacija), presumably representing the diminutive fižka (< figa). Krutit‘ fiški, incidentally, can mean ‗pull a bird‘, i.e. pick up a girl; cf. krutit‘ in the slang sense ‗carry on with‘ (‗Он крутит с девчонками‘). Another phrase euphemized with fig is mne odin xuj ‗I don‘t give a fuck‘, which can become mne odin fig, mne odin xren or even (‗Latinized‘ as a form of euphemism) mne monopenisno, mne monopenisual‘no (Mokienko and Nikitina 2000: 355)

6. Group Sex One final euphemism based on the English f-word belongs here, viz. fak-sejšen, a loanword borrowed directly from English fuck-session. Other words for group sexual activity include the following: derivatives of gruppovoj (as used in gruppovoj seks), such as gruppovičok and gruppovuxa; terms for a female‘s sex act with two males at the same time (buterbrod, because she is sandwiched between them) and with three or more males (vertolët, because of some fancied resemblance to the rotor blades and supports of a helicopter); monitor; and romaška, in which participants first sit in a circle like petals of chamomile, the daisy-like flower used to test someone‘s love (gadat‘ po romaške) by pulling off the white petals to the words ‗(s)he loves me, (s)he loves me not‘: ‗Я никогда не участвовал в попойках, которые обычно заканчивались игрой в ―ромашку‖ – когда трахаются кто с кем попало‘, Nikitina: 386. There are numerous slang terms euphemistically denoting pack rape, such as karusel‘ (shades of ‗La Ronde‘); kolxoz as in pustit‘ na kolxoz [kogo] ‗gangrape [someone]‘; smyčok, as in (pro)pustit‘ [kogo] v tri/v četyre smyčka ‗subject [someone] to gang-rape by three/by four men‘ (smyčok = penis); tramvaj, as in pobyvat‘ pod tramvaem ‗be gang-raped‘ and (pro)pustit‘ čerez/pod tramvaj ‗gang-rape‘; and xor as in pobyvat‘ pod xorom ‗be gang-raped‘, (pro)pustit‘/protjanut‘/propixnut‘ xorom/na xor ‗gang-rape‘. Similarly, xorovoe penie can denote group sex: ‗Одного мужчины ей мало, предпочитает ―хоровое пение‖‘, Nikitina: 316.

7. Homosexuality As a member of a minority sexual grouping, a homosexual is sometimes called seksmen‘ševik in youth slang: ‗Секс-меньшевики кружат в отдалении. Слышны смешки‘, Nikitina: 399. An active homosexual is often euphemistically known as an aktivist (= aktivnyj gomoseksualist): ‗Вот там и ходят ―активисты‖, ищут, к себе зовут‘, Nikitina: 12.

164

Brian Cooper

Shortenings of gomoseksualist are frequent, such as goma, gomik, gomusik. These terms are commonly used for male homosexuals (cf. English homo), as are goluboj and golubar‘, the female equivalent of which is rozovyj, presumably because blue is the traditional colour for dressing baby boys and pink for girls. Possibly the effeminacy often associated with male homosexuals results in the English use of ‗pink‘ for homosexual (e.g. pink pound), and it is interesting to note from the colour viewpoint Latin galbanus, galbinus ‗yellowish, effeminate‘. The derived Russian noun golubizna is a euphemism for homosexuality. Other words for female homosexuals include dvustvolka, although this can also mean a bisexual or simply a female, and lesba, a shortening of lesbijanka; derivatives include lesbišče ‗place where there are many lesbians‘ (inspired by the noun ležbišče ‗breeding ground, lair‘) and posadit‘ na lesbie ‗win over to lesbianism‘. Among the terms for male homosexuals are dodik, perhaps from the hypocoristic name Dodja (= David), possibly in reference to David and Jonathan in the Bible, and mužikoved, conceivably hinting at muželožstvo ‗sodomy‘. There is also čerburaška, an under-age homosexual, from the name of a popular cartoon character of diminutive size (Shlyakhov and Adler: 277). Since in Russian tradition only the passive homosexual is looked down upon, it is hardly surprising to find obižennyj and opuščennyj (here opustit‘ may have the euphemistic slang sense ‗rape‘) used to designate such a man, humiliated and despised by many, in contrast to uvažaemyj ‗active homosexual‘. Other words for a passive male homosexual are pidovka and petux (‗Теперь он – ―петух‖, он – ―опущенный‖‘, Nikitina: 319), both perhaps influenced by the very widespread terms for male homosexuals generally that are derived from shortening or otherwise modifying pederast, such as pedik, pedagog and pede (also pèdè): ‗Они все – пэдэ‘, Nikitina: 370. An active homosexual male, on the other hand, is Solidol Ivanovič (so called after solidol, a lubricant grease). In slang usage, razvesti can mean ‗seduce into homosexual activity‘: ‗Если идиот, то ―развести‖ его легче, умный – нужно повозиться‘, Nikitina: 374.

8. Masturbation and Menstruation Common euphemistic slang terms for a masturbator refer to the biblical Onan, said to have ‗spilled [his seed] on the ground‘ (Genesis 38:9); they include onan, probably a shortening of onanist, and the wordplay anonim: ‗Я там вчера ещë одного анонима видела в кустах‘, Nikitina: 14. Terms for a female masturbator in particular include balalajka, which can refer to the femate genitals too, and kovyrjalka, which also means ‗gynaecologist‘ for obvious reasons (kovyrjat‘(sja) v = pick at, poke about in). A frequent verb ‗to masturbate‘ is dročit‘; it originally meant ‗cosset, fondle‘ but this sense has been largely supplanted, and as a result the verb and its derivatives have become taboo, another example of the way in which a word first used as a euphemism becomes tainted by the word it is euphemizing. A common phrase using the verb is ‗каждый дрочит, как он хочет‘, which is roughly equivalent to English ‗each to his own‘, implying complete freedom of choice. Of the various euphemistic slang words for menstruation, boloto and podarok are perhaps self-explanatory, demonstracija is a clever pun on menstruacija, while mena and mina are shortenings of it: ‗У тебя мина, что ли?‘ (Nikitina: 251). The wordplay on colour (hint at blood) is evident in cvetnoj televizor, the slang televizor denoting the female genitals. A number of terms with similar meaning involve the word krasnyj or one of its derivatives, as might be expected, e.g. krasnuxa (‗У тебя что, краснуха?‘ – Nikitina: 207), (krasnyj) flag

A Contribution to the Study of Russian Euphemistic Youth Slang

165

vybrosit‘/dat‘/ pustit‘ (see Cooper 1993: 67-68 and cf. English slang ‗the flag is up‘) and krasnaja armija: ‗Красная армия пришла (в гости пожаловала)‘, Nikitina: 15, and cf. gosti prišli in the same sense. A similar expression to this last idea of the arrival of something or someone is kon‘ priskakal (k komu/u kogo): ‗Слава Богу, конь прискакал, я уж думала, опять залетела‘, Nikitina: 199 (zaletet‘ = fall pregnant; see section 1 above). The idea of having a (little) visitor perhaps also underlies the use of maška for menstruation; it is, after all, a diminutive of Marija, one of the commonest names for women, and occurs in slang usage in related ‗women‘ senses too, such as ‗prostitute‘ and ‗vulva‘ (see Kveselevič: 469, s.v. maška). Two final words that denote menstruation both dwell on the notion of flowing: tečka and reka. They are somewhat parallel to English slang (monthly) flowers in this sense, which is clearly a pun on ‗flow‘: flower can suggest something that flows (see Partridge: s.v. flowers).

9. Oral sex and Orgasm The various phrases used to refer to oral sex performed on a man include pripast‘ k kormuške (here kormuška = penis), sigaretku syruju dat‘ prikurit‘ (here sigaretka = penis), and exat‘ v rotterdam, a pun on Russian rot ‗mouth‘: ‗Мы на ней ехали через попенгаген в роттердам‘ (Nikitina: 387). This last quotation contains a double pun, popengagen playing on Copenhagen and the euphemism popa (for taboo žopa ‗arse‘) to hint at anal sex followed by oral sex. Another pun is evident in the expression delat‘ smoktunovskogo, used especially by homosexuals. The wordplay is between the name of the actor Smoktunovskij and the verb smoktat‘ < Ukrainian smoktaty ‗suck‘. The term čupa-čups ‗oral sex (act)‘ is derived from the name of a caramel sweet on a stick (see Mokienko and Nikitina 2000: 677). Similar playing on words lies behind a number of terms for orgasm, e.g. Ded Klim prišel, with Klim probably suggested by English climax, and končita prišla, with končita suggestive of Spanish Conchita, but also meaning ‗end‘ in slang and punning on the taboo use of končit‘ ‗come (sexually)‘: ‗Самая сладкая кончита почему-то по трезвяку‘, Nikitina: 198-199. Analogous to this use of come is the Russian slang prixod, especially in the expression pojmat‘ prixod ‗have an orgasm, come‘. Finally, mention should be made of the use of blevat‘, normally ‗vomit‘, in the sense ‗ejaculate‘: ‗Не успел вставить – уже блевать‘, Nikitina: 35.

10. Penis Euphemistic youth slang words for the penis are often based on some object of phallic shape, e.g. bazuka, balda (probably in its dialect sense ‗knob on the end of a stick‘), banan (cf. the English catch-phrase unzip a banana and now obsolete have a banana with ‗have sex with‘), bolt, vaflja (probably < German Waffel ‗ice-cream cornet‘, whence also the meaning ‗fellatio‘), vint, gradusnik, dulo, nasest (the pole on which birds perch), ogloblja, palka (‗Любовь – костëр. Она погаснет, если не подкидывать туда палок‘, Nikitina: 308), pistolet, piston, sigaretka, sosiska, svistok, fakel, xlopuška, šljapa (perhaps from the shape of the crown, as in a top-hat) and šompol. Some of them denote weapons of various sorts. Those not exactly of phallic shape are often sharp-pointed and therefore somewhat reminiscent of

166

Brian Cooper

the English vulgarism prick ‗penis‘, e.g. igla kožanaja, nožik, šip and štyk, or else knob-like and reminiscent of English taboo slang knob, sometimes nob ‗penis‘ (by synecdoche, a part for the whole), e.g. knopka, pupyryšek, šiška, piljulja: ‗А у негров пилюля тоже чëрная?‘ (Nikitina: 321). Among words for a large penis there are baldometr (cf. balda above) and gamsaxurdija. A number of words for the penis are derived with the suffix -lk(a) denoting an object producing or intended to produce the action indicated by the motivating verb, such as dolbalka, imelka, nyrjalka, podymalka, prygalka, sovalka, traxalka. Others dwell on its characteristics, e.g. generator, possibly because it generates life; gordeič, from dialectal gordej ‗proud man‘, because of its proud appearance when erect (cf. English slang pride of the morning or morning-pride ‗matutinal erection‘ – see Partridge: s.v. pride); dvižok (kožanyj), because it slides to and fro, a similar notion to that in both piston (cf. postavit‘ piston in section 3 on copulation) and smyčok (literally ‗bow‘: see section 6 on group sex); družok, conceivably with the implication that it is man‘s best friend; krasnyj bogatyr‘ , partly from its redness when engorged, partly from its being a warrior or weapon-wielder; kolokol‘čik, perhaps because it swings from side to side when flaccid, like a clapper inside a bell (cf. English slang dingaling and ding-dong in the same sense – Green: 32, and also belltop in the sense ‗a membrum virile unusually large-headed‘ – Partridge: s.v. bell-top); torčilo and torčun, because it is often erect; xobbi, perhaps because it is men‘s all-consuming enthusiasm; and šeršavyj, as in zagnat‘ šeršavogo pod kožu ‗have, take (sexually)‘, from the sense of something rough and painful – cf. Puškin‘s simile ‗Точно еловая шишка в жопе – вошла хорошо, а выйти так и шершаво‘ (letter to Vjazemskij dated 5 November 1830, from Boldino to Moscow). Some terms for the penis are much less self-explanatory than those examined so far. It is not obvious to which other senses vaučer might be linked: a token for a meal in a refectory (student slang), obviously related to English voucher, or an unreliable partner who vouches for things but cannot be trusted (see Mokienko and Nikitina 2000: 90). If the negative aspect is the main link, the same is probably true of šnjaga, which can denote a low-quality narcotic, deception, or an unpleasant or low-quality thing. Perhaps there is a similar negative view of the penis, although ironically expressed, in the euphemistic slang konsensus. The debt of much Russian youth slang to English is everywhere apparent, not least in words for the penis, such as karlson, rejting (also ‗sexual potency‘, which affects a man‘s rating: ‗С таким рейтингом жениться нельзя‘, Nikitina: 381) and sponsor, which may be related to some of its other slang senses, e.g. ‗a wealthy lover who keeps a woman‘ and ‗condom‘.

11. Prostitute One of the common euphemistic slang terms for a prostitute is babočka nočnaja, presumably a pun on the senses ‗moth‘ and ‗lady of the night‘. Other frequent euphemisms include the following: bljaxa for the taboo word bljad‘ ‗whore‘; tutka, a shortening of prostitutka; lëgkaja devočka, a young woman of easy virtue (lëgkogo povedenija); obščaja devčonka, a promiscuous young woman who has sex with anyone in her peer group; žvačka, literally ‗well-chewed‘, a girl who lost her virginity early and leads a loose style of life. Sometimes an attribute of the prostitute may be emphasized by synecdoche (a part for the whole), e.g. bublik, literally a thick ring-shaped bread roll, dyrka, skvažina, ščel‘, all

A Contribution to the Study of Russian Euphemistic Youth Slang

167

doubtless referring to her genitalia, as also do loxanka and močalka: ‗Пытался подклеить ―мочалку‖‘, Nikitina: 329. Just as a number of slang words for the female genitals, including some of those just mentioned, can denote simply ‗girl‘ and are also used to refer to prostitutes, so some words that simply refer to females may be used with the meaning ‗prostitute‘, such as biksa, čiksa, tëlka (literally ‗heifer‘: ‗Вчера запилил тëлку‘, Nikitina: 441), tëtka, šmara (and šmaren‘ – cf. the slang words čmara and čmaren‘ ‗girl‘), and šmon‘, especially the diminutive šmon‘ka. Here belongs kljuška ‗girl, prostitute‘, perhaps from kljuxa, a person who sympathizes with hippies (< KLub LJUbitelej XIppi): ‗Вон две клюшки. Подкалываем и снимаем‘, Nikitina: 189. So does ryba-xarius ‗prostitute‘ in a way, since ryba can mean ‗girl‘, while xarius can denote a woman as sex object (a pun on slang xarja ‗mug, face‘ and xarit‘ ‗have sexually‘). From baton in the slang sense of a young woman, often of easy virtue, comes pobatonit‘sja ‗spend time pleasantly having sex‘: ‗Приходи ко мне сегодня, побатонимся‘, Nikitina: 326. The various types of prostitute include, for example, the following: valjutnaja prostitutka, who deals only in foreign currency (valjuta), especially dollars, thus often being called baksil‘naja (< American English bucks = dollars); valjutnaja devočka (devuška), valjutnica or valjutčica (< valjuta); interbaba, interdevočka or interša, probably related to internacional‘nye turisty ‗international tourists‘, her main clients (‗И шикарные интерши и разномастные совпроститутки рано или поздно за решëтку попадают‘, Nikitina: 162); šubka, possibly because she can afford to wear a fur coat, though there might be a pun on šljupka, itself perhaps euphemistic for the vulgarism šljuxa (‗Выделив специализацию валютчиц мягким [...] жаргонизмом ―шубки‖, подростки для всех других проституток избрали слова по-хлеще‘, Nikitina: 534); putana, a high-class prostitute generally accepting only foreign currency < French putain ‗prostitute‘ (see Mokienko and Nikitina 2000: 491). In contrast the sovprostitutka is one working for roubles with Soviet clients, an inferior type of activity: ‗Путаны относятся к первой категории, высшему разряду, вторая категория – совпроститутки‘, Nikitina: 413-414. Other types include the following: vokzal‘naja (prostitutka) who works at the main railway stations and is otherwise known as banovaja (< German Bahn ‗station‘) or banovaja gerla (gerlá from English girl – cf. youth slang striten-gërl ‗prostitute‘, borrowed from English street-girl); šalašovka, a prostitute who especially likes to work out of town, at car parks or camping sites (šalaš is a type of tent, particularly one made of branches and straw):1 ‗Зачем ты эту шалашовку притащил?‘ (Nikitina: 520); the mnogostanočnica, who services several clients at the same time (it can also mean a woman who has many lovers < stanok ‗bed for sex, prostitute‘); and dal‘nebojščica or plečevaja, the type of prostitute serving dal‘nebojščiki, long-distance lorry drivers. Plečevaja is explained by the following quotation: ‗Оказывается, на трассе Минск–Москва определëнная часть дороги разделена на так называемые плечи. Так вот [...] ―ответственные‖ за секс-обслуживание того или иного ―плеча‖ есть ―плечевые‖‘, Nikitina: 325. Snimat‘sja can mean to attempt to initiate sexual contact with a man or engage in prostitution (cf. snimat‘ ‗pull, pick up‘ in section 3): ‗Хочешь еë заклеить? – Да она сама снимается‘, Nikitina: 411. A similar sense is given by xomutat‘ < xomut ‗horse‘s collar‘, the implication here being that the client is ‗collared‘: ‗У колонны стоит хорошо поддавший ―штатник‖ [= American] и его шикарно хомутают две интердевочки‘, Nikitina: 501. A woman given to this sort of activity, particularly one working near stalls and kiosks, is called larjušnica (cf. lar‘, larëk ‗stall‘, larëčnik ‗stall holder‘): ‗С первого курса ларюшницы за косметику снимаются‘, Nikitina: 224. A place

168

Brian Cooper

where many prostitutes gather, e.g. near hotels, is called seksodrom; for instance, seksodrom ‗Bajkonur‘ is a jocular slang sobriquet for the Hotel ‗Kosmos‘ in Moscow. Some youth slang words for a prostitute stress the sense of being a receptacle, e.g. pepel‘nica (particularly obvious when it is recalled that sigaretka can mean ‗penis‘) and sberkassa, which denotes an old and experienced prostitute. Others dwell on different negative aspects, such as švabra, which though it can denote just a female, usually implies an untidy or slovenly one, rather like English scrubber; indeed the word literally means ‗swab, mop‘ and is ultimately from Dutch swabber in the same sense (as is English swab), though perhaps via German Schwabber. Likewise, although xor‘ can denote just a girl, the diminutive xorëk in youth slang commonly denotes a prostitute, especially a young, cheap one; perhaps there is a suggestion of the slattern present, since xorëk in standard Russian refers to the polecat, especially the common or European polecat, also called the foul marten because of its odour. Not for nothing is its Latin name Mustela (or Putorius) putorius, since this last Latin word is connected with putēre ‗to stink‘, as is the French for polecat, putois (cf. also putain and pute, which both mean ‗whore‘ in French). One wonders also whether xor‘ in this sense is meant to be a pun on English whore. Tovar, although usually meaning just ‗woman, girl‘ in slang usage, obviously stresses that she is for sale, and this is especially clear in the phrase živoj tovar ‗prostitute‘. She is, as it were, just flesh, the sense underlying škura, škurëxa, which is parallel to Latin scortum, meaning both ‗skin‘ and ‗prostitute‘. She is just a grelka (vo ves‘ rost), i.e. someone who warms up her client. Moreover, she is deceptive, a zvezdataja, someone who zvezdit (< zvezdet‘ ‗deceive‘). Finally, she is flighty and frivolous, a čita, apparently named from the monkey-friend of Tarzan, Cheetah, in the popular novels of Edgar Rice Burroughs (Mokienko and Nikitina 2003: s.v. čita) – cf. the monkey at the beginning of Lev Tolstoj‘s story Prižok: ‗Посреди народа вертелась большая обезьяна и забавляла всех‘.

12. Virgin and Vulva/Vagina A number of youth slang terms for a virgin are euphemisms of the vulgarism celka, which is a bit derogatory: ‗Да оставь ты эту целку в покое‘, Nikitina: 506. It can also denote an intact hymen and is obviously derived from the adjective celyj ‗whole‘ (as in celomudrennyj ‗chaste‘). It would seem to be an example of the widespread Russian wordbuilding technique of converting an adjective + noun combination into a single noun based on the adjective with the suffix -ka, so that celaja devuška (often shortened to celaja ‗virgin‘, an adjective used as a noun) and celaja devstvennaja pleva ‗intact hymen‘ are converted to celka: ‗А если, сука, говорю, целку порвëшь, – убью!‘ (cited from Korotkov‘s Sedoj in Kveselevič: s.v. celka). Conceivably the latter sense may have been the original, from which the former developed by synecdoche (whole for the part). Euphemisms of the word include vitamin C (витамин Ц), which is rather reminiscent of vitamin X in section 3 on copulation, and Klara Celkin. Mention was made of another term for a virgin in the section on copulation: cilindr neobkatannyj, i.e. a female (cilindr) who, as it were, has not had a roll (in the hay). There may even be a play on the word cel-ka in cil-indr. Perhaps a similar notion to that implied in celka may underlie the slang term zapečatannaja, literally ‗sealed‘, since a virgin remains, so to speak, sealed until the seal (hymen) is broken.

A Contribution to the Study of Russian Euphemistic Youth Slang

169

Several words for the female genitalia have already been noted in section 3, specifically balalajka and gitara (whence igra na gitare ‗copulation‘). Another such term is loxmatka, clearly derived with the suffix -ka from the adjective + noun phrase loxmatyj sejf ‗female genitals‘ (literally ‗hairy safe‘, with reference to the pubic hair), which originated, like numerous youth slang terms, among the criminal fraternity; hence the expression vzlomat‘ loxmatyj sejf ‗rape (a woman)‘. Similar terms for the female genitalia, with the implication of a place where something of value is stored, are kassa and kopilka. Not far removed from the same type of metaphor is bjudžet, referring in slang use particularly to a prostitute‘s sexual organs, perhaps because she employs them to balance her budget. The metaphor of property appears to lie behind the use of fazenda for the female sex organs; its original sense was a house, especially a dacha with a plot of land, presumably from Portuguese fazenda ‗estate, property‘ < Latin facienda ‗things to be done‘ (the origin also of Spanish hacienda). The use of televizor to denote the female genitalia has been mentioned in section 8 on menstruation.

13. Backside and Anal Sex Youth slang words for the backside include the borrowed English term bèksajd and indeed bèk (the back side as opposed to the front side); the use of foreign words as euphemisms for taboo terms is well established, and both of these may take the place of žopa ‗arse‘ as appropriate: ‗Иди ты в бэксайд со своей попсой‘, ‗Моя бабка [...] говорит мне всегда ―нажрался в жопу‖, а ты ―удринчался2 в бэк‖‘, Nikitina: 51. Another English word, although this time invented, is used in the same way: antifejs, literally anti + face = facing away, i.e. the opposite to the front. Sometimes euphemisms are based on similar-sounding words, e.g. annaly ‗backside (usually fat)‘, perhaps inspired by anal‘nyj and anus. Yet some euphemisms can become infected by the taboo word that they replace, e.g. verzoxa < verzat‘ ‗defecate‘ (both words labelled ‗rude‘ by Shlyakhov and Adler: 311): ‗Очень просто: верзоха – задница, а верзать – ходить в туалет по-большому‘, Nikitina: 57. Verzat‘ may be linked etymologically with the verbs otverzat‘ ‗open‘ and razverzat‘ ‗open wide‘. The use of korma in relation to the backside is obviously due to its denoting the stern of a ship, while televizor, as well as denoting the female genitals, can also denote the backside, possibly implying that a lot of people watch it. Leninskie mesta can refer to the backside and the genitals, while stanok can refer specifically to a woman‘s rear. Last of all should be mentioned gaubica, its sense ‗backside‘ being an extension of the literal sense ‗howitzer‘ (< German Haubitze) because of the noise made by the backside when breaking wind: ‗Что ты ко мне своей гаубицей-то повернулся, ведь ты со мной разговариваешь, а не с унитазом‘, Nikitina: 79. The colloquial use of pukat‘ to mean ‗break wind‘ explains the slang puk zabludivšijsja (literally ‗fart that has lost its way‘) in the sense ‗belch‘. The same sense is conveyed in slang by some idioms with duša, notably duša s Bogom govorit and (privet) iz glubiny duši: ‗Ох, из глубины души. Хорошо поели‘, Nikitina: 82. The use of popengagen (a wordplay on Copenhagen and the euphemism popa ‗backside‘ for žopa ‗arse‘) in referring to anal sex has been noted above in section 9 on oral sex and orgasm; thus exat‘ v popengagen would mean ‗have anal sex‘: ‗Мы на ней ехали в попенгаген‘, Nikitina: 340. The implication of anal sex followed by oral sex is found in the term šokoladka, which for obvious reasons can be seen as analogous to English slang get some brown (sugar) ‗have anal sex‘: ‗Знаешь, что такое шоколадка? Это когда ―в

170

Brian Cooper

кукушку‖, а потом в рот‘, Nikitina : 532. Hence the phrase stučat‘ v šokoladku means ‗sodomize‘. The term v kukušku here also refers to anal intercourse; this slang use derives from kukuška in the colloquial sense ‗small steam locomotive‘, so that dat‘ [komu] v kukušku denotes ‗have anal sex [with someone]‘. An expression synonymous with this is dat‘ v sosedku, in which sosedka perhaps implies ‗neighbouring orifice‘.

14. Lavatorial Matters and Vomiting The use of verzat‘ in the sense ‗defecate‘ was mentioned in the previous section. Another slang word with the same meaning is xezat‘ < Greek khézō ‗defecate‘, and there are a number of phrases with this sense, such as bombu brosat‘/sbrosit‘, with bomba meaning ‗faeces‘ (mina is used in the same sense) – cf. ‗Опять бомба мимо унитаза‘, Nikitina: 40; kizjak metat‘, in which kizjak is literally ‗pressed dung for fuel‘; ličinku otložit‘, literally ‗deposit a maggot/grub‘: ‗Опять кто-то личинку отложил мимо кассы [мимо унитаза]‘, Nikitina: 229. It is noteworthy here that kassa is used to mean ‗lavatory pan‘, a slang sense that also belongs to sberkassa, literally a savings bank, presumably because material is deposited in it. Other slang words with the same meaning are tolčok and očko, both of which can also denote WC: ‗На толчок заполз после урока‘, ‗Захожу утром в сортир, а он, бедняга, на очке сидит и спит‘, Nikitina: 444, 306. Očko is evidently used to mean ‗lavatory pan‘ because there is a hole in it; not surprisingly, therefore, the same word can refer to the anus: ‗У меня, честно говоря, очко сыграло [= я сильно испугался], даже понос напал‘, Nikitina: 306. It is conceivably this same hint at the hole in the pan, like a zero, that explains the use of nulí (plural of nul‘ ‗zero‘) to mean WC, where a number of lavatory pans might be expected to be found: ‗Завалили в ―дабл‖ [= WC] – ―нули‖ по-нашему‘, Nikitina: 281. Incidentally, as a sort of in-joke, singl is used in youth slang with the same sense as dabl. The slang word dabl seen in the last quotation is for English double u, i.e. the letter w, and is thus used for WC (water closet): ‗Где тут дабл поблизости?‘ (Nikitina: 97). For this reason soveršit‘ dabloposeščenie and prodablit‘sja both mean ‗go to the toilet‘: ‗Пошëл в ―Гостинке‖ продаблился – уже 5 рублей‘, Nikitina: 359. The many expressions used with this same meaning include the following: sxodit‘ k Vitaliku and pozvonit‘ Vitaliku, the name Vitalik here being tantamount to ‗lavatory‘ (because it is so vital?); sxodit‘ k Džonsonu, peredavat‘ privet Džonsonu and pozvonit‘ Džonu, the choice of name probably hinting at American English the john, i.e. the lavatory; pojti posmotret‘ na Lunu and pojti k Ivan Ivanoviču (Russian for ‗go to John Johnson‘); privjazat‘ (perevjazat‘) kobylu (konja, konej); smenit‘ vodu v akvariume; pozvonit‘ po meždugorodnomu telefonu; pozvonit‘ v Pariž; pozvonit‘ Slaviku and pozvonit‘ šefu. The last phrase implies either ‗go to the loo‘ or specifically ‗go for a pee‘, while the variant prinjat‘ ot šefa faks implies a visit for defecation: ‗Пойду приму от шефа факс, а то время не терпит‘, Nikitina: 524. The key word here may be faks, which is rather suggestive of Latin faex, the singular of faeces. Other words for the WC include two abbreviations previously established in other senses, KPD (Komnata Poleznogo Dejstvija, not Koèfficient Poleznogo Dejstvija) and KPZ (Komnata Prijatnyx Zapaxov, not Kamera Predvaritel‘nogo Zaključenija). There is also rudnik, with the literal sense ‗mine, pit‘; severok, specifically a public toilet in a café or restaurant, whereas kadrovyj severok denotes a lavatory where homosexuals meet; and romaška, literally ‗chamomile‘, which has been mentioned in section 6 on group sex, but in a school

A Contribution to the Study of Russian Euphemistic Youth Slang

171

environment means also the lavatory (ironically, owing to the unpleasant smell): ‗Ну и запашок! Вашу ромашку давно пора в анашку переименовать‘, Nikitina: 386. Here anaška seems to hint at the smell of cannabis in the WC, being a conflation of anaša ‗marijuana‘ and romaška. Finally there are some expressions that suggest purely urination, which include smyt‘ uliki, literally ‗wash away the evidence‘, and fanty nacedit‘, literally ‗strain off a certain amount of Fanta‘ (Fanta being a popular soft drink) – cf. in the last (thirtieth) chapter of Viktor Docenko‘s Komanda Bešenogo: ‗Он […] начал жевать, запивая тепловатой ―фантой‖‘, Docenko: 400. As regards vomiting into the lavatory pan, a common youth slang expression for this is zvat‘ (vyzyvat‘) Ixtiandra; in fact, mention of Ixtiandr hints at any stomach upset and occurs in other phrases, such as igrat‘ v Ixtiandra ‗suffer vomiting‘: ‗У тебя от этих столовских котлет ихтиандра не будет?‘, ‗У неë своя метода была: […] в туалет – Ихтиандра вызывать. Так и похудела‘, Nikitina: 163-164. Perhaps the choice of name is meant to be suggestive of the sound of vomiting. A similar sense is achieved by using the name (Doctor) Watson, presumably as in the Sherlock Holmes stories of Arthur Conan Doyle, in the phrase vyzyvat‘ doktora Vatsona: ‗Общались ресторанным способом – весь вечер доктора Ватсона вызывали по очереди‘, Nikitina: 55. This name too may be echoic, rather like the English slang verb barf, but it may also suggest the slang noun vater ‗lavatory‘. In certain expressions with the vomiting sense there is a pun of some sort on the verb izrygat‘ ‗vomit‘, notably v Rigu poexat‘ and ariju Rigoletto (Rygoletto) ispolnit‘: ‗―Вызывать Ватсона‖ и ―поехать в Ригу‖ – это всë равно‘, Nikitina: 55. Use of the opera name alone or in combination with arija is sufficient to imply vomiting: ‗Дома попыталась что-то съесть, но со мной случилось ―риголетто‖‘, Nikitina: 383. So is use of the city name Riga (cf. mne xočetsja v Rigu, Cooper 1993:71). Other terms for vomiting include the following: podrat‘sja s tolčkom, literally ‗fight with the lavatory pan‘, over which the person involved will be bending; xarči kidat‘ (metat‘), which is analogous to English ‗throw up‘; and pugat‘ belogo druga, which may perhaps be another reference to the (white) lavatory pan. Finally there is the phrase otdavat‘ saljut: ‗Серëга весь вечер отдавал салют, а на утро есть ничего не мог‘, Nikitina: 393. This may possibly have been inspired by the sort of two-fingered salute referred to by Dal‘ (1912-14: s.v. rvat‘) in explaining the phrase sdelat‘ franc-xeraus ‗vomit‘: ‗Русский солдат с австрийским выпили за здоровье Франца, а когда второй не захотел пить за здравие нашего государя, тот первый скомандовал: Франц хераус! [= Franz heraus! ‗(Emperor) Franz out!‘] и заложил два пальца в глотку‘.

Conclusion In conclusion, it may seem that slang is mainly a young people‘s phenomenon. Like all developments in language it is notoriously difficult to control, either by prescription or by proscription; it is a favourite occupation of the young to play with words and rename things, whether by devising new words or by distorting or misusing old ones in search of novelty. Others then copy them to follow fashion. Many slang words and phrases die, but a few become established; and it is perhaps the latter group, if no other, that particularly justifies the study of slang as one of many sources by which a living language enriches itself and continues to grow and develop. It will be clear from this study that slang is concerned to an

172

Brian Cooper

appreciable extent with sex and scatology. It is often irreverent, sometimes obscene, with vocabulary emanating essentially from a conflict with the values of the dominant culture. Hence perhaps its appeal to young people in their rebellion against the conventions of the day. Before becoming part of the dominant culture, a new slang word or expression is normally widespread in a subculture, which is an especially fruitful source of slang. As conditions change, a piece of slang may change its meaning, go on being used as slang among sections of the population, or be adopted in the standard language, which is invaded gradually as words leak out of a subculture, e.g. from slang into colloquial usage and even into the written word. Some remain dormant for a long time in the dominant culture, while others are picked up and used, in many cases as fads which do not last. Some, however, survive and tend in time to become part of the standard language. Instances of this, which need not be seen as a wholly negative trend (Bauche: 27, 29), may be found among those studied. It must, however, be concluded that, even in the case of euphemistic slang words, there is always the possibility that the euphemized substitute may itself be contaminated by association with the taboo word and come to be considered in some way unacceptable or not quite decent and therefore to be avoided. It has been pointed out that ‗euphemism is a will-o‘the-wisp for ever eluding pursuit; each new word in turn becomes as explicit as its predecessors, and has to be replaced‘ (Fowler: s.v. euphemism). In the same way Murphy recognizes that ‗a euphemism may be semantically too pejorative to function for long in a positive way‘ (Murphy: 8). Thus the wheel may come full circle: after moving further and further away from the original starkness of a supposedly offensive word in an attempt to euphemize it, it is possible to be faced with an objectionable substitute. Clearly it is hard for euphemisms to preserve the delicate balance between complete neutrality and excessive explicitness. If they become free of the overtones which are part of their semantičeskaja nagruzka, they risk losing their real implication, and if they do not they risk becoming tainted. At all events, it can be concluded that verbal taboo is a powerful force for lexical change, whether by enrichment or by impoverishment of the lexis (Kany, reviewed by Widlak: 98). It can lead to the introduction of new slang usages into a language or, through euphemism, give a different nuance to a word or change the original sense of the word completely (Crnek: 21). Sometimes one sinister meaning tends to crowd out other senses, as in English intercourse and Russian dročit‘, and sometimes the sinister and other meanings coexist, as in English rubber (Steadman: 102-103) – cf. Russian rezin(k)a. Taboos of decency, which are particularly apposite in this study, do not seem especially prone to causing obsolescence of the taboo word (Bloomfield: 401), but they do tend to give rise to numerous slang variants, which can sometimes, though not necessarily deliberately, have a sort of euphemizing function. These variants may from time to time be accepted into some registers of the standard language, but often these substitutes are as offensive as the taboo word itself (Steadman: 100).

Notes 1. However, this link may only be secondary, by association with the saying ‗с милым рай и в шалаше‘ (Gračev and Mokienko: 183), and šalašovka may be etymologically linked rather with the demotic word šalava, derived from the

A Contribution to the Study of Russian Euphemistic Youth Slang

173

southern dialect word of common gender šalava < šalavyj = šal‘noj ‗silly person, halfwit‘ and according to Dal‘ also ‗slatternly woman‘ and ‗scruffy dog‘. 2. The youth slang udrinčat‘sja, like its synonym udrinkat‘sja, means napit‘sja dop‘jana (Nikitina: 462), doubtless from English drink – cf. the slang verb užrat‘sja in the same sense, and the many slang variants of English ‗drink‘: drink, drin‘k, drynk, drinč (nouns) and drinkanut‘, drinkat‘, drin‘kat‘, drinknut‘, drinčat‘ and drinčit‘ (verbs), as well as various derivatives, e.g. drinker, drinkač ‗drinker‘ (see Nikitina: 115).

References Bauche, H. Le langage populaire, Paris, 1920. Bloomfield, L. Language, New York, 1933. Budagov, R. A. ‗Javlenie tabu, èvfemizmy i ix funkcii‘, Vvedenie v nauku o jazyke, Moscow, 1958, 93-98. Cooper B. ‗Euphemism and taboo of language (with particular reference to Russian)‘, Australian Slavonic and East European Studies, 7 (2), 1993, 61-84. -------------- ‗All fingers and thumbs…‘, Australian Slavonic and East European Studies, 12 (1), 1998, 1-26. -------------- ‗The classic ―mother‖ phrase in Russian: its form and origin‘, Australian Slavonic and East European Studies, 17 (1/2), 2003, 3-26. Crnek, F. ‗Ze studiów nad eufemizmami w językach słowiańskich‘, Sprawozdania Towarzystwa Naukowego we Lwowie, 7 (1), 1927, 15-21. Dal‘, V. I. Tolkovyj slovar‘ živogo velikorusskogo jayzka, 4th edn, Moscow/St Petersburg, 1912-14. Docenko, V. Komanda Bešenogo, Moscow, 1999. Fowler, H. W. A Dictionary of Modern English Usage, 2nd edn, Oxford, 1965, reprinted 1968 with corrections. Gračev M. A., Mokienko, V. M. Istoriko-ètimologičeskij slovar‘ vorovskogo žargona, St Petersburg, 2000. Green, J. The Slang Thesaurus, London 1986, reprinted 1988. Kany, C. A. American-Spanish Euphemism, Berkeley, 1960, reviewed by S. Widlak, Kwartalnik Neofilologiczny, 12 (1), 1965, 97-100. Kveselevič, D. I. Russko-anglijskij slovar‘ nenormativnoj leksiki, Moscow, 2002. Mokienko, V. M., Nikitina, T. G. Bol‘šoj slovar‘ russkogo žargona, St Petersburg, 2000. ----------------- Slovar‘ russkoj brani, St Petersburg, 2003. Murphy, M. G. ‗Euphemism in Russian‘, Journal of Russian Studies, No. 48, Leeds, December, 1984, 3-8. Niceforo, A. ‗La magie des mots‘, Le génie de l‘argot. Essai sur les langages spéciaux, les argots et les parlers magiques, Paris, 1912, 201-277. Nikitina, T. G. Tak govorit molodëž‘: slovar‘ molodëžnogo slenga, 2nd edn, St Petersburg, 1998. Partridge, E. A Dictionary of Slang and Unconventional English, 8th edn, ed. Paul Beale, London, 2000.

174

Brian Cooper

Shlyakhov, V., Adler, E. Dictionary of Russian Slang and Colloquial Expressions, 2nd edn, New York, 1999. Steadman, J. M. ‗A study of verbal taboo‘, American Speech, 10, 1935, 93-103. Vasmer, M. Russisches etymologisches Wörterbuch, 2. unveränderte Auflage, Heidelberg, 1976-80. Widlak, S. see Kany, C. A. Department of Slavonic Studies University of Cambridge Sidgwick Avenue Cambridge CB3 9DA

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 6

WHAT DO WE LEARN ON LANGUAGE ACQUISITION FROM WILLIAMS SYNDROME? Agnès Lacroix1, Vesna Stojanovik2 and Ágnes Lukács3,4 1

Centre de Recherches sur la Cognition et l‘Apprentissage (UMR CNRS 6234), Poitiers, France 2 School of Psychology and Clinical Language Sciences, Reading, UK 3 HAS – Budapest University of Technology, Cognitive Science Research Group, Budapest, Hungary 4 Research Institute of Linguistics, Hungarian Academy of Sciences, Budapest, Hungary

Abstract Williams syndrome (WS) is a rare genetic disorder involving the deletion of approximately 27 genes on the locus 7q11.23. Work over the last twenty years has focused on delineating the distinctive neurocognitive profile that characterizes this group. The first studies identified a dissociation between cognition and language demonstrating that the verbal performance of individuals with WS was markedly better than their performance in non-verbal areas. More recent studies have shown dissociations within language (e.g., relatively good phonological memory, impaired grammatical morphology) and within non-linguistic visuo-spatial skills (good face recognition, impaired block design). At the same time with data from larger groups of individuals with WS, it has become apparent that there is heterogeneity in performance within the WS group across cognitive domains, languages and cultures. By looking at heterogeneity in language in WS from different language communities and approaching the phenomenon from multiple perspectives, we can begin to identify factors mediating their performance. Thus, we can have a better understanding of underlying processes involved in the language profile in a disorder such as WS and of the role of external and internal factors on language acquisition and development.

176

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

Introduction Williams syndrome (WS) is a rare genetic disorder (1 out of every 7 500 births) caused by a microdeletion on the long arm of chromosome 7 (7q11.23) that results in the loss of about 28 genes (Siegmüller and Bartke, 2004; Bellugi, Lichtenberger, Jones, Lai, and St George, 2000; Karmiloff-Smith, 1998). Individuals with WS show a specific neuropsychological profile which highlights dissociations in presence of a moderate mental retardation. Language seems to be relatively well-preserved while visuo-spatial skills are more impaired (Karmiloff-Smith, Thomas, Annaz, and al., 2004; Pezzini, Vicari, Volterra, Milani, and Ossella, 1999; Klein and Mervis, 1999; Bertrand, Mervis, and Eisenberg, 1997). People with WS have also been referred to as being hypersocial because of their overfriendly behaviour towards strangers. Theoretically, this uneven profile has led some researchers to argue for the cognitive modularity of the brain and mind and for the innate specification of the modules in a Fodorian sense (Fodor, 1983). In Nature, Piatelli-Palmarini (2001) argues that language is intact in people with WS supporting the idea of a modular conception of the cognitive system. However this view has been criticized because it does not take into account the process of development. An alternative view is that language in WS is surprisingly good regarding their general cognitive level, but this faculty is partially preserved. This argues in favor of a « gradual modularization» (Karmiloff-Smith, 1992): modules of language and cognition interact during development to become independent at adult-state. In this chapter, we will discuss two current questions on language in Williams‘s syndrome: (1) What are the ―real‖ language abilities of individuals with WS? and (2) How do these language skills develop?

1. Language Abilities of Individuals with WS Morpho-syntax in Williams syndrome - Research into the morpho-syntactic abilities in individuals with WS has been the most influential because of its direct applicability to the theoretical debates about language acquisition, namely the question of modularity. Investigating the morpho-syntactic abilities of a population which seems to show intact syntactic abilities in the face of severe cognitive deficits has been seen as a powerful way of providing evidence that that syntactic processing is autonomous and domain specific and therefore not affected by deficiencies in the general cognitive system. Early studies on language in Williams syndrome pointed to ‗intact‘ language with regard to morpho-syntactic abilities. Pioneering work by Bellugi and colleagues argued that despite severe cognitive impairments, individuals with WS have superior syntactic abilities (Bellugi and al., 1988; Bellugi and al., 1992; Bellugi and al., 1994). The morpho-syntactic skills of people with WS have been described as ‗perfect‘ by some (Bickerton, 1997, Piatelli-Palmarini, 2001; Pinker, 1999). A number of more recent reports have also indicated that individuals with WS show enhanced grammatical ability compared with lexical ability and better performance in grammar over lexical semantics (Clahsen and Almazan, 1998; 2001; Ring and Clahsen, 2005). These studies have argued that WS offers evidence for dissociations within the linguistic system into a computational component (concerned with rule governed operations involved in passive constructions, formation of past tense in English, binding) and a lexical component, or what has been known as ‗internal

What Do We Learn on Language Acquisition from Williams Syndrome?

177

modularity‘ (Levy, 1996). In their studies, Clahsen and colleagues showed that individuals with Ws perform better with regular grammatical inflections compared to irregular, which involved retrieving items from the lexicon. The regular/irregular issue has attracted a lot of interest and debate. A number of studies have shown that individuals with WS perform better on regulars than irregulars (Bromberg, Ullman, Marcus, Kelly and Levine, 1995; Clahsen and al., 2004; Pléh, Lukács and Racsmány, 2003) and that they may be significantly impaired on irregulars compared to controls but not on regulars (Clahsen and Almazan, 1998; Penke and Krause, 2004; Zukovski, 2004). No study has yet reported that individuals with WS outperform mental-age controls, either on regulars or irregulars. Furthermore, the results of studies which have employed larger number of participants (such as Thomas and al., 2001; Zukovski, 2001) show no interaction between group and regularity. As pointed out by Brock (2007), all the studies which have investigated regular versus irregular inflections in WS suffer from ‗ceiling‘ effects in that most of the participants in all the studies perform at ceiling on regulars, which makes it impossible for any group differences on irregulars to be detected and studies should employ much younger participants to avoid ceiling effects and to address this issue more effectively. Three studies have investigated whether individuals with WS are able to apply morphological rules to novel words. Clahsen and Almazan (1998) and Clahsen et al. (2004) found that children with WS were more successful when they had to apply a morphological ending to novel verbs which rhymed with existing irregulars; however Thomas et al. (2001) were unable to replicate these results. Thomas and colleagues reported that individuals with WS tended to omit morphological endings in novel verbs rather than over-regularising. Zukovski (2004) did not find significant group differences in individuals with WS‘s performance to morphologically mark novel nouns. A large body of research has provided evidence for impaired linguistic processing in individuals with WS as well as poor morpho-syntax (Joffe and Varlokosta, in press; Karmiloff-Smith, Tyler, Voice, Sims, Udwin, Howlin and Davies, 1998; Karmiloff-Smith, Brown, Grice and Paterson, 2003; Mervis and Klein-Tasman, 2000; Thomas, Grant, Barham, Gsödl, Laing, Lakusta, Tyler, Grice, Paterson and Karmiloff-Smith, 2001; Stojanovik, Perkins and Howard, 2001; 2004; Volterra, Caselli, Capirci, Tonucci, and Vicari, 2003). Stojanovik et al. (2004) compared the performance of five participants with WS (mean age: 9; 2 years) and five with Specific Language Impairment (mean age: 10; 0 years) on a range of receptive and expressive standardised verbal measures including measures of morphosyntax, and found similar performance across the two groups on all the measures. The children with WS were completely indistinguishable on measures of morpho-syntax from children with clinically diagnosed language impairment. Joffe and Varlokosta (2007) investigated the syntactic abilities of ten individuals with Williams Syndrome (WS) (mean chronological age: 8; 9 years; mean mental age: 4; 8 years) and ten individuals with Down‘s Syndrome (DS) (mean chronological age: 8; 7 years; mean mental age: 4; 6 years), matched individually on chronological age, mental age and performance IQ. Participants with WS and DS performed similarly on the standardised measure of grammatical ability, as well as on the experimental tasks that tapped comprehension of passives, and production and comprehension of wh-questions. Participants with DS performed significantly more poorly than both the WS cohort and typically developing (TD) controls on the repetition of wh-questions. Both the WS and DS cohorts

178

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

performed significantly more poorly on most of the syntactic tasks compared to the younger controls with the same mental age. Another recent study by Perovic and Wexler (2007) also reported deficits with complex syntactic constructions, i.e. binding and raising in two groups of children with WS (a group of 6-12 years old and a group of 12-16 years old), compared to typically developing controls matched on non-verbal mental age (MA), verbal MA, and grammar. The younger of children with WS showed a poorer performance than the older children with WS on the interpretation of personal pronouns in binding constructions, but their performance was in line with the two groups of younger TD controls. However, both groups of children with WS performed poorly on raising constructions (It seems that Paul will marry Jane) and the authors concluded that more complex aspects of grammar, those known to mature last in typically developing children, seem to be unachievable for people with Williams syndrome. Studies which have investigated morpho-syntactic abilities in older individuals with WS (adolescents and adults) also show that they perform below their vocabulary age and chronological age on a standardised test of grammar (Karmiloff-Smith, Grant, Berthoud, Davis, Howlin and Udwin, 1997). In a sentence imitation task, which required the participants to repeat relative clauses, adolescents and adults with WS performed at an age equivalent level of children aged five (Grant, Valian and Karmiloff-Smith, 2002). Lexical abilities in WS - Although WS performance generally exceeds mental age on standardized receptive vocabulary tests, it usually lags behind chronological age level and even behind MA when participants have to point to a target among several semantic distractors (Temple et al. 2002). On naming tasks, participants with WS tend to be anomic, and perform below MA level. (Temple and al., 2002; Volterra and al., 1996). Several findings suggest that besides being smaller in size, the organization of the WS lexicon can be atypical too. As early observations pointed out, individuals with WS tend to use unusual and sometimes contextually inappropriate words (Bellugi and al., 1988). Some studies of lexical skills in WS lead to the claim that frequency is not as strong a factor in organizing the WS lexicon as in TD children and individuals with WS access infrequent words just as easily as frequent ones. The bulk of results come from semantic fluency tasks, in which participants have to list as many members belonging to a given category as they can within a certain time limit. In some studies, individuals with WS produced more items of low frequency words than TD controls (Bellugi and al. 1994, Temple and al. 2002), but other studies failed to find such differences (Jarrold and al. 2000; Volterra and al. 1996, Stojanovik and van Ewijk, 2008). Atypical frequency effects are also reported in studies of list learning (Vicari and al. 1996a; 1996b), and in a similarity judgment task, but not on a free associations task to homonyms (Rossen and al. 1996). Since the evidence for atypical frequency effects has been contradictory and controversial in the literature, also indicating that performance might be influenced by the specific category (animals, a category that is generally very familiar to children) being tested, we tested Semantic Fluency with a wider range of eight categories (also differing in their familiarity to children): Food, Animals, Drinks, Musical Instruments, Occupations, Things to read, Furniture and Clothes. We also had a Picture Naming Vocabulary Task to test frequency effects in naming (see Lukács and al., 2004 and Lukács 2005 for details). In the Semantic Fluency Task, the WS group showed typical effects of frequency that were identical in every respect to those found in the receptive vocabulary controls: they did not differ in the number, frequency and rank of the items they produced. Although WS performance lagged behind that

What Do We Learn on Language Acquisition from Williams Syndrome?

179

of receptive vocabulary controls in the naming task, frequency (and category) effects were again the same in the two groups. These findings argue against reduced frequency effects as an account of peculiar language use in WS. Other proposals to explain lexical peculiarities in WS focus on semantic integration at a sentence level or on lexical access. Since they found typical priming effects for both functionally and categorically related items, Tyler et al. (1997) suggest that WS lexical organization is normal, and semantic anomalies arise from an inability to integrate word-level information into the sentence. Temple et al. (2002) in their naming study of 4 adolescents with WS (between 11; 2 and 15; 4 years) also found atypical naming errors that never occurred in the control group: naming a part instead of the whole (lamp→lampshade, reindeer→antler) or giving answers with only a few semantic features shared between target and real responses (caterpillar→antelope; spanner→corkscrew). These authors take their WS findings to reflect a selective impairment to the mental lexicon, and interpret them as resulting from faster lexical access and looser criterions for target identification, resulting in sloppy access. Lexical organization has also been tested at a different, conceptual level. Johnson and Carey (1998) studied understanding of biological concepts like animal and death, and found that in these domains, individuals with WS (between 10; 7 and 32; 1; mean age: 24; 3) had relatively strong basic conceptual knowledge. They were able to attribute bodily properties to living creatures and to project animal properties to people, but there were differences in the deeper conceptual understanding of these notions. Participants with WS were more animistic than controls in attributing life to objects, in providing a definition of death and in species transformations. The authors conclude that knowledge enrichment (adding information to existing conceptual structures) is relatively unproblematic in WS, while conceptual change (reorganizing knowledge) is difficult, resulting in deviant and less sophisticated semantic representations. Spatial language - Studies of spatial language in WS seek an answer to the following question: in the light of the severe impairment in spatial cognition and relatively good language, how do WS people perform when they have to use language to describe spatial relationship, or have to construct spatial relations by comprehending spatial language (see Bellugi and al. 2000)? Some findings suggested that there is a selective within-language deficit of spatial terms in WS. This conclusion was supported by WS children getting especially low scores on spatial items on the TROG (on the following blocks: K (longer/bigger/taller); M (in/on) and P (above and below) (eg. Clahsen and Almazan, 1998; Jarrold and al., 2001; Phillips and al., 2004). Italian children made many preposition errors in a Sentence Repetition test, that in addition were quite unlike anything seen in typically developing children: Eg. The grandchildren pick up flowers with their grandmother The grandchildren pick up flowers *on top of the grandmother. (Volterra and al., 1996). Lichtenberger and Bellugi (1998) also found WS performance on both comprehension and production of spatial terms to lag behind younger TD children. Most TD errors were giving too general descriptions, but many WS errors were atypical: for example, to describe a scene where an apple is in a bowl, WS subjects gave answers like ‗apple without the bowl‘ and ‗the bowl is in the apple‘ and ‗the apple is around the bowl‘. Even with correct use of prepositions, they sometimes reversed figure and ground (e.g. The bowl is in the apple). They also made errors with using the opposite preposition (The apple is around the bowl) or gave completely inappropriate answers. The authors conclude that WS individuals probably have a

180

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

difficulty of mapping spatial representations onto language. Based on findings showing that WS spatial representations are impaired or underspecified, it is plausible to argue these results in inappropriate descriptions even without a problem in mapping. Examining spatial language might in fact reveal little about the language-cognition interface. It is crucial for different reasons: fine-grained scrutiny in the study of impaired spatial terms might lead to findings on the more specific organization and the structure of the spatial deficit. Landau and Zukowski (2003) took this line of research and claim that ―nonlinguistic spatial deficits shown by children with Williams syndrome have, at most, limited effects on their spatial language.‖ (p. 1). Examining spatial language in its more specific organization they elicited descriptions of 80 videotaped motion events. They checked representation and linguistic encoding of all components of spatial representation of motion events (Figure and Ground, Manner of Motion and Path). WS children could represent Figure and Ground objects, their relative spatial roles, and they could map them onto their appropriate syntactic role of subject/object of preposition. They also correctly encoded, and thus perceived manner of motion. Path seemed to be the most difficult element of the motion event for WS children. But even here, WS children tended to use largely the same set of expressions for all three path types as controls, and they made errors by using an expression from one path type to describe another (Lukács and al., 2007) found the same difference between the WS and TD group in error types, and the WS group making more path type errors). In contrast to previous observations, most of their mistakes were not using inappropriate spatial terms, but using either a vague expression (like over) or omitting the path expression altogether. This tendency was strong with FROM and VIA paths, but not with TO paths. The authors hypothesize that SOURCE paths are difficult for WS because they involve a memory component missing from GOAL and STATIC scenes. They attribute the difficulty with SOURCE paths in WS to a retention deficit referring to Vicari et al.‘s (1996) demonstration that in list learning, children with WS show normal recency effect, but no primacy effect. In another detailed study of spatial language in WS, we tested comprehension and production of spatial postpositions and suffixes, and although we found that performance of the WS group was poorer than of verbal controls, they displayed the same pattern on different types of paths and spatial relations. We also argue against a selective deficit of spatial terms within language based on two findings. First, the WS group, just like controls, performed significantly better on comprehension of spatial terms than on production of the same items, showing that they find it easier to work from language to construct a spatial arrangement, than to map a scene onto the appropriate linguistic expressions. Second, results from the sentence completion task showed that in purely language task, where participants do not have to rely on describing a real-world spatial arrangement, do not show overall differences between the WS group and their verbal controls (see Lukács and al., 2007 for details). The conclusion that difficulties with spatial language in WS only reflect their problems in spatial cognition was further supported by findings from an earlier study of our research group (Racsmány and al., 2002), where performance of the WS group on the production task was at a very similar level than that of a spatial control group matched on the block design task of the WISC-R (Wechsler 1974) and another task of learning spatial arrangements tapping long-term spatial memory. This finding is complemented by the pattern observed in the production task, in which source type expressions turned out to be most difficult. These

What Do We Learn on Language Acquisition from Williams Syndrome?

181

findings are in line with Landau and Zukowski‘s hypothesis (2003), who claim that this difficulty can be explained by a problem with retaining spatial information in memory. Pragmatic abilities in WS - Studies on the pragmatic aspects of language in WS are still rare and have mainly focused on assessment of communicative, narrative, and conversational skills. The first studies of this kind investigated these abilities using questionnaires. Mervis, Klein-Tasman, and Mastin (2001) assessed the ability of 4- to 8-year-old children with WS to adapt to nearby individuals with the Vineland Adaptive Behaviour Scales (Sparrow, Balla, and Cicchetti, 1984). Semi-structured interviews with 41 parents were conducted to consider the WS children's communication skills, daily living skills, socialization level, and motor abilities. Results indicated that the children's skills were ranked as follows: socialization (strongest point), communication, daily living skills, and motor abilities (weakest point). More recently, Laws and Bishop (2004) administered the Children's Communication Checklist (CCC) to parents, teachers, and other professionals. They were asked to fill the CCC for WS, Down's syndrome (DS), children with Specific Language Impairment (SLI), and typical populations. This checklist detects communication problems by evaluating various categories such as coherence, inappropriate initiation, use of context, nonverbal communication, and social relations. The participants were 19 individuals with WS (age from 6; 5 to 25; 2), 24 individuals with DS matched on chronological age, 17 children with SLI matched on mental age, and 31 typically children matched on mental age. Although the pragmatic scores of the three clinical groups were below those of the typical group, only the WS group had a score below the impairment cut-off point. This means that the individuals with WS had some pragmatic deficits characterized by inappropriate initiation of interactions and stereotyped conversations. Concerning narrative, Losh, Bellugi, Reilly, and Anderson (2001) and Reilly, Losh, Bellugi, and Wulfeck (2004) looked at the development of narrative skills in WS children between the ages of 5 and 13. Children with WS were compared with children with language impairment and with typically children matched on chronological age, lexical development, and nonverbal development. The participants had to tell the story entitled "Frog, Where Are You?" (Mayer, 1969). The authors were interested in finding out the structure of the children's stories, measured in terms of their ability to elaborate the story schema. The results indicated that the participants with language impairment established and maintained the theme as well as the same-chronological-age typically developing children, but the WS group had more difficulty than both the language-impaired children and the same-chronological-age typical children. In addition, the authors looked at social evaluations coded in three categories: social involvement cues (sentences and exclamations to attract the listener's attention, such as sound effects and character dialogues), references to affective states and behaviours (attribution of mental states to characters), and intensifiers (repetitions or emphatic markers). The children with WS use more social evaluations than language impaired children and the typically developing children of the same chronological age or same mental age. Lacroix, Bernicot and Reilly (2007) studied the narrative skills in twelve French-speaking children and adolescents with WS (mean age: 12; 4). They were compared to children with Down's syndrome (DS), and to typical children of the same chronological age (CA) or the same mental age (MA). Results indicated that for the number of utterances, storyschema elaboration, and story-theme reiteration, the children and adolescents with WS exhibited "intermediate" performance, that is, they did better than the DS group (same IQ and same chronological age) but not as well as the children of the same chronological age.

182

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

.Moreover, the children and adolescents with WS performed atypically for the social use of language because they used many more social evaluations than did the DS, CA, and MA groups. To study conversational abilities, Stojanovik, Perkins, and Howard (2001) and Stojanovik (2006) focused on children with WS to engage in conversation. The participants in their studies were five WS children ranging in age from 7; 6 to 12; 1, eight children with SLI matched on chronological age, and nine typically developing children matched on CA. Each child was tested five times via a semi-structured conversation about a series of photographs showing everyday scenes. Exchange structure, turn-taking behaviour, information transfer, and conversation adequacy were analysed. Exchange structure was coded in terms of initiations that called for/did not call for a response from the interlocutor, short replies, long replies, and exchange continuity; turn-taking violations were categorized as no response, violating or non-violating overlap, and interruption by the adult because the child's utterance was unclear; information transfers were coded as explicit requests for information, requests for clarification, and requests for confirmation; and conversational adequacy was measured in terms of language structure and pragmatics (misunderstanding of inferences, failure to make use of the context, too much information, not enough information, and inappropriate style). Results show that the responses of the WS group were longer than those of the SLI group. Interestingly, some of the utterances produced by the WS children were inappropriate and less relevant due to semantic and syntactic problems. Compared to the typically developing children, the children with WS provided too little information. There was no difference between the groups on turn-taking violations. The WS and SLI children's replies to adult requests or clarification were not as appropriate as those of the typically developing children. It seems that the children with WS had relatively poor conversational skills: they talked less than the typically developing children and they did so in an inappropriate way. To complete these results, Lacroix, Bernicot and Reilly (2007) investigated their abilities to interact in the purpose to attain a goal. In this task, the mother and the child have to collaborate and negotiate to produce a drawing on computer, on the basis of a drawing model. In this collaborative conversation task, the WS children and adolescents spoke little. Similar to participants with DS, they produced fewer utterances than did the typically developing participants and played a weak role in the conversation compared to their mother. However, they readily expressed their psychological state (like younger children of the same mental age), producing more expressive utterances than DS and CA groups. Finally, the behaviour of the children and adolescents with WS with respect to maternal directives was atypical: they satisfied these requests less often than all other groups. This deficit and their massive social use of language let us think that they have some deficits in social cognition and more specifically, in theory of mind. Theory of mind and Williams syndrome – Since early childhood; individuals with WS are attracted by social interactions in an abnormal way (Jones and al., 2000) by focusing their attention on experimenter‘s face. Tager-Flusberg and Sullivan (2000) distinguish two subcomponents of theory of mind: social-cognitive and social-perceptual. The social-cognitive component is closely related to language and is classically tested with the false belief task. The social-perceptual component plays a main role in judgment of the other‘s mental state (based on facial expressions and gestures). The first task is a false belief task. Twenty-one children (aged between 4; 6 to 8; 7) were compared to fifteen children with Prader-Willi and fifteen children with a non-specific mental retardation. Results show that children with WS

What Do We Learn on Language Acquisition from Williams Syndrome?

183

are not better than the two other groups. It is interesting to note that there is no correlation between standardized measures of language and performance on false belief tasks. The authors attribute this to the attention deficits in WS. The second task tested the ability to explain a behavior. The stimuli were nine stories elicitating children‘s explanations about the action of a person using terms of desire, emotion, and cognition. There were three items in each category of mental state. Results show that on this task, children with WS have similar performance compared to the control groups. The third task evaluates the matching of pictures with different emotional expressions. Sixteen black-and-white photographs expressing four emotions (happiness, sadness, anger and fear) are presented. The child had to label the emotion. The performances of children with WS were similar to that of typically developing children and they did not perform better than other atypical groups. TagerFlusberg and Sullivan (2000) show that social-knowledge components are dissociable: the social-component in intact while the social-cognitive component is more impaired. Despite their strong interest in others, superficial social skills and empathic qualities, clinical reports indicate that older children and adults with WS have difficulties to maintain friendships and make low social judgments.

2. The Development of the Language Skills Unlike the abundance of research reports on morpho-syntactic abilities in older children and adults with WS, surprisingly little research has focused on early development of morphosyntactic abilities in infants in WS. Singer-Harris et al. (1997) used the McArthur Communicative Developmental Inventories of 27 infants with WS with a mean age of 45 months the infants with WS produced longer sentences than infants with Down syndrome. Mervis et al. (1999) coded the utterances of 39 children with WS (aged 2 to 12 months) using the Index of Productive Syntax (Scarborough, 1990). The syntactic complexity of the utterances of infants with WS was as it would be expected from their mean length of utterance and mental age. This was also found in a longitudinal study of two children with WS by Levy (2004) for Hebrew and in a study of Italian infants with WS by Vicari et al. (2002). In developmental disorders, constraints and mechanisms of lexical acquisition can be deviant and development can follow an atypical pathway too. Contrasting early (3;0-3;5 years) language development in children with WS and DS using the Words and Gestures scale of the CDI Singer-Harris et al. (1997) found that at younger ages there were no significant differences between the two groups in the number of words produced and comprehended, but the DS group produced significantly more gestures than the WS group. Overall, children with WS at this age were just as delayed in their language development as the DS group. The WS language advantage was clear though, in testing older individuals on the Words and Sentences scale of the CDI. WS grammatical abilities, regardless of the number of words produced, exceeded DS grammatical skills. One of the peculiarities of early word learning in WS is that naming precedes pointing, with an average of 6 months (Mervis and Bertrand 1997), inverting the typical sequence of first using pointing then naming. Besides less pointing, children with WS also produced less social referencing (Laing and al., 2002), and while in typical development and in children with DS the vocabulary spurt, fast mapping and exhaustive sorting coincided, in most children with WS the vocabulary spurt preceded exhaustive sorting and fast mapping, which

184

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

occurred together but only when vocabulary size was well over 500 words (Mervis and Bertrand, 1997). These results suggest that children with WS do not apply fast mapping, at least not in early vocabulary acquisition, and other mechanisms have to back up vocabulary growth in WS. The authors explain this as being due to a relatively good phonological shortterm memory, and attention devoted to linguistic input at the expense of other stimuli that are in the focus of normal children‘s attention. There is other evidence for atypical operation of some of the constraints of lexical acquisition. Stevens and Karmiloff-Smith (1997) studied four lexical constraints: fast mapping, mutual exclusivity, the whole object and taxonomic constraints. They found that just as typically developing children, children with WS (in two age groups, 3-4 years and 910 years) applied fast mapping (providing evidence that at older ages, fast mapping does apply in WS), i.e. they mapped a novel word onto an object which does not already have a name. They also relied on the mutual exclusivity constraint, according to which an object cannot have more that one name: subjects with WS mapped the novel word onto a part of the object, when they already had a name for the object itself and the part was unfamiliar. Differences were found, however, between the WS and typically developing groups in applying the whole object and the taxonomic constrain: namely, vocabulary acquisition in WS seemed to be less constrained by both. This could be because of their tendency for local processing (Bihrle and al., 1989), as the WS group made significantly fewer whole object responses than controls when they were presented with an unfamiliar object and a novel word. Less efficient use of the taxonomic constraint was supported by later findings of Nazzi and Karmiloff-Smith (2002) in another study investigating children with WS between 2 and 6 years of age. Children with WS were able to categorize objects if they had shared visual properties, but they could not use verbal labels as a basis of category formation, ability present in typically developing toddlers even at younger ages. So, despite the relatively large adult vocabulary sizes, some constraints on typical vocabulary acquisition do not seem to be operating in WS. Working memory and phonological abilities in WS - Many studies have found that auditory short-term memory is one of the strongest capacities of individuals with WS, and as such, is a potential engine of language development in this syndrome. Research in the last decade has shown that auditory short-term memory is a mechanism operating in typical language acquisition as well. Auditory short term memory capacity has been found to be in the normal range in most individuals with WS (Mervis and al. 1999). Several studies of WS memory have found dissociation between verbal and spatial short-term abilities (Wang and Bellugi, 1994; Jarrold and al., 1999), but this dissociation disappeared in long-term memory, where both modalities were found to be impaired (Vicari and al., 1996a). Vicari and his colleagues (1996b) also tested more specific memory effects, and found that, in a word span task, the participants with WS showed normal similarity and length effects, suggesting that phonological encoding mechanisms are relatively intact in WS. On the other hand, there was a reduced effect of frequency, interpreted by the authors as relative overdependence on phonological encoding mechanisms, and little effect of long-term memory representation of words. Strong verbal short-term memory capacities can be part of the explanation why language in WS develops to the level of sophistication not observed in other groups with mental retardation. On the other hand, suggestions have also been made that a relative overreliance on phonology can account for some of the atypicalities in WS language (Thomas and al. 2001).

What Do We Learn on Language Acquisition from Williams Syndrome?

185

Conclusion The research findings on morpho-syntactic abilities in WS clearly point that these skills are not preserved in WS in that no study has reported morpho-syntactic abilities which exceed participants‘ mental ages. Therefore, it seems that individuals with WS do not provide strong evidence for dissociations between language in terms of morpho-syntax and general cognitive abilities. Concerning lexical abilities, the profile described is more complex and atypical. Globally, it appears that the organization of the WS lexicon may be atypical and marked by a selective impairment to the mental lexicon and deviant semantic representations. Moreover, results from studies which have investigated the relationship between spatial cognition and language have shown impaired spatial language, especially when children with WS have to describe a real-world spatial arrangement. This suggests a kind of interference in the information processing when spatial and language abilities are required. In pragmatics, we observe some great difficulties in particular in interaction situations marked by the children not providing enough information for the interlocutor and not being able to always respond successfully to adult‘s requests for clarification. However, important feature of the WS profile has also been reported with regard to their social use of language. This has been observed in American-, Italian- and French-speaking children and adolescents with WS: they do more social evaluating than their typical peers (Reilly, Bernicot, Vicari, Lacroix, and Bellugi, 2005). All these data allow us now to say that language in WS is not preserved or ―intact‖ but deviant and atypical for two reasons: 1. The first studies investigating their development of language show some atypicalities, such as naming which precedes pointing, which are not in line with our knowledge on typical development. This emphasises the need to study the early stages of language development, if we are to better understand the developmental trajectories and alternate developmental pathways. 2. Language in WS is different from what we observe in people with other developmental disorders, such as DS, SLI or autism. One way forward is to compare these different disorders, in order to gain a better understanding of how language processing is related to general cognitive abilities and spatial cognition. Finally, our knowledge to date, informed by research evidence, on language in WS seems to point to the fact that individuals with WS may not provide straightfoward evidence for innate dissociations between language and other cognitive abilities.

References Bellugi, U., Marks, S., Bihrle, A., and Sabo, H. (1988). Dissociation between language and cognitive functions in Williams syndrome. In Bishop, D. and ogford, K. (eds.) Language development in exceptional circumstances. Hillsdale, NJ: Lawrence Erlbaum, pp. 177– 189.

186

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

Bellugi, U., Wang, P. P., and Jernigan, T. L. (1994). Williams syndrome: An unusual neuropsychological profile. In Broman, S. and Grafman, J. (eds.) Atypical cognitive deficits in developmental disorders: Implications for brain function. Hillsdale, NJ: Erlbaum, pp. 23–56. Bellugi, U., Lichtenberger, L., Jones, W. and Lai, Z. (2000). The neurocognitive profile of Williams syndrome: a complex pattern of strengths and weaknesses. Journal of Cognitive Neuroscience, 12:7–29. Bertrand, J., Mervis, C. B., and Eisenberg, J. D. (1997). Drawing by children with Williams syndrome: A developmental perspective. Developmental Neuropsychology, 13:41-67 Bickerton, D. (1997). Constructivism, nativism and explanatory adequacy. Behavioural and Brain Sciences, 20, 557-558 Bihrle, A. M., Bellugi, U., Delis, D., and Marks, S. (1989). Seeing either the forest or the trees: Dissociation in visual processing. Brain and Cognition, 11:37–49. Brock, J. (2007). Language abilities in Williams syndrome: a critical review. Developmental Psychopathology, 19:97-127. Bromberg, H., Ullman, M., Marcus, G., Kelly, K. and Coppola, M. (1994). A dissociation of memory and grammar: evidence from Williams syndrome. Paper presented at the Eighteenth Annual Boston University Conference on Language Development. Clahsen, H. and Almazan, M. (1998). Syntax and morphology in Williams syndrome. Cognition, 68(3):167–98. Clahsen, H. and Almazan, M. (2001). Compounding and inflection in language impairment: Evidence from Williams syndrome (and SLI). Lingua, 111, 729-757. Clahsen, H., Ring, M. and Temple, C. (2004). Lexical and morphological skills in Englishspeaking children with Williams Syndrome. In S. Bartke and J. Siegmueller (eds.), Williams Syndrome across Languages. Benjamins: Amsterdam, pp.221-244. Fodor, J. (1983). The modularity of mind. Cambridge: MA, MIT Press. Grant, J., Valian, V., and Karmiloff-Smith, A. (2002). A study of relative clauses in Williams syndrome. Journal of Child Language, 29:430-416. Jarrold, C., Baddeley, A.D., and Hewes, A. K. (1999) Genetically dissociated components of working memory: evidence from Down‘s and Williams syndrome. Neuropsychologia, 37:637–651. Jarrold, C., Hartley, S., Phillips, C., and Baddeley, A. D. (2000). Word fluency in Williams syndrome: Evidence for unusual semantic organisation? Cognitive Neuropsychiatry, 5(4):293–319. Jarrold, C., Phillips, C., and Baddeley, A. (2001) Comprehension of spatial and non-spatial language in Williams syndrome. Poster presented at the SRCD Biennial meeting, Minneapolis, April 2001. Joffe, V.L. and Varlokosta, S. (in press). Language abilities in William Syndrome: Exploring comprehension, production and repetition. Advances in Speech-Language Pathology. Joffe, V. L., and Varlokosta, S. (2007). Patterns of syntactic development n children with Williams syndrome and Down syndrome: Evidence from passives and Wh-questions. Clinical Linguistics and Phonetics, 21 (9), 705-727. Johnson, S. C., and Carey, S. (1998). Knowledge Enrichment and Conceptual Change in Folkbiology: Evidence from Williams Syndrome. Cognitive Psychology, 37:156–200. Jones, W., Bellugi, U., Lai, Z., Chiles, M., Reilly, J., Lincoln, A., and Adolphs, R. (2000). Hypersociability in Williams syndrome. Journal of Cognitive Neuroscience, 12(1):30-46.

What Do We Learn on Language Acquisition from Williams Syndrome?

187

Karmiloff-Smith, A. (1998). Development itself is the key for understanding developmental disorders. Trends in Cognitive Sciences, 2(10):389-398. Karmiloff-Smith, A., Grant, J., Berthoud, I., Davies, M., Howlin, P., and Udwin, O. (1997). Language and Williams syndrome : How intact is ‗intact‘ ? Child Development, 68:246262. Karmiloff-Smith, A., Thomas, M., Annaz, D., Humphreys, K., Ewing, S., Brace, N., et al. (2004). Exploring the Williams syndrome face-processing debate: the importance of building developmental trajectories. Journal of Child Psychology and Psychiatry, 45:1258—74. Karmiloff-Smith, A., Tyler, L. K., Voice, K., Sims, K., Udwin, O., Howlin, P., and Davies, M. (1998). Lingustic dissociations in Williams syndrome: evaluating receptive syntax in on-line and off-line tasks. Neuropsychologia, 36, 343-351. Karmiloff-Smith, A. and Thomas, M. (2003). What can developmental disorders tell us about the neurocomputational constraints that shape development? The case of Williams syndrome. Development and Psychopathology, 15, 969-990. Klein, B. P., and Mervis, C. B. (1999). Contrasting patterns of cognitive abilities of 9- and 10- years old with Williams syndrome or Down syndrome, Developmental Neuropsychology, 16 (2):177-196. Lacroix, A., Bernicot, J. and Reilly, J. (2007). Narration and collaborative conversation in French-speaking children with Williams syndrome. Journal of Neurolinguistics, 20(6), 445-461. Laing, E., Butterworth, G., Ansari, D., Gsödl, M., Longhi, E., Panagiotaki, G., Paterson, S., and Karmiloff-Smith, A. (2002). Atypical development of language and social communication in toddlers with Williams syndrome. Developmental Science (5), 2, 233246. Landau, B., and Zukowski, A. (2003). Objects, motions, and paths: Spatial language in children with Williams Syndrome. Developmental Neuropsychology, 23 (1 and 2):107– 139. Laws, G., and Bishop, D .V. M. (2004). Pragmatic language impairment and social deficits in Williams syndrome: A comparison with Down's syndrome and specific language impairment. International Journal of Language and Communication Disorders, 39(1), 45-64. Levy, Y. (1996). Modularity of language reconsidered. Brain and Language, 55, 240-263. Levy, Y. (2004). A longitudinal study of language development in two children with Williams syndrome. Journal of Child Language, 287-310. Lichtenberger, L., and Bellugi, U. (1998) Intersection of spatial cognition and language in Williams syndrome. Journal of Cognitive Neuroscience, SS:80. Losh, M., Bellugi, U., Reilly, J., and Anderson, D. (2001). The integrity and independence of evaluation in narratives: Evidence from children with Williams syndrome. Narrative Inquiry, 10(2), 1-26. Lukács, Á. (2005). Language Abilities in Williams Syndrome. Budapest: Akadémiai. Lukács, Á., Pléh, Cs. and M. Racsmány (2004). Language in Hungarian children with Williams syndrome. In: Susanne Bartke and Julia Siegmüller (eds.) Williams Syndrome across Languages. Amsterdam: John Benjamins. 187–220. Lukács, Á., Pléh, Cs., and Racsmány, M. (2007). Spatial language in Williams syndrome: evidence for a special interaction? Journal of Child Language, 34/2, 311–343.

188

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

Mervis, C. B., and Bertrand, J. (1997). Developmental relations between cognition and language: evidence from Williams syndrome. In Adamson, L. B., and Romski, M. A. (Eds.) Research on communicative and language disorders: Contributions to theories of language development. Baltimore: Paul Brookes. 75–106. Mervis, C.B. and Klein-Tasman, B.P. (2000). Williams syndrome: Cognition, personality, and adaptive behaviour. Mental Retardation and Developmental Disabilities Research Reviews, 6, 148-158. Mervis, C., Morris, C. A., Bertrand, J., and Robinson, B. F. (1999) Williams syndrome: Findings from an integrated program of research. In Tager-Flusberg, H. (Ed.) Neurodevelopmental disorders. Boston, Mass: MIT Press. 65–111. Nazzi, T., and Karmiloff-Smith, A. (2002). Early categorization abilities in young children with Williams syndrome. Cognitive Neuroscience and Neuropsychology, 13(10):1259– 1262. Penke, M., and Krause, M. (2004). Regular and irregular inflectional morphology in German Williams syndrome. In S. Bartke and J. Siegmüller (Eds). Williams syndrome across languages. Amsterdam : John Benjamins. Pezzini, G., Vicari, S., Volterra, V., Milani, L., and Ossella, M. T. (1999). Children with Williams syndrome: is there a single neuropsychological profile? Developmental Neuropsychology, 15:141-155. Perovic, A. and Wexler, K. (2007). Complex grammar in Williams syndrome. Clinical Linguistics and Phonetics, 21(9):729-745. Phillips, C. E., Jarrold, C., Baddeley, A.D., Grant, J., and Karmiloff-Smith, A. (2004). Comprehension of spatial language terms in Williams syndrome: evidence for an interaction between domains of strength and weakness. Cortex, 40:85–101. Piatelli-Palmarini, M. (2001). Speaking of learning: How do we acquire our marvellous facility for expressing ourselves in words ? Nature, 411:887-888. Pinker, S. (1999). Words and rules. London: Weidenfiled and Nicolson. Pléh, Cs., Lukács A., and Racsmány, M. (2003). Morphological patterns in hungarian children with williams syndrome and the rule debates. Brain and Language, 86:377-383. Racsmány, M., Lukács, Á. and Pléh, Cs. (2002). Munkamemória és nyelvelsajátítás Williams-szindrómában. (Working memory and language acquisition in Williams syndrome). Pszichológia, 22/3, 255-267. Reilly, J., Bernicot, J., Vicari, S., Lacroix, A., and Bellugi, U. (2005). Narratives in children with Williams syndrome: A cross-linguistic perspective. In D. Ravid and H. Batzeev Shyldkrot (Eds.). Perspectives on language and language development: Essays in honor of Ruth A. Berman. Dordrecht, Netherlands: Kluwar. Reilly, J., Losh, M., Bellugi, U., and Wulfeck, B. (2004). Frog, where are you? Narratives in children with specific language impairment, early focal brain injury and Williams syndrome. Brain and Language, 88, 229-247. Ring, M., and Clahsen, H. (2005). Distinct patterns of language impiarment in Down‘s syndrome and Williams syndrome: The case of syntactic chains. Journal of Neurolinguistics, 18, 479-501. Rossen, M. L., Klima, E. S., Bellugi, U., Bihrle, A., and Jones, W. (1996). Interaction between language and cognition: Evidence from Williams syndrome. In Beitchman, J. H. Cohen, N., Konstantareas, M., and Tannock, R. (Eds.) Language, learning and behavior

What Do We Learn on Language Acquisition from Williams Syndrome?

189

disorders: Developmental, biological and clinical perspectives. New York, NY: Cambridge University Press. 367–392. Scarborough, H. S. (1990). Index of productive syntax. Applied Psycholinguistics, 11, 1-22. Siegmüller, J., and Bartke, S. (2004). Williams syndrome from a clinical perspective. In S. Bartke and J. Siegmüller (Eds). Williams syndrome across languages. Amsterdam: John Benjamins. Singer-Harris, N. G., Bellugi, U., Bates, E., Jones, W., and Rossen, M. (1997). Contrasting profiles of language development in children with Williams and Down syndromes. Developmental Neuropsychology, 13, 345-370. Stevens T., and Karmiloff-Smith, A. (1997) Word learning in a special population: do individuals with Williams syndrome obey lexical constraints? Journal of Child Language, 24(3):737–65. Stojanovik, V. (2006). Social interaction and conversational inadequacy in Williams syndrome. Journal of Neurolinguistics, 19, 157-173. Stojanovik, V., Perkins, M., and Howard, S. (2004). Williams syndrome and specific language impairment do not support claims for developmental double dissociations and innate modularity. Journal of Neurolinguistics, 17(6):403-424. Stojanovik, V., Perkins, M., and Howard, S. (2001). Language and conversational abilities in Williams syndrome: How good is good? International Journal of Language and Communication Disorders, 36:234 — 239. Stojanovik, V., and van Ewijk. (2008). Do children with Williams syndrome have unusual vocabularies? Journal of Neurolinguistics, 21 (1):18-34. Tager-Flusberg, H., and Sullivan, K. (2000). A componential view of theory of mind: evidence from Williams syndrome. Cognition, 76:59-89. Temple, C., Almazan, M., and Sherwood, S. (2002) Lexical skills in Williams Syndrome: a cognitive neuropsychological analysis. Journal of Neurolinguistics, 15:463–495. Thomas, M. S. C., Grant, J., Barham, Z., Gsödl, M., Laing, E., Lakusta, L., Tyler, L. K., Grice, S., Paterson, S., and Karmiloff-Smith, A. (2001). Past tense formation in Williams syndrome. Language and Cognitive Processes, 16:143–176. Tyler L.K., Karmiloff-Smith, A., Voice, J.K., Stevens, T., Grant, J., Udwin, O., Davies, M., and P. Howlin (1997). Do individuals with Williams syndrome have bizarre semantics? Evidence for lexical organization using an on-line task. Cortex, 33:515–527. Vicari S., Brizzolara, D., Carlesimo, G. A., Pezzini, G., and Volterra, V. (1996a). Memory abilities in children with Williams syndrome. Cortex, 32(3):503–514. Vicari, S., Carlesimo, G., Brizzolara, D., and Pezzini, G. (1996b). Short-term memory in children with Williams syndrome: a reduced contribution of lexical-semantic knowledge to word span. Neuropsychologia, 34(9):919–925. Vicari S., Caselli M.C., Gagliardi C., Tonucci F. and Volterra V. (2002). Language acquisition in special populations: a comparison between Down and Williams Syndromes. Neuropsychologia, 40, 2461-2470. Volterra, V., Capirci, O., Pezzini, G., Sabbadini, L., and Vicari, S. (1996). Linguistic Abilities in Italian Children with Williams Syndrome. Cortex, 32:663–677. Volterra, V., Caselli, M.C., Capirci, O., Tonucci, F. and Vicari, S. (2003). Early linguistic abilities of Italian children with Williams syndrome. Developmental Neuropsychology, 23, 33-58.

190

Agnès Lacroix, Vesna Stojanovik and Ágnes Lukács

Wang, P.P., and Bellugi, U. (1994) Evidence from two genetic syndromes for a dissociation between verbal and visual-spatial short-term memory. Journal of Clinical and Experimental Neuropsychology, 16:317–322. Wechsler, D. (1974). Wechsler Intelligence Scale for Children–Revised. New York: Psychological Corporation. Zukowski, A. (2001). Uncovering grammatical competence in children with Williams syndrome. Unpublished doctoral thesis. Zukowski, A. (2004). Investigating knowledge of complex syntax: Insights from experimental studies of Williams syndrome. In M. Rice and S. Warren (Eds.), Developmental language disorders: From phenotypes to etiologies. Mahwah: NJ: Lawrence Erlbaum Associates.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 7

THE NEED FOR EXPLICIT INFERENTIAL METHODS IN LINGUISTICS Kent Johnson Department of Logic and Philosophy of Science University of California, Irvine, CA, US

Abstract Like every discipline, theorizing in linguistics crucially involves drawing (often tentative) conclusions about the relation between the available evidence and the theories under consideration. At many points in their research, linguists must decide whether the available evidence (theoretical, empirical, etc.) supports a given theory well enough to accept it, possibly over various competing theories. Surprisingly, however, virtually no attention has been paid to this aspect of linguistic theorizing, which contrasts with the standard practices of the social, behavioral and psychological sciences. In the latter fields, whole subareas – including dedicated faculty positions, journals, societies, etc. – are devoted to the quantitative, frequently statistical, details of drawing inferences from data. But in linguistics, theoretical inferences typically have the form of informal, verbal, holistic judgments made by professional linguists. I begin by characterizing the issue of linguistic inferences, showing why the current methods are cause for concern. The psychological literature on professional judgment strongly suggests that professional judgments aren‘t as accurate as the experts believe, and that they are inferior to more explicit methods. In fact, linguistic inferences are particularly (although not uniquely) suspect in this respect. I then consider several responses linguists have made to the criticism of their reliance on expert judgment. These replies, I argue, do little to mitigate the concerns about how evidence is compiled and assessed in linguistics.

1. Introduction Let‘s play a game. Consider the two sequences in (1) and (2). One of them represents the beginnings of a sequence that resulted from flipping an unbiased coin 200 times. The other one, the ―fake‖, was not generated by flipping a coin. If you correctly identify the fake, I will

192

Kent Johnson

give you $50, but I will charge you $50 if you are wrong. If you choose not to answer, I will give you $5 for your modesty. (The reader can also play the game with the two sequences of length 30 below; cf. footnote 2.) What shall you do? (1) (2)

THTHTHTTTTHHTHTTTHTHHHTTTHHTTH… HTTHHTTTHHTHHTHTTHHHTTHTTHHTTH…

Simply staring at (1) and (2) is unlikely to produce any judgments that can be confidently held. Moreover, all size 200 sequences are equally likely to be produced by a fair coin, hence from this perspective (1) and (2) are equiprobable. To make the game more interesting, I also mention that I constructed the fake sequence with an eye towards how often Heads would be in the lead1. Aha! Now you have something to work with. You recode the two sequences and discover that Heads is never in the lead in (1), but is in the lead about half the time in (2). While it is clearly possible that Heads would never have the lead in 200 flips of a fair coin, it is also clearly much less likely than the other outcome. So you now seem to be in a position to infer that (1) is the fake sequence and (2) is genuine. But how strong is this evidence? Is the case for (1)‘s being fake so strong that you should be quite comfortable drawing this conclusion? Or is the evidence only moderately strong, so that you should be cautious about drawing a conclusion? Or maybe it is quite weak, so that you should still make no guess and just take the $5? At this point, you reason that the present scenario is restricted to only two extremely opposed options – share the lead equally versus never having the lead even once. If there were more options, your confidence might be eroded, perhaps significantly. But in the present case, you stand by the logic that a fair coin would put Heads in the lead about as much as Tails is in the lead, and so you officially tender your guess. Walking away from this game $50 poorer, you wonder what happened. Was it simply a statistical fluke, and the fair coin never let Heads take the lead even once? Or could this outcome have been due to something else? It was: your assessment of the evidence was incorrect. Although it doesn‘t suit our intuitions about randomness and probability, the most likely outcomes are where either Heads or Tails is in the lead the entire time; the least likely outcome occurs when the lead is shared equally. When flipping a fair coin 200 times, there is more than an 11% chance that one side will have the lead the entire time, compared to a 1% chance that the lead will be shared equally. Similarly, there is almost a 25% chance that one side will be in the lead at least 97% of the time, but there is less than a 4% chance that the lead is shared equally within the corresponding margin of 3% (cf. Feller 77ff. for proofs and discussion).2 The relatively simple scenario just described illustrates an important phenomenon: even in this little game, it is hard to it is hard to come up with a reliable assessment of the evidence just by eyeballing the evidence. Human minds, even those of expert mathematicians and

1

More precisely, Heads is in the lead after the kth flip if either (a) there are more Heads than Tails in the k flips, or (b) there are exactly k/2 Heads (and hence k/2 Tails), and Heads was in the lead after the k–1th flip. 2 Incidentally, the actual sequences in (1) and (2) are both only of size 30, with ones in the lead not even once in (1), and 18 times in (2). Even in these truncated sequences the probabilities are telling: there‘s a 14% chance that Heads will be in the lead the entire time, and a 4% chance that one side will be in the lead exactly 18 times.

The Need for Explicit Inferential Methods in Linguistics

193

statisticians, are not built to make accurate3 on-the-fly assessments regarding how a set of evidence bears on the relevant hypotheses. This basic fact has been studied intensively for decades in the decision sciences community. The purpose of this chapter is to stress the importance to linguistics of this well-known limitation on human reasoning. Much of mainstream linguistic theorizing is built upon a practice of informally aggregating evidence in situations where we have little reason for confidence regarding the accuracy of such ―expert judgments‖. Indeed, although it is rarely acknowledged in the linguistics literature, I will argue that the particular inferential situation that the linguist finds herself in is an extremely difficult one, and far more problematic than many other areas that routinely appeal to informal expert judgments.

2.The Inferential Situation in Contemporary Linguistics To get things started, let‘s examine a fragment of actual linguistic theorizing. The example below is representative of much of the methods of mainstream linguistic theorizing. Since the purpose of the example is only to illustrate these widely used methods, I will not try to exhaustively characterize the relevant literature. For present purposes, that would only obfuscate matters. (Indeed, the precise details are unimportant enough that readers familiar with linguistic methods may wish to skim the example, and go right to §3.) In a series of papers, Norbert Hornstein (1998, 1999, 2000, 2003) argues that the phenomenon of linguistic control can be accounted for simply by allowing movement into theta positions. E.g., the relevant syntax of (3a) does not have the traditional form in (3b), where PRO is a distinct lexical item controlled by Sue. Instead, the proper form is in (3c), where Sue has moved from the lower subject position to the higher one. (Following Hornstein, I treat movement as a combination of the Minimalist operations of Copy and Merge.) (3) a. Sue wants to win; b. Suei wants [PROi to win]; c. Sue wants Sue to win. More generally, Hornstein holds that linguistic theories do not require the null pronominal element PRO or its associated control module that determines the referent of an occurrence of PRO. We don‘t need these things, Hornstein argues, because the phenomena that initially motivate positing them can be accounted for by appealing to independently motivated components of the grammar. Movement (aka Copy and Merge), Hornstein assumes, is a prevalent feature of grammar. If all the relevant facts can be accounted for without positing PRO, then ceteris paribus, linguistic theories should favor the simpler theory and reject the 3

I use the term accurate as a catchall phrase to refer to the various qualities that make for good judgments. These include most prominently validity (the tendency for judgments to center around the normatively correct answer) and reliability (the tendency for judgments to cluster relatively tightly together).

194

Kent Johnson

employment of PRO. In the development of this theory, Hornstein also notes several advantages. Here are two representative examples. Exhibit #1: de se interpretations. Consider (4a). Notice that it can only have the meaning given in (4b), and cannot have the meaning of (4c): (4) a. Only Bush remembers giving the ‗03 State of the Union Address. b. Only Bush remembers himself giving the ‘03 SOTU. c. Only Bush remembers that he gave the ‘03 SOTU. (4c) is false, since many persons remember that Bush (and only Bush) gave the address. However, since Bush (and only Bush) gave the address, he (and only he) even could remember giving it; thus, the truth of (4a) depends only on what Bush (and only Bush) remembers. What accounts for this mandatory de se reading? As Hornstein notes, if control is really a form of raising, the phenomenon in (4) is handily explained. After all, in such a case, the relevant syntax of (4a) would be as in (5a): (5) a. [IP Only Bush [VP Only Bush [remembers [IP Only Bush [VP giving the ‗03 SOTU]]]]] b. Only Bush x.[x remembers x giving the ‗03 SOTU] If the traces (i.e., results of Copy and Delete) in (5a) function as free variables bound by Only Bush, then the syntax easily supports the reflexive predicate reading in (5b), which necessitates a de se reading. Exhibit #2: Similarity to raising. As another piece of evidence, Hornstein notes some similarities between control and raising patterns. In the raising patterns in (6), we see that the raising must Merge a copy of the deleted DP in the expression, and that the raised expression must Merge in a way that respects subjacency restrictions: (6) a. John seemed John to be happy. b. *It seemed John to be happy. c. *John thought that it seemed John to be happy. An identical pattern is seen with control structures: (7) a. Cathay was expected to be happy. b. *It was expected to be happy. c. * Cathay thought that it was expected to be happy.

The Need for Explicit Inferential Methods in Linguistics

195

In short, the similarity of behaviors of control and raising structures further suggests that they may be from a common stock. Since there are, Hornstein notes, independent reasons for positing raising mechanisms, the similarities just noted provide further evidence that control is simply raising. Unsurprisingly, Hornstein‘s proposal has not gone unnoticed (e.g., Brody 1999, Culicover and Jackendoff 2001, Landau 2000, 2003, Manzini and Roussou 2000). The following two exhibits present prima facie evidence against the view that control is a type of movement. Exhibit #3: Partial Control. The first problem comes from Landau (2003), who argues that Hornstein‘s theory has problems accounting for ―partial control‖. For example, consider (8): (8) The chair of the department wanted to meet on Tuesday afternoon. As Landau notes, the most natural interpretation of (8) is that the chair of the department wanted some group that contains the chair and at least one other person to meet on Tuesday afternoon. That is, the controlling DP only partially determines the subject of the lower clause. But it is very hard to see how the remainder of a raising process, the chair of the dept., could support this ‗group‘ reading. Moreover, the lower clause can contain predicates that demand that the subject be a non-singleton group: e.g., Susan enjoyed getting together on weekends, Steve wondered whether helping one another would be productive in the long run. Worse yet, classical raising constructions do not seem to support these group readings: (9) *The chair seemed to meet on Tuesday afternoon. (9) cannot mean that some group containing the chair seemed to meet on Tuesday. In short, Landau‘s evidence regarding partial control suggests that there must be distinct elements in the relevant positions of raising and control constructions. Exhibit #4: Control and NP movement. A second bit of evidence against Hornstein‘s theory comes from Brody (1999, 218 – 19). If control and raising were a unified phenomenon, we would expect them to exhibit identical behavior; but they don‘t. If raising were simply control, then just as the control structure (10a) can express that John attempted to make it the case that John leave, so too, there should be a similar possibility of raising, allowing (10b) to express that John believed himself to have left, which is not the case: (10) a. John attempted to leave. b. *John believed to have left. In the other direction, we would also expect that just as the passive raising in (11a) is acceptable, having the meaning that it was believed that John had left, so too should (11b) be acceptable, with the meaning that some agent(s) attempted to make John leave.

196

Kent Johnson (11) a. John was believed to have left. b. *John was attempted to leave.

These disparities between raising and control structures further suggests that the latter phenomenon is not an instance of the former. Obviously there is much more to be said about the Hornstein‘s view. In addition to a great many more sources of evidence – both supporting and undermining Hornstein‘s view – there are also replies and reanalyses of the evidence, and replies and reanalyses of the replies and reanalyses, etc. However, it is enough for present purposes to consider the simple situation characterized above. In this mini example, we have four considerations: Hornstein‘s theory does well at accounting for (i) some similarities between raising and control, and (ii) some phenomena concerning de se interpretations. However, it has difficulties with (iii) partial control and with (iv) some issues concerning NP movement. Let us suppose (fictitiously, of course) that this is all the available data regarding Hornstein‘s theory. How good is the theory? Does it merit provisional acceptance? Is it promising enough that a linguist working in a related area should explore how the theory interacts with hers? All else being equal, is it better than a theory that does well with (iii) and (iv), but not so well with (i) and (ii)? If so, by how much? These are the sorts of questions that linguists regularly face and address in their research, whether tacitly or explicitly. Crucially, the dominant trend in linguistics is for the linguist to compile and assess the available evidence using informal, holistic, purely verbal methods. That is, linguists examine bodies of evidence – typically much larger and more complex than just (i) – (iv) above – and arrive at a judgment as to what it says about the competing theories. In particular, these judgments regarding the theory‘s relation to the evidence are made without the use of any ―external‖ aids, such as statistical rules or models, or other mathematical methods. The introduction to this paper presented a particular instance where our intuitions were much less accurate than a mathematical assessment of the evidence. This was so despite the fact that the evidence and theories (i.e., which sequence was the fake) were vastly simpler than what linguists typically work with. In the present section, we saw that linguists typically assess how the evidence bears on their theories solely by means of their expert judgments. The task of §3 is to argue that, in terms of their accuracy, expert judgments in linguistics are probably disturbingly similar to our faulty judgments seen in §1.

3. The Need for Explicit Inferential Methods The previous two sections raise a general issue: linguists typically assess the theory/evidence relationships with informal, holistic, verbal judgments, but this method may be unreliable. Importantly, reliance on informal judgments is widespread in linguistics. Johnson 2007 notes 40-odd articles and books in the linguistics literature where every inference, assessment, etc. has the same general form as that depicted in §2. But there is nothing special about those texts: readers of the major linguistics journals (e.g. Linguistic Inquiry, Natural Language and Linguistic Theory, etc.) know that this list could be increased

The Need for Explicit Inferential Methods in Linguistics

197

by an order of magnitude with little difficulty. Rather than being a few notable exceptions, reliance on the informal, verbal, holistic judgments of professional linguists is the dominant trend in linguistics. But what about the example in §1? That was just one particular case where our intuitions about the evidence are faulty. Could this be an isolated case? Maybe our intuitions about evidential strength – or at least the intuitions of those who are experts about the relevant topics – are generally quite good. If so, then there may not be much of a methodological problem after all. The burden of this section is to show that the inaccuracy illustrated in §1 is prevalent even in expert judgment. To begin, some terminology. The general phenomenon of interest is often called expert judgment. It concerns the abilities of professionals and other experts to aggregate diverse sources of evidence (from within their field of expertise) and arrive at high quality judgments, predictions, decisions, etc. By definition, these judgments are informal, verbal, largely holistic assessments of a body of evidence. Crucially, such judgments are produced without the aid of a statistical or mathematical analysis of the evidence. This type of ratiocination should not be conflated with human reasoning in general, which may exhibit flaws not found in expert judgments. Indeed, it is natural to suppose that expert judgments should be considerably better than those produced by typical human subjects. After all, experts are experts: they have spent many years studying the topic(s) that they are supplying judgments about. Moreover, reliance on expert judgment is a common and well-accepted cultural practice. Although one would expect that expert judgments suffer from few if any of the foibles that ordinary human reasoning does, unfortunately this is not so. This can be illustrated with a couple famous examples. (I canvass several examples in order to stress the importance of acknowledging the negative evidence regarding expert judgment; e.g. Arkes 2003. Hopefully, the concreteness of the examples will emphasize just how different experts‘ judgmental capacities are from how they are commonly perceived.) Many readers of this chapter are familiar with the task of graduate admissions: the committee needs to decide which applicants to admit, which to recommend for fellowships, assistantships, etc. Many readers may even consider themselves quite skilled at detecting which applicants have the greatest potential to become fellow professionals in their field. But such a task involves aggregating diverse sources of evidence, typically of much less complexity than the diversity of linguistic evidence. Experts and non-experts are typically quite bad at such aggregation tasks. Consider, for example, the comparison of two candidates from comparable undergraduate schools. Candidate A has a GPA of 3.3 and a GRE of 750; candidate B has a GPA of 3.7 and a GRE of 680. Which candidate is more desirable? Why? By how much? Even in this familiar comparison task, it is unclear how the candidates should be ranked. Regarding this example, Robyn Dawes writes that most judges would agree that these indicators of aptitude and previous accomplishment should be combined in some compensatory fashion, but the question is how to compensate. Many judges attempting this feat have little knowledge of the distributional characteristics of GREs and GPAs, and most have no knowledge of studies indicating their validity as predictors of graduate success. Moreover, these numbers are inherently incomparable without such knowledge, GREs running from 500 to 800 for viable applicants, and GPAs from 3.0 to 4.0. Is it any wonder that a statistical weighting scheme does better than a human judge in these circumstances?‖ (Dawes 1979, 574)

198

Kent Johnson

In another paper (1971), Dawes attempted to estimate the accuracy of the expert graduate admissions judgments about each applicant‘s potential for success. The correlation was essentially zero: r2 = .0361. Indeed, I posed these questions about combining GRE scores and GPAs to a senior colleague whose career has centered around the decision sciences. His response to them was straightforward: ―I don‘t know‖. Notice incidentally that this comparison task involving aggregating evidence is the simplest possible: two candidates measured on two differently scaled variables. Thus, this task has a structure that is similar to, but simpler than, the problem the previous section ended with. In contrast, the corresponding task in linguistics is massively harder: there are typically many disparate sources of evidence which are hardly ever precisely located on an (approximately) rational scale. Instead, the evidence is typically verbal, and often vague in terms of its exact characterization. Moreover, there are often more than two candidates (i.e., candidate theories) to be evaluated, and often some of these sources are inapplicable to some of the candidates. The second example is a famous case where Tversky and Kahneman (1971) tested the expert judgments of psychologists with a problem that was at the heart of their expertise. In a questionnaire given to members of the Mathematical Psychology Group and the American Psychological Association, they asked: Suppose you have run an experiment on 20 subjects, and have obtained a significant result which confirms your theory (z = 2.23, p < .05, two-tailed). You now have cause to run an additional group of 10 subjects. What do you think the probability is that the results will be significant, by a one-tailed test, separately for this group? (Tversky and Kahneman 1971, 105).

Here again, we would expect these expert judgments to be highly reliable. In fact, they were not. The median response of these experts was .85. However, Tversky and Kahneman argue that a much more reasonable estimate is .48 or lower, a difference of 37 percent. It is noteworthy that the respondents were asked to evaluate a fairly simple situation. All that is involved is a basic form of null hypothesis testing with a random sample drawn from a normally distributed population. Any introductory statistics course will cover the concepts employed in this question. Thus, the low quality judgments these experts made – about a question that could‘ve easily arisen in a lower-level undergraduate course that they teach – is all the more striking. Moreover, the respondents were trained academic psychologists, and statistics and probability are their main methodological tools. Thus, if anything we should expect them to be particularly good at making judgments under conditions of uncertainty (cf. Nisbett 1993). To put it mildly, these classical results are not alone. Over the decades, there has developed a mountainous literature regarding the nature of professional judgment. A consistent theme in this literature is the drawing of ―pessimistic conclusions‖ (Arkes, Dawes and Christensen 1986, 93). For example, Camerer and Johnson conclude their paper with: Our review produces a consistent, if depressing, picture of expert decision-makers. They are successful at generating hypotheses and inducing complex decision rules. The result is a more efficient search of the available information directed by goals and aided by the experts‘ superior store of knowledge. Unfortunately, their knowledge and rules have little impact on

The Need for Explicit Inferential Methods in Linguistics

199

experts‘ performance. Sometimes experts are more accurate than novices (though not always), but they are rarely better than simple statistical models. (Camerer and Johnson 1991, 211)

In particular, experts are quite bad at aggregating various sources of evidence into normatively appropriate judgments (or estimations and the like). In fact, a great deal of research has shown that even very crude statistical models outperform expert judgments (Meehl 1954, 1986, Dawes 1971, 1979, Dawes and Corrigan 1974, Lovie and Lovie 1986, Larrick 2004, Bishop and Trout 2005). As Meehl notes in a retrospective essay: When you are pushing 90 [the current count is about double this –KJ] investigations, predicting everything from the outcome of football games to the diagnosis of liver disease and when you can hardly come up with a half dozen studies showing even a weak tendency in favor of the [sc. expert over the statistical model], it is time to draw a practical conclusion (Meehl 1986, 374).

(Meehl goes on to argue that the extremely few cases where the expert outperformed the statistical model may well be due to statistical sampling error.) As part of this research, investigators have found many different causes behind the faulty judgmental capacities of experts. Here are just four well-known examples. First, experts and non-experts alike tend to give excess weight to novel information (Shafir and LeBoeuf 2004). E.g., when evaluating college applicants, people tend to assign more weight to a grade-point average that appears after the rest of the dossier has been considered than when the GPA is included in the dossier. Second, harder problems tend to increase experts‘ degree of overconfidence in judgments E.g., in one famous study, subjects tended to express more overconfidence regarding harder questions (for which the subjects‘ error rate was higher), and less overconfidence for the easier questions (Lichtenstein et al. 1982, Harvey 1997, Griffin and Brenner 2004). Third, experts with little training in normative judgmental models (e.g. statistical models) tend to show even more miscalibration with the normatively appropriate judgments than those with greater familiarity and training (Griffin and Brenner 2004). Finally, in their judgments, experts tend to reflect systematic biases in favor of their own theories, treating the evidence for their own theories (and the evidence against rival theories) as overall stronger than it actually is (e.g., Nickerson 1998, Harvey 1997, Larrick 2004, Griffin and Brenner 2004, Camerer and Johnson 1991). At this point, an obvious question arises. If things really are this bad, why is expert judgment still used in these tasks? Surely most experts and organizations would not want to expend large amounts of time and money carefully collecting evidence, only to assemble it in a way that vastly undermines its probative quality. This question too has been the subject of immense research. While there are many things to say about this issue, one stands out: experts are overconfident, often wildly so, about their abilities.4 E.g., Faust and Ziskin write of ―the immense gap between experts‘ claims about 4

Notice the difference between the confirmation bias mentioned at the end of the previous paragraph and the judgmental overconfidence under discussion. This difference can be seen by comparing psychometric curves relating expert judgment regarding the probability of some statement on the x-axis with the normatively appropriate probability on the y-axis (e.g. Harvey 1997). The diagonal line shows perfect calibration between the judged and correct probability. The convex line reflects a confirmation bias, where the judged probabilities (regarding one‘s favored theory) is systematically higher than the true probability. The sigmoidal line displays

200

Kent Johnson

their judgmental powers and the scientific findings‖ (Faust and Ziskin 1988, 35). Overconfidence about judgmental powers even appears when the evidence is highly precise and is embedded within a powerful mathematical theory. For instance, Fischoff and Henrion‘s well-known (1986) study of confidence levels of physicists‘ estimations of various physical magnitudes (e.g., the velocity of light or the rest mass of the electron) revealed a constant trend of overconfidence. In short, when it comes to the aggregation of evidence in their field, experts are typically unskilled and unaware that they are so (Kruger and Dunning 1999). Moreover, this overconfidence regarding expert judgment appears to be solidly entrenched in contemporary culture. E.g., even organizations as prestigious as the National Science Foundation have elected not to heed the cautionary tale from psychology about expert judgment (Arkes 2003). The relevance to linguistics of the literature described above is clear. There is no reason to believe that the expert judgments in linguistics are of any higher quality than the expert judgments reviewed above.5 It is not unreasonable to suspect that linguistic judgements are vulnerable to the four examples of limitations on expert judgment. It would be easy for a linguist to (tacitly) treat some newly discovered data from an exotic language as having more probative quality than it perhaps should, simply because it is ―new to the scene‖. It would also be easy for a linguist to have more confidence than is warranted in the fundamental architecture of her linguistic theory: knowing whether, e.g., Minimalist syntax is on the right right track – and in what respects – is clearly a very hard question. Moreover, the standard graduate-level training in linguistics does not include coursework (e.g., probability and statistics) which focuses on methods for objectively collecting, organizing, and analyzing the evidence so as to manage the uncertainty present in the data and assess its bearing on the theories at hand. Finally, unless linguists are unlike other experts, it is reasonable to suppose that, when ―telling a good story‖ about their theory, linguists will at times selectively recruit and overweight the evidence in favor of the theory, and underweight the evidence against it. (And of course these features could easily reappear in criticisms of a given theory, and in criticisms of the criticisms, and so forth.) In fact, linguistic judgments are in a particularly precarious position, since these judgments are not honed with the aid of feedback. It is not, for instance, part of the standard graduate-level linguistics education for students to be given evidence-theory pairs, where the theories are at or near the level of detail as Hornstein‘s and are known to be correct or judgmental overconfidence, where improbable statements are judged to be more improbable than they actually are, and probable statements are judged to be more probable than they are. (Continued) 1.0

0.8

0.6

0.4

0.2

0.2

5

0.4

0.6

0.8

1.0

Notice also that there is a skeptical facet to this difficulty. Although the evidence reviewed above give much reason for pessimism, in fact it is utterly unknown how well expert linguistic judgments typically perform. Since there is relatively little consensus regarding nontrivial issues in linguistics, it would be hard (although perhaps not impossible) to find issues which can be treated as settled, and examine the relevant inferences drawn about them in the past literature, to gauge the nature of the judgmental mechanisms.

The Need for Explicit Inferential Methods in Linguistics

201

incorrect, and told to assess how well the evidence supports the theory. If that were possible, then since the actual status of the theories in this ―training set‖ is known, the feedback would allow for the budding experts to calibrate their judgments. At the same time, mere feedback is not enough. For example, Camerer and Johnson write that: One of the main lessons of decision research is that feedback is crucial for learning. Inaccurate [rules] may persist because experts who get slow, infrequent, or unclear feedback will not learn that their rules are wrong. When feedback must be sought, inaccurate rules may persist because people tend to search instinctively for evidence that will confirm prior theories….Even when feedback is naturally provided, rather than sought, confirming evidence is more retrievable or ―available‖ than disconfirming evidence…. The disproportionate search and recall of confirming instances will sustain experts‘ faith in inaccurate [rules]. Even when evidence does disconfirm a particular rule, we suspect that the natural tendencies to construct such rules… will cause experts to refine their rules rather than discard them. (Camerer and Johnson 1991, 210)

More generally, lack of awareness of the accuracy of expert judgment in linguistics is of particular concern, since linguists, often motivated to promote or criticize a particular theory, may be led to apply inferior strategies with more determination, a phenomenon Larrick has called the ―lost pilot‖ effect (―I don‘t know where I‘m going, but I‘m making good time‖) (Larrick 2004, 321). Finally, a word should be said about group decision making. Perhaps the collective judgment of multiple experts are more accurate than that of just one expert. If so, perhaps expert judgment in linguistics can be vindicated somewhat to the extent that we focus on those cases where multiple experts agree, and very few disagree. There are, however, at least four reasons for concern about collective judgments. First, such a strategy may require ignoring a large proportion of relevant cases, where controversy exists, and yet inferences are routinely drawn via expert judgment (with any collective judgment(s) remaining unknown). Second, to the extent that group expert judgment exists in linguistics, the aggregation of the individual judgments is typically highly informal, with one expert reporting her sense of what is ―generally accepted‖ in the literature. There is much room for bias to creep into the aggregating linguist‘s judgment of how widespread a judgment is. Third, group judgment suffers from the same fundamental difficulty as individual expert judgment, namely that it is simply unknown how accurate the method is. Fourth, the evidence from the decision sciences community gives further reason for pessimism: perhaps the most insidious problem in groups is that people are unknowingly influenced by the public judgments of others. Especially under conditions of uncertainty, people are susceptible to anchoring on the judgments of others in forming their own judgments…. Shared training, shared experiences, and shared discussions all lead group members to hold a similar view of the world – and similar blind spots. (Larrick 2004, 326).

In summary, there are reasons for serious concern about the accuracy of the informal, verbal, holistic judgments of experts. This concern only increases when substantial feedback is unavailable, as it typically is in linguistics. Although they do not believe it, experts tend to misuse evidence, giving more weight to novel information, and growing increasingly miscalibrated with the facts as the issues get more difficult and as the experts‘ familiarity with quantitative methods decreases. Experts are typically unaware of these limitations, since they

202

Kent Johnson

also exhibit much overconfidence with respect to both their beliefs about the plausibility of their favored theories and their beliefs about the accuracy of their judgmental powers. Inferences in linguistics typically take the form of expert judgments. Thus, it is reasonable to expect that as linguists combine and assess evidence like that in §2, their expert judgments will substantially distort the evidence‘s actual bearing on the linguistic theories in question.

4. The Conservative Backlash In discussing the need for linguists to begin developing and using more explicit methods for analyzing and using their evidence, I have encountered a number of replies in defense of the methodological status quo. I will discuss four of these replies, and argue that they do not provide any support for this stance. Reply 1: Disbelief. The first reply is an all too familiar reaction to the literature on expert judgment (e.g., Arkes 2003, Camerer and Johnson 1991). Whether it takes the form of simply ignoring the results, insisting that linguists are good at these kinds of judgments, or overtly denying the validity of the results, the first reply amounts to an outright denial that there is any problem with the current methods of drawing inferences. However, as we saw in §3, there is enormous evidence that expert linguistic judgments are, as is typical, highly inaccurate. It may be that linguists are in fact good at making expert judgments, but without evidence that linguists are radically unlike other experts in this regard, little credence should be given to such a hope. Reply 2: Linguistics is more complicated. The second reply is also familiar. It usually takes the form of pointing out how much more complex the relevant linguistic evidence is than that used in the toy examples used to motivate the issue. It is uncontroversial that linguistic evidence is much more complicated and harder to understand than, say, two sequences of Heads and Tails. But it is utterly unclear why this fact should support the credibility of expert linguistic judgment; indeed, one would assume that it further undermines any such credibility. After all, when situations like these become more complex, human performance typically gets worse; rarely does it improve. As Dawes, Faust, and Meehl note (1989, 1672), if the rule for calculating the total at the supermarket gets more complex than simply adding prices, we wouldn‘t expect informal human judgment to get better: Suppose instead that the supermarket pricing rule were, ‗Whenever both beef and fresh vegetables are involved, multiply the logarithm of 0.78 of the meat price by the square root of twice the vegetable price‘; would the clerk and customer eyeball that any better? Worse, almost certainly.

In short, before this reply can have any force, we need reason to think that the kind of increased complexity found in linguistic data sets is likely to significantly improve expert judgment. It would be fascinating indeed if linguists turned out to be radically unlike other experts in this regard, but until reason is given for believing this surprising claim, it should not be endorsed. Reply 3: Pessimism about improvements. The third reply is to cast doubt on whether it is possible (and feasible) to find any precise, explicit methods to replace the expert judgments. If there are no alternatives in the offing, then the claim that alternative methods should be used is an empty exhortation. Two things can be said about this reply. First, the quality of a

The Need for Explicit Inferential Methods in Linguistics

203

method and the existence of alternatives to it are quite distinct issues. We would have little confidence in a report used Tarot cards to estimate the total mass of silicon on Pluto. Nor would our confidence in, or acceptance of, this method increase if the author explained that there was no known better way to construct the estimate. Barring reasons for thinking that Tarot cards could produce quality judgments on this topic, we would be faced with the choice of living without evidence regarding the silicon on Pluto, or seeking better methods. Although expert judgment enjoys a better reputation in linguistics than Tarot cards, §3 showed that there is good reason to think that the former method may also be highly inaccurate. In extreme cases it may not outperform a method based on Tarot cards. (In many of the studies cited above, experts performed at chance, or were even slightly negatively correlated with the normatively correct answers.) By making these unpalatable but plausible outcomes salient, my hope is to increase the sense of urgency in linguistics to develop and explore more explicit, analyzable methods for theorizing about human language. Second, it is not true that there are no explicit methods that can reduce the reliance on expert linguistic judgment. Elsewhere (Johnson ms.) I have shown how some familiar statistical tools can be easily modified and combined to give a measure of the degree to which one linguistic theory has better empirical coverage than another. The issues are somewhat delicate, since competing linguistic theories often don‘t overlap perfectly in terms of what phenomena they each account for. Moreover, linguistic data arrives continuously, not as a single batch, and some of these data are associated with other members of the data set, and so they cannot be treated as ―independent‖ points of empirical data. However, a considerable theoretical and mathematical grip can be gained using only the phi-coefficient and the logic of sequential probability ratio testing. The methods just mentioned do not, of course, address every issue where expert judgments are employed. Rather, they chip away at certain specific aspects of the general problem. But this ―chipping away‖ procedure is the norm in the sciences – particularly those whose methodology is still undergoing substantial improvement. E.g., journals like Biometrika, Econometrica, and The Journal of Mathematical Psychology are filled with reports of new methods for attacking specific (often very narrow!) problems in their respective fields. These journals – along with faculty positions, conferences and societies dedicated to methodological issues – have been active for decades. It is, I submit, high time linguistics began following suit. Reply 4: Analogies with other scientific methods. The final reply I consider involves appealing to methods used in other scientific disciplines. Linguists commonly compare their goals and methods to those of vision science, biology, physics, etc (cf. e.g., the papers in Chomsky, 2000, and Uriagereka 1998 esp. chap 1). In general, many of these comparisons are quite apt. Indeed, Chomsky‘s work in this area has been extremely important to the development of the theoretical foundations of linguistics. E.g., there is much to admire in Chomsky‘s discussions of why linguists need not worry about positing unobserved structure in their theories of language, reducing their theories to those of more ―basic‖ disciplines, or employing only elements that are accessible to consciousness (Chomsky 2000, chaps. 4, 5; cf. also Johnson 2007). However, such discussions should not be overinterpreted. For instance, although some aspects of linguistic methodology are shared with physics (or astronomy, biology, chemistry, etc.), it does not follow that linguistics has ―exhausted the methods of science‖ (Chomsky 1986, 252). It similarly does not follow that we are free to ignore the resulting ―mere difference in degree only, not in kind‖ between the way the various fields

204

Kent Johnson

implement these shared methodological aspects. Instead, this ―difference in degree‖ can make all the difference in evaluating linguistic methodology. To see this last point, notice that a proper assessment of the relationships between linguistic and other methodologies crucially involves assessing their differences as well as their similarities. A major difference between linguistics and the various (other) empirical sciences is that it is standard in the latter disciplines to use explicit quantitative methods for analyzing and assessing the data and its impact on various theories. Indeed, this is why calculus, algebra, and statistics are standard background training in these fields. There are many reasons why such methods are used in – and some would say are definitive of – the sciences. Not least among these reasons is the vast increase in accuracy and precision over simple human judgment. Further benefits include relative ease of analysis and insensitivity to contextual effects, biases, etc. By being explicit, the mathematical methods are themselves comparatively easy to study. By being subjected to such analyses, these method‘s strengths and weaknesses can be clarified, and the methods themselves can be improved. Moreover, by being explicit, the methods are insensitive to irrelevant features of the data or experimental situation in a way that humans are not. (For example, Dawes (1979) famously showed that a linear statistical model whose weights were randomly assigned to the variables could outperform a human expert who had access to the same information. This occurred because the model was consistent in its predictions, and did not vary from case to case, as the experts did.) In addition to the comparison with the sciences, it is also instructive to compare linguistics to history, English literature, and other fields that are typically not considered ―scientific‖. Here, too, there appear to be some rather strong similarities. These fields all gather evidence regarding various topics, and assemble it, primarily via the method of informal, holistic, verbal expert judgment, into a case that supports or undermines a theory or theories. What would happen if a linguist attempted to argue – with as much vigor as she argues that linguistics is a science – that linguistics is in fact one of the humanities? It is plausible that a very strong case could be made. What would this exercise show? The real payoff, I believe, would be to show that there is less benefit in defending linguistics as a ―science‖ or a ―humanities‖, and much more benefit in making a frank and open assessment of how the methods of contemporary linguistics work, where they are strong and where they are weak, and how they can be improved.

Conclusion There is, to my knowledge, little reason for strong confidence regarding how linguistic evidence is routinely compiled and assessed. This paper raised some doubts about the accuracy of informal, verbal, holistic expert judgments in linguistics, and stressed the importance of accuracy in assembling diverse sources of evidence. Instead of using informal judgments, linguists should seek explicit ones that can be studied, analyzed and improved. Instead of using verbal judgments, linguists should seek precise, formal ones that are incapable of ambiguity, and whose biases (if any) must be worn on their sleeves, most likely in an exact quantitative format. Instead of using holistic, all-things-considered judgments, linguists should work to adopt the spirit of decision analysis, which is to divide and conquer: ―Decompose a problem into simpler problems, get one‘s thinking straight in these simpler

The Need for Explicit Inferential Methods in Linguistics

205

problems, paste these analyses together with a logical glue, and come out with a program for action for the complex problem‖ (Raiffa 1968, 271). Hopefully, some of the examples and literature reviewed have brought out the importance to linguistics of the above recommendations. Unfortunately, however the evidence also suggests that overconfidence in expert judgments is remarkably resilient to the scientific evidence about them. It is probable that in linguistics as elsewhere, the first step – namely understanding and accepting the limited accuracy of expert judgments – will be the hardest to take.

References Arkes, H. (2003). The nonuse of psychological research at two federal agencies. Psychological science 14, 1 – 6. Armstrong, J. S (1980). Unintelligible management research and academic prestige. Interfaces 10, 80 – 86. Bishop, M., and Trout, J. D. (2005). Epistemology and the psychology of human judgment. Oxford: OUP. Camerer, C. F. and Johnson, E. J. (1991). The process-performance paradox in expert judgment: How can experts know so much and predict so badly?. In K. A. Ericsson and J. Smith (Eds.) Toward a general theory of expertise: Prospects and limits (pp. 195 – 217). Cambridge, MA: CUP. Dawes, R. (1971). A case study of graduate admissions. American psychologist 26, 180 – 188. Dawes, R. (1979). The robust beauty of improper linear models in decision making. American psychologist 34, 571 – 582. Dawes, R. and Corrigan, B. (1974). Linear models in decision making. Psychological bulletin 81, 95 – 106. Einhorn, Hillel J. (2000). Expert judgment: some necessary conditions and an example. In T. Connolly, H. Arkes, and K. Hammond (Eds.), Judgment and decision making (pp. 324 – 335). Cambridge, UK: CUP. Faust, D., and Ziskin, J. (1988). The expert witness in psychology and psychiatry. Science 241, 31 – 35. Gaeth, G., and Shanteau, J. (2000). Reducing the influence of irrelevant information on experienced decsion makers‖. In T. Connolly, H. Arkes, and K. Hammond (Eds.), Judgment and decision making (pp. 305 – 323). Cambridge, UK: CUP. Griffin, D., and Brenner, L. (2004). Perspectives on probability judgment calibration. In D. J. Koehler, and N. Harvey (Eds.), Blackwell handbook of judgment and decision making (pp. 177 – 199). Oxford, UK: Blackwell. Harvey, N. (1997). Confidence in judgment. Trends in cognitive sciences, 1, 78 – 82. Henrion, M. and Fischhoff, B. (1986). Assessing uncertainty in physical constants. American journal of physics, 54, 791 – 798. Johnson, K. (2007). The legacy of methodological dualism. Mind and language 22, 366 – 401.

206

Kent Johnson

D.J. Koehler, D. J., Brenner, L., and Griffin, D. (2000). The calibration of expert judgment. In T. Gilovich, D. Griffin, and D. Kahneman (Eds.), Heuristics and biases : The psychology of intuitive judgement (pp. 686 – 715). Cambridge: CUP. Kruger, J, and Dunning, D. 1999. Unskilled and unaware of it: How difficulties in recognizing one‘s own incompetence lead to inflated self-assessments. Journal of personality and social psychology 77, 1121 – 1134. Landau, I. (2003). Movement out of control. Linguistic inquiry, 34, 471 – 498. Larrick, R. P. (2004). Debiasing. In D. J. Koehler, and N. Harvey (Eds.), Blackwell handbook of judgment and decision making (pp. 16 – 337). Oxford, UK: Blackwell. Lichtenstein, S., Fischhoff, B., and Phillips, L. D. (1982). Calibration of probabilities: The state of the art to 1980. In D. Kahneman, P. Slovic, and A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases (pp. 306 – 334). Cambridge: CUP. Lovie, A. D., and Lovie, P. (1986). The flat maximum effect and linear scoring models for prediction. Journal of forecasting 5, 159 – 168. Meehl, P. E. (1954). Clinical versus statistical prediction: A theoretical analysis and review of the evidence. Minneapolis: University of Minnesota Press. Meehl, P. E. (1986). Causes and effects of my disturbing little book. Journal of personality assessment, 50, 370 – 375. Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of general psychology, 2, 175 – 220. Nisbett, R. E. (1993). Rules for reasoning. Hillsdale, NJ: Erlbaum. Phillips, J. K., Klein, G., and Sieck, W. R. (2004). Expertise in judgment and decision making: A case for training intuitive decision skills. In D. J. Koehler, and N. Harvey (Eds.), Blackwell handbook of judgment and decision making (pp. 297 – 315). Oxford, UK: Blackwell. Raiffa, H. (1968). Decision analysis. Reading, MA: Addison-Wesley. Shafir, E., and LeBoeuf, R. (2004). Context and conflict in multiattribute choice. In D. J. Koehler, and N. Harvey (Eds.), Blackwell handbook of judgment and decision making (pp. 341 – 359). Oxford, UK: Blackwell. Tversky, A., and Kahneman, D. (1971). Belief in the law of small numbers. psychological bulletin, 76, 105 – 110. Uriagereka, J. (1998). Rhyme and reason: An introduction to minimalist syntax. Cambridge, MA: MIT Press.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 8

„DOING‟ DISCIPLINARY POWER: MANAGER-SUBORDINATE INTERACTION IN A STAFF MEETING Jonathon Clifton University of Antwerp, Prinsstraat 13 2000 Antwerpen, Belgium

Abstract This paper attempts to provide a space for dialogue between macro-historical Foucauldian analysis of power and conversation analysis‘ fine-grained technical analysis of talk-in-interaction. Arguing that such a dialogue is possible through the use of bridging as a research tool that focuses on areas of complementarity between macro and micro analysis, this paper shows how category bound access to powerful discourse resources (notably the first part of question and answer adjacency pairs) during a staff meeting allows the management to exercise disciplinary power over their subordinates. In this way, the Foucauldian notion of disciplinary power can be seen as an emergent feature of oriented-to pre-allocated access to the first part of question and answer adjacency pairs.

Keywords: power, Foucault, conversation analysis, meetings

Introduction On the face of it, the macro-historical interest of Foucault apparent in his analysis of power has little to do with the fine-grained analysis of talk that conversation analysis (CA) has to offer. However, in debate1 over the wider issue of CA‘s ability to deal with issues raised by post-structuralist concerns, Schegloff (1997 and 1998) comes to the conclusion that CA is able to deal with such issues provided observable features of the participants‘ talk and conduct are kept to the fore and ‗technical‘ analysis precedes critical or sociopolitical  1

Email: [email protected], Tel: 00 32 3 220 41 86. For example, Billig 1999, Clegg 1987, McHoul 1987, Schegloff 1997 and 1999, and Wetherell, 1998.

208

Jonathon Clifton

analysis. Furthermore, Miller (1997) offers the framework for establishing complementarity between such diverse disciplines as Foucauldian analysis of power and CA‘s empirical analysis of talk. Firstly, he states (ibid: 36) that linking wider social theorizing with finegrained analyses of talk-in-interaction can ―provide analytical and empirical frameworks for simultaneously considering how power moves through the various activities and interaction that constitute everyday life‖. Secondly, he (ibid: 37) considers that ―agency and constraint are not mutually exclusive issues but ……. are coterminous aspects of the setting under study‖. In other words, CA can be used to investigate how actors orient to the constraints and resources of wider (Foucauldian) discourse when manoeuvring for advantage at a micropolitical level. However, after setting out a theoretical framework he does not go on to apply this framework to the analysis of data. The purpose of this article is to apply such a framework to the analysis of a transcript of naturally-occurring talk from a business meeting in order to explicate how Foucault‘s concept of disciplinary power, and its realisation in the form of ‗the exam‘, is demonstrably oriented to by participants and, secondly, how this is accomplished through the participants‘ orientation to the sequential structure of turn taking which instantiates the ‗exam‘ as an ongoing interactional achievement. More specifically, the paper focuses on the use of question and answer adjacency pairs and their expanded sequences. I shall argue that the initiation of a question and answer sequence is a category bound resource which allows the management2 to question their subordinates which talks into being ‗the exam‘, thus exercising power. Furthermore, as Foucault (1983: 225) points out, the exercise of power is open to resistance. Consequently, I will also be considering the linguistic resources available to subordinates in an attempt to challenge and subvert the management‘s exercise of power.

Methods To use Miller‘s (1997) metaphor, this paper aims to build a ―bridge‖ between Foucauldian concepts of power and CA. This does not imply that there has to be a total synthesis between Foucauldian ideas and CA, nor should it be confused with triangulation whereby several methodological approaches are used to discover indisputable facts about a single social reality. According to Miller (1997: 25), bridging ―involves providing a venue for dialogue between different interpretive frameworks‖ so that areas of compatibility between perspectives can be developed and interpretive insight may be gained. The areas of compatibility that this paper develops are a fine-grained analysis of question and answer sequences and the exercise of disciplinary power. Disciplinary power is, according to Foucault (1975), a power that creates a docile subject. This is achieved, firstly, through hierarchical observation which allows the person exercising power to observe the subject of power without being seen. Secondly, disciplinary power is exercised through normalising judgement which allows those exercising power to judge the actions of the other and sanction any deviant behaviour. These two means of exercising disciplinary power come together in the exam which allows the more powerful social actors to question the subjects of power and judge their answers. In this way the more powerful ‗normalise‘ the actions of the less 2

For the purpose of this article I shall treat the management as the director and his assistant who have coincumbency of a single identity.

‗Doing‘ Disciplinary Power

209

powerful. Foucault (1975: 172/3) sums this up when he states, ―Le succès du pouvoir disciplinaire tient sans doute à l‘usage d‘instruments simples : le regard hiérarchique, la sanction normalisatrice et leur combinaison dans une procédure qui lui est spécifique, l‘examen‖3. Moreover, Foucault (ibid: 173) goes on to say that the exam is essentially a ―jeux de question et réponses‖4. However, despite the fact that Foucault (1983) states that power is pervasive in all social interaction and that it is inherent in the relationship between individuals, his macro-historical analysis of power has no room for fine-grained textual analysis and the explication of how disciplinary power is exercised through question and answer sequences. As Findlay and Newton (2000: 225) point out, ―there is little in Foucault to explain local action. A Foucauldian framework does not give us insight into how the actions of organizational actors are constrained by operating within hierarchies that privilege certain actors‖. Conversely, as Hutchby (1996a) argues, CA is in a position to explicate how power relations are inherent in the sequential properties of turns at talk and the relationship between these turns. This, he argues (ibid), is because the interactional context and the social context are reflexively related; social structure (and its implicit power relationships) informs talk and social structure is itself informed by talk. Therefore, the process of talking social structure and power relationships into being is obseverable through a fine-grained analysis of how the participants themselves orient to the institutional setting and refrain from certain discourse activities and participate in others. Power is thus seen as a phenomenon that is enacted via access to discourse resources and can thus be seen in terms of the participants‘ orientation to the relationship between turns in sequences which accomplish ―doing‖ a business meeting. As Hutchby states (1999: 92), ―by focusing on the oriented-to structures of talk-in-interaction and their consequences in terms of the structural distribution of discourse resources, conversation analysis can succeed in making visible what for conventional sociologists is the awkwardly invisible concept of power‖.

The Data The data was recorded during a monthly staff meeting held at a local office of an organisation that deals with the promotion of cultural events. 18 participants were present. The meeting was held around a large rectangular table in the staff common room. It took the form of the office director dealing with the points on the agenda one by one and then asking specific participants for information about projects that they were working on. The data analyzed concerns the final topic dealt with during the meeting where the assistant director of the centre asks the projects officer to update the staff on a film festival that is due to be held. A full transcription can be found in appendix two. The notation used for the transcription is based on Jeffersonian conventions (see appendix one). For ease of reading the participants who spoke have been given the following abbreviations:

3

―The success of disciplinary power stems, no doubt, from the use of simple instruments: hierarchical observation, normalising judgement, and their combination in a procedure that is specific to it, the exam.‖ (Author‘s translation) 4 ―A game of question and answer.‖ (Author‘s translation)

210

Jonathon Clifton D AD ICM PO ?

Director Assistant Director Information and Communications Manager Projects Officer Unidentified participant

Analysis As Foucault states disciplinary power is exercised through the use of questions and 5 answers. Firstly, this is because the first turn in a question and answer adjacency pair is preallocated to the management. The participants‘ orientation to pre-allocation of turns can be seen in the following extract in which in the course of 11 turns the management ask four questions. Extract 1 63 D 64 PO 65 D 66 PO 67 ICM 68 69 AD 70 PO 71 AD 72 PO 73 AD 74 ICM

erm (.5) and on the third it‘s Sweet Sixteen that‘s right and then on the twenty first it‘s the gala yeah twenty first to thirty first it‘s [the core festi ]val but of all the [()] [ they‘re the ] [ they‘re the ] main sponsor events = = when did we go from seven to ten days (.2) when: the: film museum had more time [available] to fill and when [ thank you ] we got [ more films ] [ and we can ] sustain the quality yeah

Identity and access to discursive resources exist in a reflexive relationship (Zimmerman 1998) so that when participants‘ invoke discourse identities (i.e. as questioner, answerer, story teller, etc.) through access to category bound discourse resources they reflexively invoke oriented-to-situated-identities (i.e. as manager, teacher, judge, etc.). This, thus, allows participants to accomplish acts which are contingent upon oriented-to-identity. As Matoesian (1993: 100) states, in ―organisational contexts, access to sequential structure is asymmetrically distributed across social position, thereby constituting a major resource for accomplishing domination in talk.‖ In Foucault‘s terms, through category bound access to the first part of an adjacency pair the management talk themselves into being as the examiners or instruments of the exercise of disciplinary power and the subordinates are talked into being as the examined or the objects of the exercise of power.6 Secondly, as Heritage (1984: 247) says, ―the first speaker‘s production of a first pair part proposes that a second speaker should relevantly produce a second pair part which is 5

An adjacency pair can be glossed as a sequence of two turns. Each turn is spoken by different speakers and the turns are related to each other in such a way that the second turn is required to provide a relevant response to the first turn. 6 Foucault (1975 : 172) « La discipline ―fabrique‖ des individus; elle est la technique spécifique d‘un pouvoir qui se donne les individus à la fois pour objets et pour instruments de son exercice. » Discipline makes individuals. It is the specific technique of a power that considers individuals as, at the same time, both objects and instruments of its exercise (author‘s translation).

‗Doing‘ Disciplinary Power

211

accountably ‗due‘ immediately on completion of the first.‖ Garfinkel‘s (1967) breaching experiments provide a notable example of the power of adjacency pairs. He demonstrates the righteous hostility that was provoked by student experimenters failing to respond to the first pair part of a greeting. Similarly, through access to the first part of the question and answer adjacency pairs the management are in fact setting the question in the exam to which their subordinates are normatively constrained to reply. In this way, the management are able to examine the subordinates‘ knowledge and observe their actions. Moreover, as Sacks (1992 vol. 2: 340) points out, going first is a very powerful discourse resource because it enables the questioner to put off giving his or her opinion and leaves the second speaker‘s reply open to criticism or evaluation. Thus, by having category bound access to the first part of a question and answer adjacency pair the management, without having to express their views, are able to observe the actions of other participants. In this way, the questioner talks into being hierarchical observation whereby, ―le pouvoir disciplinaire, lui, s‘exerce en se rendant invisible; en revanche il impose à ceux qu‘il soumet un principe de visibilité obligatoire. Dans la discipline, ce sont les sujets qui ont à être vus‖7 (Foucault 1975 :189). Furthermore, access to the first part of an adjacency pair is a doubly powerful discourse resource because not only does it constrain the actions of the recipient but it also gives access to a third slot. As Sacks et al. (1974: 717) note, ―although they are addressed to specific parties – namely, the authors of preceding questions – answers do not select those parties to speak next, because they do not project and require the subsequent performance of an activity by them. Accordingly, if the recipients of questions confine themselves to answering, next turns are left to be allocated through self-selection.‖ Furthermore, as Sacks et al. (ibid: 712) point out, a ‗last as next‘ preference is in operation so that the speaker just prior to the current speaker generally has access to the next turn. Therefore, not only does category bound access to the first part of a pair give the management control of what can coherently be said in the second part as an answer but it also provides them with a possibility to take the floor after a minimally complete answer with a third position item, or a next question. Consequently, access to the first position and subsequent access to the third slot can be used not only to dictate what the examined can talk about but also to judge the adequacy of such responses. Returning to extract one, it can be seen how the management restrict the subordinates to minimally complete turns and thus restrict the subordinates‘ possibilities to develop their own topic. In line 65 this is done by immediately self-selecting to take a turn after a minimally complete response. In line 69 a further question is latched onto a co-authored minimally complete turn. In line 71 the assistant director self-selects at a possible turn completion point to proffer a topic closure by thanking the projects officer. However, the projects officer continues his turn only to be overlapped later by the assistant director‘s competitive incoming8 and question. In this way, the management control the adequacy of the responses and prevent the subordinates from developing topic. In Foucault‘s terms, the subordinates are thus kept ‗docile‘ to the extent that they cannot develop their own topics but must respond to the management‘s agenda.

7

Disciplinary power is exercised invisibly: on the other hand it imposes visibility on its subjects. In discipline, it is the subjects who must be seen (author‘s translation). 8 French and Local (1983: 18) define turn-competitive incoming as overlapping speech ―in which one speaker comes in clearly prior to completion of another‘s turn and can be heard as directly competing with the other for possession of the turn‖.

212

Jonathon Clifton

Significantly, as can be seen in the following extract, category bound access to the third slot can also be used to evaluate the adequacy of the response thus giving the management the ability to normalise any deviant responses and to reward non-deviant responses and so talk into being Foucault‘s normalising judgement. Extract 2 73 AD 74 ICM 75 PO 76 ICM 77 AD

[ and we can ] sustain the quality yeah yeah they‘re all good films great

In line 73 the assistant director occupies the first part of the adjacency pair. The projects officer and information and communications manager co-author the second part of the pair over three turns and the assistant director uses the third slot to evaluate. The evaluation in this instance is a one of ‗reward‘ (great). The third turn can also be used for repair or to normalise a deviant response as can be seen in the following (simplified) extract: Extract 3 120 ICM 121 D 122 123 IC 124 125 PO 126 AD 127 PO

[ is it ] [ but is this ] [some]thing [yes ] absolutely M actually worth it where we ought to be looking at bringing in an external press agency yes I just do not I dunno

In line 120 the information and communications manager self-selects to ask a question. The projects officer responds positively (line 125). The assistant director then uses the third slot as an evaluation slot (line 126). She begins her turn with the preface ‗I just do not‘. The projects officer however, orients to the ‗I don‘t know‘ as a dispreference marker (Pomerantz 1984) that prefaces disagreement. Consequently, he immediately uses his next turn to reformulate his response and he changes it to ‗I dunno‘. In so doing, he effectuates a revision of his initial response and performs a reversal (Wasson 2000). Moreover, Wasson (ibib) specifically links this kind of reversal of opinion with a Foucauldian culture of surveillance whereby organisational ideology is maintained by observation and failure to follow the organisation‘s discourse leads to a negative effect on career advancement. Thus, once speakers perceive that a suggestion is being negatively received, they have a tendency to abandon that opinion before overt disagreement is expressed. In this way, subordinates themselves orient to the hierarchical observation of the examiners and management and adapt their actions accordingly. The possibility of third turn repair is of crucial importance in the consideration of power because it is a slot that allows the management to evaluate the adequacy of the subordinate‘s response and normalise the subordinate‘s action if the response is deviant. As Schegloff (1992: 1301) notes, the third slot is ―the last subjectively provided opportunity to catch

‗Doing‘ Disciplinary Power

213

(among other troubles) such divergent understandings as embody breakdown if intersubjectivity – that is, trouble in the socially shared grasp of the talk and other conduct in the interaction‖. Through the use of the third slot for repair the management can thus normalise the participant‘s response and impose their interpretation of events and reality. In Foucault‘s (1975: 187) terms, the ability to repair allows the examiner to ―prélever et de constituer du savoir‖9. Moreover, to continue the analysis, not only does the projects officer move from a ‗yes‘ to a ‗dunno‘ but, as can be seen in the following extract, he then goes on to co-author agreement with the assistant director. Extract 4 128 AD 129 ICM 130 AD 131 132 ? 133 PO 134 ? 135 AD 136 PO

no I don‘t let‘s discuss that [ be]cause you know you can pick up certain things [ no↑ ] but I think the type of stuff we‘re doing at the moment we actually we actually need someone dedicated or we need something I agree [ we don‘t necessarily ] know what the press ( ) of our [ something something ] project I think we‘re missing out on huge opportunities at the moment it something dedicated to it anyway

In lines 128-131 the assistant director sets out her position. In lines 132-4 an unidentified participant agrees with the propositional content of the turn. In line 133 the projects officer, in a bid to anticipate and so complete the assistant director‘s utterance, repeats the assistant director‘s ‗something‘. However, he fails to gain the floor and after some overlapping speech the unidentified participant continues to completion. The assistant director then self-selects to continue her evaluation. The projects officer, in line 136, then manages to gain the floor and completes the assistant director‘s retrospectively unfinished utterance (line 131) by completing the utterance with ―something dedicated to it anyway‖. By projecting to completion the projects officer has moved from ‗yes‘, to ‗I dunno‘, to the co-construction of an evaluation in favour of a negative response. Furthermore, the third turn can be used to produce an evaluative comment in the shape of a formulation or an upshot. This can be seen in the following extract: Extract 5 108 D 109 PO 110 111 112 D 113 PO 114 AD

9

press interest yes erm lining up press interviews now for erm Martin Compton who‘s the lead actor of Sweet Sixteen and . hh possibly an article in the Bulletin which (1) Reginald Richards was very keen to do [ so we ] need to draft something for that and [ uhuh ] get back in touch with him today I think the press is something which something that really has to be discussed

―Sample and constitute knowledge.‖ (Author‘s translation)

214

Jonathon Clifton 115

[ both ] in terms of communication and in terms of that particular external press 116 PO [ yeah ] 117 AD strategy [ we‘ve ] just got to get that together and see at the moment we‘re all I 118 D [ yeah ] 119 AD mean like there are [ four ] of us phoning you know [ to individuals ]

In line 108 the director asks a question. The projects officer replies to this in lines 109113. The assistant director then uses the third turn to provide an upshot (Heritage and Watson 1979) of the reply whereby the implications of the reply are given. Thus in effect a meaning of the reply is formulated (Hutchby 1996a) and, as the backchannelling of the affiliative response token ‗yeah‘ by the projects officer and the director indicate, it is also accepted. Thus, in this way the management‘s meaning is displayed and accepted through access to the third slot.10 To summarise: Access to the third turn thus provides a category bound resource for evaluating, formulating, and repairing which can be used for ‗gatekeeping‘ (Yeung 2004(a) :122 and 2004(b) : passim) whereby deviations from, or misconceptions of, the management line are detected and are corrected in a subsequent turn. This can be seen as a ‗disciplinary‘ strategy because evaluations normalise participants‘ actions, repair ‗punishes‘ deviant actions and formulations display the normal (i.e. the management line). Consequently, the management can use category bound access to the third turn as a resource which allows the examiner ―tout en transmettant son savoir, d‘établir sur ses élèves tout un champ de connaissances‖11(Foucault 1975: 188). In short, as Linell (1990: 164) states, ―if an actor is interactionally dominant …..he or she will stand a good chance of enforcing his or her perspective or rationality on to the joint discourse‖. However, as Foucault points out, every relationship of power implies a possibility for resistance. Consequently, subordinates have access to discourse resources that can be used to resist. Firstly, they can circumvent the normative constraint of answering the question and reply with a response that does not answer the question. This can be seen in the following extract: Extract 6 1 AD 2 3 PO 4 5 AD 6 ICM 7 PO 8 AD

10

11

Roland do you want to update at all on the film festival sorry about this Graham do you want to update at all on the film festival yeah film festival is sorry perhaps I should‘ve bought through the latest programme (.1) but we lost a great film = =which one did we lose= =but we gained an even bett [er ] [yes ]but we got some good ones= =ah good =

Taylor (1995) illustrates how meaning is asymmetrically determined in father-child interaction. He explicates how a child‘s perception of events is reformulated by the father and leads to a co-constructed rewriting of an event which is then accepted by the child. ―Whilst at the same time as transmitting his knowledge, to establish on his students a field of knowledge‖. (Author‘s translation)

‗Doing‘ Disciplinary Power

215

In line one, after getting the projects officer‘s name wrong, the assistant director nominates Graham (the projects officer) as the next speaker. In line six she latches a question onto the projects officer‘s turn. The information and communications manager and the projects officer co-author a reply which does not directly answer the question. In reply to the question ―which one did we lose‖ the projects officer states that better films have been obtained but he does not say which film has been lost. However, despite not directly answering the question, the response is considered adequate and is positively evaluated (‗ah good‘). Thus, in this particular context, the risk of not directly answering a question pays off and the fact that the speakers do not provide a conditionally relevant second pair part is not sanctioned. Whereas, as Atkinson and Drew (1979: 184) or Molotch and Boden (1985: passim) point out, in a more formal setting such as a courtroom failure to reply appropriately to a question might provoke an immediate sanction. However, despite the fact that the projects officer and information and communications manager co-author a reply that is not directly relevant to the first part of the adjacency pair, the reply still orients to the ‗hierarchical observation‘ of the management. It is noticeable that the subordinates‘ turn provides a description that orients to a possible negative evaluation concerning the loss of a film by building a defence into their reply i.e. the fact that they gained an even better one. In this way, the function of the ‗exam‘ can be seen to be oriented to and constructed by all participants in the interaction. Secondly, even though access to the first part of a pair is mainly category bound, it is not as rigidly pre-allocated as in other more formal settings such as the courtroom or classroom. Consequently, categories are open to negotiation12 on a turn-by-turn basis and it is possible, though rare (at least, in the data presented here), for any participant to make a bid to selfselect and ask a question. In the following extract, the information and communications manager self-selects to ask a question: Extract 7 23 ICM 24 25 PO 26 27 28 ? 29 PO 30 31 ICM 32 AD 33

12

how did how did the businesses feel about when you when you talked about Gas Attack with the sponsors they‘ve yet to get back actually (.2) in back erm but otherwise we‘re shaping up for a great programme (.1) er I‘ve got a real mix of films erm sort of innovative themes and there‘s AKA which is a great film sort of (.1) split triptych screen = = and hopefully some directors [ and ] actors coming out to talk °it‘s great°= [ an ] =and the actors of Sweet Sixteen excellent I mean one of the things with particularly what‘s happened this morning in that erm France and Belgium blocked the [ NATO ] resolution erm

Sacks et al. (1974: 718) note, ―one of the major aspects of the flexibility of conversation is that it is compatible with multiplicites of, and changes in, the social identities of some same participants.‖ Thus, different identities can be invoked and attributed to the same participant on a turn-by-turn basis. Categorisation is not a priori but is negotiated throughout interaction.

216

Jonathon Clifton

Through the ability to negotiate identity and thus discourse rights on a turn-by-turn basis, the information and communications manager is able to exercise power over the projects officer by attributing a second pair part to him and later by evaluating this response (line 31). Moreover, by giving a response (albeit with an stepwise change of topic: ‗but otherwise‘) the projects officer accedes to this exercise of power which is also confirmed by the fact that no other participant seeks to sanction the information and communications manager for this use of category bound discourse resources. In this way, she is successful in gaining access to a discourse resource normally reserved for the management and thus she is able, at least for a turn, to align herself with the management and in so doing exercise disciplinary power over the project officer. As Ochs (1993: 288) points out, ―speakers attempt to establish the social identities of themselves and others through verbally performing social acts and verbally displaying certain stances‖. However, after the sequence, the assistant director self-selects and by speeding up her speech she fights off potential competition for the turn and ensures access to the floor (Schegloff 2000). Thus, the information and communications manager‘s claim to management identity is short lived and she is unable to continue beyond her evaluation and thus she is unable to build a chain of questions as the director and his assistant do. Thirdly, as Linell et al. (1988: 436) note, ―subordinates compensate for their inherent lack of ‗power‘ by expanding their answers‖. This can be seen in the projects officer‘s turn in the previous extract (line 25). The information and communications manager self-selects to ask the projects officer a question. However, after answering the question he uses the conjunction ‗but‘ to carry out a stepwise transition of topic and in so doing he is able to develop his own topic. However, despite such resistance, the stepwise transition is hearable as orienting to possible third turn evaluation because the topic shifts from what has not happened, and for which the projects officer could be held accountable, to the fact that the projects office has now obtained ‗good‘ films (lines 26-7: ―we‘re shaping up for a great programme‖, ―innovative themes‖, ―AKA which is a great film‖). The information and communications manager ties her turn to the new topic development and she positively evaluates the response to her question (line 31: excellent). Fourthly, the evaluation of the management can be challenged. This challenge can come by the subordinates using a fourth slot to disagree. This is illustrated in the following extract: Extract 8 135 AD 136 PO 137 ICM 138 AD 139 140 D 141 (meeting breaks up)

I think we‘re missing out on huge opportunities at the moment it something dedicated to it anyway okay maybe we can get Gerry in on this and we can talk about it later I think we‘re wasting so much time in terms of all of us following different ( ) °sit down° Yeah (.5 ) Good alright thanks very much anything to add (0.6)) anyone any other business alright thanks very much

In lines 135-6 the assistant director and the projects officer complete an evaluation. The information and communications manager then begins her turn with okay and a suggestion which normatively requires a second pair part response. As Beach (1992: 337) states, ―okay‖ can be ―understood as indicating that its producer agrees with, affirms, and/or understands

‗Doing‘ Disciplinary Power

217

what was projected prior – and perhaps even treats that talk as significant. But once one has accomplished these objectives via ―okay‖, the way is now open to what is deemed relevant through additional turn components.‖ In this instance, the ‗okay‘ prefaces a continuation of turn which makes a suggestion and pursues the discussion concerning the use of an external press agency and thus keeps open the challenge to the management. However, to counter this, the assistant director skip connects13 to her previous utterance and so avoids addressing the information and communication‘s manager‘s directly prior turn which attempted to initiate a stepwise transition of topic away from that developed by the assistant director. Consequently, there is no uptake on the information and communications manager‘s suggestion. Moreover, as Sacks (1992 vol. 2: 351) observes, ―primarily, persons are engaged in producing talk that connects to directly prior talk‖. Therefore, unless the information and communications manager also skip connects, there is little hope that her attempt to perform a stepwise transition of topic will be pursued. In this instance, the director self-selects to take the next turn and by feeding back agreement (‗yeah‘), produces talk that is connected to the immediately prior turn (i.e. the assistant director‘s). Through taking this turn he prevents the information and communications manager from skip connecting to resume her own topic development. Moreover, the director then extends his turn with a ―good‖ which, as Button (1991) states, can be hearable as an assessment which initiates a close of topic. Topic closure is indeed provided by the director as he continues his turn and thanks participants for their input which marks the topic end. He then begins a move to closure of the meeting by the use of several topic initial elicitors (anything to add (1) anyone any other business) which as Button (1984) points out are used to signal continued availability for talk but do not proffer a particular topic and if no topic is forthcoming then closing can be carried out. In this case nobody seeks to introduce another topic or come back to a topic that has been left hanging and the director closes the meeting with thanks. The closure of the meeting is confirmed by the fact that without further comment the meeting breaks up. Therefore, by using powerful discourse resources such as skip-connecting, topic closure, and the right to close the meeting, the management avoid the development of contentious topics and in so doing effectively scotch a challenge to their authority.

Conclusion In this article I have attempted to provide a bridge between Foucault‘s analysis of power and a fine-grained analysis of question and answer sequences during a staff meeting. In so doing, I hope to have created a space for dialogue between on the one hand post-structuralist Foucauldian analysis which has been accused of not demonstrating how participants orient to the analyst‘s theoretical concerns (e.g. Schegloff 1997 and 1999) and on the other hand CA which has been criticised for being too limited in its attempts to deal with power (e.g. Clegg 1987, Wetherell 1998). Taking Foucault‘s notion of the exam as an instrument of disciplinary power, I have attempted to show how participants in a business meeting demonstrably orient to category bound access to discursive resources that allow Foucault‘s notion of the exam to 13

Skip-connecting, as defined by Sacks (1992 vol. 2: 349), occurs when ―a speaker produces an utterance which is indeed related to some prior utterance, but some utterance prior to the directly prior utterance‖. It is a technique that is thus used when there is competition over topic so that a speaker, by tying his or her current utterance to his or her previous utterance and not the previous speaker, preserves his or her topic.

218

Jonathon Clifton

be talked into being. Thus, providing a fine-grained technical analysis of how category bound access to the discourse resource of going first in a question and answer sequence can explicate how power operates through language. As Hutchby (1996b: 482) states, ―the very ways in which participants design interaction can have the effect of placing them in a relationship where discourse strategies of greater or lesser power are differentially available to each of them. In this sense, power can be viewed as an ‗emergent feature‘ of oriented-to discourse practices in given settings‖. More specifically, I have attempted to show how the two instruments of disciplinary power (hierarchical observation and normalising judgement) are exercised through exam question and answer sequences. Firstly, disciplinary power is exercised because the management have category bound access to the first part of question and answer adjacency pairs which constrains their subordinates to produce conditionally relevant second pair parts. This constitutes a major resource for obliging the examined to set out their actions or ideas for observation by the examiner. Secondly, the second part of the pair (i.e. the answer) does not project a relevant next action and consequently it does not project a next speaker. Thus, the third turn is available through self-selection and, as Sacks et al. (1974) point out, it is normally the previous speaker who has access to this turn. Thus, if the management have category bound access to the first part of a pair they have access to the third slot as well. The third slot can be used to ask another question, thus continuing the question and answer sequences in a series, to formulate an upshot of the response, to evaluate it or to repair it. In this way, the management can normalise the subordinate‘s reply and author a substantive version of reality even to the extent that subordinates abandon their own world view and accept the management line. However, power can always be contested and subordinates have access to discourse resources that can be used to challenge or subvert the manager‘s power. Firstly, questions can be answered evasively. Secondly, in an informal meeting where there is room for category negotiation, any participant can self-select and ask questions. Thirdly, stepwise transitions of topic can be added to answers in an attempt to author a particular line of topic development. Fourthly, subordinates can challenge the validity of the manager‘s formulation, repair or evaluation. However, in general, subordinates have restricted access to the more powerful discourse resources and as such their challenges are usually countered effectively by the management. In sum, Foucauldian notions of disciplinary power can be firmly located in the fact that, in the context of this staff meeting at least, the management have category bound access to more powerful discourse resources which enable them to examine and normalise the actions of their subordinates.

Appendix 1. Transcript Conventions Used ˚yes˚ er: = ><   .hh ()

noticeably quieter than surrounding talk stretched sound latching between utterances and words noticeably quicker than surrounding talk rising intonation falling intonation audible in-breath untranscribable talk

‗Doing‘ Disciplinary Power (0.5) [yes ]

219 time gap in tenths of a second overlapping talk

Appendix 2. Complete Transcript 1 AD 2 3 PO 4 5 AD 6 ICM 7 PO 8 AD 9 PO 10 11 12 14 ICM 15 16 PO 17 D 18 AD 19 20 PO 21 AD 22 23 ICM 24 25 PO 26 27 28 ?

Roland do you want to update at all on the film festival sorry about this Graham do you want to update at all on the film festival yeah film festival is sorry perhaps I should‘ve bought through the latest programme (.1) but we lost a great film = =which one did we lose= =but we gained an even bett [er ] [yes ]but we got some good ones= =ah good = = which is good and this was one one thing I need to get a bit get everyone‘s opinion on for the opening (.4) gala night (.3) one for the core film festival itself on the twenty first of March er great Scottish film called Gas Attack which talks about chemical weapons in Iraq and asylum seeker dispersal program [in the UK] [we tried it] last year but [didn‘t ] [.hhh ] so extremely hot topic (.2) but [( )] [on the twenty first  ] [I think ( )] I think we may we may need to run that one past we might need to run it past [a few people first ] [I think we‘d better ] check we need to make sure we‘re getting strong support from er from headquarters how did how did the businesses feel about when you when you talked about Gas Attack with the sponsors they‘ve yet to get back actually (.2) in back erm but otherwise we‘re shaping up for a great programme (.1) er I‘ve got a real mix of films erm sort of innovative themes and there‘s AKA which is a great film sort of (.1) split triptych screen = = and hopefully some directors [ and ] actors coming out to talk °it‘s great°=

220

Jonathon Clifton 29 PO 30 31 ICM 32 AD

[ an ] =and the actors of Sweet Sixteen excellent I mean one of the things with particularly what‘s happened this morning in 33 that erm France and Belgium blocked the [ NATO ] resolution erm 34 D [ NATO yeah ] 35 PO erm 36 AD I think it‘s quite a kind of monumental I mean it really is extraordinary erm as 37 we might see [ not ]er only the beginning of the end of NATO but I think ( ) we 38 PO [ yeah ] 39 may well be seeing another nail in Tony Blair‘s coffin to be honest so it could 40 actually work (.2) very strongly in our favour but it is very risky and I think we‘ll 41 probably need to get corporate [ Chris ] Day‘s support 42 PO [ yeah ] 43 D yes (.2) .hhhh I think that‘s that fits in the [ area ]of risk management = 44 PO [ yeah ] 45 AD = yeah = 46 PO = artistically it is a very strong film as well 47 D yes I can [ I accept] that, [ I ] accept that [ but ] 48 PO [ .hhhh ] 49 AD [ but ] [ but ] that‘s the criteria but that‘s 50 the criteria = 51 PO = yeah = 52 D = I think we‘d be sailing quite close to the breeze there [ so ] we‘d have to be 53 PO [ yeah ] 54 D really really careful 55 AD and we really can‘t I mean it would be very difficult and we‘re going to be 56 inviting Brian to the big opening 57 PO? yeah 58 ICM but we should go with the criteria of the best of British films so 59 D .hhhh yes [ but we‘re not operating ] in a vacuum [ a policy vacuum] 60 AD [ yes we should but ] [ we should but ] 61 ICM I know that‘s our justification isn‘t it 62 AD yeah 63 D erm (.5) and on the third it‘s Sweet Sixteen 64 PO that‘s right

‗Doing‘ Disciplinary Power 65 D 66 PO 67 ICM 68 69 AD 70 PO 71 AD 72 PO 73 AD 74 ICM 75 PO 76 ICM 77 AD 78 ICM 79 PO 80 81 AD 82 PO 83 ICM 84 ? 85 ? 86 AD 87 ICM 88 PO 89 AD 90 PO 91 all 92 AD 93 D 94 AD 95 ? 96 D 97 PO 98 ? 99 ICM 100 D 101 PO 102 ICM 103 AD 104 PO 105 D

221

and then on the twenty first it‘s the gala yeah twenty first to thirty first it‘s [the core festi ]val but of all the [()] [ they‘re the ] [ they‘re the ] main sponsor events = = when did we go from seven to ten days (.5) when: the: film museum had more time [available] to fill and when [ thank you ] we got [ more films ] [ and we can ] sustain the quality yeah yeah they‘re all good films great the shorts are particularly strong yeah but the we‘re not showing films on (.2) the two Monday nights of the festival okay coz they‘re not (pop) nights anyway and they were (.1) It‘s also quite good to give people a day off ( ) should be good ° (.5) would be great if can get it what did we lose Dirty Pretty Things° Once Upon a Time no Dirty Pretty Things were oh did we pay for the rights no that was Love Song no but we lost Once Upon a Time in the Midlands (laughter) which was as exciting as it sounds great title great title yes Once upon a time where great title The Midlands in the Midlands we‘ve got West Midlands region support that oh roighty no we lost their advertising and their sponsorship on the back of that film company are being difficult (.2) shame (.3) but otherwise it‘s gonna be good (.5) how long is it on for

222

Jonathon Clifton 106 ICM 107 PO 108 D 109 PO 110 111 112 D 113 PO 114 AD 115 116 PO 117 AD 118 D 119 AD 120 ICM 121 D 122 123 ICM 124 125 PO 126 AD 127 PO 128 AD 129 ICM 130 AD 131 132 ? 133 PO 134 ? 135 AD 136 PO 137 ICM 138 AD 139

ten days twenty first to thirty first March press interest yes erm lining up press interviews now for erm Martin Compton who‘s the lead actor of Sweet Sixteen and . hh possibly an article in the Bulletin which (.1) Reginald Richards was very keen to do [ so we ] need to draft something for that and [ uhuh ] get back in touch with him today I think the press is something which something that really has to be discussed [ both ] in terms of communication and in terms of that particular external press [ yeah ] strategy [ we‘ve ] just got to get that together and see at the moment we‘re all I [ yeah ] mean like there are [ four ] of us phoning you know [to individuals] [ is it ] [ but is this ] [some]thing [yes] absolutely actually worth it where we ought to be looking at bringing in an external press agency yes I just do not I dunno no I don‘t let‘s discuss that [ be]cause you know you can pick up certain things [ no↑ ] but I think the type of stuff we‘re doing at the moment we actually we actually need someone dedicated or we need something I agree [ we don‘t necessarily ] know what the press ( ) of our [ something something ] project I think we‘re missing out on huge opportunities at the moment it something dedicated to it anyway okay maybe we can get Gerry in on this and we can talk about it later I think we‘re wasting so much time in terms of all of us following different ( ) °sit down°

‗Doing‘ Disciplinary Power 140 D 141 (meeting breaks up)

223

Yeah (.5 ) Good alright thanks very much anything to add (0.6)) anyone any other business alright thanks very much

References Atkinson, J. and Drew, P. (1979). Order in court. The organization of verbal interaction in judicial settings. London: Macmillan. Beach, W. (1993). Transitional regularities for casual "Okay" usages. Journal of Pragmatics, 19, 325-352. Button, G. and Casey, N. (1984). Generating the topic: the use of topic initial elicitors. In J. Atkinson and J. Heritage (Eds.). Structures of Social Action: Studies in Conversation Analysis (pp.167-190). Cambridge: Cambridge University Press. Button, G. (1991). Conversation-in-a-series. In D. Boden & D. Zimmerman (Eds.). Talk and social structure: studies in ethnomethodology and conversation analysis (pp. 251-277). Cambridge: Polity Press. Billig, M. (1999). Whose terms? Whose ordinariness? Rhetoric and ideology in conversation analysis. Discourse and Society, 10/4, 543-558. Clegg, S. (1987). The language of power and the power of language. Organisational studies, 8/1, 61-70. Findlay, F. and Newton, T. (2000). Reframing Foucault: The case of performance appraisal. In A. Mckinlay and K.Starkey (Eds.). Foucault, management and organization theory: from panopticon to technologies of self (pp. 211-230). London: Sage. French, P., & Local, J. (1983). Turn-competitive incomings. Journal of Pragmatics, 7, 17-38. Foucault, M. (1975). Surveiller et punir. Naissance de la prison. Paris: Gallimard. Foucault, M. (1983). The Subject and Power. In H. Dreyfus & P. Rainbow, Michel Foucault: Beyond structuralism and hermeneutics (pp. 208-26). Chicago: University of Chicago Press. Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall. Heritage, J. (1984). Garfinkel and ethnomethodology. Cambridge: Polity Press. Heritage, J. and Watson, D. (1979). Formulations as conversation objects. In G. Psathas, (Ed.) Everyday language: studies in ethnomethodology (pp.123-162). New York: Irvington. Hutchby, I. (1996a). Confrontation talk. Arguments, asymmetries and power on talk radio. Mahwah, New Jersey: Lawrence Erlbaum Associates. Hutchby, I.(1996b). Power in discourse: The case of arguments on British talk radio shows. Discourse and Society, 7/4, 481-497. Hutchby, I. (1999). Beyond agnosticism?: Conversation analysis and the sociological agenda. Research on Language and social interaction, 32/1&2, 85-93. Linell, P. (1990). The power of dialogue dynamics. In I. Marková & K. Foppa (Eds.), The dynamics of dialogue (pp. 147-177). Hertfordshire: Wheatsheaf. Linell, P. Gustavsson, L. and Païvi, J. (1988). Interactional dominance in dyadic communication: a presentation of initiative-response analysis. Linguistics, 26, 415-422.

224

Jonathon Clifton

McHoul, A and Clegg, S. (1987) Language and institutional reality: Reply and response. Organisational studies, 8/4, 363-374. Matoesian, M. (1993). Reproducing rape: domination through talk in the courtroom. Cambridge: Polity Press. Miller, G. (1997). Building bridges. The possibility of analytical dialogue between ethnography, conversation analysis and Foucault. In D. Silverman (Ed.), Qualitative research: theory method and practice (pp. 24-44). London: Sage. Molotch, H. and Boden, D. (1985). Talking social structure: discourse, domination and the Watergate hearings. American Sociological Review, 50, 273-88. Ochs, E. (1993). Constructing social identity: A language socialization perspective. Research on language and social interaction, 26/3, 287-306. Pomerantz, A. (1984). Agreeing and disagreeing with assessments: some features of preferred/dispreferred turn shapes. In J. Atkinson and J. Heritage (Eds.). Structures of Social Action: Studies in Conversation Analysis (pp.57-101). Cambridge: Cambridge University Press. Sacks, H., Schegloff, E., Jefferson, G., (1974). A simplest systematics for the organization of turn-taking in conversation. Language, 50, 696-735. Sacks, H. (1992). Lectures on conversation (Vols. 1 & 2). Oxford: Basil Blackwell. Schegloff, E. (1992). Repair after next turn: The last structurally provided defense of intersubjectivity in conversation. American Journal of Sociology, 98, 1295-1345. Schegloff, E. (1997). Whose text? Whose Context? Discourse and Society, 8/2, 165-187. Schegloff, E. (1999). ‗Schegloff‘s texts‘ as ‗Billig‘s data‘: A critical reply. Discourse and Society, 10/4, 558-582. Schegloff, E. (2000). Overlapping talk and the organisation of turn-taking for conversation. Language in society, 29, 1-63. Taylor, C. (1995). ―You think it was a fight?‖ Co-constructing (the struggle for) meaning, face, and family in everyday narrative activity. Research on Language and social interaction, 28/3, 283-317. Wasson, C. (2000). Caution and consensus in American business meetings. Pragmatics, 10/4, 457-481. Wetherell, M. (1998) Positioning and interpretive repertoires: conversation analysis and poststructuralism in dialogue. Discourse and Society, 9/3, 387-412. Yeung, L. (2004a). The paradox of control in participative decision-making: facilitative discourse in banks. Text, 24/1, 113-146. Yeung, L. (2004b). The paradox of control in participative decision-making: gatekeeping discourse in banks. International journal of the sociology of language, 166, 83-104. Zimmerman, D. (1998). Identity, context and interaction. In C. Antaki and S. Widdicombe (Eds.) Identities in Talk. London: Sage.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 9

SIMULATION TASKS: CAN EFL LEARNERS INTERACT AS EFFECTIVELY WITH EACH OTHER AS THEY CAN WITH NATIVE SPEAKERS? Yasuo Nakatani Tokyo University of Science, Japan

Abstract This study asked whether a simulation interactive task could facilitate EFL (English as a foreign language) learners‘ conversation in actual classrooms. A restaurant simulation task was designed for Japanese college students in order to investigate whether learners‘ interaction with other learners can be meaningful compared with their interaction with native English speakers (NSs). The database was derived from videotapes of five studentstudent (S-S) conversations and five native English speaker-student (N-S) conversations collected from the observation of the task performance. Transcripts were made and analysed by focusing on the number of turns, production rate and the quality and quantity of negotiation of meaning. The results indicate that although students in S-S use less sophisticated utterances, they can produce more language and use more turns than in N-S conversations. They frequently negotiate meaning and flexibly expand their interaction with peers. Their conversation is not the same as that of NSs but they could make use of opportunities to develop their target language communication skills in EFL contexts. Thus the two-way required information exchange tasks which facilitate each participant‘s personalised outcome can create spontaneous interaction for negotiation in classroom contexts.

Introduction It is hardly controversial to assert that EFL (English as a foreign language) students should actively participate in verbal interactions in communicative contexts in order to improve their target language (TL). However, just putting learners in a communicationcentred classroom and exposing them to the TL is not itself a sufficient condition for second or foreign language acquisition.

226

Yasuo Nakatani

According to theoretical principles concerning TL interaction emerging from numerous studies, learners‘ negotiation behaviours are believed to have significant effects on their learning. It is widely argued that comprehensible input and output through negotiated interaction enable learners to comprehend and produce messages beyond their current interlanguage receptive and expressive capacity. (e.g. Long, 1983, 1985; Long and Porter, 1985; Pica, 1987, 1996, 2002; Ellis et al., 1994). Such meaningful interaction has been observed in not only native speaker-non native speaker (NS-NNS) interaction but also NNSNNS interaction (Iwashita, 2001; Mackey, Oliver and Leeman, 2003; Pica, Lincoln-Porter, Paninos and Linnel, 1996; Varonis and Gass, 1985). Some researchers state that the most effective way to create negotiation of meaning is using specific tasks which require two-way information exchange to reach a closed outcome (e.g. Long, 1980, 1983; Pica Kanagy and Falodun, 1993). Viewed in this light, it is possible to say that EFL classrooms could create a positive learning environment through introducing tasks which facilitate negotiating interaction with peers. However, as Foster (1998) and Liebscher and Dailey-O‘Cain (2003) claim, most previous studies have investigated the performance of NNSs under research conditions. Only a few studies have examined classroom discourse as qualitative data to analyse learning processes through negotiation of meaning in actual lessons (e.g. Williams, Inscoe, and Tasker, 1997). Of course, when developing a theory of the interaction, it is important to use research settings in order to control the many variables. Yet, when it comes to designing a teaching method, it is necessary to investigate how interactions are generated according to specific classroom contexts. Furthermore, most recently Nakahama, Tyler and Lier (2001) point out that tasks which require unstructured interaction for open-ended goals can offer substantial learning opportunities through meaningful interaction. Nevertheless, only few attempts have so far made at examining learners‘ negotiation of meaning in a simulation task which requires twoway information exchange in order to achieve personalised open-ended goals. Therefore it is worth examining the effectiveness of such tasks in EFL classrooms. This article presents a preliminary attempt to investigate NNS-NNS and NS-NNS classroom interaction in a simulation task. The discourse data is examined quantitatively and qualitatively in terms of how well learners can take turns and produce TL utterances. In particular, the focus is on learners‘ negotiation of meaning to evaluate their task performance. These comparisons can have useful implications, as students have more chances to use English with their peers than with NSs in Japanese classroom contexts. This is because the class size is generally large (30-40 students in a class) and teachers have to introduce pairwork or small group activities in communicative lessons (See e.g. O‘Sullivan, 1992). Can peer students become meaningful communication partners in the TL? If so, can interaction between students create opportunities for negotiation of meaning compared to interaction with NSs? It is necessary to carefully observe how they can deal with personalised contexts and how these conditions affect their negotiation behaviours. Although this is a single case study and the sample size is small, it is suggested that simulation tasks promote negotiating situations in which students have to act as themselves in order to achieve their personalised communication goals with peers. Thus learners could obtain learning opportunities through interaction in classroom contexts. The next section reviews how negotiating interaction has been regarded as an essential component to TL development. It is also argued how previous studies have designed research

Simulation Tasks

227

tasks in order to investigate learners‘ interaction. Then it is discussed why simulation tasks are unique in enhancing TL communication in classrooms.

Background The Importance of Negotiation In Monitor theory Krashen (1982) argued that learners need comprehensible input which is slightly higher than their current level of comprehension (i + 1) in order to assist the TL acquisition process. The argument here is that students learn the TL not through memorisation of its rules and structures but through internalising these rules from input made comprehensible within an interactional context. However, in the monitor theory, Krashen does not seem to explain clearly what the essential environment is for learners to receive such comprehensible input. The role of interaction was examined more closely by Long (1983). He investigated the interaction between NSs and NNSs, and found that they both performed collaboratively to overcome communication breakdowns. In this process, NSs tended to modify their message to become comprehensible to NNS interlocutors. When NNSs indicated difficulty in understanding the meaning of an utterance, NSs repeated, reworded or used a less complex sentence until their interlocutor could understand it. Long states such NSs‘ repairing behaviour can create the ideal environment which could provide learners with comprehensible input. However, Swain (1985) points out that comprehensible input, while necessary, is not sufficient for successful second language acquisition. She investigated learners‘ performance in immersion classrooms and found their spoken production was not as developed as their other skills such as in listening, reading and writing. She argues that this is because those classrooms pay greater attention to giving learners‘ comprehensible input and require them to have little face-to-face, spoken interaction in order to produce their utterances. She claims that learners should experience meaningful interaction that leads them to convey meaning in the TL through the production of specific grammar, a progress she calls comprehensible output. It is reasonable to say that learners need to produce comprehensible output because it enables them to improve proficiency by manipulating their interlanguage in creative and complex ways. The important point to note is how learners can have enough opportunities to receive comprehensible input and to produce comprehensible output. Gass and Varonis (1985) focus more on this issue and state the importance of negotiation of meaning. Negotiation of meaning is interlocutors‘ mutual problem solving behaviour that is motivated by verbal or nonverbal requests for assistance by the hearers. The speakers need to restructure the utterance to make themselves understood. So the purpose of negotiation is to improve the comprehensibility of messages. Accordingly, it can be said that a learning environment should include opportunities for learners to produce more target-like output as well as to receive comprehensible input through engaging in meaningful social interaction. During the negotiation, learners could develop the knowledge and skills required for spontaneous communication in the TL.

228

Yasuo Nakatani

This kind of negotiation can be used even in NNS - NNS interaction in ESL (English as second language) contexts (e.g. Long, 1981; Mackey et al, 2003; Varonis and Gass, 1985) .Yet, what is important is that it might be difficult for EFL learners to become flexible comprehensible input senders because of their limited linguistic knowledge. Thus it is important to investigate whether EFL learners can use negotiation of meaning in actual classroom settings, and if so how such negotiation could affect their interlanguage development.

The Effects of Task Characteristics on Negotiation of Meaning How should a task for EFL learners be designed to elicit their negotiating behaviours? What kind of method should be introduced to evaluate their negotiation? Let us examine how previous studies have addressed these issues. Long (1980) identified some basic component features of signals for negotiation, such as comprehension checks, confirmation checks and clarification requests. Comprehension checks are when interlocutors try to acknowledge whether others have understood the preceding utterances. Confirmation checks are when interlocutors attempt to ensure their own understanding of others‘ preceding utterances, which can be realised by repeating or paraphrasing what the previous speakers said. Clarification requests are when interlocutors seek assistance in understanding others‘ preceding utterances. Varonis and Gass (1985) regard clarification requests and confirmation checks as indicators, which serve as triggers for interlocutors to produce comprehensible output in order to resolve comprehension difficulties. In this paper, I use the term modified output (See e.g. Pica et al., 1996) which is output specifically motivated by the hearer‘s indicator for negotiation. Modified output requires learners to manipulate their current interlanguage in order to make their initially unclear messages become meaningful. To maintain the discourse, interlocutors should initiate meaningful negotiation whenever communication breakdowns occur. Hence the term ―negotiation of meaning‖ can be defined as interlocutors‘ mutual problem solving and interaction maintenance behaviour, which consists of signals for negotiation and modified output. Such components have been regarded as valid schemes to evaluate the effects of tasks on ESL discourse (e.g. Liebscher and Dailey-O‘Cain, 2003; Iwashita, 2001; Pica, 1987, 1996). It is argued that task types have significant effects on the features of interaction (Pica, et al., 1993; Skehan, 1998). Doughty and Pica (1986) reported that the teacher-led task was less effective than the small group and the dyad tasks in terms of generating negotiation of meaning. However, some studies indicate that teachers could promote negotiation of meaning with students in some tasks such as vocabulary learning tasks (e.g. Ellis, Tanaka and Yamazaki, 1994; Ellis and He, 1999). On the contrary learners tend not to utilise negotiation of meaning as often as they should in some small group and dyad tasks, even in those contexts considered to be communicatively oriented. The question is how we can assess the characteristics of these tasks? Pica et al. (1993) present the useful framework for categorising communicative tasks: interactional relationship, interactional requirement, goal orientation and outcome options. The interactional relationship can be divided into two categories: the one-way and two-way direction of information flow between interlocutors. The interactional requirement focuses on

Simulation Tasks

229

whether each participant has to interact with others in order to complete a task. The goal orientation highlights whether participants should have the same goal. The outcome options concerns whether participants should reach one acceptable outcome or whether they can be allowed to reach a variety of outcomes. Pica et al. (1993) discuss that two-way information exchange and required interaction tasks are more effective. It is also said that among these tasks, closed outcome tasks such as jigsaw and information gap tasks in which participants are expected to have the same goal to achieve one outcome are preferable in order to facilitate learners‘ negotiation. The researchers argue that this is because these tasks have shared clear goals and allow learners to follow structured interaction with many turns to exchange information frequently. According to their claim, open-ended tasks such as discussion are the least effective for interaction for negotiation. Indeed, numerous studies have used jigsaw and one- or two-way information exchange tasks in order to examine the effects of negotiation of meaning on second language acquisition (e.g. Pica, 1988; Pica et al., 1996; Mackey et al, 2003). It is generally agreed that required information tasks seeking a closed outcome are effective in research settings for negotiation of meaning. However, it is not always understood how they are available to promote negotiation in classroom settings. For instance, Foster shows (1998) that her ESL students produced little negotiated interaction in the twoway required information tasks in classrooms. On the other hand, Liebscher and DaileyO‘Cain (2003) report that learners could negotiate meaning even in uncontrolled and openended tasks. Their study investigates a content-based classroom in which students learn other subjects by means of the TL, rather than learn the TL itself. Nakahama et al. (2001) suggest that when dealing with familiar topics and exchanging personalised information with a conversation partner, students can negotiate in their interaction. These results may justify the claim that it is meaningful to introduce tasks which utilises a classroom as a discourse arena (Di Pietro, 1987; Sinclair, 1984). It is assumed that language learning classrooms can be regarded as unique social contexts in which learners learn a language through use. So using the TL is both the means and a goal of learning. The classroom context could offer opportunities for public and interpersonal sharing of the content of language learning. Therefore, it is important to let learners see themselves working to achieve personally realistic outcomes in the learning events with peers. As one manifestation of such interactive tasks, an open-ended task, such as a simulation, could be useful to facilitate EFL students to use the TL. Let me briefly define the term ―simulation‖. Researchers seem to have little consensus on the terminology ―simulation‖. They use ―simulation/game‖ (Crookal and Oxford, 1990), ―simulation‖ (Ladousse, 1987), ―situational role plays‖ (Omaggio, 1993) etc. However, the distinctions are not always crucial and every term represents a teaching methodology which creates a ―real-life‖ situation in the classroom. Crookal and Oxford say, ―If simulation is regarded and treated as a reality taken for granted in its own right, then the experiences of participants become real, and the use of language becomes meaningful communication‖ (1990, p.15). In short, a simulation in language lessons should have reality of function and simulate the environment of the real world. Students are required to take initiative to think in the target language and use it through materials which are used in the target culture. The benefits of using simulations are summarised as follows:

230

Yasuo Nakatani 1. Motivating students to use the TL and helping them develop fluency through active communicative interaction among peers. 2. Building confidence by participating meaningfully in rehearsal for the real-world. 3. Reducing anxiety through an enjoyable language-using atmosphere. 4. Creating an open-ended communicative environment and training students to deal with the unpredictable nature of language which is similar to real life. 5. Giving chances to learn the target culture through created social situations in the classroom.

According to the framework of Pica et al. (1993), simulations can be regarded as twoway required information exchange tasks which promote divergent outcomes depending on each participant‘s personalised goal. On the other hand, the primary problem of the simulation is generated by one of its outstanding benefits, i.e., it needs to create similar situations to the real world and thus simulation tasks could be complex and need a lot of preparation. Students are strongly expected to take the initiative for their communication, and thus they have to get used to acting positively in a different classroom environment from their previous learning experience. These might be reasons why little attention has been given to the tasks although there is general agreement that a simulation could give students opportunities to express themselves and to be creative (e.g. Crookal and Oxford, 1990; Nakatani, 1996; Omaggio, 1993). In particular, surprisingly few studies, to date, have used actual classroom discourse data in order to examine the nature of interaction in simulation tasks. Thus it is essential to investigate whether a simulation task can be operated in EFL classroom contexts and whether such a task can enhance negotiation of meaning which has been argued to help TL development. The present study addresses these issues.

Research Design Research Questions Through quantitative and qualitative data analysis, this study examined how learners communicate effectively with other learners and with NSs in a real classroom setting. Participants‘ discourse was analysed in terms of the number of turns taken, the total amount of interaction, and the quantity and quality of negotiation. The research questions are: 1. Do student-student (S-S) and Native English teacher-student (N-S) differ in the amount of production in a simulation task? 2. Do S-S and N-S differ in the number of turns in a simulation task in terms of how often they exchange information? 3. Can Japanese EFL students negotiate meaning in a real classroom setting?

Simulation Tasks

231

Participants The students who participated in this task were 15 students in a mixed-level EFL class at a Japanese college. Participants were all female, non-English major first year students. Their ages ranged from 18 to 19. Each participant had completed 6 years of English study (which starts in the seventh grade) prior to entering the college. All of them had attended local public or private high schools that prepared them for college entrance examinations. Their English proficiency, especially listening comprehension, was low, with scores ranging from 260 to 435 on the TOEIC (Test of English for International Communication) test administered by ETS (English Testing Service). The participants formed a homogeneous group of Japanese college non-English major students. The college offered 90-minute compulsory English lessons once a week for 26 weeks. The task took place in the ninth lesson. Two male native English teachers were invited as communication partners for the task. Both of these NSs had more than 3 years of English teaching experience in Japan and they had sometimes taught the current participants in their regular lessons.

Task A simulation task dealing with ordering from a restaurant menu was used for the lesson. The 90-minute lesson consisted of three stages. First the students were given time to prepare a conversation with peers about ordering in a restaurant in order to activate their schemata of restaurant situations (10 minutes). Then a Japanese instructor explained the procedures of the simulation tasks (10 minutes). After that, the students were divided into two groups, waitresses and customers. They changed roles half way through the session (35 minutes each). Customers were to visit any three out of five different simulation restaurant sets in the classroom, and order their favourite meals using a real English menu at each place. These menus were different at each restaurant set. NSs also had a turn playing the role of waiter at a restaurant set. The waiter or waitress had to respond to the customer‘s request and take their order. Students who played each role could not guess what kind of communication would take place until they ordered.

Data Collection The database for this study was derived from videotapes of student-student (S-S), and native teacher-student (N-S) conversations collected from observation of the classroom task. Video recorders fixed in place at two of the restaurant settings (one for S-S, one for N-S) provided the data for this study. The researcher was not present during taping so that data could be collected as unobtrusively as possible. Transcriptions of the conversations were made, carefully checked and then analysed.

232

Yasuo Nakatani

Analyses The data examined were ten transcripts from the videotaped task; five from S-S interactions (Transcripts 1 to 5) and five from N-S interactions (Transcripts 6 to 10) from the two video-fixed sets. The length of talking time was slightly different according to the pairs (8-9 minutes). I transcribed the whole of each pair‘s conversation and used them for the current data analysis. First, the number of turns which indicated how often each communication pair exchanged their utterances were counted. Second, interlocutors‘ amount of speech production was measured by communication units (c-units), which are defined by Brock (1986: 52) as independent utterances that have referential or pragmatic meaning. C-unit analysis treats utterances that are ungrammatical or uncompleted as meaningful segments, as long as they have pragmatic meaning in the discourse. Thirdly, the number of signals for negotiation, and modified output defined as response triggered by an interlocutor‘s prior utterances were assessed. The features of negotiation of meaning used in the analysis of the data collected for the present study were the same as those used in current established research methods (e.g. Doughty and Pica, 1986; Pica et al., 1996). They included clarification requests, confirmation checks, comprehension checks and modified output. In making clarification requests, the listener asks for clarification when he or she does not entirely comprehend the meaning, as in the following example form the data of the present study: Customer: What is this? (She points to the item on the menu.) Waitress: Where is that? Confirmation checks occur when the listener believes he or she has understood but would like to make sure: Customer: A cup of cappuccino, please. Waitress: Hot or ice? Customer: Ice? You have ice cappuccino? In making comprehension checks, the speaker wants to be certain that the listener has understood: Customer: Jumbo hamburger. Waitress: Jumbo hamburger. It‘s very big. OK?

The number of these components indicated how often each pair engaged in negotiation of meaning.

Simulation Tasks

233

Results and Discussion By analysing the three categories above, namely, the number of turns, production rate and the quality and quantity of negotiation of meaning, I examined how well S-S and N-S demonstrated spontaneous communication to construct and maintain interaction with others. The results are presented below through statistical analysis and the excerpts of transcribed discourse. Commentaries are offered, where appropriate.

Number of Turns As shown in Table 1, S-S operated more turn taking than N-S did, and this difference was shown to be significant by an independent two-tailed t-test (p< 0.025). N-S finished their task within a smaller number of turns than S-S. NSs took more initiative in their interaction and tended to avoid sources of conversation redundancy, such as unnecessary repetition, awkward expressions and long pauses. NSs actively helped students understand better. The students in the N-S pair seemed to be more reluctant to contribute than the students in S-S pairs. This may have been due to their lack of experience in talking with NSs. Another reason may have been that the items were unfamiliar to both learners in S-S and they had to exchange more information to achieve mutual communication goals. As a whole, in S-S, students seemed to experience less anxiety and felt freer to operate ―conversation catch ball‖ with other students. Examples can be seen in Excerpts 1 and 2. Table 1. Number of turns in the restaurant simulation task and the result of independent two-tailed t-test

S-S

Number of Turns

N-S

Number of Turns

Transcript 1

57

Transcript 6

18

Transcript 2

38

Transcript 7

28

Transcript 3

33

Transcript 8

20

Transcript 4

41

Transcript 9

19

Transcript 5

29

Transcript10

18

Total

198

103

Mean

36.9

20.6

SD

10.8

4.2

t = 3.02, d f = 4

p < 0.025

Excerpt 1 below illustrates S-S negotiation with many turns.

234

Yasuo Nakatani Excerpt 1 (a part of Transcript 1) S1: a student as a customer, S2: a student as a waitress S1: I‘d like black bean cakes. S2: Which one? [looks at the menu] S1: This one. [points to the item on the menu] S2: Funny name, isn‟t it? [looks at the item on the menu] S1: Yeah. You named this? S2: …[takes the order] S1: No? S2: What? S1: You named this? S2: No. Because this is part time… part time job. [laughs]

Commentary S2 tries to make clear which item S1 orders because the name of the item sounds strange to her. She expands their interaction by using a tag question. S1 reacts to S2‘s tag question and asks whether the waitress has thought of this name herself. However, S2 concentrates on taking the order and does not realise this signal. In order to fill this communication gap, S1 checks comprehension of S2 by changing her question to a simple word, ―No?‖ Then, S2 conducts a clarification request by saying, ―What?‖ S2 simply repeats her previous question in order to confirm whether the waitress understands her question. This time, S2 receives this signal successfully. She creates spontaneous conversation with the excuse that she works part time, and that is why the item on the menu is not familiar to her, either.

[ ]: non-linguistic behaviour, …: pause In Excerpt 2, an example of a small number of turns can be seen. Excerpt 2 (a part of Transcript 6) S3: a student as a customer, N1: a NS as a waiter N1: Would you like anything on the side? S3: Can I have…split pea…? [looks at the waiter] N1: Split pear soup. [nods and takes the order] Would you like anything to drink?

[ ]: non-linguistic behaviour, …: pause

Commentary S3 has a problem understanding an item on the menu well. With rising intonation, she sends a message that she is not sure about the item but she would like to try it. The item is quite familiar to N1, and so he helps S3 by just adding the term ―soup‖ with slight stress and takes the order and then moves to the next subject. N1 successfully achieves his communication goal with simple interaction without causing S3 to lose face.

Simulation Tasks

235

Production As can be seen in Table 2, the students produced more utterances in S-S pairs than they did in N-S pairs. The mean total of utterances of S-S pairs and N-S pairs was significantly different (independent two-tailed t-test, p 20. This cutoff was determined a posteriori to exclude the least number of participants while matching Spanish and English language skills as much as possible. The resulting language-balanced grouping was comprised of 18 participants with difference scores ranging from +13 to -18 and equal numbers either side of zero. Language skills were still not quite equally balanced, as indicated by a summed group WMLS-R difference score of -27 points (i.e., 1.5 points per participant in favor of English). Language of administration at the first session was counterbalanced within the grouping.

Results WMLS-R CALP scores ranged from 2 to 5 in Spanish and 3 to 5 in English. Mean CALP scores were 3.7 (SD = .6) in Spanish and 3.8 (SD = .5) in English. Both these means are in the Limited to Fluent range. They did not differ significantly t(17) = .9, ns. Table 2 shows the mean Spanish and English WISC-IV subtest scores. They were analyzed in a 2 (language) x 10 (subtest score) repeated measures ANOVA. The main effect of language, F(1, 17) < 1, ns, η² = .03 and the language x subtest score interaction, F(9, 153) = 1.7, ns, η² = .1 were not significant. The main effect of subtest score was significant F(9, 153) = 5.9, p < .01, η2 = .3. Post hoc comparisons using the Bonferroni procedure showed the combined Spanish and English MR mean (11.0) was significantly higher than the corresponding means for VO (7.4), BD (7.6), and CO (8.3) at p < .05. No other comparison among the subtest scores was significant. Table 3 shows the mean Spanish and English WISC-IV IQ indices. They were analyzed in a 2 (language) x 4 (IQ index) repeated measures ANOVA. The main effect of language, F(1, 17) < 1, ns, η² = .01 and the language x IQ index interaction, F(3, 51) = 2.5, ns, η² = .13 were not significant. The main effect of IQ index was significant F(3, 51) = 4.8, p < .05, η2 = .22. Post hoc comparisons using the Bonferroni procedure showed the combined Spanish and English VCI mean (89.3) was significantly below the corresponding PSI mean (104.6) at p < .05. No other comparison among the IQ indices was significant.

457

Bilingualism and Hispanic American Intelligence Test Scores

Table 2. Mean (Standard Deviation) Spanish and English WISC-IV Subtest Scores Obtained by a Language-Balanced, Bilingual Grouping (N = 18). The National Mean is 10 (3) for Every Subtest Spanish

English

Similarities

8.0 (2.0)

9.5 (2.9)

SpanishEnglish Difference -1.5

Vocabulary

7.0 (2.2)

7.9 (2.7)

-0.9

7.4 *

Comprehension

8.1 (3.2)

8.6 (3.2)

-0.5

8.3 ˆ

Block Design

7.7 (2.3)

7.5 (2.3)

+0.2

7.6˚

Picture Concepts

9.7 (3.6)

9.6 (3.9)

+0.1

9.7

Matrix Reasoning

10.8 (3.9)

11.2 (3.8)

-0.4

11.0 * ˆ ˚

Digit Span

9.6 (2.1)

9.0 (2.4)

+0.6

9.3

Letter-Number-Sequencing

9.8 (3.2)

9.2 (4.5)

+0.6

9.5

Coding

11.6 (4.4)

11.3 (4.6)

+0.2

11.5

Symbol Search

9.9 (3.2)

10.2 (3.3)

-0.3

10.1

Combined Mean 8.7

*, ˆ, ˚ denotes pairs of means that differ at p < .05. Key: WISC-IV = Wechsler Intelligence Scale for Children, Fourth Edition.

Table 3. Mean (Standard Deviation) IQ Indices for the Spanish and English Versions of the WISC-IV Obtained by a Language Balanced Bilingual Grouping (N = 18). The National Mean is 100 (15) for Every IQ Index Spanish

English

Spanish – English Difference

Combined Mean

Verbal Comprehension Index

86.8 (12.5)

91.9 (15.3)

-5.1

89.3*

Perceptual Reasoning Index

96.4 (17.5)

96.7 (18.8)

-0.2

96.6

Working Memory Index

97.4 (13.4)

94.0 (17.9)

+3.4

95.8

Processing Speed Index

104.6 (19.6)

104.7 (19.9)

-0.1

104.6*

Full-Scale IQ

94.1 (14.5)

95.2 (19.7)

-1.1

94.7

IQ Index

* p < .05. Key: WISC-IV = Wechsler Intelligence Scale for Children, Fourth Edition.

Discussion Studies on adults (e.g., Gasquoine et al., 2007; Gollan et al., 2002; Harris et al., 1995; Rosselli et al., 2002) and children (e.g., Naglieri, Otero, DeLauder, and Matto, 2007) have shown that groupings of balanced bilingual participants perform equally well on cognitive tests in both languages. This finding was supported here as there were no significant differences between any pair of corresponding Spanish and English WISC-IV subtest or IQ

458

Philip G. Gasquoine, Aracely Cavazos, Juan Cantu et al.

index scores. When Spanish and English scores were combined, Vocabulary, Comprehension, and Block Design subtest scores were significantly below Matrix Reasoning and the Verbal Comprehension Index was significantly below the Processing Speed Index. Effect sizes were small in, but these results support the concept that Hispanic Americans perform better when tests emphasize visual-perceptual/visual-motor compared to language processing. Support was not unequivocal as participants scored relatively poorly on the Block Design subtest, a visual-constructional task. The visual-perceptual/visual-motor advantage occurred in both languages, suggesting it is a consequence of some aspect of the bilingual experience and is not due to limited English language proficiency. The size of the language vs. visual-perceptual/visual-motor difference was of the order of 1SD. Specifically, the largest significant differences for subtest and index scores respectively were: (a) the Matrix Reasoning subtest score (11.0) was 1.2 SD units higher than the Vocabulary subtest score (7.4); and (b) the Processing Speed Index (104.6) was 1 SD unit higher than the Verbal Comprehension Index (89.3). This is about double the size of the previously reported visual-perceptual/visual-motor advantage for Hispanic Americans in English language intelligence test standardization samples (e.g., Neisser et al., 1996; Prifitera et al., 1998; Puente and Salazar, 1998).

Conclusion Bialystok (2007) posed the question: “Are bilinguals more intelligent than monolinguals?” (p. 220). The answer appears to be “no”, at least if intelligence is defined as what is measured on popular standardized intelligence tests, a definition that is not without its critics (e.g., Sternberg, 2004). Studies investigating the effect of bilingualism on cognition have consistently shown a bilingual over monolingual advantage in response inhibition tasks, but there is a monolingual over bilingual advantage on tasks that emphasize language processing. Popular standardized intelligence tests like the Wechsler scales do not have items that require response inhibition, whereas approximately half the total number of items emphasize language processing skills. Despite low scores on intelligence tests, there are major social and economic advantages to bilingualism, especially in borderland regions where individuals speaking two different languages intermix. Analysis of two data sets of Wechsler intelligence test scores from bilingual, Hispanic American residents of the Rio Grande Valley region of Texas suggested a visualperceptual/visual-motor over language processing advantage of about 1SD. Scores on tests emphasizing visual-perceptual/visual-motor processing skills were in the average range of intellectual ability and were consistent with educational levels, but scores on tests emphasizing language processing were low. The low scores occurred with both English and Spanish language administrations and so were more likely bilingualism-related than a consequence of limited English language proficiency. The size of the visualperceptual/visual-motor over language processing advantage was about double that typically reported for Hispanic American groupings in intelligence test standardization samples (in which the extent of bilingualism is unknown). Analysis of data sets, as was done here, provides an indirect link between bilingualism and intelligence test scores of Hispanic Americans. The analysis has suggested bilingualism is a factor, but does not specifically implicate it. Other potential explanations for the visual-

Bilingualism and Hispanic American Intelligence Test Scores

459

perceptual/visual-motor over language processing advantage include relevant demographic characteristic (like SES or quality of education) unique to the Rio Grande Valley region of Texas where the participants were resident. This is a community that is evolving culturally secondary to large numbers of recent immigrants. There is no reason to assume that immigrants, as a grouping, formed a normal distribution of cognitive skills in their country of origin. In fact it is more likely that the opposite is the case. A more direct test of the relationship between bilingualism and intelligence test scores would come from studies that compared the intelligence test scores of demographically matched groupings of monolingual and bilingual Hispanic Americans. Intelligence test developers could also facilitate study of this issue by reporting the frequency of bilingualism in standardization samples as assessed by self-report. The practical implications of low scores by Hispanic Americans on intelligence test measures that emphasize language processing are not limited to special education placement in K through 12. Intelligence test scores tend to correlate highly with achievement test scores such as the Scholastic Assessment Test (SAT) and the Graduate Record Examination (GRE) that are used to determine admission to college. Bilingual vs. monolingual group differences in these achievement test scores could result in disproportionate numbers of Hispanic Americans being excluded from college, an admission policy that would violate the 14th Amendment. Solutions to this problem are complex, involving variants of affirmative action (Gasquoine, 2008). Why bilingualism might lead to low scores on language based IQ/achievement test measures remains a matter of conjecture. The most plausible explanation developed to date is that parallel activation of both languages causes interlanguage interference thereby slowing processing time and increasing the possibility of errors under certain stressful environmental conditions such as those involved with IQ/achievement testing (Bialystok et al., 2008).

Author Note This research was supported by a Faculty Research Grant from the University of Texas – Pan American to the first author. Correspondence should be addressed to Philip G. Gasquoine, Ph.D., Graduate Psychology Program Director, Department of Psychology and Anthropology, University of Texas – Pan American, 1201 W. University Drive, Edinburg, Texas 78541. Electronic mail may be sent to [email protected].

References Ardila, A., Rosselli, M., Ostrosky-Solis, F., Marcos, J., Granda, G., and Soto, M. (2000). Syntactic comprehension, verbal memory, and calculation abilities in Spanish-English bilinguals. Applied Neuropsychology, 7, 3-16. Artiola i Fortuny, L., Hermosillo, D., Heaton, R.K., and Pardee, R.E. (1999). Manual de normas y procedimientos para la bateria neuropsicologica en Espanol. Tucson, AZ: m Press.

460

Philip G. Gasquoine, Aracely Cavazos, Juan Cantu et al.

Bialystok, E. (1999). Cognitive complexity and attentional control in the bilingual mind. Child Development, 70, 636-644. Bialystok, E. (2002). Acquisition of literacy in bilingual children: A framework for research. Language Learning, 52, 159-199. Bialystok, E. (2007). Cognitive effects of bilingualism: How linguistic experience leads to cognitive change. International Journal of Bilingual Education and Bilingualism, 10, 210-223. Bialystok, E., Craik, F., and Luk, G. (2008). Cognitive control and lexical access in younger and older bilinguals. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34, 859-873. Bialystok, E., Martin, M. M., and Viswanathan, M. (2005). Bilingualism across the lifespan: The rise and fall of inhibitory control. International Journal of Bilingualism, 9, 103-119. Bialystok, E., and Senman, L. (2004). Executive processes in Appearance-Reality tasks: The role of inhibition of attention and symbolic representation. Child Development, 75, 562579. Boake, C. (2002). From the Binet-Simon to the Wechsler-Bellevue: Tracing the history of intelligence testing. Journal of Clinical and Experimental Neuropsychology, 24, 383-405. Brown, L., Sherbenou, R. J., and Johnson, S. K. (1990). TONI-2. Test of Nonverbal Intelligence. Examiner’s manual. Austin, TX: Pro-edn. Costa, A., Hernandez, M., and Sebastian-Galles, N. (2008). Bilingualism aids conflict resolution: Evidence from the ANT task. Cognition, 106, 59-86. Crawford et al. v. Honig et al. (1994). 37 F. 3d 485, 487. Diana v. State Board of Education. (1973, June 18). C-70-37 RFP (Northern District of California) (Stipulated settlement). Dickens, W. T., and Flynn, J. R. (2006). Black Americans reduce the racial IQ gap: Evidence from standardization samples. Psychological Science, 17, 913-920. Farah, M. J., Shera, D. M., Savage, J. H., Betancourt, L., Giannetta, J. M., Brodsky, N. L., et al. (2006). Childhood poverty: Specific associations with neurocognitive development. Brain Research, 1110, 166-174. Fernandez, K., Boccaccini, M. T., and Noland, R. M. (2007). Professionally responsible test selection for Spanish-speaking clients: A four-step approach for identifying and selecting translated tests. Professional Psychology: Research and Practice, 38, 363-374. Flaugher, R. L. (1978). The many definitions of test bias. American Psychologist, 33, 671679. Gasquoine, P. G. (1999). Variables moderating cultural and ethnic differences in neuropsychological assessment: The case of Hispanic-Americans. Clinical Neuropsychologist, 13, 376-383. Gasquoine, P. G. (2001). Research in clinical neuropsychology with Hispanic American participants: A review. Clinical Neuropsychologist, 15, 2-12. Gasquoine, P. G. (2008). Psychology of multicultural competence: Critical issues. New York: Linus Publishing. Gasquoine, P. G., Croyle K. L., Cavazos-Gonzalez, C., and Sandoval, O. (2007). Language of administration and neuropsychological test performance in neurologically intact Hispanic American bilingual adults. Archives of Clinical Neuropsychology, 22, 991-1001. Goetz, P. J. (2003). The effects of bilingualism on theory of mind development. Bilingualism: Language and Cognition, 6, 1-15.

Bilingualism and Hispanic American Intelligence Test Scores

461

Gollan, T. H., and Acenas, L. A. (2004). What is a TOT? Cognate and translation effects on tip-of-the-tongue states in Spanish-English and tagalong-English bilinguals. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30, 246-269. Gollan, T. H., Fennema-Notestine, C., Montoya, R. I., and Jernigan, T. L. (2007). The bilingual effect on Boston Naming Test performance. Journal of the International Neuropsychological Society, 13, 197-208. Gollan, T. H., Montoya, R. I., Fennema-Notestine, C., and Morris, S. K. (2005). Bilingualism affects picture naming but not picture classification. Memory and Cognition, 33, 12201234. Gollan, T. H., Montoya, R. I., and Werner, G. (2002). Semantic and letter fluency in SpanishEnglish bilinguals. Neuropsychology, 16, 562-576. Gregory, R. J. (2007). Psychological testing: History, principles, and applications (5th ed.). Boston: Allyn and Bacon. Grosjean, F. (1998). Studying bilinguals: Methodological and conceptual issues. Bilingualism: Language and Cognition, 1, 131-149. Harris, J. G., Cullum, C. M., and Puente, A. E. (1995). Effects of bilingualism on verbal learning and memory in Hispanic adults. Journal of the International Neuropsychological Society, 1, 10–16. Harris, J. G., Wagner, B., and Cullum, C. M. (2007). Symbol vs. digit substitution task performance in diverse cultural and linguistic groups. Clinical Neuropsychologist, 21, 800-810. Heaton, R. K., Chelune, G. J., Talley, J. L., Kay, G. G., and Curtiss, G. (1993). Wisconsin Card Sorting Test manual: Revised and expanded. Odessa, FL: Psychological Assessment Resources. Hosp, J. L., and Reschly, D. J. (2004). Disproportionate representation of minority students in special education: Academic, demographic, and economic predictors. Exceptional Children, 70, 185-199. Individuals with Disabilities Education Act. (2004). 20 U.S.C., 1400. Jensen, A. R. (1969). How much can we boost IQ and scholastic achievement? Harvard Educational Review, 39, 1-123. Kaplan, R. M., and Saccuzzo, D. P. (2005). Psychological testing: Principles, applications, and issues, (6th ed.). Belmont, CA: Thompson. Kovacs, A. M. (2007). Beyond language: Childhood bilingualism enhances high-level cognitive functions. In I. Keckes and L. Albertazzi (Eds.), Cognitive aspects of bilingualism. New York: Springer. Larry P. v. Wilson Riles. (1979). 343 F. Supp. 1306 (Northern District California 1972), affd 502 F. 2d 963 (9th Circuit). Lezak, M. D., Howieson, D. B., and Loring, D. W. (2004). Neuropsychological assessment (4th ed.). New York: Oxford University Press. Miller, G. A., and Chapman, J. P. (2001). Misunderstanding analysis of covariance. Journal of Abnormal Psychology, 110, 40-48. Naglieri, J.A., Otero, T., DeLauder, B., and Matto, H. (2007). Bilingual Hispanic children’s performance on the English and Spanish versions of the Cognitive Assessment System. School Psychology Quarterly, 22, 432-488. National Research Council. (2002). Minority students in special and gifted education. Washington, DC: National Academy Press.

462

Philip G. Gasquoine, Aracely Cavazos, Juan Cantu et al.

Neisser, U., Boodoo, G., Bouchard, T. J., Boykin, A. W., Brody, N., Ceci, S. J., et al. (1996). Intelligence: Knowns and unknowns. American Psychologist, 51, 77-100. Noble, K. G., McCandliss, B. D., and Farah, M. J. (2007). Socioeconomic gradients predict individual differences in neurocognitive abilities. Developmental Science, 10, 464-480. Overton, T., Fielding, C., and Simonsson, M. (2004). Decision making in determining eligibility of culturally and linguistically diverse learners: Reasons given by assessment personnel. Journal of Learning Disabilities, 37, 319-330. Prifitera, A., and Saklofske, D. (1998). WISC-III clinical use and interpretation: Scientistpractitioner perspectives. San Diego, CA: Academic Press. Prifitera, A., Weiss, L. G., and Saklofske, D. H. (1998). The WISC-III in context. In A. Prifitera and D. H. Saklofske (Eds.) WISC-III: Clinical use and interpretation: Scientistpractitioner perspectives. San Diego, CA: Academic Press, pp. 1-38. Puente, A. E., and Salazar, G. D. (1998). Assessment of minority and culturally diverse children. In A. Prifitera and D. H. Saklofske (Eds.) WISC-III clinical use and interpretation: Scientist-practitioner perspectives. San Diego, CA: Academic Press, pp. 227-248. Rhodes, R. L., Ochoa, S. H., and Ortiz, S. O. (2005). Assessing culturally and linguistically diverse students: A practical guide. New York: Guilford Press. Roberts, P. M., Garcia, L. J., Desrochers, A., and Hernandez, D. (2002). English performance of proficient bilingual adults on the Boston Naming Test. Aphasiology, 16, 635-645. Roid, G. H., and Miller, L. J. (1995). Leiter International Performance Scale – Revised. Wood Dale, IL: Stoelting. Rosselli, M. A., Ardila, A., Santisi, M. N., Del Rosario Arecco, M., Salvatierra, J., Conde, A., et al. (2002). Stroop effect in Spanish-English bilinguals. Journal of the International Neuropsychological Society, 8, 819–827. Rushton, J. P., and Jensen, A. R. (2006). The totality of available evidence shows that the race IQ gap still remains. Psychological Science, 17, 921-922. Schmidt, F. L., and Hunter, J. E. (1974). Racial and ethnic bias in psychological tests: Divergent implications of two definitions of test bias. American Psychologist, 29, 1-13. Sternberg, R. J. (2004). Culture and intelligence. American Psychologist, 59, 325-338. Strickland, B. R. (2000). Misassumptions, misadventures, and the misuse of psychology. American Psychologist, 55, 331-338. Stuss, D. T., Gallup, G. G., Jr., and Alexander, M. P. (2001). The frontal lobes are necessary for ‘theory of mind’. Brain, 124, 279-286. U.S. Bureau of the Census. (2000). Population reports: 2000. Washington, DC: Author. Valdes, G., and Figueroa, R. A. (1996). Bilingualism and testing: A special case of bias. Norwood, NJ: Ablex. van de Vijver, F. J. R., and Phalet, K. (2004). Assessment in multicultural groups: The role of acculturation. Applied Psychology: An International Review, 53, 215-236. Wechsler, D. (1991). Wechsler Intelligence Scale for Children - Third Edition. San Antonio, TX: Psychological Corporation. Wechsler, D. (1997). Wechsler Adult Intelligence Scale – Third Edition. San Antonio, TX: Psychological Corporation. Wechsler, D. (2003). Wechsler Intelligence Scale for Children – Fourth Edition. San Antonio, TX: Psychological Corporation.

Bilingualism and Hispanic American Intelligence Test Scores

463

Wechsler, D. (2005). Wechsler Intelligence Scale for Children – Fourth Edition, Spanish. San Antonio, TX: Psychological Corporation. Williams, W. M., and Ceci, S. J. (1997). Are Americans becoming more or less alike? Trends in race, class, and ability differences in intelligence. American Psychologist, 52, 11261235. Woodcock, R. W., Munoz-Sandoval, A. F., Ruef, M. L., and Alvarado, C. G. (2005). Woodcock-Munoz Language Survey – Revised. Itasca, IL: Riverside Publishing.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 20

LANGUAGE DEVELOPMENT THROUGH A BILINGUAL LENS Eswen Fava and Rachel Hull∗ Department of Psychology, Texas A and M University College Station, TX, USA

Abstract Bilingualism is a fertile resource for studying facets of language development and brain plasticity that may not be apparent in monolinguals. This chapter will summarize historical and contemporary findings in the literature, discuss methodological issues that influence their interpretation, and suggest future directions for examining the neural substrates of language development and the consequences of having two languages in one brain.

Introduction Historical Context: Where Have We Come From? The notion that language “lives” in the left hemisphere (LH) emerged from the classic observations of Broca (1861) and Wernicke (1911), who found that patients with (LH) focal lesions showed consistent and systematic behavioral deficits for language. Over time, additional lesion deficit data have largely supported the early findings of Broca and Wernicke, and the assumption that language is typically a left-lateralized processing event has become a generally accepted tenet. However, there has been a noteworthy subset of cases that do not follow the predicted pattern. For instance, numerous studies have found that LH lesions sustained in early childhood do not typically result in language deficits or pronounced speech problems (e.g., Thal, 1997; Vargha-Khadem, O'Gorman, and Watters, 1985; Woods and Carey, 1979, 2000; Woods and Teuber, 1978). Other studies have found the same pattern ∗

E-mail address: [email protected]. 979-845-7146 (phone), 979-845-4727 (fax),

466

Eswen Fava and Rachel Hull

in young (but not older) patients and have suggested that reorganization of language representation from the LH to the right hemisphere (RH) is more likely if injury occurs before the age of five (Duncan et al., 1997; Muller et al., 1999; Muller, Rothermel, Behen et al., 1998; Muller, Rothermel, Muzik et al., 1998). Although it was initially assumed that early LH injury caused reorganization only when lesions were extensive or impinged upon language areas, some studies have indicated that a reorganization of left lateralized language processing to the RH can occur in children even when lesions are remote from classic language areas. For example, children with motor deficits (stemming from LH motor cortex damage) showed an atypical left ear advantage during a dichotic listening language task, indicating that the RH was more involved in their language processing (Isaacs, Christie, Vargha-Khadem, and Mishkin, 1996). There has also been some neuroimaging evidence that the brain may bring about a change in lateralization for language even when damage does not involve areas traditionally thought to be critical for language processing. For instance, an fMRI study by DeVos et al., (1995) found an association between RH localization for language processing and childhood onset tumors in the LH (see also Liégeois et al., 2004; Springer et al., 1999). At present, the reason for the association between remote lesions and profound changes in language localization remains unclear, although the involvement of frontal areas in redistribution of language processing has been suggested (Dehaene et al., 1997; Hahne and Friederici, 2001; Hernandez, Martinez, and Kohnert, 2000; Liégeois et al., 2004; Weber-Fox and Neville, 2001; Weber-Fox and Neville, 1999). What is clear, however, is that the brain has the capacity to respond to different kinds of LH damage by redistributing language function to healthy RH areas, particularly when the damage occurs during early development (Booth et al., 2000; Staudt et al., 2002). Thus, even if one accepts that the LH is dominant for language processing in most brain-intact individuals, there is certainly evidence to suggest that neural plasticity can operate to reorganize language representation in response to brain damage. Moreover, quite apart from pathology, it is known that the degree of hemispheric dominance varies with differences in genetic predispositions, developmental experiences, gender, and environment (Springer et al., 1999). Given that early brain damage can alter brain organization for language, it is important to consider what other substantial early experiences could have a similar effect. An obvious experience to examine would be early exposure to more than one language system. That is, what might the neural repercussions be of fitting two language systems into one developing brain? It is critical to note that the lesion-deficit outcomes discussed thus far have almost exclusively involved monolingual patients; indeed, the vast majority of patients whose data contributed to the established belief that language is constrained to the left hemisphere were monolingual (see review by Martin, 2000). Given that bilinguals can be expected to share all of the variations found among monolinguals, as well as additional variations arising from circumstances unique to bilingualism (e.g., the age of second language learning, proficiency in the second language, similarity of the first and second language), it seems reasonable that we might expect more variation in brain organization for language in bilinguals, particularly when the onset of bilingualism occurs during early development. Indeed, hints of a difference between monolinguals and bilinguals in terms of the neural architecture of language were suggested when lesion deficit studies began to appear with bilingual aphasics. Both early (Borod, Koff, Lorch, and Nicholas, 1985; Geschwind and Galaburda, 1985; Rasmussen and Milner, 1977) and more recent studies of bilinguals with

Language Development through a Bilingual Lens

467

brain damage (Booth et al., 2000; Liégeois et al., 2004; Naeser et al., 2004) seem to indicate evidence of brain plasticity for language beyond that suggested in monolingual lesion studies. For instance, numerous case studies of bilingual aphasics have shown differential post-morbid recovery of the two languages (Abutalebi, Miozzo, and Cappa, 2000; Fabbro, 2001; Vaid, 2002; Vaid and Hull, 2002). In addition, some bilingual aphasics can access one language (but not the other) on a given day, then show the reverse pattern later (Paradis, 1998). There is also some limited evidence to suggest that bilinguals show a higher frequency of crossed aphasia (language deficits following damage to the RH) relative to monolinguals (Karanth and Rangamani, 1988; Vaid, 2002). Although lesion deficit studies can point toward the operation of brain plasticity for language, they leave unclear whether specific language deficits are the result of trauma to a specialized brain component at the lesion site, or if the damaged area is simply part of a larger neural network that mediates a given component of language (Abutalebi, Cappa, and Perani, 2001; Szaflarski, Holland, Schmithorst, and Byars, 2006; Vaid and Hull, 2002). Moreover, although lesion deficit studies can provide time-slice glimpses of static circumstances that are valuable in refining theories and models of language representation, the literature is poorly positioned to analyze language development as a process (Brown et al., 2005; Holland et al., 2001; Schlaggar et al., 2002; Springer et al., 1999; Szaflarski et al., 2002). Nevertheless, these studies taken together suggest specific directions for investigating how brain plasticity may give rise to differential development of language representation in the face of profoundly different language experiences, and identify bilingualism as a unique circumstance to explore the brain’s full potential for developing a neural architecture for language.

Contemporary Context: Where Are We Now? In response to the compelling evidence from the lesion deficit literature, studies of the bilingual brain have escalated dramatically in the last three decades. The theoretical approaches that have dominated bilingualism research have centered on either the age of second language acquisition or proficiency in the second language, and several hypotheses have been introduced to explain findings in the literature. One of the first was the Age Hypothesis (Genesee, Hamers, Lambert, Mononen, Seitz, and Starck, 1978; Vaid and Genesee, 1980), which advocates cognitive and developmental states as the principle mechanisms that drive functional brain organization for multiple languages. In general, the Age Hypothesis states that the more similar the cognitive and brain maturational states are when the first and second languages are acquired, the more similar will be their functional organization. Thus, the Age Hypothesis posits that the age of second language (L2) acquisition will predict brain functional organization for language. Another of the earliest hypotheses, the Stage Hypothesis (Albert and Obler, 1978; Galloway and Krashen, 1980; Obler, 1981; Schneiderman, 1986), focused instead on L2 proficiency as the principle mechanism driving bilingual language organization. In this view, low L2 proficiency is thought to be associated with a heavier reliance on RH strategies that involve using contextual cues to meaning, whereas increased proficiency yields a shift to LH dominance as syntactic and phonological processing become more automated (see also Ullman, 2004). Thus, the Stage Hypothesis says that L2 proficiency will predict brain functional organization for language.

468

Eswen Fava and Rachel Hull

More recent hypotheses have attempted to reconcile the Age and Stage approaches. Based on outcomes from two comprehensive meta-analyses of the behavioral bilingual laterality literature, Hull and Vaid (2007) have introduced the Anchoring Hypothesis, which states that the neural architecture of language is established early in development, during a period when the brain is especially plastic (i.e., before the age of six), that this pattern persists into adulthood, and that it anchors later-learned languages to the same pattern. More specifically, for people who learn only one language during early childhood (i.e., monolinguals and late bilinguals who learn only one language before the age of 6), the prevailing pattern is LH dominance, and consequently any languages learned after age 6, regardless of proficiency, will also be LH dominant. However, the degree of LH dominance in late bilinguals is predicted to vary as a function of L2 proficiency; that is, low-proficiency late bilinguals are expected to show more LH involvement relative to their high-proficiency counterparts. In contrast, those who learn two (or more) languages during early childhood (i.e., before age 6) are expected to demonstrate bilateral involvement during the processing of both languages, and also for any languages learned later in life. Hull and Vaid argue that the increased involvement of the RH in early bilinguals may be a consequence of the precocious development of RH-specialized pragmatic strategies to support dual-language processing (e.g., monitoring social usage of the two languages; see Beeman and Chiarello, 1998; Boatman, 2004; Genesee, 1980; Hull and Vaid, 2006; Obler, 1981); thus, the relatively early establishment of and reliance upon these RH pathways is thought to yield a pattern of bilateral involvement for language processing that persists through the lifespan. Two additional hypotheses that predict language organization differences in monolinguals and early bilinguals have recently emerged. The Signature Hypothesis (Kovelman, Baker, and Petitto, 2008) suggests that early, proficient bilinguals will recruit different neural tissue than monolinguals across all language processing contexts. This hypothesis is based largely on behavioral language mastery differences in brain-intact bilinguals and takes the theoretical approach that the languages of early, proficient bilinguals are subserved by neural substrates distinct from those of monolinguals. In contrast, the Functional Switching Hypothesis (Kovelman et al., 2008), which is based on differences in language impairment or recovery patterns among bilingual aphasics, assumes that early, proficient bilinguals will demonstrate recruitment of neural tissue similar to that of monolinguals, but only when bilinguals are processing language in a monolingual context (i.e., while suppressing the unused language). In a bilingual context (e.g., switching between languages), bilinguals are predicted to recruit additional and distinct neural tissue. This hypothesis is consistent with theoretical accounts that suggest bilinguals preferentially rely on an attentional, language switching mechanism that exists outside the language systems per se (Green, 1998; Paradis, 2000). In a test of the Signature and Functional Switching hypotheses, Kovelman et al. (2008) used Near Infrared Spectroscopy (NIRS) to examine whether sustained, early exposure to two languages altered the brain’s organization of processing language. They found that in a monolingual context, classic language areas such as Broca’s Area showed similar recruitment in monolinguals and early bilinguals, but only the bilinguals also showed bilateral recruitment of frontal areas associated with working memory, namely, dorsolateral prefrontal cortex (DLPFC) and inferior frontal cortex (IFC). In a bilingual context, early bilinguals demonstrated a similar pattern as during the monolingual context, but they also showed relatively increased recruitment of the RH working memory areas. The authors concluded that

Language Development through a Bilingual Lens

469

the outcomes of their studies were consistent with their Signature Hypothesis (but not their Functional Switching Hypothesis). The results of these studies were also consistent with the Anchoring Hypothesis. In what follows, we summarize other findings from the contemporary bilingualism literature that support the Anchoring and Signature hypotheses. Behavioral evidence. Data from behavioral bilingualism studies, as a general rule, have the benefit of well-established paradigms and tasks that demonstrate validity in their measurement of subtle constructs such as pragmatics, lexico-semantics, and syntax. The three most commonly used behavioral paradigms for inferring the lateralization of languages have been dichotic listening, tachistoscopic viewing and dual task, where each method typically relies on a different response measure (accuracy, reaction time, and interference size, respectively). In dichotic listening studies, participants typically hear words in one or the other ear, and accuracy of recall by ear is the dependent variable. In visual half-field studies, words are presented to one or the other visual field, and the dependent variable is either recognition speed or speeded word judgments of different sorts for stimuli varying on visual, phonetic, semantic or syntactic dimensions. Finally, dual task studies typically measure how much interference finger tapping with the left- or right-hand finger (hence engaging the right or left hemisphere, respectively) produces on a concurrent language task, such as reading for comprehension. Behavioral studies and reviews of hemispheric involvement for language in brain-intact individuals have consistently suggested differences in laterality between monolinguals and at least some types of bilinguals (Bialystock and Ryan, 1985; Birdsong and Molis, 2001; Goral, Levy, and Obler, 2002; Hull and Vaid, 2005; Johnson and Newport, 1989; Long, 1990; Newport, Bavelier, and Neville, 2001; Vaid, 1984, 1987; Vaid and Frenck-Mestre, 1991). Especially in light of outcomes from the bilingual aphasia literature, the behavioral results suggest that language representation differences might arise from developmental sources (consistent with the Anchoring and Signature hypotheses). In order to identify and quantify any systematic differences in underlying patterns of language representation in the brain, two recent meta-analyses examined the behavioral language laterality literature and identified several consistent patterns across experimental studies. Hull and Vaid (2006) specifically compared behavioral studies of monolinguals and bilinguals, coding also for variables such as testing paradigm (dichotic listening, visual hemifield, dual task), age of L2 acquisition (early, late), and level of L2 proficiency (proficient, non-proficient). They observed that late bilinguals and monolinguals reliably showed LH language dominance, regardless of proficiency level, whereas early bilinguals reliably showed bilateral involvement for language. A subsequent meta-analysis by Hull and Vaid (2007) focused specifically on disentangling laterality differences among bilinguals. Consistent with findings for bilinguals in their earlier meta-analysis ( Hull and Vaid, 2006), the outcomes demonstrated reliably different patterns of language lateralization for bilinguals with early (bilaterally organized) versus late bilingualism onset (LH dominant). In addition, lateralization within each bilingual subgroup overlapped across first and second languages; that is, both the L1 and L2 of early bilinguals were organized bilaterally (consistent with the Anchoring and Signature hypotheses), whereas both languages were LH dominant in late bilinguals (consistent with the Anchoring hypothesis). Moreover, the 2007 meta-analysis showed increased LH dominance for non-proficient late bilinguals relative to proficient ones (consistent with the Anchoring Hypothesis).

470

Eswen Fava and Rachel Hull

Taken together, outcomes from the behavioral bilingual laterality literature clearly point to differences in brain organization for language associated with systematic differences in both the developmental stage during which the L2 is acquired and the amount of experience (i.e., proficiency) with the L2. However, while partial answers to questions about language representation in the bilingual brain have emerged from behavioral sources, such measures are by their nature limited in what they can reveal about neural pathways. Neuroimaging evidence. Several alternative (i.e., non-behavioral) techniques for inferring hemispheric organization for language have more recently emerged, namely electroencephalographic (EEG) recording, event related brain potentials (ERPs), positron emission tomography (PET), functional magnetic resonance imaging (fMRI), and nearinfrared spectroscopy (NIRS). These techniques typically utilize an entirely different form of participant responses relative to behavioral paradigms. Generally speaking, measures of the onset and duration of electrical activity in response to a language stimulus (EEG, ERP) have been used to identify the timing and character of neural activity for language processing, whereas differences in the amount of regional cerebral blood flow during a language task relative to some baseline, non-language task (fMRI, PET, NIRS) have been used to infer the location of neural structures and pathways that support language processing. The remainder of this chapter will focus primarily on neuroimaging evidence about language from NIRS and fMRI (but see Hull and Vaid, 2005, for a review of the relevant EEG/ERP literature). Numerous neuroimaging studies have lent support to the long-standing notion that language in monolinguals is LH dominant (Dehaene-Lambertz, 2000; Hickok and Poeppel, 2000; Holowka and Petitto, 2002; Knecht et al., 2001; Paus et al., 1996; Pujol, Deus, Losilla, and Capdevila, 1999; Scott and Wise, 2004). For instance, a specific role for phonological processing has been indicated in the left IFG (Devlin, Matthews, and Rushworth, 2003; Gabrieli, Poldrack, and Desmond, 1998; Klein, Milner, Zatorre, Meyer, and Evans, 1995), and left superior temporal gyrus (STG) (Hickok, 2001; Hughdal et al., 1999; Karbe, 1995; Parker et al., 2005). In addition, activity in the most posterior and medial part of the STG at the temporo-parietal junction has been associated with speech production rather than perception (Wise et al., 2001), whereas both overlapping and unique areas of neural activation have been identified in frontal and posterior areas during speech production and perception, (Hickok, Buchsbaum, Humphries, and Muftuler, 2003; Hickok and Poeppel, 2000; Parker et al., 2005). Although neuroimaging evidence with late bilinguals has typically demonstrated the traditional LH dominance for language seen in monolinguals (e.g., Klein et al., 1994; Vingerhoets et al., 2003; for reviews, see Indefrey, 2006 and Vaid and Hull, 2002), the outcomes with early bilinguals, in contrast, have been markedly less consistent with the notion of a unilateral LH dominance for language. For instance, one fMRI study of multilingual speakers with varying ages of non-native language acquisition and different levels of non-native language proficiency (Briellman et al., 2004) demonstrated that late multilingual speakers consistently showed LH dominance for processing in all languages, with increased LH activation for less proficient languages. However, the lone early multilingual in this study appeared to show overlapping, bilateral activation patterns for all languages, including those for which the early bilingual was not proficient. These outcomes are consistent with the Anchoring Hypothesis. Another fMRI study used a covert picture naming task to reveal that early bilinguals showed equal activation bilaterally in the left and right STG while switching between L1 and

Language Development through a Bilingual Lens

471

L2 (Hernandez et al., 2000). In addition, an fMRI study involving deaf and hearing signlanguage users showed that both hearing and deaf American Sign Language (ASL)-English early bilinguals demonstrated bilateral activation during sign language processing, whereas late ASL-English bilinguals and ASL or English monolinguals displayed primarily LH activation (Neville and Bavelier, 1998). Finally, a structural MRI study found increased gray matter density in bilinguals relative to monolinguals in the inferior parietal lobe near the temporo-parietal junction, and significantly more so for early than late bilinguals, suggesting that the onset age of bilingualism and/or the length of experience as a bilingual might alter the very structure of the brain (Mechelli et al., 2004). This outcome suggests that variations in hemispheric dominance for language might be explained in terms of different amounts of gray matter volume and density. Another possibility is that the mechanism of increased dominance or density might be length of exposure and the associated automatization of language processing (Lucas, McKhann, and Ojemann, 2004), although the intrinsic link between exposure and proficiency may be difficult to ultimately disentangle. Taken together, data from behavioral and neuroimaging studies provide converging support for the Anchoring Hypothesis and the Signature Hypothesis. However, innovative approaches are needed to produce new evidence to refine and connect these hypotheses.

Where Are We Moving? Like behavioral measures, neuroimaging measures have some notable limits on what they can reveal about language and the brain. PET studies involve a radioactive tracer that makes it impractical (even unethical) to test individuals on multiple tasks and/or trials, or even to test certain populations at all. fMRI is temporally sluggish, expensive, unavailable to most researchers, and limited in the paradigms it can test (e.g., overt speech production). That is, extreme sensitivity to motion artifacts often requires modifying tasks for fMRI (e.g., covertly “thinking of” words instead of physically responding), making it impossible to behaviorally validate participants’ responses and quite possibly not capturing the full range of activation associated with actual speech production. NIRS represents an alternative to fMRI in such cases, because it is more robust to motion artifacts and can therefore be used with the same kinds of tasks used in behavioral paradigms (e.g., overt speech production, handwriting, etc.). In addition, NIRS can be safely used with protected populations (e.g., healthy infants) previously restricted in traditional neuroimaging paradigms. However, NIRS is limited to tracking activation in the cortex and thus is not appropriate for measuring subcortical activity, for which fMRI remains the most informative paradigm at present. We suggest that the future of language research lies in the marriage of behavioral and neuroimaging techniques. In studies with adult bilinguals, this strategy of combining techniques has already begun to produce promising outcomes. One recent study used NIRS with an established behavioral measure (i.e., picture-naming) to demonstrate that cortical activity associated with overt speech production was localized to the left (but not right) STG in monolingual adults (Hull, Bortfeld, and Koons, 2009). A study with late bilinguals combined methodologies from a behavioral picture-naming task with a neuroimaging paradigm (language mapping using electronic stimulation of the cortex) and found that left temporal regions were shared for L1 and L2 processing (Lucas, McKhann and Ojemann, 2004). However, the authors further reported that L1 processing was also related to activation

472

Eswen Fava and Rachel Hull

in the frontal cortex, whereas L2 processing was related to additional sites in mid- to posterior temporal areas, thereby suggesting a functional separation of languages. Importantly, this study suggests that while a classic left lateralization is present in late bilinguals (consistent with the Anchoring Hypothesis), there may still be potential for the two languages to be separated within the LH. Other evidence for a functional LH separation for L1 and L2 in late bilinguals includes an innovative study by Marian, Spivey, and Hirsch (2003), who used both eye-tracking (a behavioral measure) and fMRI to test late Russian-English bilinguals on a picture-recognition task using both phonetic and visual within- and between-language distractors. The results pointed toward common activation across languages for lower levels of language processing (e.g. phonetic), whereas higher-level processing (e.g. lexical) was supported by distinct areas that appeared to be preferentially associated with one language or the other. In studies with infants, Dehaene-Lambertz and colleagues (Dehaene-Lambertz, Dehaene, and Hertz-Pannier, 2002; Dehaene-Lambertz et al., 2006) used fMRI to demonstrate left lateralized hemodynamic responses to speech in 2-4 month old infants. In a NIRS study with neonates from monolingual environments, newborns showed LH dominance for their native language relative to non-language stimuli (Peña et al., 2003). A separate NIRS study with older infants (aged 6-9 months) also demonstrated preferential activation in the left temporal region in response to native language stimuli by Bortfeld, et al., (2009). In contrast, neonates in an EEG study showed RH dominance for processing the native language, and the same infants showed bilateral activation during subsequent testing at 2.5 months (Radicevic, Vujovic, Jelicic, and Sovilj, 2008). Zaramella et al., (2001) also found bilateral activation when using NIRS to measure newborn infants processing of tonal frequency sweeps (although this stimulus may not capture exactly the same activity as true linguistic input). Thus, although there is growing evidence for a left lateralized preference for native language processing in infants, at least those from monolingual environments, the data are limited and there is no clear consensus on how or when hemispheric specializations may develop in either monolingual or bilingual babies. We suggest that this debate could be addressed using NIRS and other imaging and behavioral techniques concurrently, or at least on the same infants in temporal proximity as well as longitudinally. In addition, future studies combining NIRS and overt speech production paradigms in older children and adults could shed new light on the notion a frontal-posterior network of auditory-motor integration, which Hickok et al. (2003) have suggested guides speech development and supports speech production in monolinguals. It would be interesting and valuable to test the hypothesis using NIRS and behavioral measures of overt speech production, and to extend it to include bilingual participants.

Conclusion The language development literature is at a critical crossroads. Forward progress in understanding exactly how, when, and why the critical junctures emerge hinges on refining how researchers frame and test experimental questions. Ample evidence from behavioral and neuroimaging methods suggests the brain’s developmental sensitivity to the acquisition of multiple language systems, and therefore warrants this as a promising frontier to be explored. We suggest the most effective studies will combine existing techniques to further illuminate

Language Development through a Bilingual Lens

473

the ways in which early language experience influences brain structural and functional organization. Moreover, the emergence of the NIRS technique will allow us to design unprecedented studies that access previously underexplored populations (namely infants and children). Such studies are anticipated to provide an improved means to address the complexity and range of language organization in the intact, developing brain, and to facilitate necessary connections between existing knowledge and theoretical explanations across research domains.

References Abutalebi, J., Cappa, S. F., and Perani, D. (2001). The bilingual brain as revealed by functional neuroimaging. Bilingualism: Language and Cognition, 4, pp 179-190. Abutalebi, J., Miozzo, A., and Cappa, S. F. (2000). Do subcortical structures control language selection‚ in polyglots? Evidence from pathological language mixing. Neurocase, 6(1), 51 - 56. Albert, M., and Obler, L. K. (1978). The bilingual brain: Neuropsychological and neurolinguistic aspects of bilingualism. New York: Academic Press. Beeman, M., and Chiarello, C. (1998). Complementary left and right hemisphere language comprehension. Current Directions in Psychological Science, 7, 2-8. Bialystock, E., and Ryan, E. B. (1985). Toward a Definition of Metalinguistic Skill. MerrillPalmer Quarterly, 31(3), 229-251. Birdsong, D., and Molis, M. (2001). On the evidence for maturational constraints in secondlanguage acquisition. Journal of Memory and Language, 44, 235-249. Boatman, D. (2004). Cortical bases of speech perception: Evidence from functional lesion studies. . Cognition, 92, 47-65. Booth, J., MacWhinney B., Thulborn K.R., Sacco, K., Voyvodic, J.T., and HM, F. (2000). Developmental and lesion effects in brain activation during sentence comprehension and mental rotation. Developmental Neuropsychology, 18(2), 139-169. Borod, J. C., Koff, E., Lorch, M. P., and Nicholas, M. (1985). Channels of Emotional Expression in Patients With Unilateral Brain Damage. Arch Neurol, 42(4), 345-348. Bortfeld, H., Fava, E., and Boas, D. A. (2009). Identifying cortical lateralization of speech processing in infants using near-infrared spectroscopy. Developmental Neuropsychology, 34(1), 52-65. Briellmann, R. S., Saling, M. M., Connell, A. B., Waites, A. B., Abbott, D. F., and Jackson, G. D. (2004). A high-field functional MRI study of quadri-lingual subjects. Brain and Language, 89, 531-542. Broca, P. (1861). Remarques sur le siege de la faculte du langage articule; suivies d’une observation d’aphemie. Paper presented at the Bulletin de la Société Anatomique de Paris, Paris, France. Brown, T. T., Lugar, H. M., Coalson, R. S., Miezin, F. M., Petersen, S. E., and Schlaggar, B. L. (2005). Developmental Changes in Human Cerebral Functional Organization for Word Generation. Cereb. Cortex, 15(3), 275-290. Dehaene, S., Dupoux, E., Mehler, J., Cohen, L. G., Paulesu, E., and Perani, D. (1997). Anatomical variability in the cortical representation of first and second language. NeuroReport, 8, 3809-3815.

474

Eswen Fava and Rachel Hull

Dehaene-Lambertz, G. (2000). Cerebral Specialization for Speech and Non-Speech Stimuli in Infants. Journal of Cognitive Neuroscience, 12(3), 449-460. Dehaene-Lambertz, G., Dehaene, S., and Hertz-Pannier, L. (2002). Functional Neuroimaging of Speech Perception in Infants. Science, 298. Dehaene-Lambertz, G., Hertz-Pannier, L., Dubois, J., Meriaux, S., Roche, A., Sigman, M., et al. (2006). Functional organization of perisylvian activation during presentation of sentences in perverbal infants. Proceedings of the National Academy of Sciences, 103, 14240-14245. Devlin, J. T., Matthews, P. M., and Rushworth, M. F. S. (2003). Semantic Processing in the Left Inferior Prefrontal Cortex: A Combined Functional Magnetic Resonance Imaging and Transcranial Magnetic Stimulation Study Journal of Cognitive Neuroscience, 15(1), 71-84. DeVos, K., Wyllie, E., Geckler, C., Kotagal, P., and Comair, Y. (1995). Language dominance in patients with early childhood tumors near left hemisphere language areas. Neurology, 45(2), 349-356. Duncan, J., Moss, S., Bandy, D., Manwaring, K., Kaplan, A., and Reinian, E. (1997). Use of positron emission tomography for presurgical localization of eloquent brain areas in children with seizures. Pediatric Neurosurgery, 26, 144-156. Fabbro, F. (2001). The Bilingual Brain: Cerebral Representation of Language. Brain and Language, 79(2), 211-222. Gabrieli, J. D. E., Poldrack, R. A., and Desmond, J. E. (1998). The role of left prefrontal cortex in language and‚Äâmemory. Proceedings of the National Academy of Sciences of the United States of America, 95(3), 906-913. Galloway, L. (1980). Cerebral Organization in Bilingualism and Second Language. Paper presented at the Research in Second Language Acquisition: Selected Papers of the Los Angeles Second Language Acquisition Research Forum. Genesee, F., Hamers, J., Lambert, W. E., Mononen, L., Seitz, M., and Starck, R. (1978). Language processing strategies in bilinguals: A neuropsychological study. Brain and Language, 5, 1-12. Geschwind, N., and Galaburda, A. M. (1985). Cerebral lateralization: Biological mechanisms, associations, and pathology: III. A hypothesis and a program for research. Archives of Neurology, 42(7), 634-654. Goral, M., Levy, E., and Obler, L. (2002). Neurolinguistic aspects of bilingualism. International Journal of Bilingualism, 6(4), 411-440. Green, D. W. (1998). Mental control of the bilingual lexico-semantic system. Bilingualism: Language and Cognition, 1, 67-81. Hahne, A., and Friederici, A. (2001). Processing a second language: late learners' comprehension mechanisms as revealed by event-related brain potentials. Bilingualism: Language and Cognition, 4(2), 123-141. Hernandez, A. E., Martinez, A., and Kohnert, K. (2000). In Search of the Language Switch: An fMRI Study of Picture Naming in Spanish-English Bilinguals. Brain and Language, 73(3), 421-431. Hickok, G. (2001). Functional anatomy of speech perception and speech production: Psycholinguistic implications. Journal of Psycholinguistic Research, 30, 225-234.

Language Development through a Bilingual Lens

475

Hickok, G., Buchsbaum, B., Humphries, C., and Muftuler, T. (2003). Auditory-Motor Interaction Revealed by fMRI: Speech, Music, and Working Memory in Area Spt. Journal of Cognitive Neuroscience, 15(5), 673-682. Hickok, G., and Poeppel, D. (2000). Towards a functional neuroanatomy of speech perception. TRENDS in Cognitive Sciences, 4(4), 131-138. Holland, S. K., Plante, E., Weber Byars, A., Strawsburg, R. H., Schmithorst, V. J., and Ball, W. S. (2001). Normal fMRI Brain Activation Patterns in Children Performing a Verb Generation Task. NeuroImage, 14(4), 837-843. Holowka, S., and Petitto, L.-A. (2002). Left Hemisphere Cerebral Specialization for Babies While Babbling. Science, 297(5586), 1515. Hughdal, K., Bronnick, K., Kyllingsbrk, S., Law, I., Gade, A., and Paulson, O. B. (1999). Brain activation during dichotic presentations of consonant-vowel and musical instrument stimuli: a 15O-PET studyfn2. Neuropsychologia, 37(4), 431-440. Hull, R., Bortfeld, H., and Koons, S. (2009). Near-infrared spectroscopy and cortical responses to speech production. The Open Neuroimaging Journal, 3, 26–30. Hull, R., and Vaid, J. (2005). Clearing the cobwebs from the study of the bilingual brain: Toward converging evidence laterality and ERP. In J. Kroll and A. M. B. deGroot (Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 480-496). London: Oxford University Press. Hull, R., and Vaid, J. (2006). Laterality and language experience. Laterality: Asymmetries of Body, Brain and Cognition, 11(5), 436 - 464. Hull, R., and Vaid, J. (2007). Bilingual language lateralization: A meta-analytic tale of two hemispheres. Neuropsychologia, 45, 1987-2008. Indefrey, P. (2006). A Meta-analysis of Hemodynamic Studies on First and Second Language Processing: Which Suggested Differences Can We Trust and What Do They Mean? In M. Gullberg and P. Indefrey (Eds.), The cognitive neuroscience of second language acquisition. (pp. 279-304). Malden, MA: Blackwell. Isaacs, E. B., Christie, D., Vargha-Khadem, F., and Mishkin., M. (1996). Effects of hemispheric side of injury, age at injury, and presence of seizur disorder on functional ear and hand asymmetries in hemiplegic children. Neuropsychologia, 34, 127-137. Johnson, J. S., and Newport, E. L. (1989). Critical Period Effects in Second Language Learning: The Inluence of Maturational State on the Acquisition of English as a Second Language. Cognitive Psychology, 21, 60-99. Karanth, P., and Rangamani, G. N. (1988). Crossed aphasia in multilinguals. Brain and Language, 34(1), 169-180. Karbe, H. (1995). Long-term prognosis of poststroke aphasia studied with positron emission tomography. . Archives of Neurology, 52(2), 186-190. Klein, D., Milner, B., Zatorre, R., Meyer, E., and Evans, A. C. (1995). The neural substrates underlying word generation: a abilingual functional-imaging sutdy. PNAS, 92(7), 28992903. Knecht, S., Dräger, B., Flöel, A., Lohmann, H., Breitenstein, C., Deppe, M., et al. (2001). Behavioural relevance of atypical language lateralization in healthy subjects. Brain:, 124(8), 1657-1665. Kovelman, I., Baker, S. A., and Petitto, L.-A. (2008). Bilingual and Monolingual Brains Compared: A Functional Magnetic Resonance Imaging Investigation of Syntactic

476

Eswen Fava and Rachel Hull

Processing and a Possible and#x201C;Neural Signatureand#x201D; of Bilingualism. Journal of Cognitive Neuroscience, 20(1), 153-169. Liégeois, F., Connelly, A., Cross, J. H., Boyd, S. G., Gadian, D. G., Vargha-Khadem, F., et al. (2004). Language reorganization in children with early- onset lesions of the left hemisphere: an fMRI study Brain, 127, 1229-1236. Long, M. (1990). Maturational constraints on language development. Studies in Second Language Acquisition, 12, 251-285. Lucas, T. H., McKhann, G. M., and Ojemann, G. A. (2004). Functional sep- aration of languages in the bilingual brain: A comparison of electrical stimulation language mapping in 25 bilingual patients and 117 monolin- gual control patients. Journal of Neurosurgery, 101, 449–445. Marian, V., Spivey, M., and Hirsch, J. (2003). Shared and separate systems in bilingual language processing: Converging evidence from eyetracking and brain imaging. Brain and Language, 86(1), 70-82. Martin, R. C. (2000). Contributions from the neuropsychology of language and memory to the development of cognitive theory. Journal of Memory and Language, 43, 149-156. Mechelli, A., Crinion, J. T., Noppeney, U., O'Doherty, J., Ashburner, J., Frackowiak, R. S., et al. (2004). Neurolinguistics: Structural plasticity in the bilingual brain. Nature, 431(7010), 757-757. Muller, R.-A., Behen, M., Rothermel, R., Muzik O, Chakraborty, P., and Chugani, H. (1999). Brain organization for language in children, adolescents, and adults with left hemisphere lesion: a PET study. Progressive Neuropsychopharmacology and Biological Psychiatry, 23, 657-668. Muller, R.-A., Rothermel, R., Behen, M., Muzik, O., Mangner, T., and Chakraborty, P. (1998). Brain organization of language after early unilateral lesion: a PET study. Brain and Language, 62, 422-451. Muller, R.-A., Rothermel, R., Muzik, O., Becker, C., Fuerst, D., and Behen, M. (1998). Determination of language dominance by [15O]-water PET in children and adolescents: a comparison with the Wada Test. Journal of Epilepsy, 11, 152-161. Naeser, M. A., Martina, P. I., Bakera, E. H., Hodgea, S. M., Sczerzeniea, S. E., Nicholasa, M., et al. (2004). Overt propositional speech in chronic nonfluent aphasia studied with the dynamic susceptibility contrast fMRI method NeuroImage, 22(1), 29-41. Neville, H. J., and Bavelier, D. (1998). Neural organization and plasticity of language. Current Opinion in Neurobiology, 8(2), 254-258. Newport, E. L., Bavelier, D., and Neville, H. J. (2001). Critical thinking about critical periods: Perspectives on a critical period for language acquisition. In E. Dupoux (Ed.), Language, brain, and cognitive development: Essays in honor of Jacques Mehler (pp. 481-502). Cambridge, MA: MIT Press. Obler, L. K. (1981). Right Hemisphere Participation in Second Language Acquisition. In K. Diller (Ed.), Individual Differences and Universals in Language Learning Aptitude (pp. 53-64). Rowley, MA Newbury Press. Paradis, M. (1998). The other side of language: Pragmatic competence Journal of Linguistics, 11(1-2), 1-10. Paradis, M. (2000). Language lateralization in bilinguals: enough already! In L. Wei (Ed.), The Bilingualism Reader (pp. 365-372): Routledge.

Language Development through a Bilingual Lens

477

Parker, G. J. M., Luzzi, S., Alexander, D. C., Wheeler-Kingshott, C. A. M., Ciccerelli, O., and Lambon Ralph, M. A. (2005). Lateralization of ventral and dorsal auditory-language pathways in the human brain NeuroImage, 24(3), 656-666. Paus, T., Tomaiuolo, F., Otaky, N., MacDonald, D., Petrides, M., Atlas, J., et al. (1996). Human Cingulate and Paracingulate Sulci: Pattern, Variability, Asymmetry, and Probabilistic Map Cerebral Cortex, 6, 207-214. Peña, M., Maki, A., Kovacić, D., Dehaene-Lambertz, G., Koizumi, H., Bouquet, F., et al. (2003). Sounds and silence: an optical topography study of language recognition at birth. Proceedings of the National Academy of Sciences of the United States of America., 100(20), 11702-11705. Pujol, J., Deus, J., Losilla, J. L., and Capdevila, A. (1999). Cerebral lateralization of language in normal left-handed people studied by functional MRI Neurology, 52, 1038. Radicevic, Z., Vujovic, M., Jelicic, L., and Sovilj, M. (2008). Comparative findings of voice and speech: language processing at an early ontogenetic age in quantitative EEG mapping. Experimental Brain Research, 184(4), 529-532. Rasmussen, T., and Milner, B. (1977). The role of early left-brain injury in determining lateralization of cerebral speech functions. Annals of the New York Academy of Science, 299(September), 355-369. Schlaggar, B. L., Brown, T. T., Lugar, H. M., Visscher, K. M., Miezin, F. M., and Petersen, S. E. (2002). Functional Neuroanatomical Differences Between Adults and School-Age Children in the Processing of Single Words. Science, 296(5572), 1476-1479. Schneiderman, E. I. (1986). Right Hemisphere Processing and Foreign Language Aptitude. ITL: Review of Applied Linguistics, 71, 43-64. Scott, S., and Wise, R. (2004). The functional neuroanatomy of prelexical processing in speech perception. Cognition, 92(1-2 ), 13-45. Springer, J. A., Binder, J. R., Hammeke, T. A., Swanson, S. J., Frost, J. A., Bellgowan, P. S. F., et al. (1999). Language dominance in neurologically normal and epilepsy subjects: A functional MRI study. Brain, 122(11), 2033-2046. Staudt, M., Lidzba, K., Grodd, W., Wildgruber, D., Erb, M., and Kr‰geloh-Mann, I. (2002). Right-Hemispheric Organization of Language Following Early Left-Sided Brain Lesions: Functional MRI Topography. NeuroImage, 16(4), 954-967. Szaflarski, J. P., Binder, J. R., Possing, E. T., McKiernan, K. A., Ward, B. D., and Hammeke, T. A. (2002). Language lateralization in left-handed and ambidextrous people: fMRI data. 59(2), pp 238-244. Szaflarski, J. P., Holland, S. K., Schmithorst, V. J., and Byars, A. W. (2006). An fMRI study of language lateralization in children and adults. Human Brain Mapping, 27(3), 202-212. Thal, D. J. (1997). Origins of language disorders. Developmental Neuropsychology, 13(Special Issue: Origins of language disorders), 233-237. Ullman, M. T. (2004). Contributions of memory circuits to language: the declarative/procedural model. Cognition, 92(1-2), 231-270. Vaid, J. (1984). On Readings on Aphasia in Bilinguals and Polyglots. Journal of the History of the Behavioral Sciences, 20, 372-374. Vaid, J. (1987). Visual field asymmetries for rhyme and syntactic category judgements in monolinguals and fluent early and late bilinguals. Brain and Language, 30, 263-277. Vaid, J. (2002). Bilingualism. In V. S. Ramachandran (Ed.), Encyclopedia of the human brain (Vol. 1, pp. 417-434). San Diego: Elsevier.

478

Eswen Fava and Rachel Hull

Vaid, J., and Frenck-Mestre, C. A. (1991). Incidental memory for format of presentation of number stimuli: Evidence from monolinguals and bilinguals. Brain and Cognition, 17(2), 272-284. Vaid, J., and Genesee, F. (1980). Neuropsychological approaches to bilingualism: A critical review. . Canadian Journal of Psychology(34), 417-445. Vaid, J., and Hull, R. (2002). Re-envisioning the bilingual brain using functional neuroimaging: Methodological and interpretive issues. In F. Fabbro (Ed.), In the Neurolinguistics of Bilingualism: Essays in Honor of Michel Paradis. (pp. 315-355). Udine Forum: Udine University Press. Vargha-Khadem, F., O'Gorman, A. M., and Watters, G. V. (1985). Aphasia and Handedness in Relation to Hemisphere Side, Age at Injury and Severity of Cerebral Lesion during Childhood. Brain, 108(3), 677-696. Vingerhoets, G., Van Borsel, J., Tesink, C., van den Noort, M., Deblaere, K., Seurinck, R., et al. (2003). Multilingualism: an fMRI study. NeuroImage, 20, 2181-2196. Weber-Fox, C., and Neville, H. J. (2001). Sensitive Periods Differentiate Processing of Openand Closed-Class Words: An ERP Study of Bilinguals. J Speech Lang Hear Res, 44(6), 1338-1353. Weber-Fox, C.-M., and Neville, H.-J. (1999). Functional neural subsystems are differentially affected by delays in second language immersion: ERP and behavioral evidence in bilinguals. In D. E. Birdsong (Ed.), Second language acquisition and the Critical Period Hypothesis. Second language acquisition research. (pp. 23-38). (pp. 191 pp). Mahwah, NJ, US: Lawrence Erlbaum Associates, Inc. Wernicke, C. (1911). The symptom of complex aphasia. In C. AE (Ed.), Diseases of the nervous system (pp. 265–324). New York: Appleton. Wise, R. K., Scott, S., Blank, C. S., Mummery, C. J., Murphy, K., and Warburton, E. A. (2001). Separate neural subsystems within `Wernicke's area' Brain, 124(1), 83-95. Woods, B. T., and Carey, S. (1979). Language deficits after apparent clinical recovery from childhood aphasia. Annals of Neurology, 6(5), 405-409. Woods, B. T., and Teuber, H.-L. (1978). Changing patterns of childhood aphasia. Annals of Neurology, 3(3), 273-280. Zaramella, P., Freato, F., Amigoni, A., Salvadori, S., Marangoni, P., Suppjei, A., et al. (2001). Brain auditory activation measured by near-infrared spectroscopy (NIRS) in neonates. Pediatric Research, 49, 213-219.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 21

IMPROVING READING SKILLS FOR ESL LEARNERS USING SOUNDSPEL Michael D. Young, Michelle L. Wilson and Alice F. Healy University of Colorado, USA

Abstract This study examined the effects of using a revised, transparent spelling system SoundSpel, a phonetic reading tool, with learners of English as a Second Language. During 6 training sessions, 12 participants used unaltered material and 12 used SoundSpel texts, in parallel with standard English, when reading American elementary school material. They then answered multiple-choice comprehension questions. Both groups were pre-tested and post-tested on comprehension tests of similar elementary school material without SoundSpel. No group differences were found across tests or training (in quiz performance or reading time), suggesting no beneficial or harmful effects from using SoundSpel. A post hoc analysis suggested that SoundSpel would be most beneficial for students who learn to speak English before they learn to read it.

Introduction Learning to read is an important goal of elementary education. Specifically students need to move from sounding out words to sight reading, except, perhaps, for low-frequency words. Children typically have a larger speaking vocabulary than reading vocabulary, so a challenge for them is to match their speaking vocabulary to printed words. Towards this end, young readers often make use of dictionaries with phonetic notation. Mastering the spelling system, which includes 26 letters and approximately 42 sounds, can be a difficult task, as evident in the following example: “Though the rough cough and the hiccough plough me through, I ought to cross the lough” (Mole, 2003, p. 4). To read this sentence successfully, the pronunciation of each of these instances of the letter string -ough would need to be memorized. In this way, English differs from other, more transparent languages that have better letter-to-sound correspondence, and hence would require less memorization.

480

Michael D. Young, Michelle L. Wilson and Alice F. Healy

Illustrating this fact, Patel, Snowing, and de Jong (2004) showed that Dutch children (ages 611) reading Dutch, which is a phonetically transparent language, were more accurate and faster at pronouncing visually presented words and nonwords than were comparable American children reading English, especially at younger ages. Reading skill development has been shown to be highly associated with phonological processing skills. Evidence that phonological awareness is related to reading comes from longitudinal studies that have found that differing degrees of phonological awareness among children (before they even learn to read) can predict reading skill through at least age 6 (e.g., Bryant, MacLean, Bradley, and Crossland, 1990). Some phonological awareness develops early, although awareness of phonemes as units develops later than an awareness of syllables, onsets, and rimes. Phonological awareness can develop without reading instruction, but in the absence of reading an alphabetic system, this capacity seems to be retarded (Mann, 1986). However, it has been shown that specific training in phonological awareness can help with remediation of reading problems, particularly for poor adult readers learning an alphabetic system (Rayner, Foorman, Perfetti, Pesetsky, and Seidenberg, 2001). For example, training with “talking computers” that pronounce words in text has been effective at improving word recognition, phonological decoding, and phonological awareness in children with reading difficulties (Olson, Wise, Ring, and Johnson, 1997). In the present study, phonological processing is investigated by using a program called SoundSpel, developed by the American Literacy Council (Mole, 2003), which converts English text into a more predictable spelling system. Specifically, SoundSpel is a phonetic spelling system with unambiguous, easy to learn, letter-to-sound correspondences, including improved vowel representations and—unlike standard dictionaries—no additional characters or diacritical marks. English text can be written using a format called DoubleLine, in which every line of text appears in parallel with the same line converted directly below into SoundSpel, with the SoundSpel version of a given standard English word placed immediately under it. If a reader encounters an unfamiliar word in English, using DoubleLine, he or she can refer to the line below, which in effect contains a transcription (i.e., a respelling) of the word according to its pronunciation. For instance, the word “glad” is pronounced as it is spelled, relying on no special rule or exception for proper pronunciation. In SoundSpel, it would be written “glad” as in English. Yet the word “give” is an instance when the phonological representation of a word lacks congruency with its graphemic representation because of the terminal silent letter “e”. That word in SoundSpel would be written “giv” (without the superfluous final letter “e”). Thus, DoubleLine would be useful because it could function as a point of reference, representing words according to their phonological features, thereby providing a systematic link between written and spoken language. In the present study, we explore the possibility of using DoubleLine to improve reading comprehension with adults learning English as a second language (ESL). Two groups of nonnative English speakers were compared: The SoundSpel group was instructed in SoundSpel and read English text written in DoubleLine during training, and the control group was not exposed to SoundSpel or DoubleLine but read the same English text written so that every line of text appeared in parallel with the same line directly below it (i.e., duplicated). Before and after training, both groups of participants were given tests. The pretest was equivalent to the posttest, and neither test included DoubleLine or duplicated lines; instead the text was printed in ordinary English. During training and testing, each English passage that was read was followed by multiple-choice comprehension questions. Two measures of performance were

Improving Reading Skills for ESL Learners Using SoundSpel

481

collected: number correct on the reading comprehension test of a given passage and total time to read the passage and answer the comprehension questions. The two passages used in testing were both at the fourth grade level, whereas the six passages used in training included three at the third grade level followed by three at the fourth grade level. These reading levels were selected to match the current reading levels of the ESL participants as closely as possible. If DoubleLine effectively promotes phonological awareness in adults, then we should find that performance on both measures is better during training for the SoundSpel group than for the control group. Most importantly, if the phonological training derived from SoundSpel transfers to reading normal English, then any improvement in performance from the pretest to the posttest should be larger for the SoundSpel group than for the control group. Thus, a comparison of SoundSpel and control groups during training would reveal any immediate benefits of DoubleLine, whereas a comparison of the two groups on the posttest would reveal any long-term benefits of prior exposure to DoubleLine.

Method Participants Participants for this study were recruited from the International English Center ESL program at the University of Colorado and the InterCambio de Comunidades ESL program in the Boulder community. This population was limited to non-native English speaking adults with a reading level ranging from approximately third to fourth grade. A total of 24 participants were included in the analyses. An additional participant was tested, but the data from that participant were excluded because that participant did not follow directions in one of the training sessions. The 24 participants could be divided by native language background into the following groups: Arabic (2), French (2), Korean (6), Japanese (4), Nepali (1), Spanish (8), and Ukrainian (1). Recruitment methods included informational fliers and wordof-mouth. Upon completion of the entire eight-part study, participants were compensated $100 for their time.

Design For the tests, one between-subjects independent variable was the experimental condition (control, SoundSpel). An additional independent between-subjects variable was the order of the pretest and posttest (A then B, B then A). Test type (pretest, posttest) was a withinsubjects independent variable. Other within-subjects variables that pertained to training but not the tests were grade level of material (third or fourth) and story number (first, second, or third) within a given grade level. The primary dependent variable of this study was reading comprehension, measured by the number of correct responses on the quizzes. Time taken to read each text and respond to the multiple-choice comprehension questions constituted a second dependent variable.

482

Michael D. Young, Michelle L. Wilson and Alice F. Healy

Materials and Apparatus Materials used for this study were from American third and fourth grade reading passages taken from Collections: In Good Company and Collections: Sometimes I Wonder (Allington, Cortez, Cunningham, Sebesta, and Tierney, 1989b, 1989c). In the test portion of the study, occurring at the beginning and end of the experiment, participants read the unmodified fourth grade Texts A and B in a counterbalanced order. Text A was 12 pages long (1838 words) with a picture on each page. Text B was 20 pages long (5766 words) with 9 small illustrations. The texts used in the training portion of this study varied according to condition. The control condition training texts were double-spaced paragraphs with an identical line written below each line of the paragraph. Texts for the SoundSpel group were also in the format of doublespaced paragraphs; however, below each line that appeared in standard English spelling was that same line converted to SoundSpel DoubleLine (see Appendix A). All documents for the SoundSpel group had been converted using a Microsoft Word macro, provided by the American Literacy Council, and printed in Courier font. The first three texts were at the third grade level, and the last three texts were at the fourth grade level. Training Text 3-1 was 42 pages long (4217 words) with 21 illustrations. Text 3-2 was 22 pages (2192 words) with 14 pictures, and Text 3-3 was 28 pages (3154 words) with 9 pictures. Training Text 4-1 was 58 pages (6965 words) and had 9 illustrations, Text 4-2 was 54 pages (6631 words) with 8 pictures, and Text 4-3 was 52 pages (6714 words) and had 14 illustrations. Multiple-choice questions (four questions per text) were used to quiz reading comprehension. These multiple choice questions were developed by the experimenters and tested for clarity using native speaking undergraduates in a preliminary experiment (see Appendix B). Each multiple-choice question had three different answers to choose from. Most of the participants in the SoundSpel condition (except the first four tested) used a standard ruler to minimize any distraction of the double lines in reading these passages. (They were told to hide the SoundSpel line when reading the standard English text unless it was needed to decipher the standard English text.) Participants in the control condition were not given the ruler. Time to complete reading each passage and answering the comprehension questions was recorded in minutes by the experimenter using a watch or wall clock.

Procedure This study consisted of eight sessions: pretest, six training sessions, and posttest. With the exception of 4 participants (whose sessions were all on 8 separate days), sessions were paired over the course of 4 days—both were administered on a given day with a short break in between sessions (e.g., 5 min). Participants were assigned to one of two conditions based on a fixed rotation depending on their time of arrival for the first session: the control condition or the SoundSpel condition. Because there was no systematic relation between time of arrival and language background, assignment of participants to conditions was essentially random. All participants were then handed an unmodified pretest text (A or B), followed by a comprehension quiz including four multiple-choice questions. Participants in the experimental condition then received approximately 5 min of training on the SoundSpel spelling system from the experimenter.

Improving Reading Skills for ESL Learners Using SoundSpel

483

Instructing participants in SoundSpel involved introducing every sound-to-letter(s) representation for the consonants (which remained relatively unchanged). Instruction proceeded by then introducing the less familiar, but systematic, SoundSpel representation of short and long vowels. Participants in the SoundSpel group saw all the consonants and then all the vowels on PowerPoint slides or printouts of the slides (two for the consonants followed by two for the vowels), with a sample word or words containing a given letter to illustrate that letter’s pronunciation. The experimenter pronounced aloud for the participant each letter and many of the sample words (all of those for the vowels). After participants in the SoundSpel condition were tutored on SoundSpel, they were instructed to read aloud a short passage transcribed using SoundSpel (see Appendix C). The task of reading aloud the transcribed passage equipped the participant with (a) some familiarity in using the SoundSpel system, (b) an opportunity to receive feedback from the experimenter, and (c) a chance to demonstrate that he or she was, in fact, able to comprehend and use this new phonological tool. Training sessions followed the initial SoundSpel instruction and testing session. There were six training sessions, which were broken up by grade level. The first three training session texts involved third grade stories. The subsequent three training session texts involved fourth grade stories. After reading each text, the experimenter administered a multiple-choice quiz consisting of four multiple-choice questions regarding its corresponding story. Participants completed the quiz without the aid of the text. Generally, the experimenter started timing upon distribution of the passage and ended upon completion of the passage’s corresponding comprehension questions, although for some or all sessions of 4 of the participants (2 in each condition) only the reading of the passage was timed. Following these six training sessions was the final test session, in which the posttest was given. If participants were initially administered Text A, they received Text B following the six training sessions (or vice-versa). Time needed to complete this experiment was estimated to be a total of four 2-hour sessions (or eight 1-hour sessions for 4 participants), although a given session often took somewhat less time than estimated.

Results Tests A mixed factorial analysis of variance was conducted on the number of correct responses, out of four total questions per test, for comprehension questions on tests before and after training, both of which were based on fourth grade materials (see Figure 1). The analysis included the between-subjects factor of condition (control, SoundSpel) and the withinsubjects factor of test type (pretest, posttest). There were no significant main effects or interaction, F(1, 22) < 1 in each case. Specifically, there was no significant improvement from the pretest to the posttest for either the control condition (pretest = 3.000, posttest = 3.000) or the SoundSpel condition (pretest = 3.167, posttest = 3.000). A second mixed factorial analysis of variance, including the same factors, was conducted on the response time (in minutes) to complete reading each passage and responding to the four multiple-choice questions. Again, there were no significant main effects or interaction; F(1, 22) < 1 both for the main effect of condition and for the interaction of condition and test type; F(1,22) = 2.331, MSE = 151.042, p = .1411 for the main effect of test type. The means,

484

Michael D. Young, Michelle L. Wilson and Alice F. Healy

which are provided in Figure 2, show a small decrease in response time across tests of comparable magnitude for the control condition (pretest = 33.750 min, posttest = 28.333 min) and the SoundSpel condition (pretest = 33.750 min, posttest = 28.333 min).

Mean Number Correct (out of 4)

4

3

Pretest Posttest

2

1

0

Control

SoundSpel

Subject Group

Figure 1. Mean number correct (out of 4 questions total per test) on pretest and posttest for two conditions. Error bars show standard errors of the mean.

Mean Reading Time (in minutes)

40

30

Pretest Posttest

20

10

0

Control

SoundSpel

Subject Group

Figure 2. Mean reading time (in minutes) on pretest and posttest for two conditions. Error bars show standard errors of the mean.

Improving Reading Skills for ESL Learners Using SoundSpel

485

Thus, participants in the SoundSpel condition did not differ significantly from those in the control condition in terms of their performance on the tests before and after the training phase with respect to both accuracy and speed.

Training With respect to training, a mixed factorial analysis of variance was conducted on the number of correct responses (out of four total questions per story) on the comprehension questions given after each passage (see Figure 3). The analysis included the between-subjects factor of condition (control, SoundSpel) and the within-subjects factors of passage grade level (third, fourth) and story number within grade level (1, 2, 3). The analysis revealed a significant main effect of story number, F(2, 44) = 11.932, MSE = 0.514, p < .0001, reflecting a decline in the number of correct responses from Story 1 (3.667) to Story 2 (2.958), with Story 3 (3.229) in between. Also, there was a significant main effect of grade level, F(1, 22) = 6.459, MSE = 0.474, p = .0186, reflecting a lower mean number of correct responses for the fourth grade stories (3.139) than for the third grade stories (3.431). In addition, there was a significant interaction of story number and grade level, F(2, 44) = 4.421, MSE = 0.457, p = .0178. For the third grade stories, the mean number of correct responses was highest for the first story (3.875), lowest for the second story (2.875), and intermediate for the third story (3.542). For the fourth grade stories, the mean number of correct responses decreased across stories (first = 3.458; second = 3.042; third = 2.917). Control SoundSpel

Mean Number Correct (out of 4)

4

3

2

1

0

First

Second

Third Grade

Third

First

Second

Third

Fourth Grade

Figure 3. Mean number correct (out of 4 questions total per story) during training session for two conditions as a function of grade level and story number. Error bars show standard errors of the mean.

486

Michael D. Young, Michelle L. Wilson and Alice F. Healy 60

Mean Reading Time (in minutes)

Control SoundSpel 50

40

30

20

10

0

First

Second

Third Grade

Third

First

Second

Third

Fourth Grade

Figure 4. Mean reading time (in minutes) during training session for two conditions as a function of grade level and story number. Error bars show standard errors of the mean.

These differences across stories may be due to the content, length, and nature of the stories themselves, which were not counterbalanced across participants, because the stories were distributed in a fixed order within each grade level. There were no other significant effects; in particular, there was not a main effect of condition, F(1, 22) < 1, an interaction of condition and grade level, F(1, 22) < 1, an interaction of condition and story number, F(2, 44) = 1.095, MSE = 0.514, p = .3436, or a three-way interaction of condition, grade level, and story number, F(2, 44) < 1. Thus, there were essentially no differences between the control and SoundSpel conditions in accuracy during training. A mixed factorial analysis of variance was also conducted on response time (in minutes) needed to complete reading a given passage and responding to the four multiple choice questions during the training phase. As seen in Figure 4, there were three significant effects in this analysis: the main effect of grade level, F(1, 22) = 64.019, MSE = 125.124, p < .0001, the main effect of story number, F(2, 44) = 3.537, MSE = 50.484, p = .0376, and the interaction of grade level and story number, F(2, 44) = 6.419, MSE = 70.919, p = .0036. Third grade stories took less time to complete (34.250 min) than fourth grade stories (49.167 min). This result can be directly attributed to the overall longer length of the fourth grade passages relative to the third grade passages. Also, the first passage of the third grade stories had a longer response time (39.167 min) than the subsequent third grade stories (second = 29.208 min, third = 34.375 min). However, there was less change across the three fourth grade stories (first = 48.333 min, second = 50.625 min, third = 48.542 min). This pattern might reflect the fact that participants were particularly slow at reading the very first story they encountered in training, which was always the initial third grade story. There were no other significant effects; in particular, there was not a main effect of condition, F(1, 22) < 1, an interaction of condition and grade level, F(1, 22) < 1, an interaction of condition and story number, F(2, 44)

Improving Reading Skills for ESL Learners Using SoundSpel

487

< 1, or a three-way interaction of condition, grade level, and story number, F(2, 44) < 1. Thus, there were no differences at all between the control and SoundSpel conditions in reading time during training, despite the differences in the material read.

Discussion With respect to both accuracy and speed, participants in the SoundSpel condition did not differ significantly in their performance from those in the control condition on the tests before and after training. These results do not support the hypothesis that SoundSpel participants would demonstrate greater improvement from pretest to posttest in their reading comprehension. Although SoundSpel did not enhance performance at the posttest, it also did not depress performance, even though the testing conditions did not match the training conditions. This observation is noteworthy because of former studies showing that test performance is depressed whenever testing and training conditions do not correspond, in accordance with both the principle of transfer appropriate processing (e.g., Morris, Bransford, and Franks, 1977; Roediger, Weldon, and Challis, 1989) and the principle of procedural reinstatement (e.g., Healy, Wohldmann, and Bourne, 2005). Hence, there were no facilitative or detrimental effects of training with SoundSpel. One reason why we might not have found a significant advantage of training with SoundSpel concerns the language background of the participants. It seems most likely that participants would benefit from SoundSpel to the extent that they can understand spoken English better than they can understand written English. For participants with a Far East Asian language background (viz., Korean and Japanese), reading English typically precedes listening to English, so it seems unlikely for them that comprehension of spoken English would be superior to that of written English. In contrast, for participants with other language backgrounds, it is more likely that they encountered spoken English before written English. Thus, to determine whether SoundSpel would have a training advantage for individuals who speak English better than they read it, we conducted a post hoc analysis on the training data in which we divided the participants into two groups: Far East Asian and other. This analysis yielded the expected pattern of results; the interaction between native language category and condition was significant by a one-tailed test, F(1, 20) = 3.297, MSE = 0.938, p = .0422. As shown in Figure 5, the participants with a Far East Asian native language showed better comprehension accuracy during training in the control condition than in the SoundSpel condition. In contrast, the other participants showed better comprehension accuracy during training using SoundSpel than without using SoundSpel. This result suggests that future studies investigating SoundSpel be restricted to individuals who learn to speak English before they learn to read it, such as children and ESL students with a language background other than Far East Asian. Past studies have demonstrated that the phonology of the first language exerts a very strong influence on the acquisition of second language vocabulary (Feldman and Healy, 1998). Problems then occur for ESL students whose native language has different phonological rules (i.e., different sound-to-spelling rules) than English. SoundSpel can help such ESL students better understand written English by simplifying the sound-to-spelling mapping rules.

488

Michael D. Young, Michelle L. Wilson and Alice F. Healy

Mean Number Correct (out of 4)

4

Control SoundSpel

3

2

1

0

Far East

Other

Subject Group Figure 5. Mean number correct (out of 4 questions total per story) during training session for two conditions as a function of native language background. Error bars show standard errors of the mean.

Thus, the ESL students can use SoundSpel to identify those words that are in their spoken vocabulary but have an orthography with less familiar sound-to-spelling correspondences. However, if the particular words or syllables encountered during training with SoundSpel do not include those that appear on the test, the ESL students will not show any benefit from SoundSpel training. A subsequent study should, therefore, ensure that the words encountered in the tests are also included among the training passages. We found that story grade level had a significant impact on comprehension accuracy during training, with third grade stories more accurate overall than fourth grade stories. This finding, although not unexpected, is important for two reasons. First, it provides a manipulation verification because the third grade materials were assumed to be easier to read than the fourth grade materials. Second, it verifies that the material selected was appropriate for the reading abilities of the participants (i.e., the material was not so simple that comprehension was at the ceiling, as it was for the college undergraduates in the preliminary experiment; see Appendix B). There was also a significant impact of story grade level on response time. However, this finding is difficult to interpret because the fourth grade stories were longer (i.e., included more pages) than the third grade stories, and the fourth grade stories had a smaller number of pictures per page than the third grade stories. Also, story grade level was confounded with position in the sequence, with all third grade stories appearing before fourth grade stories. However, this confounding only works against the finding that performance was worse on fourth grade stories than on third grade stories because the additional practice should probably improve performance. We also found significant effects of story number for both comprehension and reading time during training. The effect of story number is difficult to interpret because the stories were not counterbalanced across sessions. Also, the effect of story number on comprehension

Improving Reading Skills for ESL Learners Using SoundSpel

489

is difficult to understand because there was neither a consistent increase nor a consistent decrease in performance across stories; of the three stories performance was best overall on Story 1 and worst on Story 2. For reading time there was little difference for the fourth grade stories, but for the third grade stories participants were disproportionately slow in reading the first story. That finding could be due either to specific aspects of the story itself or to the fact that participants need some minimal practice with the unusual format in which lines are repeated before they can read the stories efficiently. A suggestion for improving performance on SoundSpel would be to make the SoundSpel instruction more congruent with its actual use. In the present experiment, SoundSpel instruction did not include DoubleLine although DoubleLine was used during the training session. Practice with DoubleLine during SoundSpel instruction may lead to better performance during training for the SoundSpel group. Although we used SoundSpel in this experiment to promote the skill of silent reading, it may prove to be a more effective tool for promoting the skill of reading aloud. The advantage of SoundSpel is that by simplifying the spelling-to-sound mapping it allows readers to go from the orthographic representation of a word to its phonological (i.e., spoken) representation. Thus, reading aloud, which requires the phonological representation, should be facilitated by the use of SoundSpel, whereas reading silently might not require access to the phonological representation, so should not show as large a benefit from SoundSpel. Therefore a future study might compare the SoundSpel and control conditions in a situation where both training and testing require the participants to read passages aloud, followed by multiple-choice comprehension quizzes, as in the present experiment. In such a study, any advantages of SoundSpel should be magnified.

Author Notes This research was supported in part by a grant from the American Literacy Council, contract DASW01-03-K-0002 from the Army Research Institute, and grant W911NF-05-10153 from the Army Research Office, all to the University of Colorado (Alice Healy, Principal Investigator). Michelle Wilson’s work on this project was also supported in part by a summer undergraduate research fellowship (SURF) award from the University of Colorado. We are indebted to Kathleen Shea for her help in constructing the experimental materials and in developing the multiple-choice comprehension questions. Correspondence concerning this report should be sent to Alice F. Healy, Department of Psychology and Neuroscience, 345 UCB, University of Colorado, Boulder, CO 80309-0345.

Appendix A. Excerpt of SoundSpel Version of a Passage Used in the Experiment A Raccoon’s Life Julia Cunningham’s story about a talking raccoon is a A Raccoon’s Lief Julia Cunningham’s story about a tauking raccoon is a fantasy, a make-believe story. Even though the talking raccoon in Macaroon fantasy, a maek-beleev story. Eeven tho th tauking raccoon in Macaroon is not true to life, the environment the author describes could be real. In the

490

Michael D. Young, Michelle L. Wilson and Alice F. Healy is not troo to life, th envieronment th author descriebs cuud be reel. In th spring and summer a leafy, green forest is a comfortable place for raccoons. spring and sumer a leefy, green forest is a cumfortabl plaes for raccoons.

Appendix B. Preliminary Experiment Prior to testing ESL students, 48 native English-speaking University of Colorado undergraduates participated in a preliminary experiment to ensure the adequacy of the materials. These students, who were tested individually, received credit in an introductory psychology course for their participation, which lasted approximately 1 hour. Eighteen passages were selected from third, fourth, and fifth grade readers (Allington, Cortez, Cunningham, Sebesta, and Tierney, 1989a, 1989b, 1989c), with six passages from each grade level. Each passage was accompanied by four three-alternative multiple-choice comprehension questions written by the experimenters. Each student was given three passages to read, one from each grade level, and each passage was read by 8 students. The texts were written in the same format as the pretest and posttest; that is, they did not contain DoubleLine but were written double-spaced with no repeated lines or SoundSpel text. Immediately after reading a given passage, students answered the multiple-choice questions for that passage, which occurred on a separate page given to them by the experimenter. The overall accuracy was very high for all of the passages; the mean proportion correct was over .95. Nevertheless, the passages used in the experiment were selected from the third and fourth grade passages using the criterion that all selected passages had the highest proportion of correct responses by the 8 students who read them.

Appendix C. Passage Read Aloud by Participants Learning SoundSpel System It was about th midl of a laet spring morning when th Hors caem inside frum th far sied of th bair rij and stuud for a whiel on th hieest part. He apeerd to be foer or five years oeld, compactly bilt and with a ruf coet sumwherr between bloo-grae and mous culor. He wor no owner’s brand, and th oenly distinktiv mark on him was th straengj mask-like pach of darker culor covering his foerhed and upper muzl. Th untiedy tangles and mats of cockleburs and mud in his long tael and maen markt him as a raenj hors, and not wun that had simply straed off frum sum ranch or farm.

References Allington, R. L., Cortez, J., Cunningham, P. M., Sebesta, S. L., and Tierney, R. J. (1989a). Collections: Between times. Glenview, IL: Scott Foresman and Company. Allington, R. L., Cortez, J., Cunningham, P. M., Sebesta, S. L., and Tierney, R. J. (1989b). Collections: In good company. Glenview, IL: Scott Foresman and Company.

Improving Reading Skills for ESL Learners Using SoundSpel

491

Allington, R. L., Cortez, J., Cunningham, P. M., Sebesta, S. L., and Tierney, R. J. (1989c). Collections: Sometimes I wonder. Glenview, IL: Scott Foresman and Company. Bryant, P. E., MacLean, M., Bradley, L. L., and Crossland, J. (1990). Rhyme and alliteration, phoneme detection, and learning to read. Developmental Psychology, 26, 429-438. Feldman, A. and Healy, A. F. (1998). Effect of first language phonological configuration on lexical acquisition in a second language. In A. F. Healy and L. E. Bourne, Jr. (Eds.), Foreign language learning: Psycholinguistic studies on training and retention (pp. 5576). Mahwah, NJ: Erlbaum. Healy, A. F., Wohldmann, E. L., and Bourne, L. E., Jr. (2005). The procedural reinstatement principle: Studies on training, retention, and transfer. In A. F. Healy (Ed.), Experimental cognitive psychology and its applications (pp. 59-71). Washington, DC: American Psychological Association. Mann, V. A. (1986). Phonological awareness: The role of reading experience. Cognition, 24, 65-92. Mole, A. (Ed.) (2003). Stories in reformed spelling. Boulder, CO: American Literacy Council. Morris, C. D., Bransford, J. D., and Franks, J. J. (1977). Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behavior, 16, 519-533. Olson, R. K., Wise, B., Ring, J., and Johnson, M. (1997). Computer-based remedial training in phoneme awareness and phonological decoding: Effects on the posttraining development of word recognition. Scientific Studies of Reading, 1, 235-253. Patel, T. K., Snowling, M. J., and de Jong, P. F. (2004). A cross-linguistic comparison of children learning to read in English and Dutch. Journal of Educational Psychology, 96, 785-797. Rayner, K., Foorman, B. R., Perfetti, C. A., Pesetsky, D., and Seidenberg, M. S. (2001). How psychological science informs the teaching of reading. Psychological Science in the Public Interest, 2, 31-74. Roediger, H. L., III, Weldon, M. S., and Challis, B. H. (1989). Explaining dissociations between implicit and explicit measures of retention: A processing account. In H. L. Roediger, III, and F. I. M. Craik (Eds.), Varieties of memory and consciousness: Essays in honour of Endel Tulving (pp. 3-41). Hillsdale, NJ: Erlbaum.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 22

A NOVEL TRANSLITERATION APPROACH IN AN ENGLISH-ARABIC CROSS LANGUAGE INFORMATION RETRIEVAL SYSTEM Ghita Amor-Tijani∗ and Abdelghani Bellaachia Department of Computer Science, The George Washington University Washington DC, USA

Abstract One of the main issues facing Cross Language Information Retrieval (CLIR) is untranslatable words, i.e., words not found in dictionaries, which are usually referred to as Out Of Vocabulary (OOV) words. Bilingual dictionaries in general do not cover most proper nouns (e.g., names of places, people, countries, etc.), which constitute a large proportion of OOV words. As they are often primary keys in a query, their correct translation is often necessary to maintain a good retrieval performance. Because they are spelling variants of each other in most languages, an approximate string matching technique against the target database index is usually used to find the target language correspondents of the original query key. The n-gram technique has proven to be the most effective among other approximate string matching techniques. A more complicated issue arises when the languages dealt with have different alphabets. The approach usually taken is transliteration. It is applied based on phonetic similarities between the languages involved. However, transliteration by itself cannot guarantee the exact spelling of the transliterated words as found in the document collection. There are a variety of ways that a transliterated word can be spelled despite conventions that might exist. The fact that there is no one correct way of spelling a transliterated word shows the need for a technique that is capable of generating the different spellings found in the document collection. In this study, we chose to combine both transliteration and the n-gram technique in an English-Arabic CLIR system, in which Arabic documents were searched using English queries. We evaluated the effectiveness of this approach and compared it with other transliteration approaches. Experimental results showed the retrieval improvement gained using our transliteration approach over other existing approaches. ∗

E-mail address: [email protected], [email protected]

494

Ghita Amor-Tijani and Abdelghani Bellaachia

Introduction Issues related to ambiguity and Out Of Vocabulary (OOV) words in Cross Language Information Retrieval (CLIR) have been addressed by several studies. Using a dictionarybased approach in information retrieval, queries are translated by replacing source language query keys with their target language equivalents provided by online dictionaries. This approach, however, presents several problems: previous studies have demonstrated that automatic Word by Word (WBW) translation of queries using Machine Readable Dictionaries (MRD) results in a 40-60% drop in effectiveness compared to monolingual retrieval (Hull and Grefenstette, 1996). The main factors that cause this loss in effectiveness are 1) OOV query keys, most of which are not covered by general dictionaries; 2) the processing of inflected words; 3) phrase identification and translation; and 4) lexical ambiguity in source and target languages. The problems associated with OOV words include compound words and cross lingual spelling variants, specifically proper nouns and technical terms (Pirkola, Hedlund, Keskuslato, and Järvelin, 2000; Pirkola, Hedlund, Keskuslato, and Järvelin, 2001). Around 50% of OOV words were observed to be proper nouns (Davis and Ogden, 1998). Studies presented at the 2002 Text REtrieval Conference (TREC2002) on English-Arabic cross language track noted that performance drops by more than 50% when proper nouns are not covered by the dictionaries used (Larkey, AbdulJaleel, and Connell, 2003). Their correct translation is therefore crucial to the overall retrieval performance of a CLIR system. General dictionaries only include the most commonly used proper nouns and technical terms. Those proper nouns and technical terms not found in dictionaries are subsequently considered untranslatable by the CLIR system processing them. As they are usually primary keys in the query, their correct translation might be necessary to maintain a good retrieval effectiveness. A common method for handling OOV words is to carry them over unchanged to the target query. This can, however, lead to deterioration in retrieval effectiveness. As OOV words have spelling variants in most languages, an approximate string matching technique against the target database index could be used to find equivalents of the original query key in the target language. Transliteration could also be determined based on the phonetic similarities of the languages in question. The n-gram technique is a language-independent technique which has been reported to be even more effective than other string matching techniques such as Soundex or Phonix, which are based on phonetic similarity (Zobel and Dart, 1995). In n-gram matching, query keys and terms in the index of the document set are split into sub-strings of length n. The n-gram sets of the query key and the document's indexed terms are compared, and the best matching words are used as the key’s correspondents. The number of matching words retrieved is based on the threshold used. The n-gram technique is effective because proper nouns as well as technical terms are usually spelling variants of each other in such languages (Pirkola, Keskustalo, Leppänen, Känsälä, and Järvelin, 2002). Sometimes, an approximate string matching technique misses the correct target word as the best match; a transformation rule based translation (TRT) was applied to generate intermediate forms (Pirkola, Toivonen, Keskustalo, Visala, and Järvelin, 2003). However, the TRT approach may be useless if it just generates a set of translation equivalent candidates but is not able to identify the one correct equivalent for a source word. Using regular frequency patterns of generated word forms for a source word was the alternative method to recognize the correct target word (Pirkola,

A Novel Transliteration Approach in an English-Arabic Cross …

495

Toivonen, Keskustalo, and Järvelin, 2006). Those were obtained using the TRT (Pirkola et al., 2003). This approach was demonstrated to effectively identify the target language equivalent of an OOV word. As we are dealing with a CLIR system involving English and Arabic, which are two languages with different alphabets, a string matching technique cannot be applied. Different versions of transliteration were used as the alternative methods to render the query term orthography into that of the target language (Darwish, Doermann, Jones, Oard, and Rautiainen, 2002; Larkey, Connell, and AbdulJaleel, 2003). Experiments showed that generating multiple spelling variants for Arabic transliterated words improves effectiveness (Larkey, AbdulJaleel, and Connell, 2003). One of the transliteration models that was shown to improve retrieval effectiveness is the n-gram based statistical transliteration technique studied by the University of Massachusetts (UMass) (AbdulJaleel and Larkey, 2003). This model generates multiple Arabic transliterations based on a mapping derived from a character-level alignment of a list of Arabic/English word pairs. There are other standard transliterations like the library of congress scheme (Library of Congress, 1998); which mainly follows the transliterations used by the best known Arabic-English dictionary, the Hans Wehr dictionary of Modern Written Arabic edited by J.M. Cowan. Our approach was to extract the spelling variants of the transliterated words from the document collection, rather than to generate them automatically. An approximate string matching technique seemed to be the best technique to be used for that purpose. In our transliteration approach, which we refer to as Transliteration N-Gram (TNG), we combine transliteration with an approximate string matching technique. We first generate one transliteration to have the OOV word in the same alphabet as our document collection. Then, we use the n-gram as it was shown to be the most effective among other approximate string matching techniques, such as Soundex and Phonix (Zobel and Dart, 1995). Unlike fuzzy translation (Pirkola et al., 2003), our technique focuses on two languages of different orthographies and the two steps of data transformation are used for different purposes. This paper is organized as follows. In Section 2, our system is illustrated and the TNG technique, which is our transliteration approach, is defined. Section 3 shows the experimental results and the effectiveness of our transliteration approach on the TREC 2002 (English-Arabic) cross language track. A comparison of our work with UMass statistical transliteration and the Arabic transliteration model used by the library of congress is also presented. Finally, Section 4 concludes the paper.

TNG Framework In this section, we describe our approach of transliteration and the main components of our system. We first present the architectural model that illustrates the steps of query processing and then give a detailed description of those steps.

Architectural Model The flow chart illustrated in Figure 1 shows the steps taken to process the query file. We used the English query file from TREC2002. The English query file used consisted of fifty queries.

496

Ghita Amor-Tijani and Abdelghani Bellaachia

Figure 1. TNG Framework.

Each query is processed individually by first running it through a tokenizer to extract the different tokens that are checked against an English stop word list. After stop words are removed from the query, all terms are stemmed using the Porter stemmer (Porter, 1980). Using an English-to-Arabic dictionary, stemmed terms are translated. In case there is an entry for each term in the query, the Arabic query is formed and fed to the search engine. Otherwise, terms without an entry in the dictionary are transliterated using a modified version of UMass statistical transliteration (AbdulJaleel and Larkey, 2003). Either one transliteration is generated for each OOV query term to form the final query, or spelling variants of that transliteration are extracted from the list of terms in the index of our document collection using the n-gram technique. The main difference using TNG is the new approach to generate multiple transliterations in a CLIR system involving two languages of different alphabets; more specifically, the Arabic language. TNG is a combination of transliteration and n-gram string matching technique. Once the OOV words are processed using either transliteration or TNG, the final Arabic query is formed and fed to the search engine. Examples of the data transformation that occurs in the transliteration and the ngram processing steps of OOV words are given later in Table 1. Table 1. Examples of Spelling Variants Extracted Using Ngram0 OOV word Clinton

Transliteration ‫( نﻮﺗﻧﻳﻟﻜ‬Clinton)

Spelling variants extracted using Ngram0 ‫( نﻮﺗﻧﻟﻜ‬Clnton) ،‫(نﻮﺗﻨﻳﻳﻟﻜ‬Cliinton) ،‫(نﻮﺗﻨﻟﻜ‬Clnton) ‫(نﺗﻧﻳﻟﻜ‬Clintn) ،‫( نﻮﺗﻳﻧﻴﻟﻜ‬Cliniton) ،‫( نﻮﺗﻴﻨﻟﻜ‬Clniton)

Query Processing Using the TNG technique, both transliteration and an n-gram mapping are applied to get a set of possible transliterations. One transliteration is first generated and the n-gram string matching technique is then applied on the transliterated term and the list of terms in the index

A Novel Transliteration Approach in an English-Arabic Cross …

497

of the document collection. The transliteration model followed is similar to the one used by UMass, known as the statistical transliteration. Their transliteration model is based on a character-level alignment performed on a list of Arabic/English word pairs. Segments of English characters that mapped to a single Arabic letter were extracted and given a probability based on how many Arabic segments an English segment corresponded to and how often the mapping occurred. Instead of generating multiple transliterations, we only produce one transliteration using the equivalent Arabic letters with the highest probability. As for the n-gram technique, it is used on the transliterated term and the list of stemmed words in the index of the document collection. Five different approaches were used and compared. The basic idea of the n-gram technique is to consider the set of pairs of adjacent characters of words to be compared and calculate the degree of similarity between them. Similarity is based on the number of similar n-grams and the total number of unique n-grams. Digrams (sets of two characters) were used as they proved to give better performance than trigrams (sets of three characters) for this specific data collection. Five types of character combinations were considered in the calculation of the similarity value: −

Ngram0: only adjacent characters (combination of two letters) are considered. Word abcd

−

Ngram1: adjacent and non-adjacent characters separated by one character are considered. Word abcd

−

−

Digram set for ngram1 {ab,ac,bc,bd,cd}

Ngram2: adjacent and non-adjacent characters separated by one character and by two characters are considered. Word abcd

−

Digram set for ngram0 {ab,bc,cd}

Digram set for ngram2 {ab,ac,ad,bc,bd,cd}

Ngram0-1: adjacent (first set) and non-adjacent (second set) characters separated by one character are considered. Only digrams belonging to the same category set are compared. Word Digram sets for ngram0-1 abcd {ab,bc,cd}{ac,bd} Ngram0-2: adjacent (first set) and non-adjacent (second set) characters separated by one character and by two characters are considered. Only digrams belonging to the same category set are compared. Word abcd

Digram sets for ngram0-2 {ab,bc,cd}{ac,ad,bd}

498

Ghita Amor-Tijani and Abdelghani Bellaachia

The five different approaches of the n-gram technique were indeed used to extract the different spellings of OOV words using a transliterated form, which can be one of the different ways the word is spelled. Examples of spelling variants of transliterated words extracted using ngram0 are illustrated in Table 1. Those approaches were compared and experimental results showed that using the TNG technique that combines transliteration with the ngram0 function showed the best retrieval performance (Bellaachia and Amor-Tijani, 2008). This is the TNG approach that will be used in the next comparisons.

Experimental Results The Indri search engine from the Lemur1 project was used as a tool for information retrieval. It is a cooperative effort between the University of Massachusetts and Carnegie Mellon University. The document set used is known as the "Arabic Newswire Part 1". It was obtained from LDC. It contains approximately 383,872 news articles taken from the “Agence France Presse” (AFP) Arabic newswire dated from May 13, 1994 through December 20, 2000. The collection is about 896MB. The documents are represented in Unicode and encoded in utf-8. The category of the Arabic language used is the Modern Standard Arabic (MSA). This collection consists of 76 Million tokens; among them are 666,094 unique words. This number is reduced to 240,823 after the collection is stemmed and stop-words are removed. This corpus was indexed using the application “buildindex” available in Indri. TREC 2002 topics2 in both English and Arabic were used to search the Arabic document set. Relevance judgments3 corresponding to the topics and the document set were used to evaluate the effectiveness of our retrieval system since they give information about which documents are relevant to what query. Fifty queries were used in our experimental approach. Only the topic queries which are typical user’s queries were considered in order to evaluate our techniques. Below is an example of an English query:

Number:AR26 Kurdistan Independence Description: How does the National Council of Resistance relate to the potential independence of Kurdistan? Narrative: Articles reporting activities of the National Council of Resistance are considered on topic. Articles discussing Ocalan's leadership within the context of the Kurdish efforts toward independence are also considered on topic.

The English topics were processed before they were fed to the search engine. The TREC2002 query file consists of a title, description and a narration field for each of the 50 queries. Only the topic field, which corresponds to short queries, was considered in our study as it represents typical user queries. The queries were first tokenized and stop-words were 1

http://www.lemurproject.org/indri/ http://trec.nist.gov/data/topics_noneng/index.html 3 http://trec.nist.gov/data/qrels_noneng/index.html 2

A Novel Transliteration Approach in an English-Arabic Cross …

499

removed. These are common words with no content-specific information. The resulting query terms were then stemmed by the Porter stemmer. The stemmed query terms were subsequently translated using a corpus generated from a set of parallel UN documents provided by the LDC. It provides a varying set of translation probabilities for the plausible translations of each entry term. Its English entries are processed with the Porter stemmer and the Arabic translations are stemmed using Al-Stem provided by Kareem Darwish from the University of Maryland. For each query term, the translation evaluated at the highest probability was considered to be the best translation. The final translated query was formed using these translations. Previous research showed that considering translations with probability higher than 0.15 gave the best performance for this dictionary, referred to as UNdict (Larkey and Connell, 2003). We decided to use only the best translation as we are not testing the efficiency of this dictionary, but rather other issues more specifically related to word-by-word translation. In case no entry was found in the dictionary, the query term was passed unchanged; unless further processing was considered. Once the translated query was ready, the application “runquery” from Indri was used to search the Indri repositories built from the TREC documents. The indexed documents were searched and a list of ranked documents was retrieved reflecting relevant documents to different queries. The following results are for 50 topic queries. All proper nouns used in topic queries were detected manually and their translation was disconnected from the dictionary for two reasons. The effectiveness of the TNG technique can be better reflected in the results. Also, typical dictionaries usually do not include translations for proper nouns. Thirty one queries included untranslatable words, among which 29 included about 25 proper nouns (e.g. Sadeq AlSadr, Lebanon) and 9 adjectives (e.g. Iranian). Variants of the OOV term were chosen based on a Similarity Value (SIM) calculated using Equation 1. A SIM threshold was chosen after comparing the performance of different runs. All terms with a SIM higher than or equal to 0.4 were considered in the translated query. Those terms were combined with the synonym operator “#wsyn” from Indri using the SIM value as the weight of the term. Equation 1: Similarity Value SIM (N1, N2)= | N1∩N2 | / | N1UN2 | where, N1 and N2 are digram sets of two words. | N1∩N2 |: number of similar digrams | N1UN2 |: number of unique digrams Precision and recall are two metrics used to evaluate retrieval performance in an IR system: Precision is the measure of the system’s capacity to retrieve relevant documents at the top ranking. It is the ratio of relevant documents retrieved over the total number of all documents retrieved. A high precision is achieved when most of the retrieved documents are relevant.

Ghhita Amor-Tijaani and Abdelgghani Bellaachhia

5000 Equation 2: Precision

he measure off the system’s capacity to reetrieve all relevvant documennts. It is the Recall is th raatio of relevan nt documents that t are retrievved over the tootal number of o all relevant documents avvailable. A peerfect recall ratio r of 1.0 iss achieved when w the systeem retrieves all a relevant doocuments. Equation 3: Recall

To computte both precisiion and recall,, the documennts obtained inn the search reesults using thhe IR system are a compared against the reelevance judgment file whiich specifies thhe relevant doocuments for each query. The T latter is usually u predeffined for a speecific set of queries q and doocument set. A group of people assesss the collectioon according to specific criteria c and deetermine whicch documents are relevant too which queriees and which ones o are not. This T data is sttored in the relevance r judggment file. When W the querries are run using u the IR system, s the vaariables needeed to computee both precisioon and recall are a identified by b comparingg the results obbtained using Indri with thee ones in the reelevance judggment file. Thee relevant doccuments are thhose in the relevance judgm ment file, and the t retrieved documents d aree those obtaineed with the IR R system. Mean averrage precision (MAP) is useed to measure how relevant the retrieved documents arre. Precision at a any recall level l is calcullated as the avverage precisiion of all queeries at that reecall value. A standard tooll used by the TREC commuunity known as a “trec_eval”” is used in orrder to get thee system’s perrformance datta. It takes botth the results file from the application “rrunquery” from m Indri and thhe standard seet of judged reesults from TR REC and utilizzes them as paarameters to generate g tabless reflecting perrformance as a measure of precision p and recall. For the ev valuation of each e run, thee number of retrieved doccuments, the number of reelevant docum ments, and the number of reetrieved docum ments that are relevant is geenerated. A taable of precisio on averages att different recaall values (0.00, 0.1,…,1.0) is generated foor each run. The average prrecision valuees given are thhe percent of retrieved docuuments that are a relevant p off all relevant documents d is retrieved. r In other o words, precision p at affter a certain percentage reecall 0.1 is prrecision after 10% of relevvant documennts are retrieveed. Values are averaged ovver all queriess to determinee the MAP of the run. The statistical s signnificance of peerformance im mprovement iss evaluated using the Wilcoxxon signed rannks test with the t p-level set at 0.5. Different runs r were caarried out annd compared for all 50 queries q to coompare the peerformance off the different transliterationn approaches: −

Mono: In the monolingual run, Arrabic queries were w used to search s the doccument set. Al-Stem stemmer was w used to sttem both the Arabic A queriees and the doccument set.

A Novel Transliteration Approach in an English-Arabic Cross …

−

− − −

501

Stemming was included in the monolingual run as well, because our interests do not lie in evaluating the stemmer. Transliterated: English queries were used to search the document set. First, stopwords were removed. The queries were then stemmed with the Porter stemmer. Translations for query terms were looked up in the UNdict. OOV words were then transliterated generating one transliteration corresponding to the one with the highest probability using UMass transliteration model. Congress: English queries were translated and OOV words were transliterated using the congress transliteration model. UMass: English queries were translated and OOV words were transliterated using UMass statistical transliteration model. TNG: English queries were translated and OOV words in translated English queries were first transliterated, then the ngram0 function was applied on the transliterated term and the indexed stemmed terms. Different possible spelling variations of the transliterated word were derived using the SIM value as the lower bound of possible correct spellings. Those were grouped a weighted synonym “#wsyn” operator in the final query.

Retrieval Performance A comparison was made between our TNG approach, the Library of Congress transliteration scheme (Library of Congress, 1998), and the statistical transliteration used by the University of Massachusetts (UMass) (AbdulJaleel and Larkey, 2003). The difference between the three transliteration approaches is explained earlier in Section 1 and Section 2.1. The main difference between the models is that the Library of Congress scheme is based on a one-to-one mapping generating one transliteration for each word, whereas the other two models generate multiple transliterations based on two different approaches. As for the TNG technique, the ngram0 function was used in the comparison as it is explained earlier in Section 2.2. Table 2 and Table 3 illustrate the effectiveness of all three approaches. Table 2. Interpolated Average Precision, Averaged over 50 Queries Recall 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Mono 0.5968 0.3845 0.3192 0.2831 0.2494 0.2306 0.1839 0.1567 0.1298 0.0651 0.0134

Translated 0.273 0.1621 0.1391 0.1276 0.1028 0.0965 0.0897 0.0787 0.065 0.0532 0.0028

TNG 0.4379 0.2793 0.243 0.2055 0.1744 0.1509 0.1264 0.1039 0.0722 0.0555 0.0054

502

Ghita Amor-Tijani and Abdelghani Bellaachia Table 3. Interpolated Average Precision, Averaged over 50 Queries Transliterated 0.3211 0.203 0.1802 0.1601 0.135 0.1228 0.1112 0.0951 0.0755 0.0535 0.0074

Recall 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

UMass 0.3544 0.2221 0.1971 0.1721 0.1458 0.1256 0.1128 0.0978 0.0794 0.0567 0.0074

Congress 0.323 0.2016 0.1801 0.1524 0.1228 0.1031 0.0914 0.0792 0.0646 0.0488 0.0032

TNG 0.4379 0.2793 0.243 0.2055 0.1744 0.1509 0.1264 0.1039 0.0722 0.0555 0.0054

It displays the 11-point interpolated average precision values for both runs, averaged over the 50 TREC 2002 queries used. The MAP values were also obtained and compared. Precision of the monolingual run is first calculated by using the Arabic translated TREC queries. Precision using transliterated queries is then calculated using both the UMass transliteration with the highest probability and the congress transliteration model. Finally, precision using the TNG technique is calculated and compared to precision using transliteration only. Table 2 and Figure 2 show the benefit of handling untranslatable words, which are mostly proper nouns in this case. Using the TNG technique significantly improves average precision. 0.7 Mono Translated TNG

0.6

Precision

0.5 0.4 0.3 0.2 0.1 0 1

2

3

4

5

6 Recall

7

8

Figure 2. Precision-Recall Graph of the Mono, Translated, and TNG Runs.

9

10

11

503

A Novel Transliteration Approach in an English-Arabic Cross …

0.5 Transliterated Congress UMass TNG

0.45 0.4 Precision

0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 0

0.1

0.2

0.3

0.4

0.5 0.6 Recall

0.7

0.8

0.9

1

Figure 3. Precision-Recall Graph of the Transliterated, Congress, UMass, and TNG Runs.

As explained in Section 2, adjacent characters of the transliterated word are compared to the set of stemmed words in the index of the document collection. Those words with similarity value higher or equal to 0.4 are considered in the translated query. The improvement gained using the TNG technique on OOV words shows the importance of correctly translating query terms, specifically proper nouns that are usually primary keys in a query. The significant improvement also emphasizes the need of having multiple spellings of those words in the translated query. Table 3 and Figure 3 show the difference in performance between generating one and multiple transliterations for proper nouns. We notice that using TNG, which combines transliteration and the n-gram technique, gives a better performance than the statistical transliteration used by UMass.The difference in performance can be explained by the fact that transliteration by itself, although it can generate different spellings of a word, cannot cover all the variations of a word spelling used in the document collection. The n-gram technique compensates this shortcoming by allowing a wider range of existing spelling variations to be used in the search query. Both those approaches give a better performance than the transliterated run, the modified version of the statistical transliteration approach, and the library of congress transliteration model. These observations emphasize the importance of generating multiple transliterations to maintain a good retrieval performance.

Summary of Results Table 4 and Table 5 summarize the results obtained at different runs. These results were obtained when processing 50 topic queries. The percent improvement in effectiveness is given as a percentage of transliteration alone in which one transliteration is produced for each OOV word (%Transliterated). The different approaches used are also compared to the

504

Ghita Amor-Tijani and Abdelghani Bellaachia

monolingual run (%Mono). This comparison represents how effective each technique is, as the monolingual run reflects a CLIR system without the effect of translation ambiguity. Table 4 shows the improvement in effectiveness of the TNG and UMass techniques over the transliterated run. A significant improvement of 28% and 7% was obtained respectively. The difference in performance between the transliteration scheme used in the transliterated run and by the Library of Congress is insignificant. Both approaches generate one possible transliteration, which limits the retrieval performance of the system. Table 5 depicts the performance of all runs compared to the monolingual run. The Library of Congress scheme yields about the same performance as the transliterated run. Both approaches generate one transliteration for each OOV word. The transliteration generated might not necessarily be used in the document collection. The statistical transliteration used by UMass gives a better performance than the transliterated run as multiple transliterations as generated. An effectiveness of 72% of that of the monolingual run is achieved when TNG is applied. Table 4. Percent Improvement of the Congress, UMass, and TNG Runs over the Transliterated Run Transliterated Congress UMass TNG

MAP 0.1209 0.1123 0.1294 0.1553

%Transliterated ---7% 7% 28%

Table 5. Percent Effectiveness of the Translated, Transliterated, Congress, UMass, and TNG Runs Compared to the Monolingual Run Mono Translated Transliterated Congress UMass TNG

MAP 0.2168 0.0989 0.1209 0.1123 0.1294 0.1553

%Mono --46% 56% 52% 60% 72%

Table 6. Percent Improvement of the TNG and UMass Techniques over The Library of Congress Transliteration Scheme Congress UMass TNG0

MAP 0.1123 0.1294 0.1553

%UMass --15% 38%

This improvement is significant as effectiveness is only 46% of that of the monolingual run when OOV words are not processed. These results demonstrate the significant improvement in effectiveness gained in our CLIR system when OOV words are processed,

A Novel Transliteration Approach in an English-Arabic Cross …

505

and specifically when variations of transliterations are extracted from the document set rather than generated automatically. The results in Table 6 show that TNG and UMass statistical transliteration give a significant improvement in MAP of 38% and 15%, respectively, over the Library of Congress transliteration model. This is because using the Library of Congress transliteration scheme only generates one transliteration, while it has been shown in this study as well as previous studies that generating multiple transliteration spellings is necessary for a good retrieval performance.

Conclusion In this study we have compared retrieval performance using three transliteration schemes: the Library of Congress, UMass, and TNG. Both the UMass and TNG techniques gave better results than the standard transliteration model used by the Library of Congress. The difference in retrieval effectiveness is due to the fact that the Library of Congress scheme uses a one-toone mapping to produce the corresponding transliteration, thus generating one spelling possibility. It has been shown in previous studies as well as in this study that generating multiple transliteration spellings enhances retrieval performance. Indeed, the TNG technique and UMass statistical transliteration model gave good results. They yield a retrieval performance better than the Library of Congress model by 15% and 38%, respectively. Experimental results also show that the TNG based approach significantly improves retrieval effectiveness and gives a MAP that is 20% better than the UMass statistical transliteration, which also produces different transliterations. The improvement in effectiveness is due to the use of different spelling variants of transliterated terms used in the existing document collection, in the translated Arabic query. We can conclude that generating multiple transliterations indeed enhances performance and correctly choosing the possible spelling variants is important to further improve it.

References AbdulJaleel, N., and Larkey, L.S. (2003). Statistical transliteration for English-Arabic cross language information retrieval. Proceedings of the twelfth international conference on information and knowledge management (CIKM 2003) (pp. 139-146). New Orleans, LA: ACM Press. Bellaachia, A. and Amor-Tijani, G. (2008). Proper nouns in English-Arabic cross language information retrieval. Journal of the American Society for Information Science and Technology, 59(12), 1925-1932. Darwish, K., Doermann, D., Jones, R., Oard, D., and Rautiainen, M. (2002). TREC-10 experiments at University of Maryland CLIR and video. Proceedings of Text Retrieval Conference TREC10 (TREC2001) (pp. 549-562). Gaithersburg, MD. Davis, M.W. and Ogden, W.C. (1998). Free resources and advanced alignment for crosslanguage text retrieval. Proceedings of the sixth text retrieval conference (TREC-6) (pp. 385-394). Gaithersburg: NIST Special Publication.

506

Ghita Amor-Tijani and Abdelghani Bellaachia

Diab, M. (2004). An unsupervised approach for bootstrapping Arabic word sense tagging. Proceedings of the Workshop on Arabic Based Script Languages at the International Conference on Computational Linguistics (COLING 2004) (pp. 43-50). Stroudsburg, PA: ACL. Diab, M., Hacioglu, K., and Jurafsky, D. (2004). Automatic tagging of Arabic text: From raw text to base phrase chunks. Proceedings of Human Language Technology and the North American Chapter of the Association for Computational Linguistics (HLT-NAACL 2004) (pp.149-152). Stroudsburg, PA: ACL. Hull, D.A., and Grefenstette, G. (1996). Querying across languages: a dictionary-based approach to multilingual information retrieval. Proceedings of the 19th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (pp. 49-57). New York City: ACM Press. Larkey, L.S., AbdulJaleel, N. and Connell, M. (2003). What's in a name?: Proper names in Arabic cross language information retrieval. CIIR Technical Report, IR-278. [Available at http://ciir.cs.umass.edu/pubfiles/ir-278.pdf]. Larkey, L.S., and Connell, M. (2003). Structured Queries, Language Modeling, and Relevance Modeling in Cross-Language Information Retrieval. Information Processing and Management Special Issue on Cross Language Information Retrieval. 41, 457-473. Larkey, L.S., Connell, M., and AbdulJaleel, N. (2003). Hindi CLIR in thirty days. ACM Transactions on Asian Language Information Processing (TALIP), 2(2), 130-142. New York City: ACM Press. Library of Congress. (1998). ALA-LC Romanization tables: transliteration schemes for non_roman scripts. [Available at http://www.loc.gov/catdir/cpso/roman.html]. Pirkola A, Hedlund T, Keskuslato H, and Järvelin K. (2000). Cross-Lingual Information Retrieval Problems: Methods and findings for three language pairs. ProLISSa Progress in Library and Information Science in Southern Africa. First biannual DISSAnet Conference. Pretoria, 26-27. Pirkola, A., Hedlund, T., Keskustalo, H., and Järvelin, K. (2001). Dictionary-based cross language information retrieval: Problems, methods, and research findings. Information Retrieval, 4(3/4), 209-230. Hingham, MA: KAP. Pirkola, A., Keskustalo, H., Leppänen, E., Känsälä, A.P., and Järvelin, K. (2002). Targeted sgram matching: A novel n-gram matching technique for cross- and monolingual word form variants. Information Research, 7(2) [Available at http://InformationR.net/ir/72/paper126.html]. Pirkola, A., Toivonen, J., Keskustalo, H., and Järvelin, K. (2006). FITE-TRT: A high quality translation technique for OOV words. Proceedings of the 21st Annual ACM Symposium on Applied Computing (SAC 2006) (pp. 1043-1049). New York City: ACM Press. Pirkola, A., Toivonen, J., Keskustalo, H., Visala, K., and Järvelin, K. (2003). Fuzzy translation of cross-lingual spelling variants. Proceedings of the 26th annual international ACM SIGIR conference on Research and development in information retrieval (pp. 345352). New York City: ACM Press. Porter, M.F. (1980). An algorithm for suffix stripping, Program 14(3), 130-137. Zobel, J., and Dart, P. (1995). Finding approximate matches in large lexicons. SoftwarePractice and Experience, 25(3), 331-345.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 c 2012 Nova Science Publishers, Inc. Editor: Kimberly L. Meinken

Chapter 23

M ETHODS FOR C ROSS -L ANGUAGE I NFORMATION R ETRIEVAL Kazuaki Kishida School of Library Information Science, Keio University, Japan

Abstract This article reviews technical methods for enhancing effectiveness of cross-language information retrieval (CLIR), in which target documents are written in different languages from that used for representing a search request. As the Internet has spread since the 1990s, the importance of CLIR has grown, and the research community of information retrieval has been tackling various CLIR problems. The purpose of this article is to overview exhaustively CLIR techniques developed in the research efforts. The following research issues on CLIR are covered: (1) strategies for matching the query and documents written in different languages, e.g., automatic translation or transliteration techniques, (2) techniques for solving the problem of translation ambiguity, (3) formal retrieval models for CLIR such as application of the language modeling, (4) methods for searching a multilingual document collection in which two or more languages are used for writing documents, etc.

1.

Introduction

Cross-language information retrieval (CLIR) is a special type of information retrieval (IR) that enables a set of documents written in one language to be searched using queries made in another language. For example, it would be convenient for Japanese people to be able to find documents in English by entering a search request in Japanese into the retrieval system. Especially, as the Internet has spread since the 1990s, the importance of CLIR for allowing users to access information resources written in various languages on the Internet has grown. Some search engines already allow users to use a translation function for searching and viewing web sites in foreign languages. Another area in which CLIR plays an important role is patent retrieval, by enabling searchers to locate easily related patents registered in foreign countries. The IR research community has therefore been tackling a wide range of CLIR problems. The Workshop of Cross-Linguistic Information Retrieval held in August 1996 during the

508

Kazuaki Kishida

ACM SIGIR Conference marked a turning point for research on CLIR. Since then, many retrieval experiments have been attempted, and have yielded unexpected enhancements in CLIR techniques. Intuitively, it would seem that machine translation (MT) could easily solve CLIR problems, because CLIR tasks can be reduced to the execution of standard monolingual IR if a given search query or its target documents are automatically translated into the other language. However, as yet there is no perfect MT system that always returns the correct results of translation, and it is suspected that such a system will be developed in the near future. Thus, in the IR field, various unique methods of language processing or document ranking have been explored for improving CLIR performance. Such research has also yielded deeper insights on some previously unconsidered aspects of IR. More specifically, CLIR includes two types of retrieval: 1. bilingual information retrieval (BLIR) 2. multilingual information retrieval (MLIR) In the case of BLIR, the documents are written in a single language different from the query language as in the above example (i.e., a Japanese-English bilingual search). Meanwhile, the target in MLIR is a heterogeneous set of documents written in two or more languages. For example, when Japanese, French and English documents are concurrently searched for a Japanese query, it is MLIR (see Figure 1). In general, MLIR is more complicated than BLIR as discussed later.

BLIR query: Japanese

document collection English

MLIR query: Japanese

document collection French English Japanese

Figure 1. BLIR and MLIR. This article explains in detail the principal techniques of CLIR by extending a previous review [Kishida, 2005]. Specifically, it discusses matching strategies, translation techniques, term disambiguation in the process of translation, formal CLIR models, and MLIR methods.

Methods for Cross-Language Information Retrieval

2.

509

Matching Strategies

2.1.

Matching Operation in CLIR

The most basic operation in IR is to compare the subject representation of a given search request (i.e., ‘query’) with that of each target document, and to measure topical similarity between them. If the degree of similarity (or relatedness) vi between the query q and a document di (i = 1, ..., N) were calculated by a method, the value vi would enable us to generate a search output in which documents are ranked in descending order of their similarity with the given query. This similarity is often called the document score or retrieval status value (RSV), and is operationally defined in a retrieval model such as a vector space model or a probabilistic model. For example, in the framework of language modeling (LM), vi is computed as the probability that a set of query terms is generated from a given document di, i.e., vi = P (Ωq |di) where Ωq is a set of terms included in q. Several formulae for computing this probability have been proposed, and a simple one of which is as follows [Hiemstra, 1998a]. P (Ωq |di ) =

Y

αP (t|di ) + (1 − α)P (t),

(1)

t∈Ωq

where t is a term in the query and α is a mixing parameter (0 ≤ α ≤ 1). Actually, P (t|di ) in the formula is easily estimated as the relative frequency of term occurrence in the document di and P (t) as the proportion of documents in which the term t appears. Inevitably, when the query and documents are written in different languages, it is not possible to estimate P (t|di ) without matching correctly a query term t with the corresponding term s in each document. Such kind of matching operation is important for CLIR. There are four types of strategies for matching a query with a set of documents in the context of CLIR [Oard and Diekema, 1998]: • Translation 1. Query translation 2. Document translation 3. Interlingual techniques • No translation – Cognate matching

2.2.

Query Translation and Document Translation

Query translation is the most widely used matching strategy for CLIR due to its tractability, i.e., the retrieval system does not have to change inherent components (e.g., index files) at all in response to queries in any language if an external translation module that can convert the text of the query into the document language is incorporated. This is a remarkable advantage of query translation in practice.

510

Kazuaki Kishida

However, it is difficult to resolve term ambiguity arising in the process of query translation because “queries are often short and short queries provide little context for disambiguation” [Oard and Diekema, 1998]. Suppose that a given query is “post service”. The word “post” has some different senses related to “mail”, “position” and so on, and this short query does not provide sufficient information for selecting a correct translation. Many empirical analyses of transaction log data in search engines on the Internet have shown that actual queries usually consist of about two terms as in this example (e.g., see [Markey, 2007]). In such a situation, the effectiveness of query translation is often limited. Therefore, a few researchers have attempted to translate target documents into the query language (i.e., document translation) due to the fact that sentences included in documents tend to be more complete and to be translated correctly [Oard and Hackett, 1998, Franz et al., 2000, Braschler and Sch¨auble, 2001]. In fact, an experiment [Oard and Hackett, 1998] showed that document translation using commercial MT software outperforms query translation in BLIR from German to English. This result suggests the potential of document translation while the short query problem remains unsolved even if documents were perfectly translated. Furthermore, it is possible to combine the results from query translation and document translation to form a hybrid approach (see Figure 2), e.g., we can use the average (or weighted average) of two document scores which were computed from query translation and document translation respectively in order to rank documents for final output [Scott McCarley, 1999, Braschler, 2004, Kang et al., 2005, Kishida and Kando, 2006]. An advantage of the hybrid approach is that it increases the possibility of correctly identifying documents having the same subject content with the query. Suppose that a term t is included in a given query and its corresponding term in the language of documents is s. If a tool for translating from the query language to the document language can not translate t into s correctly, the system will fail to find documents containing term s by this query translation. However, if another tool for translation in the reverse direction, i.e., the document language into the query language, can identify term t from term s, matching between the query and documents including term s becomes successful. search Japanese query

Japanese docs

query translation

document translation

doc list merge final list

search English query

English docs

doc list

Figure 2. Hybrid of query and document translation. For implementing document translation or the hybrid approach, it is important to solve the problem that document translation is a very cost-intensive task, i.e., it would take too

Methods for Cross-Language Information Retrieval

511

long to translate all documents by commercial MT software. One possible solution is to employ a high-speed algorithm with low complexity at the cost of translation quality such as a ‘fast document translation’ algorithm [Franz et al., 2000], which is based on a statistical approach developed by IBM research group [Brown et al., 1993]. Also, simple replacement of each term in documents with its translation using a bilingual dictionary is often used as a convenient technique for document translation. Of course, such dictionary-based method is also used for query translation when appropriate MT software is not available as discussed later.

2.3.

Interlingual Techniques

In interlingual techniques, an intermediate space of subject representation into which both the query and the documents are converted is used to compare them. There are two main categories of this technique [Oard and Diekema, 1998]: 1. Matching within a space generated by latent semantic indexing (LSI) 2. Matching via multilingual thesauri 2.3.1.

LSI-based Technique

Suppose that there is a document-aligned parallel corpus in which N documents are included and each document has a pair of equivalent texts in two different languages. If the two languages correspond to the query and document languages respectively, this corpus can be used to produce an intermediate space for BLIR. We denote an M1 × N term-document frequency matrix in a language by X1 , where M1 is the distinct number of terms of this language contained in the corpus and each component xji of this matrix (j = 1, ..., M1; i = 1, ..., N ) is the frequency (or normalized frequency) of j-th term tj within the text of a document di. Similarly, X2 is assumed to be an M2 × N term-document frequency matrix for another language. Then we constitute an M × N matrix such that   X1 X= , (2) X2 where M = M1 + M2 . According to the theory of linear algebra, the matrix X can be broken down such that X = UΛVT where U is an M × r orthogonal matrix, Λ is an r × r diagonal matrix, V is an N × r orthogonal matrix and r is rank of X. This decomposition is called singular value decomposition (SVD). Latent semantic indexing (LSI) theory [Deerwester et al., 1990, Landauer and Littman, 1990] extracts b principal diagonal elements from Λ (b < r) and interprets that they represent ‘latent’ meanings included in the original X. As a result, a new indexing space Ub ΛbVTb is constructed for conceptual retrieval where Ub and Vb are matrices in which only b columns are extracted from original ones, respectively, and Λb is a diagonal matrix including only b principal elements. Since X is derived from a parallel corpus, Ub ΛbVTb can be interpreted as a multilingual indexing space in which each meaning is represented independent of its language expressions. Therefore, if we can project a query and a document in different languages into this

512

Kazuaki Kishida

indexing space, it is possible to compare them within the space. One of the methods is to −1 T T compute Λ−1 b Ub di and Λb Ub q where di and q are subject representations (e.g., termfrequency vectors) of di and q, respectively [Rehder et al., 1998, Landauer et al., 2007]. Note that di is an M -dimensional vector and its (M1 + 1)-th to M -th elements are zero if di is written in the first language. Similarly, the first to M1 -th elements are zero in the vector of q in the second language. Therefore, an inner product of the two vectors dTi q T T −1 T is always zero, but (Λ−1 b Ub di ) Λb Ub q > 0 if di has semantic similarity with q in the multilingual indexing space. This means that we can use a similarity measure based on T T −1 T the inner product (Λ−1 b Ub di ) Λb Ub q as vi for ranking documents. Similar approaches were employed by some researchers (e.g, see [Berry and Young, 1995, Dumais et al., 1996, Littman et al., 1998]). 2.3.2.

Thesaurus-Based Techniques

Another type of interlingual approach is to map a term in a language into the corresponding term in another language via linguistically neutral labels of concepts registered in bilingual thesauri such as WordNet [Diekema et al., 1999, Bian and Lin, 2001], ULMS (Unified Medical Language System) [Eichmann et al., 1998] and so on. For example, we can use sense labels of ‘synsets’ (sets of synonymous words) provided in WordNet. The English word “train” has the sense label “train/1” implicating a line of railway cars, and it is mapped to an Italian synset including “convoglio” and “treno” in MultiWordNet1. The label “train/1” can be employed as an interlingua for English to Italian BLIR in the case that “train” implicating a line of railway cars is included in an English query (logically, the matching operation with this interlingua is equivalent to using directly such translations as “convoglio” and “treno” listed in the multilingual thesaurus). It is not so easy, however, to find interlingual concepts relevant to a given query in such a multilingual thesaurus. One possibility is to search interlingual concepts (i.e., labels) for the given query using their descriptions given in the thesaurus, and to rank interlingual concepts according to a retrieval model. For example, we may be able to identify correctly the relevant concept “train/1” for the query “trains on a railroad” if an English text in the explanation of “train/1” is searched for the query and this label has the highest score for ranking (see [Eichmann et al., 1998] for details).

2.4. 2.4.1.

Cognate Matching Fuzzy Matching

For BLIR between two similar languages (e.g., English and French), it is possible to identify document terms equivalent to a given query term by using a fuzzy matching technique without any MT. This method is sometimes called cognate matching, and in the most na¨ıve case, untranslatable terms such as proper nouns or technical terminology are left unchanged in the stage of translation. The unchanged terms are expected to match successfully a corresponding term in another language if the two languages have a close linguistic relationship. 1

http://multiwordnet.itc.it/english/home.php

Methods for Cross-Language Information Retrieval

513

A useful device for more effective matching cognates of two different languages is edit distance, which measures similarity between two character strings [Davis, 1997]. For example, Levenshtein distance is defined as the minimum number of deletions, insertions, or substitutions required to transform one string into another string [Galvez and Moya-Aneg´on, 2007]. That is, since the English word “family” can be converted into the French word “famille” by inserting “l” and substituting “e” for “y”, the distance between them is 2. If two terms between which such distance is less than a threshold are heuristically regarded as identical terms, term matching in BLIR can be operated with no translation. Interestingly, Buckley et al. pointed out that “English query words are treated as potentially misspelled French words”, and attempted to treat English words as variations of French words according to lexicographical rules [Buckley et al., 1998]. Furthermore, a kind of rule for transformation, e.g., if a Spanish term starts with “es”, “es” should be replaced with “e” for converting it into an English word, would be helpful for fuzzy matching with edit distance. Such rules may be automatically generated from statistical analysis of language resources (e.g., see [Pirkola et al., 2003] for details). An alternative approach to fuzzy matching is to decompose each word in both the query and documents into n-grams (more specifically, character-based overlapping n-grams), and to perform matching operations between the two sets of n-grams [Hedlund et al., 2002, McNamee and Mayfield, 2002b, McNamee and Mayfield, 2004]. For example, when n = 2 (i.e., bi-gram), “family” and “famille” are decomposed into { fa, am, mi, il, ly } and { fa, am, mi, il, ll, le }, respectively, and the similarity between them is computed. Since they have four common bi-grams, the Dice coefficient is calculated as (2×4)/(5+6) = 0.727..., which can be used as a metric to identify corresponding terms. It is also possible to extract and compare tri-grams (n = 3), quad-grams (n = 4) and so on. In order to enhance matching probabilities, it may be effective to transform the term beforehand according to a heuristic rule such as the Spanish-English example described above (see [Toivonen et al., 2005]). 2.4.2.

Machine Transliteration

When two languages are very different, e.g., English and Japanese, techniques based on edit distance or n-gram inevitably will not work well. However, in such cases, phonetic transliteration from English words may be effective for cognate matching. Gey stated that “...we can often find that many words, particularly in technology areas, have been borrowed phonetically from English and are pronounced similarly, yet with phonetic customization in the borrowing language” [Gey, 2001]. Accordingly, we can use machine transliteration [Knight and Graehl, 1998] for the operation of matching terms in very different languages. There are two methods of machine transliteration: modeling of transliteration and extracting transliterations from a parallel corpus [Kuo et al., 2008]. In a typical transliteration modeling, phonetic coincidence between two terms is examined. For example, the English word “America” corresponds to a Japanese word consisting of four Katakana characters pronounced “a”, “me”, “ri” and “ka”, respectively (see Figure 3). This means that the English term “America” can be identified from the combination of four sound representations “ame-ri-ka” in Japanese if a heuristic rule for converting “ka” to “ca” is introduced (see [Fujii and Ishikawa, 2001, Qu et al., 2003a] for details).

514

Kazuaki Kishida

Japanese word:

phonetic elements:

a

me

ri

ka

English word:

A

me

ri

ca

Figure 3. Example of machine transliteration. On the other hand, it is possible to estimate a conditional probability P (t|s) from parallel corpora as discussed later, where s is a transliteration and t indicates an original term. Such kinds of probabilities or alternative statistics indicating the relationship between t and s enable us to produce a list of transliterations [Kuo et al., 2008]. A combination of the rule-based method using phonetic elements and corpus-based method would be a promising strategy for correctly detecting transliterations in CLIR.

3.

Translation Methods It is widely recognized that there are three main approaches to translation in CLIR: • Machine translation • Translation by bilingual machine-readable dictionary • Parallel or comparative corpora-based methods

In addition, some researchers have recently attempted to use Internet resources for obtaining translation equivalents.

3.1. 3.1.1.

MT and Dictionary-Based Methods Machine Translation System

As already stated, intuitively, the MT system seems to be a fine tool for CLIR, and if good MT software were available, the CLIR task would become easier. However, in query translation, the MT approach has not always shown better performance than simpler dictionary-based translation. For example, an experiment [Ballesteros and Croft, 1998] indicated the dominance of dictionary-based techniques over a popular commercial MT system (of course, which method is dominant depends highly on the quality of the dictionary and MT system used for each experiment). One of the reasons is that queries are often short and do not provide sufficient contextual information for translation. In particular, a query is often represented as only a set of terms, and it may be difficult to expect MT systems to work well with such poor representation. Also, MT systems usually try to select only one translation from many candidates that each source word may have, which may lead to removing synonyms or related terms from the set of translations, and to missing relevant documents [Nie et al., 1999].

Methods for Cross-Language Information Retrieval 3.1.2.

515

Dictionary-Based Methods

Simple replacement of source terms using a bilingual machine-readable dictionary (MRD) is a general approach for CLIR when no MT system with an established reputation is available. In the world, numerous languages are spoken, and there are many pairs of languages for which effective MT software has not yet been developed. Therefore, it is still important to explore CLIR methods based on bilingual MRD because such dictionaries are easier to prepare than MT software. Most retrieval systems are still based on the so-called ‘bag-of-words’ architecture, in which both the query and document texts are decomposed into a set of words (or phrases) through a process of indexing. Thus we can translate a query easily by replacing each query term with its translation equivalents appearing in a bilingual dictionary or a bilingual term list. Due to its convenience, the dictionary-based method is also used in document translation as described above. Unlike standard MT systems, it is not difficult in dictionary-based methods to remain multiple translations for each source term. If they contain synonyms or effective related terms relevant to the given query, search performance may be improved by entering them into the IR system (of course, this strategy does not always have positive effects). However, there are some difficulties when executing this method as follows [Ballesteros and Croft, 1997]: • Specialized vocabulary not contained in the dictionary will not be translated. • Dictionary-based translation is inherently ambiguous and may add extraneous information. • Failure to translate multi term concepts such as phrases reduces effectiveness. These defects are the main reasons for degradation of CLIR performance in comparison with that of monolingual retrieval. Hull and Grefenstette stated that “...we learn that translation ambiguity and missing terminology are the two primary sources of error...” [Hull and Grefenstette, 1996]. Also, they reported that manual translation of multi word noun phrases improves retrieval performance. This suggests the importance of translation of multi term concepts. For example, a combination of independent translations from “hot” and “dog” would produce highly inappropriate translations of “hot dog” in many cases. Various methods have been developed for solving problems of out-of-vocabulary, term disambiguation and phrasal translation, as will be discussed later. Practically, it is necessary for implementing dictionary-based translation to compare a string of each source term with that of headwords in the dictionary. A stemming algorithm that removes a suffix from each string is often employed before the matching operation (e.g., “libraries” is automatically converted into “librar”) in order to find successfully the corresponding terms in the dictionary. For enhancing the performance of this process, it may be effective to apply backoff translation in which both the surface form and its stems are taken into account [Oard et al., 2001, Levow et al., 2005]. For example, in four-stage backoff translation, four different matching operations are performed: (1) matching of the surface form of a source term to the surface form of headwords in the dictionary, (2) matching of the stem of a source term to the surface form of headwords, (3) matching of the

516

Kazuaki Kishida

surface form of a source term to the stems of headwords and (4) matching of the stem of a source term to the stems of headwords. Since there is no completely perfect stemmer, it is important to use a technique such as backoff translation to increase the possibility of finding the corresponding term in the dictionary.

3.2.

Parallel Corpora-Based Method

Parallel or comparable corpora are useful resources enabling us to extract beneficial information for CLIR. As described already, the cross-lingual LSI approach uses this kind of corpus to construct a multidimensional indexing space. We can also obtain translation equivalents directly from a parallel or comparable corpus by the following methods: • Use of search results from parallel corpus • Construction of bilingual term lists 3.2.1.

PRF-based Method Using Parallel Corpus

Suppose that we execute English to French bilingual searches using a documentaligned parallel corpus of English and French languages. The first approach [Davis and Dunning, 1995, Yang et al., 1998] tries to extract French terms appearing frequently in French documents obtained from the parallel corpus by searching for the given English query (see Figure 4). That is, since each French document is aligned with an English document in the corpus, we can identify French documents corresponding to English documents searched for the given English query. Naturally, such French documents are expected to include search terms relevant to the query.

Figure 4. Example of BLIR using parallel corpus. This approach can be regarded as a kind of pseudo-relevance feedback (PRF), which is often used for query expansion (i.e., adding new search terms to the original query) in IR experiments. Its basic assumption is that top-ranked documents searched for a given query tend to be relevant and contain effective terms other than those included in the initial query.

Methods for Cross-Language Information Retrieval

517

Table 1. Term co-occurrence frequency in parallel corpus

s appears s does not appears Total

t appears nts nt − nts nt

t does not appears ns − nts N − ns − nt + nts N − nt

Total ns N − ns N

According to a standard PRF technique, such effective terms can be identified from a list of terms ranked by scores computed for each term t such that rt × log

(rt + 0.5)(N − R − nt + rt + 0.5) , (R − rt + 0.5)(nt − rt + 0.5)

(3)

where N is the total number of documents in the collection, nt is the number of documents including the term t (nt ≤ N ), R is the presupposed number of top-ranked documents (e.g., R = 30) and rt is the number of top-ranked documents including the term t (rt ≤ R). By using this score, it is possible to select some top-ranked French terms from the result of searching the parallel corpus. 3.2.2.

Estimation of Association between Terms

In the second approach, a bilingual term list is generated from parallel or comparable corpora based on empirical term association computed from co-occurrence statistics as shown in Table 1. The generated list can be used as a bilingual MRD. Suppose that t indicates a French word and s an English word. In this case, nts is the number of alignments including both t and s in a parallel corpus. From the data in Table 1, it is possible to calculate association τts between t and s as the Jaccard’s coefficient τts = nts /(nt + ns − nts ) [Adriani, 2002] or mutual information (MI) [McNamee and Mayfield, 2002a] τts = log

P (t, s) N nts = log P (t)P (s) nt ns

(4)

in which it is assumed that P (t, s) = nts /N , P (t) = nt /N and P (s) = ns /N . Also the logarithm of a likelihood ratio −2 log λ [Dunning, 1993] such that −2 log λ = 2 log

ζ(p1 , nts , ns )ζ(p2 , nt − nts , N − ns ) ζ(p, nts, ns )ζ(p, nt − nts , N − ns )

(5)

can be used as association τts where ζ(x, y, z) ≡ xy (1 − x)y−z Cp1 = nts /ns Cp2 = (nt − nts )/(N − ns )Cand p = nt /N . Furthermore, it is possible to employ other metrics such as the Dice coefficient, χ2 statistics and so on, which are computed from a contingency table like Table 1. Otherwise, a method for constructing a so-called similarity thesaurus can be used to compute term association between t and s if a concatenation of two aligned documents in different languages is assumed to be a single document [Sheridan and Ballerini, 1996, Braschler and Sch¨auble, 2000]. According to the standard vector space model for IR, it is

518

Kazuaki Kishida (t)

possible to define an N -dimensional term vector of which i-th element wi is the weight of the term in di, e.g., (t)

wi = xi (t) × log N/nt,

i = 1, ..., N,

(6)

where xi (t) is the frequency of term t in document di , and nt is the number of documents including t in this case 2 . If term similarity is computed as a cosine measure between the two vectors of t and s, query terms in the target language can be selected for each source term s according to their cosine values 3 . Also, as another method for constructing a bilingual term list from a parallel or comparable corpus, a more complicated technique based on cognate matching and morpho-semantic analysis has been proposed by [Mark´o et al., 2005]. 3.2.3.

Estimation of Translation Probability

Some researchers in the CLIR field (e.g., [Nie et al., 1999]) have attempted to estimate translation probability P (t|s) from parallel corpora according to a well-known algorithm developed by a research group at IBM [Brown et al., 1993]. By executing the algorithm for a set of sentence alignments included in a parallel corpus, a bilingual term list with a set of probabilities that a term is translated into equivalents in another language is automatically generated. The algorithm includes five models, Model 1 through Model 5, of which Model 1 is the most basic and is often used for CLIR. In particular, researchers employing the language modeling approach for CLIR (see below) have used the IBM Model 1 for computing translation probabilities (e.g., [Xu et al., 2001, Kraaij, 2002]). The fundamental idea of Model 1 is to estimate each translation probability so that the probability defined as Pr(t|s) =

m X l Y ε P (tj |sk ), (l + 1)m

(7)

j=1 k=0

is maximized, where t is a sequence (usually a sentence) of terms t1 t2 . . . tm , s is the corresponding sequence of terms s1 s2 . . . sl and ε is a parameter (s0 indicates an empty in s). That is, the translation probability P (tj |sk ) in Equation (7) is determined so that Pr(t|s), which is the probability that sentence s is translated into t, takes a maximum value. Actually, each P (tj |sk ) is estimated by iterative computation as an EM algorithm (see the appendix for details). The algorithm based on the IBM Model has become widely available with the release of a software package, the GIZA++ toolkit [Och and Ney, 2003], incorporating it as a component, but other models for statistical machine translation have also been developed. For example, another statistical translation model by [Melamed, 2000] was used for generating Japanese-English bilingual thesauri from bilingual corpora [Tsuji and Kageura, 2006]. In addition, estimation techniques based on EM algorithms have been proposed by some researchers (e.g., [Hiemstra, 1998b, Koehn and Knight, 2000, Cao and Li, 2002]). 2

Note that log N/nt is a typical idf (inverse document frequence) factor, which is generally assumed to be higher for more effetive index terms. Equation (6) is a kind of tf-idf weighting, where ‘tf’ indicates term frequency (for details, see IR textbooks, e.g., [Manning et al., 2008]) p 3 In general, the cosine measure of two vectors x and y is defined as cos(x, y) = xT y/ xT x × yT y.

Methods for Cross-Language Information Retrieval 3.2.4.

519

Availability of Parallel Corpus

Terms used in current affairs and people’s names are often not registered in MRDs or dictionaries incorporated into MT systems, in which case the correct translation can not be acquired. If a parallel corpus consisting of current news articles is available, it may be possible to construct a bilingual term list including current terminology. This is a remarkable advantage of parallel corpus-based methods. However, it would not always be possible to obtain such parallel corpus relevant to each CLIR situation, thus preventing the generalized application of parallel corpus-based methods. Even if a parallel corpus in two languages corresponding to a BLIR task is available, the search performance will inevitably decrease when the subject domains covered by the parallel corpus do not match the BLIR situation [Rogati and Yang, 2004]. One means of overcoming this problem may be to generate automatically parallel corpora from Internet resources. For example, official web sites of organizations or institutions often have English and non-English pages with almost the same contents, from which parallel or comparable corpora could be produced (see [Nie et al., 1999, Resnik, 1999, Nie, 2000] for details). Also, automatic construction of document alignments from independent collections in two different languages has been explored by [Talvensaari et al., 2006]. The fundamental operation for detecting such alignments is BLIR using an MT system or MRD, i.e., we have to specify corresponding documents in one collection by searching for translations of key terms extracted from a document in another collection. If appropriate pairs of documents are detected by the BLIR, it may be possible to find useful target terms other than translations of the key terms used as a query from the aligned documents.

3.3. 3.3.1.

Out-of-Vocabulary Problem Detecting Unknown Translations from Web Documents

If the dictionary used for translation does not include the target entries, i.e., the outof-vocabulary (OOV) situation, a special device is needed if cognate matching or machine transliteration can not be applied. Suppose that entries of a Japanese-English dictionary do not contain a particular Japanese term represented by Kanji characters, for which machine transliteration does not work. A promising method for resolving the OOV problem is to extract translations from Japanese Web documents, in which English equivalents in parentheses are often provided for proper nouns or technical terms [Chen and Gey, 2003]. It is possible to identify such English equivalents by examining frequencies of cooccurrence with the un-translated term in top-ranked documents fetched automatically from a search engine (e.g., Google) via its application program interface (API) by searching for the un-translated term [Zhang and Vines, 2004]. A simple heuristic rule is to select the English term co-occurring most frequently in a presupposed range of word sequence (i.e., text window) with the un-translated term. Of course, we can measure the association between two terms by using a metric such as χ2 statistics and so on, and select translations based on its values [Cheng et al., 2004]. Another approach for measuring the term association is to compare context vectors of two terms [Cheng et al., 2004]. We denote the context vector of term t by wt , of which

520

Kazuaki Kishida

the j-th element represents the weight of the j-th term appearing in the top-ranked Web documents. For example, the j-th element can be defined as x(t, tj ) × log N/ntj where x(t, tj ) means the number of times that term tj appears with t in a text window of fixed size. Since the un-translated term and its translation possibly share common contextual terms in the top-ranked Web documents, the similarity of their context vectors is expected to be relatively higher. The similarity can be computed as a cosine measure of two vectors wt and ws according to the standard vector space model for IR. 3.3.2.

Combining Multiple Language Resources

A method for reducing the possibility of the OOV problem occurring in the translation process is to combine multiple translation resources. For example, we can merge translation results from distinct types of resources such as MRD, MT system and parallel corpus (e.g., [Xu et al., 2001]). For merging the results, it is possible to employ techniques of data fusion or query combination [Jones and Lam-Adesina, 2002]. Suppose that we have results from two distinct MT systems. In the case of data fusion, two document scores computed from outputs by the two MT systems respectively were summed for each document. On the other hand, in query combination, before the estimation of document scores, a single query was formed by taking the unique terms from two outputs.

3.4.

Pivot Language Approach

As already mentioned, it is not always possible to obtain bilingual resources for a particular pair of languages in a BLIR task. A promising technique to circumvent the problem of limited availability of linguistic resources would be the pivot language approach, in which an intermediate language acts as a mediator between two languages for which no bilingual resources are available. Suppose that a BLIR task between Japanese and Dutch is requested by a user. Even if any language resource between Japanese and Dutch is not available, it would be easier to find Japanese-English and English-Dutch resources since English is widely used as an international language. Thus BLIR between Japanese and Dutch can be performed via English as an intermediary without direct bilingual resources between Japanese and Dutch. The pivot language approach would also alleviate the problem of explosive combinations of languages, i.e., if we have to execute BLIR tasks between each pair of n languages, O(n2 ) resources are needed. However, the pivot language approach enables us to handle the complex job with only O(n) resources [Gey, 2001]. A basic pivot language approach is transitive translation of a query using two bilingual dictionaries [Ballesteros, 2000]. In the case of searches from Japanese to Dutch via English, if Japanese-English and English-Dutch MRDs are available, CLIR can be performed by replacing Japanese query terms with the corresponding English equivalents and successively substituting the English equivalents with the Dutch equivalents. Of course, if JapaneseEnglish and English-Dutch MT systems can be used, a similar transitive translation is also feasible. In dictionary-based transitive translation, translation ambiguity becomes a more serious problem. Suppose that a Japanese source query consists of three terms, and every term

Methods for Cross-Language Information Retrieval

521

has three English equivalents. If every English equivalent has three Dutch equivalents, simple replacements will produce 27 (= 33 ) search terms in total from only three source terms, and the final set of search terms will inevitably contain some irrelevant translations [Kishida and Kando, 2004] (see Figure 5). Therefore, it is important to apply a translation disambiguation technique (discussed later) to the set of intermediary or target terms obtained from the bilingual dictionaries. Actually, an experiment [Kishida et al., 2005] reported that translation disambiguation for intermediary terms did not have a remarkable effect in German-Italian bilingual searches via English and it was enough to disambiguate only final Italian terms. Japanese query terms (source) relevant translation

irrelevant translation

…

…

…

irrelevant translations

…

…

…

…

…

…

…

Dutch query terms (target)

… irrelevant translation

: English query term (intermediary)

Figure 5. Example of dictionary-based query translation via pivot language. Retrieval experiments on the dictionary-based query translation via a pivot language have been conducted for various combinations of languages, e.g., English > French > German [Franz et al., 1999], French > English > German [Gey et al., 1999], German > English > Italian [Hiemstra and Kraaij, 1999], Japanese > English > Chinese [Lin and Chen, 2003], Chinese > English > Japanese [Chen and Gey, 2003], and so on (these are only a portion of many experiments). In particular, Franz et al. proposed some interesting techniques for searching German documents with English queries [Franz et al., 1999] as follows (the intermediary is French). • Convolution of translation probability: Estimating translation probability from an English term s to a German term t through French terms f such that P (t|s) = P f P (t|f )P (f |s).

• Automatic query generation from the intermediate language corpus: Generating French queries automatically by simply merging all non-stopwords in the top-ranked French documents searched by the English-French BLIR system, and entering the French query into the French-German BLIR system.

3.5.

Translation Quality

The quality or correctness of the translation greatly affects CLIR performance. For example, experiments [McNamee and Mayfield, 2002a, Xu and Weischedel, 2005] have shown that the size or lexical coverage of MRD has an influence on the effectiveness of

522

Kazuaki Kishida

CLIR, i.e., as the number of entry terms in the dictionary is reduced and as its translation quality becomes lower, search performance gradually deteriorates. CLIR performance can be represented as a regression model with two independent variables, translation quality z1 and ‘ease of searching for query’ z2 , i.e., y = a + b1 z1 + b2 z2

(8)

where y denotes CLIR performance, and a, b1 , b2 are parameters [Kishida, 2008]. The ‘ease of searching for query’ is an inherent nature of a given query, which is independent of the translation process. For example, if the query does not contain any specific concept that explicitly designates its content, it will be difficult to find relevant documents even if the translation is perfectly correct. Therefore, it is necessary to take the ‘ease of searching for query’ into consideration in the regression model [Kishida, 2008]. In retrieval experiments in the laboratory, the ‘ease of searching for query’ z2 can be measured by evaluation indicators (e.g., average precision) representing the performance of monolingual searches for which correct translations of queries are given by human experts. Similarly, it is possible to gauge translation quality z1 by a metric for assessing automatically translation results such as well-known BLEU [Papineni et al., 2002] based on the correct translations. In particular, a metric called WAMU (weighted average for matching unigrams) that is specifically designed to evaluate translations in CLIR situations was developed [Kishida, 2008]. A simpler formula of WAMU is:  −1 m m X X min(cj , ˜cj ) W AM U =  ωj  ωj , cj j=1

(9)

j=1

where cj is the total number of the j-th unigram (i.e., a word) in the translation, c˜j is the total number of the j-th unigram in an answer given by a human expert (i.e., a correct translation), ωj is the weight of the j-th unigram, and m is the number of unigrams in the translation. The weight ωj is calculated such that ωj = log(N/ntj ) where ntj is the number of documents including the j-th unigram tj (in CLIR situations, ntj can be obtained from the target document collection). Suppose that two translations are obtained with an answer as follows. - Translation 1: Database management system issue - Translation 2: Database administration system problem - Answer: Database management system problem In either of the translations, only one term is different from the corresponding term in the answer, i.e., “issue” in Translation 1 and “administration” in Translation 2. If these translations are to be used as a query, Translation 1 is clearly better because it represents “database management system” correctly. If the numbers of documents including each term are as shown in Table 2 and N = 1000, the WAMU scores for Translations 1 and 2 are computed as follows.

Methods for Cross-Language Information Retrieval

523

Table 2. Sample data for calculating WAMU j 1 2 3 4

1 2 3 4

Terms tj Translation 1 database management system issue Total Translation 2 database administration system problem Total

min(cj , ˜cj )

ntj

log(N/ntj )

1 1 1 0 -

5 50 50 100 -

5.30 3.00 3.00 2.30 13.59

1 0 1 1 -

5 50 50 100 -

5.30 3.00 3.00 2.30 13.59

Translation 1: (5.30 + 3.00 + 3.00)/13.59 = 0.83 Translation 2: (5.30 + 3.00 + 2.30)/13.59 = 0.78 As this example indicates, WAMU differentiates adequacy of translation between specific and non-specific terms, i.e., mistranslation of a specific term reduces its WAMU score more largely than non-specific terms due to idf factor log(N/ntj ). An experiment in [Kishida, 2008] showed that the regression model containing two independent variables, ‘ease of searching for query’ and translation quality, can explain approximately 60% of the variation in CLIR performance. In this result, translation quality has a statistically significant effect, which means that translation quality is crucial for enhancing CLIR effectiveness, as would be expected.

4. 4.1.

Term Disambiguation Techniques Translation Ambiguity

For improving translation quality, it is indispensable to select a correct translation from a set of candidates by disambiguating the sense of a given source term. In general, word sense disambiguation (WSD) is a key element in various applications such as machine translation, information retrieval and hypertext navigation systems, content and thematic analysis, grammatical analysis, speech processing, text processing and so on [Ide and V´eronis, 1998]. For CLIR, it is often necessary to disambiguate translations enumerated under each headword in a bilingual MRD or a term list generated from a parallel corpus. If all translations listed in bilingual resources are straightforwardly adopted as search terms, extraneous or superfluous terms irrelevant to the original query usually diminish the effectiveness of the search. Thus it is desirable that only relevant terms will be automatically

524

Kazuaki Kishida

or semi-automatically selected from a set of translation candidates. A simple method is to take only translations corresponding to the first sense listed in the dictionary. Alternatively, we could investigate the frequency of each translation within a corpus and use only the most frequent translation. However, such heuristic strategies would be insufficient to resolve the ambiguity of words that are highly homonymous. Several more sophisticated methods have been explored in the field of CLIR as follows. 1. Use of part-of-speech (POS) tags 2. Use of parallel corpora 3. Use of co-occurrence statistics in the target corpus 4. Use of query expansion technique It should be noted that the term ‘corpus-based disambiguation’ is often used in literature for collectively referring to techniques 2 through 4. As another solution, the ‘structured query model’ has also been investigated for improving search performance in cases where multiple translations are obtained from a bilingual dictionary. This model will be discussed in another section.

4.2.

Use of Part-of-Speech Tags

The basic idea of using part-of-speech (POS) tags for translation disambiguation is to select only translations having the same POS tag as that of the source query term [Davis, 1997, Davis and Ogden, 1997, Ballesteros and Croft, 1998]. For example, in the case of applying it to the BLIR task from English to Spanish, the translation is selected as a search term only if the POS tag of a Spanish equivalent listed in an English-Spanish dictionary coincides with that of the English query term. This technique requires POS tagging tools for both languages.

4.3.

Parallel Corpus-Based Methods

It is possible to determine the best translations among the candidates according to the result of searching a parallel corpus for the original query [Davis, 1997, Davis, 1998, Boughanem et al., 2002]; specifically, a typical procedure is as follows (see also Figure 6). 1. Identify a set of translations for each term in a given source query by using an MRD. 2. Search a part of the parallel corpus written in the target language for each translation respectively, and save each set of documents in the target language. 3. Search the other part of the parallel corpus written in the source language for the source query. 4. Select a translation for which the set of documents is the closest to the set of documents searched by the source query (e.g., choose a translation for which documents are mostly included in the set of search results by the source term). This procedure is repeated for each query term until a final set of the best translations is obtained.

Methods for Cross-Language Information Retrieval Search results

A parallel corpus

comparison

525

A source query term

translation

: source language

: target language

Figure 6. Outline of parallel corpus-based disambiguation.

4.4.

Disambiguation Based on Term Co-occurrence Statistics

A basic assumption underlying disambiguation techniques based on term co-occurrence is that “the correct translations of query terms should co-occur in target language documents and incorrect translations should tend not to co-occur” [Ballesteros and Croft, 1998]. For example, in the case of a French to English search, if the source query includes a concept of “database management system”, the combination of translations, “database”, “management” and “system” is reasonably expected to appear more frequently in the target document collection than that of “database”, “administration” and “system”. Thus, if we appropriately employ statistics on co-occurrence frequencies, “management” would be able to be identified as a correct translation in the context 4 . This kind of methods based on term co-occurrences does not need any extra corpus because the statistics can be obtained from the target documents. This is a remarkable merit in practice. 4.4.1.

Best Pairs Selection

When a given query includes many terms, the frequency of their combination may become very low or zero. Thus, the query is usually decomposed into a set of term pairs, and co-occurrence frequencies of the corresponding two translations are used for determining final query terms {t˜1 , ..., ˜tm} in the target language. For example, we can select them such that [ t˜j = arg max sim(t, t0 ), t0 ∈ Tk ; j = 1, ..., m, (10) t∈Tj

k6=j

where Tj indicates a set of translations in the target language for j-th source term sj (j = 1, ..., m) and sim(t, t0 ) is a term association or similarity between the t and t0 computed from term co-occurrences. Equation (10) simply means that we choose repeatedly a pair of translations with the highest similarity from those excluding pairs that have already been selected [Bian and Chen, 1998], i.e., if a pair with the highest similarity includes a translation of the source term for which a translation was previously determined, this translation is ignored. As a variation of this method, we may be able to consider only similarities with the translations selected in the previous step [Jang et al., 1999]. 4

Moreover, it is possible to incorporate term co-occurrence statistics into the process of estimating translation probabilities (see [Liu et al., 2005] for details).

526

Kazuaki Kishida

The term similarity sim(t, t0 ) can be measured as MI, the Dice coefficient and so on, which are computed from term occurrence and co-occurrence statistics in the target document collection. For example, if we employ MI in Equation (4), its probabilities are operationally defined such that P (t, s) = nts /N , P (t) = nt /N and P (s) = ns /N where nt and ns are the numbers of sentences including term t and s, respectively, nts is the frequency of appearance of both terms in the same sentence, and N is the total number of sentences included in the target collection. While the number of observations is normalized by the size of the corpus according to [Church and Hanks, 1990] in these formulae, it is possible to adopt other definitions (e.g., see [Gao et al., 2001]). The term co-occurrence statistics required for computing the similarity can be compiled in a process of constructing index files for searching the target collection, or obtained from a search engine on the Internet via API [Maeda et al., 2000]. Fortunately, the computational complexity of selecting the best pairs based on Equation (4) is not so high. If the similarity measure used in this process is symmetric such as MI, the number of pairs to be measured for the selection, Mp , amounts to Mp =

m−1 X

m X

|Tk | × |Th |,

(11)

k=1 h=k+1

(see [Kishida, 2007]). It should be noted that Mp does not exceed b2 m(m − 1)/2 where m(m − 1)/2 corresponds to the number of all combinations of two source terms and b indicates the maximum number of translations for a source term, i.e., b = maxj=1,...,m |Tj |. For example, if there are five source query terms and every set of translations includes five words respectively, we obtain Mp = (5 × 5) × (5 × 4/2) = 250. 4.4.2.

Best Sequence Selection

However, this algorithm may yield erroneous translations because it checks only a local relationship between just two translations at each step for selecting a pair. Suppose that, in the above example, the pair of “administration” and “system” has the highest degree of similarity in the target document collection even though the collection contains some documents on “database management system”. In this case, the correct translation, “management”, is never obtained by the algorithm based on Equation (10). Since this algorithm looks at only a very limited range (i.e., a span of just two terms) in the query, a few pairs having strong relationships out of the context of the query tend to impact excessively on the result. A straightforward solution to this problem is to focus on all relationships between the query terms [Seo et al., 2005]. We denote a sequence of translations by τ = {t1 , ..., tj , ..., tm} where tj ∈ Tj , i.e., τ is constituted by taking arbitrarily a term from each translation set, respectively (see Figure 7). If the sum of degrees of similarity between all translations included in such a sequence, i.e., U (τ ) =

m−1 X

m X

k=1 h=k+1

sim(tk , th ),

tk , th ∈ τ,

(12)

Methods for Cross-Language Information Retrieval

527

can be computed, the sequence with the highest value of the sum should be selected as the final set of translations (similar strategies were adopted by [Maeda et al., 2000, Qu et al., 2003b]). Since Equation (12) contains all relationships between possible pairs of translations in each sequence, we can avoid errors caused by the locality of range in Equation (10). For example, the possibility that “management” is correctly selected remains even if the pair of “administration” and “system” has the highest similarity. Source terms

Sequence Translations Pair

Figure 7. Sequences and pairs of translations. However, the computational complexity of processing the algorithm based on Equation (12) is very high. The total number of sequences to be processed is given by Mτ = |T1 | × ... × |Tm| =

m Y

|Tj |,

(13)

j=1

and the upper limit of Mτ amounts to bm . In addition, each sequence includes m(m − 1)/2 pairs as expressed in Equation (12), and therefore, in total, we need to deal with bmm(m − 1)/2 pairs at most in the process of disambiguation (note that a single pair is repeatedly counted). If we have five source query terms and every set of this translation includes five words respectively, the number of pairs amounts to 31,250 (= 55 × 5 × 4/2). As this example shows, it takes much longer to select translations when using Equation (12). 4.4.3.

Approximation for Best Sequence Selection

One way to reduce the computational difficulty of the best sequence method is to apply the maximal value of similarity between a given translation and those of another source query term, i.e., C(t, Tk ) = max sim(t, t0 ), t 6∈ Tk , (14) 0 t ∈Tk

which was employed in some experiments (e.g., [Adriani, 2000, Gao et al., 2001]). C(t, Tk ) is often called ‘cohesion’. In order to solve ambiguity of translations using this quantity, for each term t, it is necessary to compute the sum of C(t, Tk ) over all sets Tk except the set that includes the term t itself, and then to select the term with the highest score from each set of translations respectively [Adriani, 2000, Gao et al., 2001], i.e., X ˜tj = arg max C(t, Tk ), j = 1, ..., m. (15) t∈Tj

k6=j

528

Kazuaki Kishida

The number of translation pairs to be processed in this algorithm for determining the P translation of the j-th source term is |Tj | × k6=j |Tk |, and therefore, the total number amounts to   m X X |Tj | × Mp = |Tk | , (16) j=1

k6=j

which does not exceed m×b×(m−1)b = b2 m(m−1). If we have five source query terms and every set of translations includes five words respectively, we obtain Mp = 5×5×4×5 = 500. Although its degree of computational complexity is almost the same as that of the method in Equation (10), the algorithm using C(t, Tk ) avoids making local judgments in the selection of translations to some degree unlike Equation (10). That is, a translation of a given source term sj is chosen using information on relationships with all other source terms through the C(t, Tk ) in Equation (14), not with only a single source term. In our example, even if the similarity score between “administration” and “system” is the highest one, it is possible that “management” is correctly selected as a translation, i.e., the final sum of C(t, Tk ) for “management” may become greater than that for “administration” due to the contribution of a high degree of similarity between “management” and “database”. One experiment [Kishida, 2007] has reported that the search performance of the disambiguation technique based on Equation (15) is compatible with that of the best sequence method using Equation (12).

4.5.

PRF-based Techniques for Disambiguation

We can apply PRF techniques to translation disambiguation [Ballesteros and Croft, 1997, Ballesteros and Croft, 1998]. In CLIR situations, two kinds of PRF are feasible: • Pre-translation feedback • Post-translation feedback Suppose that we have a corpus in the source language which is independent of the target document collection. First, this corpus is searched for a given source query, and resulting documents are analyzed prior to translation for CLIR in order to add a set of new terms to the source query (pre-translation feedback). The new terms can be selected based on term weights, e.g., Equation (3). Second, after translation, standard PRF can be applied using the target document collection (post-translation feedback). Inevitably, for executing the pre-translation feedback, an extra corpus in the source language is needed unlike the post-translation feedback working on the target collection. The pre-translation feedback may improve the precision (see [Ballesteros and Croft, 1997]), because PRF is basically executed using the entire query, not each source term respectively. That is, synonyms or related terms corresponding to the correct meaning of each source term within the context of the query are expected to be automatically added through the PRF process. An experiment [McNamee and Mayfield, 2002a] has suggested that the pre-translation query expansion is useful when lexical coverage of

Methods for Cross-Language Information Retrieval

529

translation resources is poor. On the other hand, the post-translation feedback is just a standard PRF, and therefore, the recall ratio would increase by applying it as many IR experiments have shown. It is also possible to determine explicitly a final translation t˜j for each source term (j = 1, ..., m) based on frequencies of term occurrence in output from the initial search of post-translation feedback [Kishida and Kando, 2004, Kishida, 2007]. In the first stage of this process, the target document collection is searched for a set of all translation candidates, and the number of documents including each translation candidate in the list of top-ranked documents by this search is calculated. Finally, ˜tj = arg max rt ,

j = 1, ..., m

(17)

t∈Tj

is selected as a final translation for the j-th source term where rt indicates the number of documents including a translation candidate t within the top-ranked documents. As Yamabana et al. pointed out, unexpected false combinations of translations may be generated by disambiguation techniques based on term co-occurrences discussed in the previous section because it is possible that two translations having no relation within the context of a given source query tend to co-occur frequently in the whole document collection [Yamabana et al., 1998]. That is, suggestions from macro-statistics compiled using the whole collection are not always valid in the sense implied by a particular query. A similar problem occurs when applying query expansion techniques to general IR situations. It is widely known that query expansion techniques using statistical thesauri generated through term co-occurrence statistics cannot achieve better performance than PRF in which new search terms are locally identified from a restricted set of top-ranked documents searched for the given query. Similarly, in the context of translation disambiguation for CLIR, local analysis of the target document collection such as Equation (17) may yield better results than global analysis. In addition, it should be noted that the technique based on Equation (17) is easier to implement in IR systems having a standard PRF function. This is a practical advantage of this technique.

4.6.

Disambiguation for Phrasal Translation

As Ballesteros and Croft pointed out that “...failure to translate multi term concepts as phrases reduces effectiveness” [Ballesteros and Croft, 1997], phrasal translation is certainly significant for CLIR. The basic technique is to search a bilingual dictionary or a term list including phrases or compound words as its headwords. That is, they can be automatically identified in the source query by simple matching operations against headwords of such a language resource. Also, if a part-of-speech tagger is available in this process, a word combination of ‘noun-noun’ or ‘adjective-noun’ would be reasonably assumed to be a compound word. Inevitably, the coverage of lexical resources to be used will not always be sufficient, i.e., all phrases or compound words are not always found in the resources as headwords. When an untranslatable phrase is included in a given source query, there is no way other than to attempt word-by-word translation, which may cause a term ambiguity problem

530

Kazuaki Kishida

[Ballesteros and Croft, 1997, Ballesteros and Croft, 1998]. Therefore, one of the disambiguation techniques discussed above is required in this word-by-word translation process. However, it should be noted that there are two kinds of compound words as follows [Pirkola et al., 2001]. • Compositional compounds: The meaning can be derived from meanings of the component words, e.g., “database management system”. • Non-compositional compounds: The meaning cannot be derived from meanings of the component words, e.g., “hot dog”. In many cases, disambiguation methods based on term co-occurrences may have limitations for detecting correctly translations of non-compositional compounds. It is thus indispensable to augment the coverage of bilingual dictionaries in order to enhance the quality of phrasal translations. This could be done by some techniques for detecting OOV from the web (see the above section), for extracting phrasal representations from parallel or comparable corpora (see [Lopez-Ostenero et al., 2005]), or for disambiguating directly noun phrases not at the level of distinct words (see [Gao and Nie, 2006]).

4.7.

Other Disambiguation Techniques

In a technique called bi-directional translation, backward translations in which translation results are automatically re-translated into the original language are used for ranking translation candidates [Boughanem et al., 2002]. For example, when French terms are translated into English ones, first a set of English equivalents for each French term is extracted from a French-English bilingual dictionary. Next, using an English-French dictionary, each English equivalent is reversely translated into a set of French terms. Basically, if the set includes the original source term, the English translation equivalent is chosen as a preferred translation. When a pivot language approach to translation is used, we can apply lexical triangulation to translation disambiguation [Gollins and Sanderson, 2001, Lehtokangas et al., 2004]. In this technique, two pivot languages are used independently, and an attempt is made to remove erroneous translations by taking only translations in common obtained from both ways of transitive translation using the two pivot languages respectively. Also, a disambiguation technique via dynamic clustering of search results has been explored by [Lee et al., 2004]. In the first stage of this method, all translation candidates are used as the search query without any disambiguation like the PRF-based method, and in the next stage, top-ranked documents searched for the initial query are divided into some clusters by a clustering algorithm. Finally, the degree of similarity between the query and a cluster to which each document belongs is reflected in its document score for ranking, and documents are re-ranked according to the new scores. If each cluster correctly corresponds to one of the multiple senses and appropriate clusters for the given query can be identified, search performance may be enhanced by the re-ranking.

Methods for Cross-Language Information Retrieval

5.

531

Formal Models for CLIR

If representations in two different languages of a given query and documents become comparable after the translation process, the BLIR problem reduces to that of standard monolingual IR. That is, after the translation is completed, the remaining task is how to compute document scores for ranked output, in which a retrieval model such as Equation (1) is applied in a standard manner of monolingual IR. On the other hand, some researchers have tried to incorporate directly the translation process into a retrieval model.

5.1.

Language Modeling for CLIR

It is relatively easy to incorporate translation probabilities P (t|s) into Equation (1), which is a typical formula derived from language modeling for IR. Figure 8 shows the basic idea of the incorporation, i.e., we assume that P (t|di ) =

X

P (t|sk )P (sk |di),

(18)

k

where sk indicates a term included in document di . It should be noted that t is a query term in a different language from that of documents in this case. Equation (18) means that P (t|di ), which is the probability that query term t is generated from document di , is decomposed into term occurrence probabilities P (sk |di) and translation probabilities P (t|sk ), and that its value is computed by summing the products of them for all sk . By directly substituting Equation (18) into Equation (1), we finally obtain: P (Ωq |di) =

Y

α

t∈Ωq

X

P (t|sk )P (sk |di) + (1 − α)P (t),

(19)

k

according to [Miller et al., 1999, Xu et al., 2001, Xu and Weischedel, 2005]. translation prob. P(t|s)

P(t|d)

t is a query term.

term occurrence prob. P(s|d)

t

information

information

t

report

report

t

intelligent

intelligent

……………... information.. …report….... …………….. …intelligent.. …………..

Document

Figure 8. Language modeling with translation probabilities. In Equation (1), for preventing P (Ωq |di) from automatically becoming zero when P (t|di ) = 0 for term t, a general probability P (t) is added to P (t|di ) with a mixing parameter α. This modification is generally called Jelinek-Mercer smoothing, in which the term αP (t|di ) + (1 − α)P (t) is considered to be the ‘true probability’ that t appears in di .

532

Kazuaki Kishida

If translation probabilities are applied to the true probability that term sk appears in di , we obtain another formula: YX P (Ωq |di) = P (t|sk )[αP (sk |di ) + (1 − α)P (sk )], (20) t∈Ωq

k

for language modeling in BLIR [Hiemstra and Kraaij, 1999]. While P (t) in Equation (19) needs to be estimated using an extra corpus in the query language (e.g., Pˆ (t) = nt /N ), an estimation of P (sk ) in Equation (20) can be obtained from the target document collection. As discussed above, translation probability P (t|sk ) can be estimated by the IBM model if an appropriate parallel corpus is available. Fortunately, the GIZA++ toolkit including a module for the IBM model has been developed, and we can use it without writing source code. If no parallel corpus is available, other methods have to be employed. A simple one is to count the number of translations for each source term in a bilingual dictionary. For example, if a source term s has ms translations t1 , ..., tms , the translation probabilities can be assumed such that P (tj |s) = 1/ms (j = 1, ..., ms) uniformly [Xu et al., 2001]. As a more sophisticated method, an EM algorithm computing iteratively the translation probabilities based on term co-occurrence statistics has been proposed by [Monz and Dorr, 2005] (the initial values of translation probabilities are estimated as the above P (tj |s) = 1/ms from a bilingual dictionary). Since term co-occurrence statistics can be compiled from the target document collection, this algorithm allows us to estimate translation probabilities without any parallel corpus.

5.2.

Relevance Model

The relevance model (RM) developed by [Lavrenko and Croft, 2001] can be theoretically applied to CLIR situations [Lavrenko et al., 2002, Larkey and Connell, 2005] since essentially RM is a variation of LM. In RM, the document score is often calculated as relative entropy E(R k di) between P (t|di ) in Equation (1) and P (t|R), which is the average probability that t is included in a relevant document, i.e., X P (t|R) E(R k di ) = P (t|R) log (21) P (t|di ) t∈ΩD

where ΩD indicates a set of all terms appearing in the database D. If a set of relevant documents is unknown, P (t|R) can be approximately estimated by P (t|Ωq ) = P (t, Ωq )/P (Ωq ), and we can write: X Y P (t, Ωq ) = P (di )P (t|di ) P (t0 |di), (22) i:di ∈D

t0∈Ωq

under a assumption [Lavrenko and Croft, 2001] 5 . Also, X P (Ωq ) = P (t, Ωq ).

(23)

t∈ΩD

5

P First, P (t, Ωq ) = di ∈D P (di )P (t, Ωq |di ). If we assume that all terms are sampled independently and Q identically from di , P (t, Ωq |di) = P (t|di) t0 ∈Ωq P (t0 |di ). By substituting it into the first equation, Equation (22) is obtained.

Methods for Cross-Language Information Retrieval

533

In order to apply RM to CLIR, it is necessary to distinguish terms in documents from those in queries by denoting a document term by sk in Equations (21) and (22), i.e., sk ∈ ΩD . For example, Equation (22) is re-written as X Y P (sk , Ωq ) = P (di )P (sk |di ) P (t|di ), (24) i:di ∈D

t∈Ωq

Q

in which P (t|di ) can be estimated using Equation (19) and usually P (di ) = 1/N . While only query terms appearing in the document contribute to computation of the document score in standard IR models including LM, RM takes all terms in the database D into consideration as suggested by Equation (21). This means that much more information on term occurrences is used in RM than other IR models, which may enhance search performance. However, more computational time is required for calculating document scores based on RM.

5.3.

Hidden Markov Model for CLIR

Suppose that s is a sequence of terms included in a source query and t is its translation, and that the number of source terms is equal to that of translations (e.g., m = l in Equation (7)). From the definition of conditional probability, we can write Pr(t, s) such that Pr(t, s) = Pr(t1 , ..., tm, s1 , ..., sm) = Pr(s1 , ..., sm|t1 , ..., tm)Pr(t1 , ..., tm),

(25)

where Pr(t1 , ..., tm) = Pr(tm |tm−1 , ..., t1) × ... × Pr(t3 |t2 , t1 ) × Pr(t2 |t1 ) × Pr(t1 ), and if the translation process is interpreted as a Markov model, Pr(t1 , ..., tm) = P (tm |tm−1 )...P (t3 |t2 )P (t2 |t1 )P (t1 ).

(26)

Similarly, Pr(s1 , ..., sm|t1 , ..., tm) = P (sm |tm ) × ... × P (s2|t2 ) × P (s1 |t1 ), and therefore, Pr(t, s) = P (t1 )P (s1 |t1 )

m Y

P (sj |tj )P (tj |tj−1 ).

(27)

j=2

This is a query translation model based on HMM (Hidden Markov Model) proposed by [Federico and Bertoldi, 2002, Bertoldi and Federico, 2004], in which ‘reverse’ translation probability P (sj |tj ) and transitive probability P (tj |tj−1 ) of translations (target terms) are included. The transitive probability in this model may work as a device for translation disambiguation. For example, P (“management”|“database”) × P (s2 |“management”) is expected to be greater than P (“administration”|“database”) × P (s2 |“administration”) where s2 is the second term of a query representing “database management system” in a language other than English.

5.4.

Structured Query Model

The INQUERY system developed by a research group at Massachusetts University [Callan et al., 1995a, Turtle and Croft, 1991] provides an important function for CLIR. In

534

Kazuaki Kishida

principle, INQUERY is a retrieval model based on a probabilistic Bayesian network, in which a ‘belief score’ B(q|di ) measuring the degree to which a document di is relevant to a given query q is estimated. Although some variations can be derived dependently on assumptions to be selected, B(q|di) is basically computed from B(t|di ) of each search term included in q, i.e.,   xi (t) log[(N + 0.5)/nt] B(t|di ) = 0.4 + 0.6 , (28) × log(N + 1) xi (t) + 0.5 + 1.5li/¯l where li is the length of di , ¯l is the average of li in the database and xi (t) is the frequency of t in the document di . For example, Pmwhen ‘#sum( )’ operator is used such as #sum(t1 , t2 , ..., tm), B(q|di ) is calculated as j=1 B(tj |di). INQUERY has several operators like #sum( ), which clearly distinguish this system from other retrieval models. Among the operators, the #syn( ) operator for dealing with synonyms is often used in CLIR tasks. Suppose that there are three source terms s1 , s2 , s3 , and translations of these terms are obtained from a bilingual dictionary such that T1 = {t11 , t12, t13 }, T2 = {t21 , t22 } and T3 = {t31 , t32 }. We can enter a structured query, #sum ( #syn (t11 t12 t13 ) #syn (t21 t22 ) #syn (t31 t32 )), into the INQUERY system without any translation disambiguation [Pirkola, 1998]. This is often called Pirkola’s method, in which the belief score of a set of translations Tk for the source term sk is computed based on the #syn operator such that   x0i (Tk ) log[(N + 0.5)/n0(Tk )] B(Tk |di) = 0.4 + 0.6 × , (29) log(N + 1) x0i (Tk ) + 0.5 + 1.5li/¯l P 0 where x0i (Tk ) = t∈Tk xi (t) and n (Tk ) indicates the number of documents including at least one translation in Tk [Kek¨al¨ainen and J¨arveline, 1998, Sperer and Oard, 2000, Lehtokangas et al., 2004, Hedlund et al., 2004]. Pirkola’s method has been slightly modified by some studies [Darwish and Oard, 2003, Wang and Oard, 2006]. For example, in an experiment [Darwish and Oard, 2003], translation probabilities are incorporated into P the computation of x0i (Tk ) and n0 (Tk ) such as x0i (Tk ) = t∈Tk P (t|sk ) × xi (t) and P n0 (Tk ) = t∈Tk P (t|sk ) × nt in order to increase infulence of probable target terms with higher translation probabilities.

6.

Method for Multilingual Information Retrieval

6.1.

Approaches to MLIR

Suppose that we have a multilingual document collection in which two or more languages are mixed (not a parallel corpus), and that a user wishes to search the collection for a query expressed in a single language (see Figure 1). This MLIR task is more complicated than simple BLIR. Basically, there are two strategies for MLIR as follows [Lin and Chen, 2003]. • Distributed architecture in which the document collection is separated by language, and each part is indexed and retrieved independently.

Methods for Cross-Language Information Retrieval

535

• Centralized architecture in which the document collection in various languages is viewed as a single document collection and is indexed in one huge index file.

indexing

search query

multilingual collection

index

result

translation

merge final list

Figure 9. Distributed architecture for MLIR.

6.2.

Merging Technique

In the distributed architecture, a standard bilingual search is repeatedly performed for each separate language sub-collection respectively, and several ranked document lists are generated by each run (see Figure 9). Then the problem becomes how to merge the results of each run into a single ranked list so that all relevant documents in any language are successfully located on upper ranks. Essentially, the merging strategy is a general research topic of IR when searching distributed resources (i.e., distributed IR), in which it is necessary to merge ranked lists obtained from each resource. In CLIR, the following merging strategies have been investigated 6 . • Raw score: straightforwardly using document scores estimated in each run • Round robin: interleaving each document list in a round robin fashion by assuming that the distribution of relevant documents is identical among the lists • Normalized score: normalizing document scores of each run in order to remove the effects of collection-dependent statistics used for estimating the scores • Rank-based score: mathematically converting ranks in each run into scores by assuming a relationship between probabilities of relevance and the ranks • Modified score: modifying raw scores in each run so as to reduce the effects of collection-size dependency, translation ambiguity, and so on If the retrieval model employed for each run can estimate the relevance probability of each document correctly, it would be reasonable to re-rank all documents together according 6

Experimental comparison of the search performance of these methods has been attempted by [Savoy, 2004].

536

Kazuaki Kishida

to values of the probability (i.e., raw score strategy). For example, we can use the logistic regression model for IR as such a retrieval model (see [Chen and Gey, 2003] for details). However, in most cases, it would be too difficult to interpret each document score as a pure probability of relevance even though a probabilistic retrieval model was actually used because the probability is often approximately estimated for convenience of calculation in the model. In such cases, if it can be assumed that relevant documents are distributed in the same way within every separate language sub-collection, we can employ round robin-based merging in which only the rank of each document is taken into account. Otherwise, an alternative method is to use normalized document scores such that v˜i = (vi − vmin )/(vmax − vmin ),

(30)

where vi is the raw score of document di, and vmin and vmax are the minimum and maximum in each search run respectively [Powell et al., 2000]. One widely-known method for normalization in distributed IR is the CORI (collection retrieval inference) algorithm [Callan et al., 1995b]. In this algorithm, first, the score of a sub-collection for the given query is computed based on Equation (28) by considering each sub-collection Dk (k = 1, ..., L) as a single huge document where L is the number of sub-collections included in the multilingual collection (i.e., the number of languages). We denote the score by B(q|Dk ). Second, the score vi (= B(q|di )) of a document di in a sub-collection Dk (i.e., di ∈ Dk ) is converted such that  ¯  B(q|Dk ) − B(q) v˜i = vi × 1 + L × , (31) ¯ B(q) P ¯ where B(q) = L−1 L k=1 B(q|Dk ). For more details on applying the CORI algorithm to MLIR, see [Savoy, 2002] or [Moulinier and Molina-Salgado, 2003]. Another possibility is to predict the relevance probability of a document ranked in a position using training data sets [Franz et al., 1999, Kraaij et al., 2000, Savoy, 2003]. For example, we may calculate this probability such that Pˆ (R|di) = a + b log(ρi )

(32)

where P (R|di) indicates the relevance probability of document di , ρi is its rank in the output list, and a and b are parameters to be estimated from training data (a more complicated regression model was used in [Savoy, 2003]). Meanwhile, researchers have explored other techniques of modifying raw scores so as to remove the effect of collection-size dependency (e.g., [Hiemstra et al., 2001]), or of reducing them based on the degree of translation ambiguity according to the assumption that a good translation may give much more relevant documents (e.g., [Lin and Chen, 2003]).

6.3.

Searching Heterogeneous Collections

On the other hand, in the centralized architecture, the set of multilingual documents is not divided into sub-collections for each language. In order to search such a heterogeneous collection we need either

Methods for Cross-Language Information Retrieval

537

1. to translate a given source query into all languages included in the document collection and to merge all translations into a single query, or 2. to translate all documents in a single language used in the query. In general, the first method has been adopted (e.g., [Gey et al., 1999, Nie and Jin, 2003]). Since documents in a language having fewer documents may take advantage of weighting by document frequency, it may be necessary in this method to adjust the idf factor [Lin and Chen, 2003]. For example, in Equation (28), {log[(N + 0.5)/nt]}/ log(N + 1) is an idf factor, and its value becomes larger as nt is smaller. Meanwhile, in the second method, it is not necessary to add such adjustment of parameters because a single index registering query language words is created although the document translation is a very time-consuming task. This method was attempted by [Chen and Gey, 2004], and the search performance of various MLIR techniques discussed in this section was experimentally compared.

7.

Concluding Remarks

As is widely recognized, research on developing CLIR techniques can be traced back to G. Salton’s paper in 1970 [Salton, 1970]. Since then, many research results have been published in various journals and conferences in different fields such as IR and NLP, especially since the mid-1990s. This review has not covered all the techniques reported in such publications, and excludes studies focusing on only a particular language (e.g., Chinese, Japanese, etc.). Other review articles [Oard and Hackett, 1998, Peters and Sheridan, 2001, Kishida, 2005] provide further useful information on CLIR research efforts. The rapid development of CLIR techniques since the mid-1990s is largely due to CLIR experiment workshops such as TREC [Harman, 2005], CLEF [Braschler and Peters, 2004], and NTCIR [Kishida et al., 2007], in which research programs on CLIR have been incorporated and many researchers worldwide have participated. Furthermore, test collections for CLIR constructed in these projects allow a wider range of researchers including those not participating in the projects to test new ideas on CLIR techniques. New applications on the Internet have been continually appearing, and accordingly, IR techniques tailored to them have been required (e.g., blog searching). CLIR techniques will need to be developed in parallel in order to satisfy users’ emerging and diverse needs.

Appendix: IBM Model 1 Let a = am 1 ≡ a1 a2 . . . am be an alignment, where aj indicates mapping of a target word to one or more source words. For example, aj = k means that the word in position j of the target string is connected to the word in position k of the source string. The IBM Model [Brown et al., 1993] is Pr(t|s) =

X a

Pr(t, a|s),

(33)

538

Kazuaki Kishida

where Pr(t, a|s) = Pr(m|s)

m Y

j−1 j j−1 Pr(aj |aj−1 1 , t1 , m, s)Pr(tj |a1 , t1 , m, s).

(34)

j=1

In the simplest Model 1, we put the following assumptions: • Pr(m|s) is independent of s j−1 −1 • Pr(aj |aj−1 1 , t1 , m, s) depends only on l, and must be (l + 1)

• Pr(tj |aj1 , tj−1 depends only 1 , m, s) j j−1 Pr(tj |a1 , t1 , m, s) = P (tj |saj )

on

tj

saj ,

and

and

therefor,

Hence, Equation (34) becomes Pr(t, a|s) =

m Y ε P (tj |saj ), (l + 1)m

(35)

j=1

where ε ≡ Pr(m|s). By substituting it into Equation (33), we have l l m X X Y ε Pr(t|s) = ··· P (tj |saj ), (l + 1)m a1 =0

(36)

am =0 j=1

where aj = 0 means that there is no corresponding term in the source string. Since these sums are interchangeable with the product in Equation (36), Equation (7) can be obtained. The probability P (t|s) can P be maximized by introducing the Lagrange multipliers λs . That is, under the condition t P (t|s) = 1, we set m X l Y X X ε H(P, λ) = P (t |s ) − λ ( P (t|s) − 1), j k s (l + 1)m s t

(37)

j=1 k=0

and ∂H(P, λ) = ∂P (t|s) =

Qm Pl m X l X P (th |su ) ε δ(t, tj )δ(s, sk ) h=1 − λs Pl u=0 m (l + 1) u=0 P (t|su ) j=1 k=0

ε (l + 1)m

Q m Pl h=1

m

u=0

Pl

u=0

P (th |eu ) X

P (t|su )

δ(t, tj )

j=1

l X

δ(s, sk ) − λs,

k=0

where δ(·, ·) is the Kronecker’s delta. From the equation ∂H(P, λ)/∂P (t|s) = 0, it follows that P (t|s) = λ−1 s

m X l m l Y X X ε P (t|s) δ(s, si) P (t |s ) δ(t, t ) P h u j l (l + 1)m u=0 P (t|su ) h=1 u=0

= λ−1 s Pr(t|s) Pl

P (t|s)

u=0

P (t|su )

j=1

m X j=1

δ(t, tj )

l X k=0

δ(s, sk ).

i=0

(38)

Methods for Cross-Language Information Retrieval

539

If we replace λs by λs Pr(t|s), and introduce the expected number of times that s connects to t in the translation (t|s) as m

l

j=1

k=0

X X P (t|s) c(t|s; t, s) = δ(t, tj ) δ(s, sk ), P (t|s0 ) + · · · + P (t|sl )

(39)

then the Equation (38) can be written compactly as P (t|s) = λ−1 s c(t|s; t, s)

(40)

In practice, we have N translations, (t(1)|s(1)), ..., (t(N )|s(N )) as training data, so this equation becomes N X P (t|s) = λ−1 c(t|s; t(i), s(i)). (41) s i=1

We can estimate P (t|s) iteratively using Equation (39) and (41) as follows. 1. Choose initial values for P (t|s). 2. Compute Equation (39) for each sentence (1 ≤ i ≤ N ). 3. For each s, estimate λs as λs =

N XX t

c(t|s; t(i), s(i))

(42)

i=1

4. For each t, use Equation (41) to obtain a new value for P (t|s) 5. Repeat steps 2 to 4 until the values of P (t|s) have converged to the desired degree.

References [Adriani, 2000] Adriani, M. (2000). Using statistical term similarity for sense disambiguation in cross-language information retrieval. Information Retrieval, 2(1):71–82. [Adriani, 2002] Adriani, M. (2002). English-Dutch CLIR using query translation techniques. In et al., C. P., editor, Evaluation of Cross-Language Information Retrieval Systems, pages 219–225. Springer-Verlag. (LNCS 2406). [Ballesteros and Croft, 1997] Ballesteros, L. and Croft, W. B. (1997). Phrasal translation and query expansion techniques for cross-language information retrieval. In Proceedings of the 20th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 84–91. [Ballesteros and Croft, 1998] Ballesteros, L. and Croft, W. B. (1998). Resolving ambiguity for cross-language retrieval. In Proceedings of the 21st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 64–71.

540

Kazuaki Kishida

[Ballesteros, 2000] Ballesteros, L. A. (2000). Cross-language retrieval via transitive translation. In W. B. Croft et al., editor, Advances in Information Retrieval: Recent Research from the Center for Intelligent Information Retrieval, pages 203–234. Kluwer. [Berry and Young, 1995] Berry, M. W. and Young, P. G. (1995). Using latent semantic indexing for multilanguage information retrieval. Computers and the Humanities, 29(6):413–429. [Bertoldi and Federico, 2004] Bertoldi, N. and Federico, M. (2004). Statistical models for monolingual and bilingual information retrieval. Information Retrieval, 7(1-2):53–72. [Bian and Chen, 1998] Bian, G. W. and Chen, H. H. (1998). Integrating query translation and document translation in a cross-language information retrieval system. In D. Farwell et al., editor, Machine Translation and the Information Soup, pages 250–265. Springer. (LNCS 1529). [Bian and Lin, 2001] Bian, G. W. and Lin, C. C. (2001). Trans-EZ at NTCIR-2 : synset cooccurrence method for English-Chinese cross-lingual information retrieval. In Proceedings of the Second NTCIR Workshop on Research in Chinese & Japanese Text Retrieval and Text Summarization. Available from: http://research.nii.ac.jp/ntcir/workshop/. [Boughanem et al., 2002] Boughanem, M., Chrisment, C., and Nassr, N. (2002). Investigation on disambiguation in CLIR: aligned corpus and bi-directional translation-based strategies. In et al., C. P., editor, Evaluation of Cross-Language Information Retrieval Systems, pages 158–168. Springer-Verlag. (LNCS 2406). [Braschler, 2004] Braschler, M. (2004). Combination approaches for multilingual text retrieval. Information Retrieval, 7(1-2):183–204. [Braschler and Peters, 2004] Braschler, M. and Peters, C. (2004). Cross-Language Evaluation Forum: objectives, results, achievements. Information Retrieval, 7(1-2):7–31. [Braschler and Sch¨auble, 2000] Braschler, M. and Sch¨auble, P. (2000). Using corpusbased approaches in a system for multilingual information retrieval. Information Retrieval, 3(3):273–284. [Braschler and Sch¨auble, 2001] Braschler, M. and Sch¨auble, P. (2001). Experiments with the Eurospider retrieval system for CLEF 2000. In Peters, C., editor, Cross-Language Information Retrieval Evaluation, pages 140–148. Springer-Verlag. (LNCS 2069). [Brown et al., 1993] Brown, P. F., Della Pietra, S. A., Della Pietra, V. J., and Mercer, R. L. (1993). The mathematics of statistical machine translation: parameter estimation. Computational Linguistics, 19(2):263–311. [Buckley et al., 1998] Buckley, C., Walz, J., Mitra, M., and Cardie, C. (1998). Using clustering and superconcepts within SMART: TREC-6. In Proceedings of TREC-6. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html.

Methods for Cross-Language Information Retrieval

541

[Callan et al., 1995a] Callan, J. P., Croft, W. B., and Broglio, J. (1995a). TREC and TIPSTER experiments with INQUERY. Information Processing & Management, 31(3):327– 343. [Callan et al., 1995b] Callan, J. P., Lu, Z., and Croft, W. B. (1995b). Searching distributed collections with inference networks. In Proceedings of the 18th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 21–28. [Cao and Li, 2002] Cao, Y. and Li, H. (2002). Base noun phrase translation using web data and the EM algorithm. In Proceedings of the 19th International Conference on Computational Linguistics, volume 1, pages 1–7. [Chen and Gey, 2003] Chen, A. and Gey, F. C. (2003). Experiments on cross-language and patent retrieval at NTCIR-3 workshop. In Proceedings of the Third NTCIR Workshop on Research in Information Retrieval, Automatic Text Summarization and Question Answering. Available from: http://research.nii.ac.jp/ntcir/. [Chen and Gey, 2004] Chen, A. and Gey, F. C. (2004). Multilingual information retrieval using machine translation, relevance feedback and decompounding. Information Retrieval, 7(1-2):149–182. [Cheng et al., 2004] Cheng, P.-J., Teng, J.-W., Chen, R.-C., Wang, J.-H., Lu, W.-H., and Chien, L.-F. (2004). Translating unknown queries with Web corpora for cross-language information retrieval. In Proceedings of the 27th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 146–153. [Church and Hanks, 1990] Church, K. W. and Hanks, P. (1990). Word association norms, mutual information, and lexicography. Computational Linguistics, 16(1):22–29. [Darwish and Oard, 2003] Darwish, K. and Oard, D. W. (2003). Probabilistic structured query methods. In Proceedings of the 26th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 338–344. [Davis, 1997] Davis, M. (1997). New experiments in cross-language text retrieval at NMSU’s Computing Research Lab. In Proceedings of TREC-5. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Davis and Dunning, 1995] Davis, M. and Dunning, T. (1995). A TREC evaluation of query translation methods for multi-lingual text retrieval. In Proceedings of TREC-4. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Davis, 1998] Davis, M. W. (1998). On the effective use of large parallel corpora in crosslanguage text retrieval. In Grefenstette, G., editor, Cross-Language Information Retrieval, pages 12–22. Kluwer. [Davis and Ogden, 1997] Davis, M. W. and Ogden, W. C. (1997). QUILT: implementing a large-scale cross-language text retrieval system. In Proceedings of the 20th Annual

542

Kazuaki Kishida

International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 92–98. [Deerwester et al., 1990] Deerwester, S., Dumais, S. T., Furnas, G. W., Landauer, T. K., and Harshman, R. (1990). Indexing by latent semantic analysis. Journal of the American Society for Information Science, 41(6):391–407. [Diekema et al., 1999] Diekema, A., Oroumchian, F., Sheridan, P., and Liddy, E. D. (1999). TREC7 evaluation of conceptual interlingua document retrieval (CINDOR) in English and French. In Proceedings of TREC-7. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Dumais et al., 1996] Dumais, S. T., Landauer, T. K., and Littman, M. L. (1996). Automatic cross-linguistic information retrieval using latent semantic indexing. In Proceedings of the 19th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 16–23. [Dunning, 1993] Dunning, T. (1993). Accurate methods for the statistics of surprise and coincidence. Computational Linguistics, 19(1):61–74. [Eichmann et al., 1998] Eichmann, D., Ruiz, M. E., and Srinivasan, P. (1998). Crosslanguage information retrieval with the UMLS metathesaurus. In Proceedings of the 21st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 72–80. [Federico and Bertoldi, 2002] Federico, M. and Bertoldi, N. (2002). Statistical crosslanguage information retrieval using N-best query translations. In Proceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 167–174. [Franz et al., 1999] Franz, M., Scott McCarley, J., and Roukos, S. (1999). Ad hoc and multilingual information retrieval at IBM. In Proceedings of TREC-7. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Franz et al., 2000] Franz, M., Scott McCarley, J., and Todd Ward, R. (2000). Ad hoc, cross-language and spoken document information retrieval at IBM. In Proceedings of TREC-8. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Fujii and Ishikawa, 2001] Fujii, A. and Ishikawa, T. (2001). Japanese/English crosslanguage information retrieval: exploration of query translation and transliteration. Computers and the Humanities, 35:389–420. [Galvez and Moya-Aneg´on, 2007] Galvez, C. and Moya-Aneg´on, F. (2007). Approximate personal name-matching through finite-state graphs. Journal of the American Society for Information Science and Technology, 58(13):1960–1976.

Methods for Cross-Language Information Retrieval

543

[Gao and Nie, 2006] Gao, J. and Nie, J.-Y. (2006). A study of statistical models for query translation: finding a good unit of translation. In Proceedings of the 29th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 194–201. [Gao et al., 2001] Gao, J., Nie, J.-Y., Zhang, J., Xun, E., Su, Y., Zhou, M., and Huang, C. (2001). Improving query translation for cross-language information retrieval using statistical models. In Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 96–104. [Gey, 2001] Gey, F. C. (2001). Research to improve cross-language retrieval: position paper for CLEF. In Peters, C., editor, Cross-Language Information Retrieval Evaluation, pages 83–88. Springer-Verlag. (LNCS, 2069). [Gey et al., 1999] Gey, F. C., Jiang, H., Chen, A., and Larson, R. R. (1999). Manual queries and machine translation in cross-language retrieval and interactive retrieval with Cheshire II at TREC-7. In Proceedings of TREC-7. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Gollins and Sanderson, 2001] Gollins, T. and Sanderson, M. (2001). Improving cross language information retrieval with triangulated translation. In Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 90–95. [Harman, 2005] Harman, D. (2005). Beyond English. In Voorhees, E. M. and Harman, D. K., editors, TREC: Experiment and Evaluation in Information Retrieval, pages 153– 181. MIT Press. [Hedlund et al., 2004] Hedlund, T., Airio, E., Keskustalo, H., Lehtokangas, R., Pirkola, A., and J¨arveline, K. (2004). Dictionary-based cross-language information retrieval: learning experiences from CLEF 2000-2002. Information Retrieval, 7(1-2):99–119. [Hedlund et al., 2002] Hedlund, T., Keskustalo, H., Pirkola, A., Airio, E., and J¨arvelin, K. (2002). Utaclir@CLEF2001: effects of compound splitting and n-gram techniques. In et al., C. P., editor, Evaluation of Cross-Language Information Retrieval Systems, pages 118–136. Springer-Verlag. (LNCS 2406). [Hiemstra, 1998a] Hiemstra, D. (1998a). A linguistically motivated probabilistic model of information retrieval. In Nikolaou, C. and Stephanidis, C., editors, Research and Advanced Technology for Digital Libraries (LNCS 1513), pages 569–584. Springer. [Hiemstra, 1998b] Hiemstra, D. (1998b). Multilingual domain modeling in Twenty-One: automatic creation of a bi-directional translation lexicon from a parallel corpus. In Proceedings of the 8th CLIN Meeting, pages 41–58. [Hiemstra and Kraaij, 1999] Hiemstra, D. and Kraaij, W. (1999). Twenty-one at TREC7: ad-hoc and cross-language track. In Proceedings of TREC-7. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html.

544

Kazuaki Kishida

[Hiemstra et al., 2001] Hiemstra, D., Kraaij, W., Pohlmann, R., and Westerveld, T. (2001). Translation resources, merging strategies, and relevance feedback for cross-language information retrieval. In Peters, C., editor, Cross-Language Information Retrieval Evaluation, pages 102–115. Springer-Verlag. (LNCS 2069). [Hull and Grefenstette, 1996] Hull, D. A. and Grefenstette, G. (1996). Querying across languages: a dictionary-based approach to multilingual information retrieval. In Proceedings of the 19th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 49–57. [Ide and V´eronis, 1998] Ide, N. and V´eronis, J. (1998). Introduction to the special issue on word sense disambiguation: the state of the art. Computational Linguistics, 24(1):1–40. [Jang et al., 1999] Jang, M.-G., Myaeng, S. H., and Park, S. Y. (1999). Using mutual information to resolve query translation ambiguities and query term weighting. In Proceedings of the 37th Annual Conference of the Association of Computational Linguistics, pages 223–229. [Jones and Lam-Adesina, 2002] Jones, G. J. F. and Lam-Adesina, A. M. (2002). Experiments with machine translation for bilingual retrieval. In et al., C. P., editor, Evaluation of Cross-Language Information Retrieval Systems, pages 59–77. Springer-Verlag. (LNCS 2406). [Kang et al., 2005] Kang, I.-S., Na, S.-H., and Lee, J.-H. (2005). Combination approaches in information retrieval: words vs. n-grams and query translation vs. document translation. In Proceedings of the Fourth NTCIR Workshop on Research in Information Access Technologies Information Retrieval, Question Answering and Summarization. Available from: http://research.nii.ac.jp/ntcir/workshop/OnlineProceedings4/. [Kek¨al¨ainen and J¨arveline, 1998] Kek¨al¨ainen, J. and J¨arveline, K. (1998). The impact of query structure and query expansion on retrieval performance. In Proceedings of the 21st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 130–137. [Kishida, 2005] Kishida, K. (2005). Technical issues of cross-language information retrieval: a review. Information Processing & Management, 41(3):433–455. [Kishida, 2007] Kishida, K. (2007). Term disambiguation techniques based on target document collection for cross-language information retrieval: an empirical comparison of performance between techniques. Information Processing & Management, 43(1):103– 120. [Kishida, 2008] Kishida, K. (2008). Prediction of performance of cross-language information retrieval using automatic evaluation of translation. Library and Information Science Research, 30(2):138–144. [Kishida et al., 2007] Kishida, K., Chen, K.-H., Lee, S., Kuriyama, K., Kando, N., and Chen, H.-H. (2007). Overview of CLIR task at the Sixth NTCIR Workshop. In Proceedings of the 6th NTCIR Workshop Meeting on Evaluation of Information Access Tech-

Methods for Cross-Language Information Retrieval

545

nologies: Information Retrieval, Question Answering and Cross-Lingual Information Access. Available from: http://research.nii.ac.jp/ntcir/workshop/OnlineProceedings6/. [Kishida and Kando, 2004] Kishida, K. and Kando, N. (2004). Two-stage refinement of query translation in a pivot language approach to cross-lingual information retrieval: an experiment at CLEF 2003. In C. Peters et al., editor, Comparative Evaluation of Multilingual Information Access Systems, pages 253–262. Springer-Verlag. (LNCS 3237). [Kishida and Kando, 2006] Kishida, K. and Kando, N. (2006). A hybrid approach to query and document translation using a pivot language for cross-language information retrieval. In C. Peters et al., editor, Accessing Multilingual Information Repositories, pages 93–101. Springer-Verlag. (LNCS 4022). [Kishida et al., 2005] Kishida, K., Kando, N., and Chen, K. H. (2005). Two-stage refinement of transitive query translation with english disambiguation for cross-language information retrieval: an experiment at CLEF2004. In C. Peters et al., editor, Multilingual Information Access for Text, Speech and Images, Lecture Notes in Computer Science, pages 135–142. Springer-Verlag. (LNCS 3491). [Knight and Graehl, 1998] Knight, K. and Graehl, J. (1998). Machine transliteration. Computational Linguistics, 24(4):599–612. [Koehn and Knight, 2000] Koehn, P. and Knight, K. (2000). Estimating word translation probabilities from unrelated monolingual corpora using the EM algorithm. In Proceedings of the Seventeenth National Conference on Artificial Intelligence and Twelfth Conference on Innovative Applications of Artificial Intelligence, pages 711–715. [Kraaij, 2002] Kraaij, W. (2002). TNO at CLEF-2001: comparing translation resources. In et al., C. P., editor, Evaluation of Cross-Language Information Retrieval Systems, pages 78–93. Springer-Verlag. (LNCS 2406). [Kraaij et al., 2000] Kraaij, W., Pohlmann, R., and Hiemstra, D. (2000). Twentyone at TREC-8: using language technology for information retrieval. In Proceedings of TREC-8. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Kuo et al., 2008] Kuo, J., Li, H., and Yang, Y. (2008). Active learning for constructing transliteration lexicons from the Web. Journal of the American Society for Information Science and Technology, 59(1):126–135. [Landauer and Littman, 1990] Landauer, T. and Littman, M. L. (1990). A statistical method for language-independent representation of the topical content of text segments. In Proceedings of the 6th Annual Conference of the UW Centre for the New Oxford English Dictionary and Text Research. Available from: http://www.cs.duke.edu/˜mlittman/docs/x-lang.ps. [Landauer et al., 2007] Landauer, T. K., McNamara, D. S., Dennis, S., and Kintsch, W., editors (2007). Handbook of Latent Semantic Analysis. Lawrence Erlbaum Associates.

546

Kazuaki Kishida

[Larkey and Connell, 2005] Larkey, L. S. and Connell, M. E. (2005). Structured queries, language modeling, and relevance modeling in cross-language information retrieval. Information Processing & Management, 41(3):457–473. [Lavrenko et al., 2002] Lavrenko, V., Choquette, M., and Croft, W. B. (2002). Crosslingual relevance models. In Proceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 175–182. [Lavrenko and Croft, 2001] Lavrenko, V. and Croft, W. B. (2001). Relevance-based language models. In Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 120–127. [Lee et al., 2004] Lee, K.-S., Kageura, K., and Choi, K.-S. (2004). Implicit ambiguity resolution using incremental clustering in cross-language information retrieval. Information Processing & Management, 40(1):145–159. [Lehtokangas et al., 2004] Lehtokangas, R., Airio, E., and J¨arveline, K. (2004). Transitive dictionary translation challenges direct dictionary translation in CLIR. Information Processing & Management, 40(6):973–988. [Levow et al., 2005] Levow, G.-A., Oard, D. W., and Resnik, P. (2005). Dictionary-based techniques for cross-language information retrieval. Information Processing & Management, 41(3):523–547. [Lin and Chen, 2003] Lin, W. C. and Chen, H. H. (2003). Description of NTU approach to NTCIR3 multilingual information retrieval. In Proceedings of the Third NTCIR Workshop on Research in Information Retrieval, Automatic Text Summarization and Question Answering. Available from: http://research.nii.ac.jp/ntcir/. [Littman et al., 1998] Littman, M. L., Dumais, S., and Landauer, T. K. (1998). Automatic cross-language information retrieval using latent semantic indexing. In Grefenstette, G., editor, Cross-Language Information Retrieval, pages 51–62. Kluwer. [Liu et al., 2005] Liu, Y., Jin, R., and Chai, J. Y. (2005). A maximum coherence model for dictionary-based cross-language information retrieval. In Proceedings of the 28th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 536–543. [Lopez-Ostenero et al., 2005] Lopez-Ostenero, F., Gonzalo, J., and Verdejo, F. (2005). Noun phrases as building blocks for cross-language search assistance. Information Processing & Management, 41(3):549–568. [Maeda et al., 2000] Maeda, A., Sadat, F., Yoshikawa, M., and Uemura, S. (2000). Query term disambiguation for Web cross-language information retrieval using a search engine. In Proceedings of the 5th International Workshop on Information Retrieval with Asian languages, pages 25–32. [Manning et al., 2008] Manning, C. D., Raghavan, P., and Sch¨utze, H. (2008). Introduction to Information Retrieval. Cambridge.

Methods for Cross-Language Information Retrieval

547

[Markey, 2007] Markey, K. (2007). Twenty-five years of end-user searching, part 1: research findings. Journal of the American Society for Information Science and Technology, 58(8):1071–1081. [Mark´o et al., 2005] Mark´o, K., Schulz, S., Medelyan, O., and Hahn, U. (2005). Bootstrapping dictionaries for cross-language information retrieval. In Proceedings of the 28th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 528–535. [McNamee and Mayfield, 2002a] McNamee, P. and Mayfield, J. (2002a). Comparing cross-language query expansion techniques by degrading translation resources. In Proceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 159–166. [McNamee and Mayfield, 2002b] McNamee, P. and Mayfield, J. (2002b). JHU/APL experiments at CLEF: translation resources and score normalization. In et al., C. P., editor, Evaluation of Cross-Language Information Retrieval Systems, pages 193–208. SpringerVerlag. (LNCS 2406). [McNamee and Mayfield, 2004] McNamee, P. and Mayfield, J. (2004). Character n-gram tokenization for European language text retrieval. Information Retrieval, 7(1-2):73–97. [Melamed, 2000] Melamed, I. D. (2000). Models of translational equivalence among words. Computational Linguistics, 26(2):221–249. [Miller et al., 1999] Miller, D. R. H., Leek, T., and Schwartz, R. M. (1999). A hidden Markov model information retrieval system. In Proceedings of the 22nd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 214–221. [Monz and Dorr, 2005] Monz, C. and Dorr, B. J. (2005). Iterative translation disambiguation for cross-language information retrieval. In Proceedings of the 28th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 520–527. [Moulinier and Molina-Salgado, 2003] Moulinier, I. and Molina-Salgado, H. (2003). Thomson Legal and Regulatory experiments for CLEF2002. In et al., C. P., editor, Advances in Cross-Language Information Retrieval, pages 155–163. Springer-Verlag. (LNCS 2785). [Nie, 2000] Nie, J.-Y. (2000). CLIR using a probabilistic translation model based on Web documents. In Proceedings of TREC-8. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Nie and Jin, 2003] Nie, J.-Y. and Jin, F. (2003). A multilingual approach to multilingual information retrieval. In et al., C. P., editor, Advances in Cross-Language Information Retrieval, pages 101–110. Springer-Verlag. (LNCS 2785).

548

Kazuaki Kishida

[Nie et al., 1999] Nie, J.-Y., Simard, M., Isabelle, P., and Durand, R. (1999). Crosslanguage information retrieval based on parallel texts and automatic mining of parallel texts from the Web. In Proceedings of the 22nd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 74–81. [Oard and Diekema, 1998] Oard, D. W. and Diekema, A. R. (1998). Cross-language information retrieval. Annual Review of Information Science and Technology, 33:223–256. [Oard and Hackett, 1998] Oard, D. W. and Hackett, P. (1998). Document translation for cross-language text retrieval at the University of Maryland. In Proceedings of TREC-6. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Oard et al., 2001] Oard, D. W., Levow, G. A., and Cabezas, C. I. (2001). CLEF experiments at Maryland: statistical stemming and backoff translation. In Peters, C., editor, Cross-Language Information Retrieval Evaluation, pages 176–187. Springer-Verlag. (LNCS 2069). [Och and Ney, 2003] Och, F. J. and Ney, H. (2003). Systematic comparison of various statistical alignment models. Computational Linguistics, 29(1):19–51. [Papineni et al., 2002] Papineni, K., Roukos, S., Ward, T., and Zhu, W.-J. (2002). BLEU: a method for automatic evaluation of machine translation. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pages 311–318. ACL. [Peters and Sheridan, 2001] Peters, C. and Sheridan, P. (2001). Multilingual information access. In M. Agosti, et al., editor, Lectures on Information Retrieval, pages 51–80. Springer-Verlag. (LNCS 1980). [Pirkola, 1998] Pirkola, A. (1998). The effects of query structure and dictionary-based cross-language information retrieval. In Proceedings of the 21st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 55–63. [Pirkola et al., 2001] Pirkola, A., Hedlund, T., Keskustalo, H., and J¨arvelin, K. (2001). Dictionary-based cross-language information retrieval: problems, methods, and research findings. Information Retrieval, 4(3-4):209–230. [Pirkola et al., 2003] Pirkola, A., Toivonen, J., Keskustalo, H., V., K., and J¨arvelin, K. (2003). Fuzzy translation of cross-lingual spelling variants. In Proceedings of the 26th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 345–352. [Powell et al., 2000] Powell, A. L., French, J. C., Callan, J., Connell, M., and Viles, C. L. (2000). The impact of database selection on distributed searching. In Proceedings of the 23rd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 232–239.

Methods for Cross-Language Information Retrieval

549

[Qu et al., 2003a] Qu, Y., Grefenstette, G., and Evans, D. A. (2003a). Automatic transliteration for Japanese-to-English text retrieval. In Proceedings of the 26th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 353–360. [Qu et al., 2003b] Qu, Y., Grefenstette, G., and Evans, D. A. (2003b). Resolving translation ambiguity using monolingual corpora. In et al., C. P., editor, Advances in CrossLanguage Information Retrieval, pages 223–241. Springer. (LNCS 2785). [Rehder et al., 1998] Rehder, B., Littman, M. L., Dumais, S., and Landauer, T. K. (1998). Automatic 3-language cross-language information retrieval with latent semantic indexing. In Proceedings of the TREC6. National Institute of Standards and Technology. Available from: http://trec.nist.gov/pubs.html. [Resnik, 1999] Resnik, P. (1999). Mining the web for bilingual text. In Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics, pages 527–534. [Rogati and Yang, 2004] Rogati, M. and Yang, Y. (2004). Resource selection for domainspecific cross-lingual IR. In Proceedings of the 27th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 154–161. [Salton, 1970] Salton, G. (1970). Automatic processing of foreign language documents. Journal of the American Society for Information Science, 21(3):187–194. [Savoy, 2002] Savoy, J. (2002). Report on CLEF-2001 experiments: effective combined query-translation approach. In et al., C. P., editor, Evaluation of Cross-Language Information Retrieval Systems, pages 27–43. Springer-Verlag. (LNCS 2406). [Savoy, 2003] Savoy, J. (2003). Report on CLEF-2002 experiments: combining multiple sources of evidence. In et al., C. P., editor, Advances in Cross-Language Information Retrieval, pages 66–90. Springer-Verlag. (LNCS 2785). [Savoy, 2004] Savoy, J. (2004). Combining multiple strategies for effective monolingual and cross-language retrieval. Information Retrieval, 7(1-2):121–148. [Scott McCarley, 1999] Scott McCarley, J. (1999). Should we translate the documents or the queries in cross-language information retrieval? In Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics, pages 208–214. [Seo et al., 2005] Seo, H.-C., Kim, S.-B., Rim, H.-C., and Myaeng, S.-H. (2005). Improving query translation in English-Korean cross-language information retrieval. Information Processing & Management, 41(3):507–522. [Sheridan and Ballerini, 1996] Sheridan, P. and Ballerini, J. P. (1996). Experiments in multilingual information retrieval using the SPIDER system. In Proceedings of the 19th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 58–64.

550

Kazuaki Kishida

[Sperer and Oard, 2000] Sperer, R. and Oard, D. W. (2000). Structured translation for cross-language information retrieval. In Proceedings of the 23rd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 120–127. [Talvensaari et al., 2006] Talvensaari, T., Laurikkala, J., J¨arveline, K., and Juhola, M. (2006). A study on automatic creation of a comparable document collection in crosslanguage information retrieval. Journal of Documentation, 62(3):372–387. [Toivonen et al., 2005] Toivonen, J., Pirkola, A., Keskustalo, H., Visala, K., and J¨arveline, K. (2005). Translating cross-lingual spelling variants using transformation rules. Information Processing & Management, 41(4):859–872. [Tsuji and Kageura, 2006] Tsuji, K. and Kageura, K. (2006). Automatic generation of Japanese-English bilingual thesauri based on bilingual corpora. Journal of the American Society for Information Science and Technology, 57(7):891–906. [Turtle and Croft, 1991] Turtle, H. R. and Croft, W. B. (1991). Evaluation of an inference network-based retrieval model. ACM Transactions on Information Systems, 9(3):187– 222. [Wang and Oard, 2006] Wang, J. and Oard, D. W. (2006). Combining bidirectional translation and synonymy for cross-language information retrieval. In Proceedings of the 29th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 202–209. [Xu and Weischedel, 2005] Xu, J. and Weischedel, R. (2005). Empirical studies on the impact of lexical resources on CLIR performance. Information Processing & Management, 41(3):475–487. [Xu et al., 2001] Xu, J., Weischedel, R., and Nguyen, C. (2001). Evaluating a probabilistic model for cross-lingual information retrieval. In Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 105–110. [Yamabana et al., 1998] Yamabana, K., Muraki, K., Doi, S., and Kamei, S. (1998). A language conversion front-end for cross-language information retrieval. In Grefenstette, G., editor, Cross-Language Information Retrieval, pages 93–104. Kluwer. [Yang et al., 1998] Yang, Y., Carbonell, J., Brown, R. D., and Frederking, R. E. (1998). Translingual information retrieval: learning from bilingual corpora. Artificial Intelligence, 103:323–345. [Zhang and Vines, 2004] Zhang, Y. and Vines, P. (2004). Using the web for automated translation extraction in cross-language information retrieval. In Proceedings of the 27th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 162–169.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 24

LANGUAGE AND THOUGHT: LINGUISTIC INFLUENCE ON DEVELOPMENTAL NEURAL BASIS OF ‘THEORY OF MIND’ Chiyoko Kobayashi Department of Psychology, Cornell University, Uris Hall, Ithaca, NY, US

Abstract Theory of mind (ToM) – ability to perceive, interpret and predict behaviors or actions of others based on their underlying mental states – has been explored with a plethora of paradigms in normally and atypically developing children by many researchers. ToM is fundamental for human social interaction universally. Among ToM researchers it has long been debated whether or not language development constrains ToM development. Recent results from neurological studies suggest that some aspects of language (e.g., grammar) may function merely as a ‘co-opted’ system, but other aspects of language (e.g., pragmatics and reading communicative intentions) may profoundly affect ToM throughout the development. Neuroimaging studies of ToM development are still scarce. However, results from a few studies that explored neural correlates of ToM and related socio-cognitive functions revealed age-related differences in ToM/social-cognitionspecific brain activity in several language regions. Despite the increasing evidence that supports the developmental relationship between language and various cognitive capacities ToM, current developmental theories of ToM (except one) discount the linguistic effects on ToM. There are four main-stream theories of ToM development; 1) Modular ToM hypothesis, 2) Theory, theory of mind hypothesis, 3) Simulation ToM hypothesis, and 4) Linguistic determinism of ToM hypothesis. In the first part of the chapter, I will discuss both behavioral and neurological evidence that supports or negates the linguistic influence on ToM. In the second part of this chapter, I will discuss those four prominent theories of ToM development and whether or not evidence from the most current neuroimaging studies of ToM in children and adults (including ours) support these theories. Finally, based on the most current results, I will present a new model for the developmental mechanism of ToM.

552

Chiyoko Kobayashi

Introduction Are our thoughts same as our language? Or, are they different? Whorf (1956) hypothesized that our language constrains our thoughts and reflects our culturally unique world view. Later, Vygotsky (1967) elaborated this hypothesis, positing that human consciousness (or thought) has its basis in linguistic or historical contexts and is enabled only through the internalization of culture-specific symbols (i.e., language). According to the Vygotskyan view, human consciousness never develops independently of language, which is acquired only through intersubjective perspective-taking and communication with other individuals in the society [Tomasello, 2001; Valsiner, 1989; Vygotsky, 1967]. From the late 1960s to mid 1990s, the so-called Sapir-Whorf hypothesis was discredited by cognitive scientists and linguists who emphasized universal and veridical ways of perceiving our world. Throughout these decades a view, which posited that universal linguistic [Chomsky, 1980] and cognitive developmental [Piaget, 1962; Sinclair, 1970] principles determined individuals’ thoughts and cognition, dominated. Following the Chomskyan tradition (and counter to the Whorfian hypothesis), Fodor (1975) argued that humans possess a rich, propositional system which we share with nonhuman primates and infants. Fodor named this innate representational system “language of thought.” This primitive computational and symbolic system, he has theorized, exists independently from natural language and governs our preverbal theory of mind (ToM): our ability to perceive, interpret and predict the behaviors or actions of others in terms of their underlying mental states [Fodor, 1975; the definition is by Scholl & Leslie, 1999]. A few years later, Premack and Woodruff (1978) tested this ability in a chimpanzee and found evidence for ToM in this nonhuman primate. Although the validity of their tests and the very capacity of ToM in nonhuman primates are still debated [Cheney & Seyfarth, 1990; de Waal, 2001; Hayes, 1998)], no one questions that the ability to understand others’ desires and intentions is important for humans [Frith & Frith, 2003]. Since the first experiment with the chimpanzee [Premack & Woodruff, 1978], various paradigms have been devised to test ToM in humans [Baron-Cohen, 2000]. Among those ToM tasks, a false-belief (FB) task has been the most commonly used for testing normally developing [Wimmer & Perner, 1983] as well as atypically developing children [BaronCohen, Leslie, & Frith, 1985; see Baron-Cohen, 2000 for a review]. In a typical FB task, two characters appear (e.g., Sally and Ann) in a scene. When one character, Sally, is present, Ann, the other character, puts a toy into a basket. Sally then disappears from the scene. While Sally is away, Ann takes the toy out of the basket and puts it into a box. The experimenter then asks the child the critical false-belief question, “Where will Sally look for the toy?” Nearly universally observed results are that adults and children over 4 years of age correctly answer “basket” whereas younger children (as well as older children and adolescents with autism) fail the task by answering “box” [Baron-Cohen, Leslie, & Frith, 1985; 1986]. These failures reflect their lack of understanding that Sally’s belief about the location of the toy is different from Ann’s [Frith, 2003; Happé, 1993]. Several brain imaging studies have examined the neural correlates of ToM using the FB style paradigm in adults [Brunet et al., 2000; Fletcher et al., 1995; Gallagher et al., 2000; Goel et al., 1995; Happé et al., 1996; Kobayashi, Glover, & Temple, 2006; Sabbagh & Taylor, 2000; Saxe & Kanwisher, 2003; Vogeley et al., 2001]. Many of these studies have found

Language and Thought

553

significant activity in medial prefrontal cortex (mPFC) during false-belief conditions [Flethcer et al., 1995; Gallagher et al., 2000; Goel et al., 1995; Happé et al., 1996] (Figure 1). In addition, the temporo-parietal junction (TPJ) has also been suggested to be important for ToM processing. This area was found to become active during both true- and false-belief conditions and not during false representations in a non-social control condition [Saxe & Kanwisher, 2003; Saxe & Wexler, 2005]. Other brain regions implicated in these and other ToM brain imaging studies include the anterior cingulate cortex (ACC), the middle frontal gyrus, the precuneus/posterior cingulate cortex, the superior temporal sulcus, the orbitofrontal cortex, and the temporal pole. However, these areas have not been implicated as consistently as the mPFC and the TPJ [see Frith & Frith, 2003, and Saxe, Carey, & Kanwisher, 2004 for reviews]. Although brain imaging studies of ToM in children are still scarce, several studies found similar medial frontal and/or TPJ activity using a variety of ToM or related social/cognitive tasks [Dapretto et al., 2006; Kobayashi, Glover, & Temple, 2007a; 2007b; Liu, 2005; Liu et al., 2005; Moriguchi et al., 2007; Ohnishi et al., 2004; Wang et al., 2006a; 2006b] (Table 1).

Figure 1.

554

Chiyoko Kobayashi

Table 1. Developmental neuroimaging studies of ToM and related social/ cognitive functions (selected studies in ascending Order by Year of Publication) Authors (Year) Ohnishi, et al. (2004)

Imaging Method fMRI

Subject

Task

Main Findings

11 children (7-13 yrs-old)

Video of hands grasping some objects (imitation). Animation of intentional movement of geometric figures (ToM).

ERP

24 adults and 41 children (5-7 yrsold)

Animation based firstorder TB and FB task

Dapretto, et al. (2006)

fMRI

10 TD children and 10 children with ASD (12-13 years-old)

Facial imitation task

Wang, et al. (2006a)

fMRI

12 Adults (23-33 yrs-old) and 12 children (9-14 yrs-old)

Cartoon based Irony task

Wang, et al. (2006b)

fMRI

Story-based irony task

Kobayashi, et al. (2007a)

fMRI

18 children/ adolescents with ASD (7-17 yrsold) and 18 TD children/ adolescents (8-16 yrs-old) 16 Adults (18-40 yrs-old) and 12 children (8-12 yrs-old)

Kobayashi, et al. (2007b)

fMRI

ToM condition activated right mPFC, bilateral STG, right SMG, right MTG, right TP, right FG, bilateral MOG, and left cerebellum. Both imitation and ToM conditions activated bilateral MTG, right STG, bilateral MOG, and bilateral FG. Increasing localization in the left frontal region from child-passers to adults. More diffused bilateral frontal activity in children than in adults. TD children activated more right precentral gyrus, right ACC, bilateral IFG, insula, amygdala, hippocampus, caudate, putamen, and thalamus than children with ASD. Children activated Right STG, bilateral IFG, right MFG, right STS, and left mPFC more than adults. Adults activated more posterior brain regions (e.g., visual cortex) than children. TD children activated more right IFG (when contextual cues were available) and bilateral STS (when both types of cues were available) than children with ASD. Children with ASD activated temporal regions more when only prosodic cues were available. Overall more brain activity in children than adults. Both Adults and children activated right IPL and TPJ. Adults activated ToM areas (e.g., TPJ) during the story condition, but children activated these areas during the cartoon ToM condition. Both American children and Japanese bilingual children activated bilateral vmPFC.

Liu (2005)

Moriguchi, et al. (2007)

fMRI

12 American and 12 Japanese children (8-12 yrs-old) 16 children/ adolescents (9-16 yrs old)

Cartoon and story based second-order FB task

Cartoon and story based second-order FB task

Animation of intentional movement of geometric figures.

Age related positive correlation in the dorsal mPFC, and negative correlation in the ventral mPFC.

Abbreviations: ACC = anterior cingulate cortex, FG = fusiform gyrus, IFG = inferior frontal gyrus, IPL = inferior parietal lobule, MFG = middle frontal gyrus, MOG = middle occipital gyrus, mPFC = medial prefrontal cortex, MTG = middle temporal gyrus, OFC = orbito-frontal cortex, SMG = supramarginal gyrus, STG = superior temporal gyrus, STS = superior temporal sulcus, TP = temporal pole, TPJ = temporo-parietal junction, vmPFC = ventro-medial prefrontal cortex; AS = Asperger’s syndrome, ASD = autism spectrum disorder, HFA = high functioning autism, TD = typically developing

Despite a long period of obloquy, the Whorfian hypothesis has recently been revived following several new findings from cross-cultural/linguistic studies that have shown some

Language and Thought

555

influences of culture/language on people’s representations of conceptual properties [e.g., Boroditsky, 2001; Bowerman & Choi, 2003; Brown & Levinson, 1993; Choi & Bowerman, 1991; Lucy, 1992]. Although certain aspects of cognition may precede language, as shown in studies of prelinguistic infants [McDonough, Choi, & Mandler, 2003; Hespos & Spelke, 2004], our cognition or thoughts may be influenced strongly by the language we speak throughout development. However, the biological underpinnings of the linguistic influence on thoughts have been unknown. A few recent neuroimaging studies exploring linguistic influence on arithmetic ability were the first to present evidence that language may influence one’s cognition at the neural level [Dehaene et al., 1999; Venkatraman et al., 2006]. Likewise, our group’s findings on American and Japanese adults and children have supported the Whorfian hypothesis by demonstrating some influence of language on the neural bases of ToM [Kobayashi, Glover, & Temple, 2006; 2007b]. In addition, several recent neuroimaging studies in children found activity in several “language regions” [Brauer & Friederici, 2007; Price, 2000] including the superior temporal gyrus (STG) and/or inferior frontal gyrus (IFG) while children processed ToM or related social/cognitive tasks [Dapretto et al., 2006; Kobayashi, Glover, & Temple, 2007a; 2007b; Moriguchi et al., 2007; Ohnishi et al., 2004; Wang et al., 2006a; 2006b]. For instance, in our group’s study, children activated the STG while they processed verbal FB task, yet adults activated the same region while they processed nonverbal FB task [Kobayashi et al., 2007a]. These results suggest that language influences ToM throughout development. Despite the increasing evidence that supports the developmental relationship between language and various cognitive capacities (including ToM), current developmental theories of ToM (except one) discount the linguistic effects on ToM and its neural basis. In this chapter, I first discuss some of the prominent theories of ToM development and whether evidence from the neuroimaging studies of ToM (including ours) supports these theories. Especially, my focus is on “Linguistic determinism” hypothesis of ToM [de Villiers, 2000]. I then review literature on the developmental relationship between language and ToM and its neural basis. I then discuss the possible interactions between brain areas associated with language processing and ToM candidate regions. Finally, I present a new model for the developmental mechanism of ToM.

Four Theories of ToM Development The developmental mechanism of ToM has been investigated with a plethora of paradigms by many researchers. There are four primary theories of ToM development; “Modular”, “Theory-theory”, “Simulation”, and “Linguistic determinism”. Each of the four theories was developed mainly to account for the results of the FB task performance; failure in normally-developing 3-year-old children and older children with autism, and successful performance in 4-year-old children.

Modular ToM Theory Since Fodor’s highly influential book, “The Modularity of Mind” (1983) was published, the ‘modular’ view has become influential in cognitive science. According to Fodor’s view,

556

Chiyoko Kobayashi

the architecture of each of our sensory (and some of our cognitive) functions can be best described as modular in the sense that it is: (1) informationally encapsulated, (2) unconscious, (3) fast, (4) has shallow outputs, (5) involves obligatory firing, (6) localized, (7) domain specific, and (8) ontogenetically and pathologically universal. This highly nativistic point of view has been supported by several ToM researchers, though not in a purely Fodorian sense. A main advocate of modular ToM, Leslie describes ToM as maturing serially in three steps during the course of development. The first one is termed ‘theory of body mechanism’ (ToBy), which matures around 4 months of age and embodies the infant’s theory of physical objects. At this stage, the infant formulates a primary theory about the mechanical movement of objects, but is not capable of associating the animation with the ‘Agent’, nor can he/she distinguish the intentionality of the ‘Agent’ from mere physical force. The next one, ‘theory of mind mechanism 1’ (ToMM1), matures around 8 months of age and confers upon the infant a new knowledge that the ‘Agent’ has to have some intention to move the objects. The sequence completes when a more adult-like ToM, dubbed ‘theory of mind mechanism 2’ (ToMM2), matures around 18 months of age and enables the toddler to formulate ‘propositional attitudes’ such as “she believes/pretends/thinks that …” [Leslie, 1992; 1994]. According to Leslie, the reason why 3-year-olds fail the FB task is not due to any impairments in the core ToMM module, but due to immature ‘selection processing’ (SP) mechanism which is domain-general (unlike ToMM which is domain-specific) and functions like an inhibitory control. He argues that older children with autism are impaired in ToM not because of any damage in the SP part, but because of some dysfunction in the ToMM module [Roth & Leslie, 1998] and this component of ToM comprises the specific innate basis of ToM. Modular hypothesis has recently found some robust support from a series of ‘habituation’ experiments in infants. These experiments have shown that non-verbal FB tasks can be passed even by 15 month-old infants [Onishi & Baillargeon, 2005, Southgate, Senju, & Csibra, 2007; Surian, Caldi, & Sperber, 2007]. These studies support the main tenet of modular hypothesis; innate basis of ToM. They, however, were criticized because they failed to show why the reality-bias seen in 3 years-olds was not seen in the infants that they tested and that infants may be doing three-way association among agent, object, and place without any FB understanding [Perner & Ruffman, 2005]. In terms of neural bases of ToM, the modular hypothesis of ToM predicts that ToM is represented in a highly circumscribed brain region [Gallagher & Frith, 2003]. However, neuroimaging studies of ToM and autism have presented mixed results regarding this prediction. On the one hand, many ToM imaging studies utilizing a variety of ToM paradigms have consistently implicated the mPFC [Brunet et al., 2000; Fletcher et al., 1995; Gallagher et al., 2000; Gallagher et al., 2002; Goel et al., 1995; Happé et al., 1996; Kobayashi et al., 2006; Vogeley et al., 2001; see also Frith & Frith, 2003, for a review], indicating some modularity in ToM. On the other hand, structural brain imaging studies on autism alone found abnormalities (usually increases/decreases in volume) in more than 14 brain regions that are widely distributed [Brambilla et al., 2003]. Although the amygdala is most often found to be structurally abnormal in individuals with autism [Schultz, Romanski, & Tsatsanis, 2000], only a few functional brain imaging studies of ToM implicated this region for ToM (e.g., Baron-Cohen, Ring, et al., 1999). The increasing evidence suggests that autism is not a modular but a distributed disorder involving functionally connected multiple brain regions [Müller, 2007]. Moreover, it has become increasingly likely that none of the candidate ToM areas are specialized for ToM or mentalizing [Frith & Frith, 2003] per se. For instance, it has

Language and Thought

557

been shown that the medial prefrontal area is also involved in reading emotional cues [Gusnard et al., 2001; Lane et al., 1997], social norm transgression [Berthoz et al., 2002], and finding coherence in stories [Ferstl & von Cramon, 2002]. Another prediction of the modular ToM hypothesis is that since ToM is relatively unchanging throughout the development (after 18 months), children’s neural bases of ToM would be very similar to adults’ [Fodor, 1983; Scholl & Leslie, 1999]. However, evidence from several recent brain imaging studies of developmental neural basis of ToM argues against this prediction. For instance, Liu (2005) found more diffused bilateral frontal activity in children than adults who showed more localized left frontal activity during the animationbased FB task. Similarly, in our study, children activated many more brain regions than adults during the ToM condition relative to the baseline [Kobayashi et al., 2007a]. Moriguchi et al. (2007) has also found an age related positive correlation in the dorsal mPFC area, but a negative correlation in the ventral mPFC area. Moreover, examining neural basis of irony in adults and children, Wang et al. (2006a) found more robust activity in the prefrontal areas in children than in adults, who activated posterior brain regions more during the irony relative to the control conditions. These results seem to argue against the main prediction of the ‘modular’ hypothesis of ToM: ToM is innate and relatively unchanging throughout development, and so, there should be little variation between adults and children in the neural bases of ToM. Thus, the results from the developmental neuroimaging studies of ToM do not seem to lend support to the strict ‘modular’ theory of ToM. Nonetheless, one thing that the ‘modular’ ToM theory (but not others) explains well is the relationship between the executive function (or inhibitory control) and ToM. Although the neuro-functional relationship between the executive function and ToM is still controversial [see Saxe, Carey, and Kanwisher, 2004 and Kain & Perner, 2005 for reviews], several brain imaging studies of ToM (including ours) have implicated the DLPFC (i.e., a brain region often implicated in the neuroimaging studies of executive function) [Baron-Cohen, Ring, et al., 1999; Brunet et al., 2000; Kobayashi, Glover, & Temple, 2006; Sanfey et al., 2003]. Thus, neurological evidence may at least support the domain-general SP part of this hypothesis.

Theory-Theory ToM The ‘Theory-theory’ hypothesis of ToM has been developed as an alternative to the ‘modular ToM’ view, which Wellman and his colleagues describe as being ‘antidevelopmental’ (Wellman, Cross, & Watson, 2001). This hypothesis posits that the adults’ version of ToM, folk psychology, is drastically different from the children’s version because we revise our ToM theory many times throughout life just as a scientist revises his/her theory based upon alternative empirical evidence. They further argue that ToM is essentially a dynamic process: it is subject to revision depending upon our individual experiences [Gopnic and Wellman, 1992]. Although the theory-theorists agree that some innate modules or ‘core’ structures exist for ToM, they contend that these structures are subject to extensive revisions throughout one’s life, whereas the modularists insist that those ‘core’ modules do not change [Wellman, Cross, & Watson, 2001]. According to this theory, the reason why 2-year-olds fail the FB task is that they do not have the concept of ‘belief,’ which develops only sometime after the third birthday when other epistemic concepts (e.g., ‘knowing’ and ‘thinking’) become available along with lexical expansion [Bartsch & Wellman, 1995]. Evidence from a

558

Chiyoko Kobayashi

meta-analysis of more than 100 ToM studies seems to support the ‘theory-theory’ hypothesis: it has failed to show early ToM competence in children younger than 3 years of age, but has shown instead a significant covariation between age and ToM performance. Until recently, there has been no neurological evidence to support or reject this hypothesis, as there has been no brain imaging studies of ToM development. The results of a few recent developmental brain imaging studies seem to support this hypothesis because these studies have found clear age-related differences in several brain regions involved in ToM and related socio-cognitive functions [Liu, 2005; Kobayashi et al. 2007a; Moriguchi et al., 2007; Wang et al., 2006a]. These results support the theory-theorists’ main tenet: some major differences exist between adults’ ToM (a.k.a., folk psychology) and children’s ToM, which develops throughout life. These studies provide some initial evidence for significant changes in the neural basis of ToM and related social cognition between the first 10 years of childhood and adulthood [Liu, 2005; Kobayashi et al., 2007a; Moriguchi et al., 2007; Wang et al., 2006a]. Interestingly, the majority of these studies [Kobayashi et al., 2007a; Moriguchi et al., 2007; Wang et al., 2006a] did not find any behavioral differences between the age groups in the FB/irony task performance. This implies that the changes that may occur in the neural basis underlying ToM throughout development may not directly affect ToM performance but that children’s way of understanding ToM may be different from adults’. In addition, Wang et al. (2006a) suggested that the developmental change reflects increasing automatization of ToM/irony understanding as people age. One notable difference between adults and children in the ToM related brain activity found in our study is that even though both adults and children activated the mPFC, children activated more ventral mPFC areas than adults [Kobayashi et al., 2007b]. Similarly, an event related-potential (ERP) experiment found more activity in ventral prefrontal area in 6 yearsold children during the animation-based FB than the control condition [Liu et al., 2005]. Moreover, in Moriguchi et al’s (2007) study, a significant positive correlation between ToM related dorsal mPFC activity and age was found. In the vmPFC the relationship was reversed: the activity decreased as age progressed. It has been suggested that the dorsal cingulate area is primarily dedicated to cognitive aspects of behaviors while the ventral cingulate area is more dedicated to emotional aspects of behaviors [see Bush, Luu, & Postner, 2000, for a review]. In line with these results, an ERP study [Sabbagh, 2004] found vmPFC/orbito-frontal activity while their subjects encoded others’ emotions from eye gazes, but dorsal mPFC activity when they engaged in the cognition-based standard ToM task. These results suggest that ToM may require more emotional processing for children but more cognitive processing for adults. These results also seem to fit well with the Theory-theorists’ prediction that more emotionladen ‘desire’-based ToM understanding precedes more cognition-laden ‘belief’-based ToM understanding in development [Bartsch & Wellman, 1995].

Simulation ToM A third theory, ‘simulation’ ToM theory, posits that people do not use any naïve theories of psychology when predicting and explaining the behaviors of others [Goldman, 1989; Harris, 1992]. The Simulationists agree that ToM development depends upon conceptual development, but they argue that the concepts are derived from a child’s own direct experience of such states, rather than through some abstract theorizing. When explaining

Language and Thought

559

others’ mental states, the child uses his/her own mental states as a model, very much like putting his/her mind into others’ shoes [Harris, 1991, 1992]. Simulation theory has recently been embraced with enthusiasm by neurologists and cognitive scientists following the discovery of the ‘mirror neuron’ system [Fogassi et al., 2005; Iacoboni, 2005]. Robust activity in the mirror regions has been found in several studies that tested imitation [Decety et al., 1997; Iacoboni, 2005], discrimination of ‘self’ from ‘other’ [Decety & Chaminade, 2004], and reading of others’ intentions [Burgess, Quayle, & Frith, 2001; Iacoboni et al., 2005). For example, significant brain activity in these areas was found when monkeys engaged in a task in which they had to infer an experimenter’s intentions [Fogassi et al., 2005). Several brain imaging studies on human adults also found significant activity in the inferior parietal regions (i.e., a part of the mirror neuron system) while their subjects engaged in imitation tasks [Decety et al., 1997; Chaminade & Decety, 2002; Nakamura et al., 2004]. It has been suggested that these lower-levels of ToM processing (e.g., understanding intentions of others and imitation) form bases for higher-order ToM (e.g., inferring from others’ beliefs) [Meltzoff & Brooks, 2001]. However, whether or not the ‘mirror neuron’ system is involved in higher-order ToM such as FB reasoning is still controversial. The major reasons are that the mirror neuron regions (the inferior parietal lobule [IPL] and inferior frontal gyrus [IFG]) are not commonly implicated as ToM candidate areas, and that the FB task requires subjects to identify reasons for the behavior and to predict the next action. ‘Reasoning’ is more than just ‘decoding’ in that it involves prediction of behavior based on the person’s past mental states in addition to the capacity to attribute current mental states to others [Nichols & Stich, 2003; Sabbagh, 2004]. This reasoning might require some abstract theorizing, not only the detection of intentions through the ‘simulation’. Another criticism of the simulation theory is that simulation theory cannot account for errors that people often make about their own mental state judgment [see Saxe, 2005, for a review]. A recent neuroimaging study that tested social norm violation within the person’s own culture or others’ culture did not support this theory [Saxe & Wexler, 2005]. Subjects employed the right TPJ more when the stories were about norm-violation within their own cultures than when they were about the same norm-violation in other cultures. These results argue against the main prediction of ‘simulation’ theory: other minds are represented fundamentally in terms of their similarity to the person’s own mind. Our study with adults and children, however, found some support for this theory by showing convergent activity in the IPL for both cartoon- and story-based ToM tasks [Kobayashi et al., 2007a]. Similarly, Wang et al. (2006a) found a greater activity in the IFG area in children than in adults. One explanation of these results may be that mirror neuron regions are important for children (more than for adults) because these regions are involved in lower and implicit level of ToM processing. Taken together, these results may indicate that the human mirror neuron system may be involved in lower-level ToM processing (e.g., detection of intentions) that does not require higher-level abstract theorizing or reasoning behind the action. Another point of contention involves the definition of ToM and empathy. While simulation process is often thought of as synonymous with the empathy (which is defined as an ability to identify another person’s emotions and thoughts, and to respond to these with an appropriate emotion [Baron-Cohen, 2003; Wheelwright et al., 2006]), ToM and empathy may be different functionally as well as neuro-psychologically [Singer, 2006]. ToM has been conceptualized as the cognitive component of an empathizing system, the residual component

560

Chiyoko Kobayashi

being affective empathy [Chakrabarti & Baron-Cohen, 2006; Davis, 1994]. It is clear that affective empathy is not required in the type of ToM tasks used in many neuroimaging studies of ToM. It has been suggested that affective empathy and ToM tasks, especially FB tasks, may tap different psychological capacities. Specifically, affective empathy may be more related to the ability to infer and react to the emotions of others whereas ToM may be more narrowly related to reading the intentions and beliefs of others [Singer, 2006]. However, neuroimaging research using paradigms that tapped empathy has found brain activity in the candidate ToM regions (i.e., mPFC and ACC) [Jackson et al., 2006; Jackson et al., 2005; Lamm et al., 2007; Völlm et al., 2006]. These results suggest that there may not be a clear psychological boundary between empathizing and ToM and that at least in some brain regions these two have functional overlaps [see Oberman & Ramachandran, 2007]. Future research will have to investigate further details of these different levels of processing and neural networks involved in each. A final but important point is that there is no direct evidence of the human mirror neurons that are homologous to the machaque brain. While it is easy to assume that humans have the same mirror neurons as the monkeys, there is no direct evidence of human mirror neurons that respond to action [Agnew, Bhakoo, and Puri, 2007]. Thus, the association among human mirror neuron, simulation and ToM remains speculative until further anatomical as well as functional characterizations are done.

Linguistic Determinism of ToM Despite the observation that the FB reasoning seems to rely upon some verbal ability, all of the above theories of ToM (except the ‘theory-theory’ to a limited extent) downplay the contribution of linguistic ability to ToM development. The last of the four main theories of ToM emphasizes this component of ToM and is called ‘linguistic determinism’ [de Villiers, 2000; de Villiers and de Villiers, 2000]. These theorists follow the Whorfian theory of language and cognition, positing that ToM develops as language develops in children. A strong form of this hypothesis proposes that linguistic (primarily syntactic and semantic) ability constrains ToM ability. Main advocates of this hypothesis, de Villiers and her colleagues maintain that a sophisticated command of syntax, or more precisely, complement, is necessary for FB task understanding [de Villiers, 2000]. They argue that 3-year-olds fail the FB task because their syntactic commands are not fully developed to handle the propositions embedded in the sentences of the FB task. De Villiers and her colleagues tested 3-4 years-old children for FB and syntactic ability, and found correlation between the two [de Villiers, 2000]. Moreover, a longitudinal study [Astington & Jenkins, 1999] has found that earlier language (syntax and semantics) ability predicts later ToM performance and not vice versa, indicating that language competence is a prerequisite for competitive ToM performance. In addition, Morgan and Kegl’s (2006) study on Nicaraguan Sign Language (NSL) signers supported the linguistic determinism in that the early signers (who had been exposed to the NSL before the first 10 years of age) did better in the ToM task than late signers (who had been exposed to the NSL after the first 10 years of age). Another source of evidence for the linguistic determinism hypothesis comes from studies on children with Asperger syndrome (AS) and high-functioning autism (HFA). These children are known to have relatively intact language skills (despite being impaired in social skills as much as children with low-

Language and Thought

561

functioning autism) and they both pass the standard first-order FB task somewhat easily compared to children with low-functioning autism [Landa, 2000]. Others have found that the understanding of ‘belief’ terms predicts the performance level of FB tasks in children with autism [Ziatas, Durkin, & Pratt, 1998]. As described above, strong linguistic determinists emphasize the contribution of the command in syntax to ToM. However, there is another version of ‘linguistic determinism’ which emphasizes the contribution of communicative development and the role of sociocultural experience for ToM. This alternative version stresses non-theoretical, direct interpersonal or social knowledge for ToM understanding [Hobson, 1991; Tomasello, 2003]. Although the proponents of this weaker version of ‘linguistic determinism’ are in agreement with the idea that language plays a major role in ToM, they argue that language does not constrain ToM development. Their focus is more on culture-specific socio-communicative aspects than constitutive aspects of language and therefore follows the Vygotskyan tradition more closely. A few studies have supported this second version by demonstrating that performance on a FB task is improved by discussing perspectives on the same objects or events with others without any use of the sentential complements [Lohman & Tomasello, 2003; Harris, 2005]. It has been found that children/adolescents with HFA and AS pass the FB task [BaronCohen, O’Riordan, et al., 1999], presumably being helped by their more advanced language skills than children/adolescents with low functioning autism [Tager-Flusberg, 2000]. But the fact that children/adolescents with HFA/AS show little understanding of Faux Pass sentences [Baron-Cohen, O’Riordan, et al., 1999] and irony (Leekam & Prior, 1994] indicates that the help of syntax for ToM in HFA/AS population is limited to non-inferential reasoning. Given that understanding of ToM requires inferential skill to go beyond the literal meaning and to read communicative intent of others [Sperber & Wilson, 2002], the higher-order ToM task performance in children/adolescents with HFA/AS may be constrained by the pragmatic aspects (i.e., reading communicative intentions and inferential reasoning) of language. Taken together, these results seem to support the second version of linguistic determinism suggesting that language is important for ToM not because of the syntactic aspects but because of the inferential/pragmatic aspects.

Relationship Between Language and Tom Development: Evidence from Neurological Studies of Tom Neurological studies that examined the relationship between neural correlates of ToM and those of language have obtained mixed results similar to what behavioral studies have found. On the one hand, a severe aphasic patient, who had a wide-range of left hemisphere damage, showed intact performance in some nonverbal ToM tasks, despite failing all other syntax-related tasks [Varley & Siegal, 2000]. On the other hand, evidence suggests that processing of pragmatically coherent sentences also recruits the mPFC area primarily [Fersl & von Cramon, 2002]. Ours is the first study to compare the effects of language/culture on ToM development, and our results are consistent with this hypothesis. We found clear cultural/linguistic effects on the neural bases of ToM [Kobayashi et al., 2006; 2007b], and at least the cultural effects had little to do with syntax (as both cultural groups saw exactly the

562

Chiyoko Kobayashi

same cartoons). Thus, both behavioral and brain imaging results seem to support the notion that language is important for ToM not because of its constitutive aspects (i.e. syntax and semantics) but because of its pragmatic aspects.

Do We Use “Language” Regions of the Brain for Tom? Consistent with the second version of ‘linguistic determinism’ of ToM, in several recent developmental studies of ToM and related social/cognitive functions, children employed some of the “language” regions for processing the story-based [Kobayashi et al., 2007a; 2007b; Wang et al., 2006b] and/or the cartoon/animation-based tasks [Dapretto et al., 2006; Kobayashi et al., 2007a; 2007b; Moriguchi et al., 2007; Ohnishi et al., 2004; Wang et al., 2006a]. In addition, a study found some interference on the FB reasoning by verbal shadowing in adults [Newton & de Villiers, 2007]. These results suggest that adults process ToM more verbally than children and ToM develops as “language regions” develop. The classical language regions encompass Broca’s area or Brodmann area (BA) 44/45 in the IFG, Wernicke’s area or BA 22 in the STG, and the angular gyrus and supramarginal gyrus (SMG) [Carter, 1998]. Thanks to new findings from brain imaging studies on various linguistic processing, there is now a broad consensus that syntactic processes are subserved by the left STG and the IFG (specifically, BA 44 and frontal operculum [FO]) in adults [Bornkessel et al., 2005; Friederici et al., 2003; Moro et al., 2001]. Semantic processes, in contrast, are supported by the left middle temporal gyrus (MTG), the SMG, and BA 45/47 in the IFG in adults [Kotz et al., 2002; Poldrack et al., 1999]. Brain imaging studies of language in children or on development are still too scarce to reach a broad agreement on the specialization of those linguistic processes in the different areas of the brain. But increasing evidence suggests that language is less lateralized and involves broader regions in children than in adults [Brauer & Friederici, 2007]. Thus, in this chapter, I consider right hemisphere homologues of the aforementioned brain regions as language regions (in addition to the left hemisphere language regions). In what follows, I will discuss some of the major functions of the language regions, and whether or not ToM studies indicate involvement of those language regions in ToM development.

Inferior Frontal Gyrus Traditionally, IFG or the Broca’s area (see Figure 1) has been thought to be specialized for syntactic processing and speech [see Price, 2000, for a review]. In neuroimaging studies, the anterior portion of the left IFG (BA 47 and FO) has been implicated in semantic processing [see Poldrack et al., 1999, for a review] and the posterior left IFG (BA 44/45) for syntactical processing [Bookheimer, 2002; Dapretto & Bookheimer, 1999] as well as phonological processing, such as phoneme monitoring or rhyme judgments [Poldrack et al., 1999; Roskies et al., 2001; Temple et al., 2003]. In addition, the bilateral IFG (BA 45/47) activity was found when children processed semantics [Brauer & Friederici, 2007; Chou et al., 2006] and syntax [Brauer & Friederici, 2007] in sentential level. In terms of the involvement of the IFG area in the ToM development, several neuroimaging studies in children found activity in the IFG while the child participants

Language and Thought

563

engaged in facial imitation [Dapretto et al., 2006], story- and cartoon-based irony [Wang et al., 2006a; 2006b] and ToM [Kobayashi et al., 2007a; Moriguchi et al., 2007] tasks. We found a three-way interaction in the left IFG (BA 45) for children and adults [Kobayashi, Glover, & Temple, 2007a]. Children employed this area more for the cartoon ToM condition, yet the adults used this area more for the story ToM condition. Similarly, in Wang et al.’s (2006a) study on irony processing, children recruited the left IFG (BA 44 and 45) more than adults. Moriguchi et al. (2007) also found activity in the right IFG (BA 45) when children/ adolescents processed animation-based ToM task. These results may suggest that the IFG is important for ToM processing especially during childhood because of its role as a language center. However, the fact that some of the developmental ToM brain imaging studies found the IFG activity during the nonverbal tasks [Kobayashi, Glover, & Temple, 2007a; Moriguchi et al., 2007] may indicate that IFG is important for ToM development because of its involvement in inhibitory control. Increasing evidence suggests that bilateral IFG is selectively involved in working memory-related response inhibition [Bunge et al., 2002; Garavan et al., 1999; see also Aron, Robbins, & Poldrack, 2004, for a review]. The greater activity in this area may represent a greater effort to inhibit the immediate and more salient (but wrong) responses in the ToM/irony stories/cartoons in children than in adults. Another alternative interpretation is that the IFG plays an important role in processing ToM during childhood because of the human mirror neurons that are localized in this area [Agnew, Bhakoo, & Puri, 2007; Oberman & Ramachandran, 2007; Saxe, 2005]. In line with this conjuncture, several adult brain imaging studies implicated this area in nonverbal imitation [Buccino et al., 2004; Grèzes, Frith, & Passingham, 2004; Heiser et al., 2003] and ToM [Brunet et al., 2000; German et al., 2004; Kobayashi et al., 2007a]. The IFG activity in children during the nonverbal ToM [Kobayashi et al., 2007a; Moriguchi et al., 2007] and related social cognitive tasks [Dapretto et al., 2006; Wang et al., 2006b] may support this hypothesis given that nonverbal tasks do not usually involve language processing. However, since adults especially seem to process nonverbal ToM tasks verbally [Kobayashi et al., 2007a; Newton & de Villiers, 2007] and since the mirror neuron system is closely associated with language system in humans [Gallese, 2007; Oberman & Ramachandran, 2007; Siegal & Varley, 2002], future studies need to examine precisely to how much extent the human mirror neuron system is involved in ToM and language development.

Middle Temporal Gyrus The left MTG has been implicated in processing visually-processed semantic knowledge [Binder et al., 1997; Kotz et al., 2002; Vandenverghe et al., 1996; see also Price, 2000, for a review]. The right homologous area has also been implicated in semantic and lexical analyses [Newman et al., 2001; Sevostianov et al., 2002; Tracy et al., 2003]. Moreover, this region has been suggested to be involved in a process of converting pictures to words [Sevostianov et al., 2002]. In addition, a recent neuroimaging study in children implicated the left MTG in sentential semantic processing [Chou et al., 2006]. Several recent neuroimaging studies of ToM and related social cognition studies in children implicated this area [Kobayashi, 2007; Moriguchi et al., 2007; Onishi et al., 2004; Wang et al., 2006a]. A few studies found bilateral [Onishi et al., 2004] or left MTG activity

564

Chiyoko Kobayashi

[Moriguchi et al., 2007] during animation-based ToM conditions in children/adolescents relative to control conditions. In our study, Japanese children showed more activity in the left MTG during the cartoon ToM condition, yet Japanese adults showed more activity in the same area during the story ToM condition [Kobayashi, 2007]. Similarly, in Wang et al. (2006a), adults used this area more for processing the cartoon-based irony task than children. Thus, children may use the MTG area more for the visual-based ToM and other related social cognition tasks because they try harder to convert the meanings of the cartoons/animations into words and sentences.

Superior Temporal Gyrus The STG at/around Wernicke’s area (see Figure 1) has been originally associated with speech comprehension [Wernicke, 1874]. Sitting anterior to Wernicke’s area, the primary auditory cortex (BA 22) processes various auditory input including music [Koelsch, 2005; Koelsch et al., 2005; Limb et al., 2006] and speech or heard word comprehension [Pekkola et al., 2006; Patel, Bowman, & Rilling, 2006; Rimol, Specht, & Hugdahl, 2006]. There is now a broad consensus that in the adult brain, the STG is primarily involved in syntactic processing [Bornkessel et al., 2005; Friederici et al., 2003; Moro et al., 2001]. In the child brain, however, the STG activity was found for both syntactic and semantic aspects of the language [Brauer & Friederici, 2007]. The STG is not among the ToM regions (see Figure 1; see also Frith and Frith [2003], and Saxe, Carey, and Kanwisher [2004] for reviews). However, a few recent imaging studies have implicated the right STG area for some functions that may be related to ToM: i.e., empathy mapping through facial and hand-gesture imitations [Leslie, Johnson-Frey, & Grafton, 2004] and reading eye-gaze directions [Akiyama et al., 2006]. Also, increasing evidence from neurobiological studies suggests that the STG has reciprocal connections with both the mPFC area and the parahippocampal gyrus [see Price, 2005 for a review]. Onishi et al. (2004) found activity in the bilateral STG area when 7-13 year-old children processed animation-based ToM tasks. In our study, we found story and cartoon task-specific interaction in the left STG between the American adult and child groups: adults used this area more for the cartoon ToM condition, while children used this area more for processing the story ToM condition [Kobayashi, 2007a]. Similarly, Wang et al. (2006a) found a greater activity in the right STG in children while they processed the cartoon-based irony task than in adults. It has been shown that language processing tasks normally recruit the left hemisphere, but processing of pragmatics employs the right hemisphere [Paradis, 1998]. Also, results of a few studies have indicated that processing of story-based FB tasks is like processing pragmatically coherent sentences [Ferstl & von Carmon, 2002; Siegal, Carrington, & Radel, 1996]. Thus, our speculation is that children use this area more than adults for understanding pragmatics in the ToM (and related socio-cognitive) stories more than adults do.

Supramarginal Gyrus and Angular Gyrus The SMG and angular gyrus are adjacent to each other (see Figure 1). Both of these regions have been implicated in some aspects of language processing. Although the 19th

Language and Thought

565

century language model has described these areas as being specialized for visual word form processing, they are now considered as a part of a more general semantic analysis network [see Price, 2000]. For instance, in some silent reading studies these areas became activated when the subjects read meaningful sentences [Bottini et al., 1994; Bavelier et al., 1997]. In children, however, activity in the SMG was found during the syntactic (rather than semantic) violation in the sentences [Brauer & Friederici, 2007]. The TPJ area that has been implicated in the more recent brain imaging studies of ToM (Saxe & Kanwisher, 2003; Saxe & Wexler, 2005; Kobayashi, Glover, & Temple, 2007a] may include these areas if we consider it as a region with 10 or more square-centimeter surface as suggested by Saxe (2006). Most of the ToM neuroimaging studies that have implicated the TPJ area have used story-based tasks [Gallagher et al., 2000; Kobayashi, Glover, & Temple, 2007a; Saxe & Kanwisher, 2003; Saxe & Wexler, 2005]. Therefore, it is possible that the TPJ-SMG/angular gyrus network is recruited for ToM processing especially when the task requires some sentential semantic/syntactic analyses. Alternatively, as the TPJ, these areas may be involved in ‘self’ and ‘other’ distinction. The SMG area, especially the right SMG, has been implicated in anosognosia, asomatognosia, or somatoparaphrenia,that occur after the brain injury [Feinberg, 2000; Berlucchi & Aglioti, 1997]. A hallmark of these psychiatric conditions is the lack of selfawareness and body knowledge [see Decety & Grèzes, 2006, for a review]. Besides our study, Ohnishi et al. (2004) implicated this area in nonverbal ToM in children using a geometric figures task. Since, in their experiment children were asked to indicate what they thought the central figures (e.g., triangles) were doing, their task might have tapped the distinction between ‘self’ (i.e., the central figure) and ‘other’ (i.e., other figure) rather than ToM per se. Thus, the TPJ-SMG/angular gyrus network may also be recruited ToM for the ‘self’ and ‘other’ distinction.

Development of ToM neural Bases Several recent neuroimaging studies of ToM and related social and cognitive function suggested that children’s neural bases of ToM are not entirely the same as adults’ [Kobayashi et al., 2007a; 2007b; Liu, 2005; Moriguchi et al., 2007; Wang et al., 2006]. Moreover, results from several neuroimaging studies have suggested that maturation of the prefrontal cortex continues throughout childhood and late adolescence [Giedd et al., 1999; Gogtay et al., 2004; Sowell et al., 2003]. These results imply that development of ToM (and possibly other sociocognitive functions) is a more gradual process than had been previously thought, and that it continues until even late adolescence. In the remainder of the chapter, I present a developmental model of neural bases of ToM based, in part, on the evidence from those recent developmental ToM brain imaging studies. This model I present here describes a universal developmental trajectory of ToM neural bases.

566

Chiyoko Kobayashi

Limitations of Four Theories of Tom Development and a New Model of Neural Bases of Tom Development As I have described above, several recent developmental neuroimaging studies of ToM suggest that children’s neural bases of ToM are different from adults’ [Kobayashi, Glover, & Temple, 2007a; 2007b; Liu, 2005; Moriguchi et al., 2007; Wang et al., 2006] universally [Kobayashi, 2007]. The modality-specific age-related interactions found in our study are especially interesting. Children activated ToM regions (i.e., bilateral TPJ) more during the cartoon ToM condition, yet adults activated the same regions more during the story ToM condition [Kobayashi et al., 2007a]. These results suggest children’s ToM neural correlates are more tied to a visual modality while adults’ ToM neural correlates are more tied to a verbal modality. These results further imply language may affect ToM throughout development. As described earlier, there are four main hypotheses of ToM development; ‘modular’, ‘linguistic determinism’, ‘simulation’, and ‘theory,-theory’ hypotheses of ToM. Among these, the ‘modular’ and ‘simulation’ theories have been suggested to have support from neurological and neuroscience studies (see Frith and Frith [2003] for the ‘modular’ theory, and Saxe, Carey, and Kanwisher [2004] for the ‘simulation’ theory). The ‘modular’ theory predicts a highly circumscribed brain region that is specialized for ToM. A strong modular theory would predict that this region would not be involved in other tasks, that no other regions would be involved in ToM, there would be no change in this region across development, and that only this region would be damaged in autism. However, as described above, increasing evidence suggests that autism involves multiple and distributed brain regions [Müller, 2007]. In addition, also as described earlier, a few studies of ToM development found differences in ToM specific activity between adults and children, indicating that ToM module (if it exists) changes throughout development. The ‘simulation’ theory predicts that the mirror neuron system is the most important for ToM. The mirror neuron system, however, seems to play only a limited role for ToM (see above). The proponents of the other two developmental theories of ToM have had no neurological evidence to support or refute their theories, primarily because of scarcity of neurological studies of ToM in children. A few neuroimaging studies of ToM and related social and cognitive functions in children begin to address some of the predictions these theories might make about the neural bases of ToM. Our [Kobayashi, Glover, & Temple, 2007a] and several others’ [Liu et al., 2005; Moriguchi et al., 2007; Wang et al., 2006a] results support some aspects of the ‘theory-theory’ hypothesis. The ‘theory-theory’ (in opposition to the modular theory) would predict major differences between adults’ folk psychology and children’s ToM (Wellman, Cross, & Watson, 2001]. Consistent with that prediction, age-related differences in the neural bases of ToM and related social and cognitive functions have been found [Kobayashi, Glover, & Temple, 2007a; 2007b; Moriguchi et al., 2007; Wang et al., 2006a]. The ‘theory-theory’ would also predict that adults’ ToM is based more on a cognition-laden ‘belief’ concept, whereas children’s ToM is based more on an emotion-laden ‘desire’ concept [Bartch & Wellman, 1995]. Results from a few studies have also supported this notion, finding that children recruit the vmPFC more than the dorsal mPFC [Kobayashi, Glover, and Temple, 2007a; Liu et al., 2005; Moriguchi et al., 2007]. As described above, the ventral aspect of the ACC and mPFC has been hypothesized to be

Language and Thought

567

involved in emotional processing and the dorsal aspect in cognitive processing. As described above, recent findings from brain imaging studies in children have also supported aspects of the second version of the ‘linguistic determinism’ hypothesis of ToM, showing differential recruiting of language regions during ToM processing depending on age of the subjects and modality of the task (see Table 1). Thus, I have constructed a neural developmental model of ToM incorporating aspects of both of these hypotheses of ToM development. Although this model has limitations due to the limitations of our study (and of a few other developmental brain imaging studies on ToM and/or related social cognition) and scarcity of pediatric data in the ToM neuroimaging literature overall, it attempts to incorporate the findings that have been presented here. I hypothesize that during childhood (before the age of 12), ToM understanding is more closely linked to understanding of emotions, and that young children employ the vmPFC or the ventral part of PFC to process the highly emotion-laden ToM (Figure 2). At this earlier stage, especially younger children may understand ToM more in terms of the ‘desire’ concept than through the ‘belief’ concept. During late childhood through adolescence, our linguistic ability continues to develop as various language brain regions (e.g., the IFG, MTG, STG, and angular gyrus) mature and language continues to influence ToM development during this time. Finally, in adulthood, ToM is understood in terms of the ‘belief’ concept, and the dorsal aspect of the mPFC is employed to understand the cognition-laden ToM. However, this model is subject to further empirical testing. It has been suggested that the dorsal mPFC activity is associated with various social-cognitions, while the ventral mPFC is associated with self-referential process [Mitchell, Banaji, & Macrae, 2005]. Thus, alternatively, the ventral-emotion versus dorsal-cognition dichotomy is a bi-product of socio-cognitive versus self-referential dichotomy. Future neuroimaging studies that will systematically test both children (especially children of ages between 3 and 5 during when ToM performance dramatically improves) and adults with either internally- or externally-directed ‘desire’- and ‘belief’-related tasks will be useful for examining how the activity in the different brain regions is associated with the development of different concepts related to ToM and the selfreferential process. Further testing of children and adults with various language processing tasks in combination with those tasks will be helpful for examining precisely what aspects of language play the quintessential role for the conceptual development and ToM.

Figure 2.

568

Chiyoko Kobayashi

Conclusion To our knowledge, this is the first review to examine the neural correlates of ToM and other socio-cognitive functions in human development based on the most recent neuroimaging studies in children and adults for these functions. Increasing evidence suggests that neural correlates of ToM in children are different from those in adults and that ToM develops even throughout pre-pubertal childhood. These findings may also argue against ‘modular’ developmental hypothesis of ToM because this hypothesis predicts relatively unchanging innate ToM neural basis. ‘Simulation’ hypothesis of ToM has only a limited support from these recent results because of the limited involvement of the human mirror neuron regions (e.g., the IFG and IPL) in ToM processing. The new evidence also suggests that ToM development involves general (not only syntactic aspects of) language development. Thus, it supports the other two of the four major theories of ToM; ‘theorytheory’ and the weaker version of ‘linguistic determinism’ of ToM. ToM development may be strongly associated with linguistic (especially pragmatics) and conceptual development. These new findings may demand a rethink of ‘modularity’ in ToM (and other cognitive) development. This is not to say that specialized neural networks are not involved in the development of ToM, only that the neural correlates of ToM are neither innate nor informationally encapsulated, but subject to dramatic developmental changes based on the experience. Moreover, they indicate that the development of ToM may go hand-in-hand with language and other socio-cognitive development. These new findings support the Whorfian/Vygotskian hypothesis overall and imply that future neuro-psychological research will be benefited by emphasizing the socio-cultural influence on cognitive development more strongly.

Acknowledgments I thank Dr. Elise Temple, Dr. Barbara C. Lust, and Dr. Michael J. Spivey for discussion. I also thank Frank Randall for assistance.

References Agnew, Z. K., Bhakoo, K. K., & Puri, B. K. (2007). The human mirror system: A motor resonance theory of mind-reading. Brain Research Reviews, 54, 286-293. Akiyama, T., Kato, M., Muramatsu, T., Saito, F., Nakachi, R., & Kashima, H. (2006). A deficit in discriminating gaze direction in a case with right superior temporal gyrus lesion. Neuropsychologia, 44(2),161-170. Aron, A.R., Robbins, T.W., & Poldrack, R.A. (2004). Inhibition and the right inferior frontal cortex. Trends in Cognitive Sciences, 8(4), 170-177. Astington, J.W., & Jenkins, J.M. (1999). A longitudinal study of the relation between language and theory-of-mind development. Developmental Psychology, 35(5), 13111320.

Language and Thought

569

Baron-Cohen, S. (2000). Theory of mind and autism: A fifteen years. In S. Baron-Cohen, H. Tager-Flusberg, & D. Cohen (Eds.), Understanding other minds: Perspective from autism (2nd edition, pp. 3-20). Oxford: Oxford University Press. Baron-Cohen, S., 2003. The essential difference: men, women and the extreme male brain. London: Penguin. Baron-Cohen, S., Leslie, A.M., & Frith, U. (1985). Does the autistic child have a “theory of mind”? Cognition, 21, 37-46. Baron-Cohen, S., Leslie, A.M., & Frith, U. (1986). Mechanical, behavioral and intentional understanding of picture stories in autistic children. British Journal of Developmental Psychology, 4, 113-125. Baron-Cohen, S., O’Riordan, M., Stone, V., Jones, R., & Plaisted, K. (1999). Recognition of Faux Pas by normally developing children and children with Asperger syndrome or highfunctioning autism. Journal of Autism and Developmental Disorders, 29, 407-418. Baron-Cohen, S., Ring, H. A., Wheelwright, S., Bullmore, E. T., Brammer, M. J., Simmons, A., & Williams, S. C. (1999). Social intelligence in the normal and autistic brain: an fMRI study. European Journal of Neuroscience, 11, 1891-1898. Bartsch, K., & Wellman, H. (1995). Children talk about the mind. New York: Oxford University Press. Bavelier, D., Corina, D., Jessard, P., Padmanabhan, S., Prinster, A., Braun, A., et al. (1997). Sentence reading: A functional MRI study at 4 tesla. Journal of Cognitive Neuroscience, 9, 664-686. Berlucci, A., & Aglioti, S. (1997). The body in the brain: Neural bases of corporeal awareness. Trends in Neuroscience, 20, 560-564. Berthoz, S., Armony, J.L., Blair, R.J.R., & Dolan, R.J. (2002). An fMRI study of intentional and unintentional (embarrassing) violation of social norms. Brain, 125, 1696-1708. Binder, J. R., Frost, J. A., Hammeke, T. A., Cox, R. W., Rao, S. M., & Prieto, T. (1997). Human brain language areas identified by functional magnetic resonance imaging. Journal of Neuroscience, 17, 353-362. Bookheimer, S. (2002). Functional MRI of language: new approaches to understanding the cortical organization of semantic processing. Annual Review of Neuroscience, 25, 151188. Bornkessel, I., Zysset, S., Friederici, A. D., von Cramon, D. Y., & Schlesewsky, M. (2005). Who did what to whom? The neural basis of argument hierarchies during language comprehension. Neuroimage, 26, 221-233. Boroditsky, L. (2001). Does language shape thought? Mandarin and English speakers’ conception of time. Cognitive Psychology, 43, 1-22. Bottini, G., Corcoran, R., Sterzi, R. Paulesu, E., Schenone, P., Scarpa, P., et al. (1994). The role of the right hemisphere in the interpretation of figurative aspects of language: A position emission tomography activation study. Brain, 117, 1241-1253. Bowerman, M., & Choi, S. (2001). Shaping meanings for language: Universal and languagespecific in the acquisition of spatial semantic categories. In M. Bowerman, & S. C. Levinson (Eds.), Language acquisition and conceptual development (pp. 475-511). Cambridge: Cambridge University Press. Brambilla, P., Harden, A. Y., di Nemi, S. U., Caverzasi, E., Soares, J. C., Perez, J., & Barale, F. (2004). The functional neuroanatomy of autism. Functional Neurology, 19(1), 9-17.

570

Chiyoko Kobayashi

Brauer, J., & Friederici, A. D. (2007). Functional neural networks of semantic and syntactic processes in the developing brain. Journal of Cognitive Neuroscience, 19(10), 16091623. Brown, P., & Levinson, S. C. (1993). Linguistic and nonlinguistic coding of spatial arrays: Exploration in Mayan cognition. Working Paper 24. Nijmegen: Max Planck Institute for Psycholinguistics, Cognitive Anthropology Research Group. Brunet, E., Sarfati, Y., Hardy-Bayle, M-C., & Decety, J. (2000). A PET investigation of the attribution of intentions with a nonverbal task. Neuroimage, 11, 157-166. Buccino, G., Vogt, S., Ritzl, A., Fink, G. R., Zilles, K., Freund, H. J., & Rizzolatti, G. (2004). Neural circuits underlying imitation learning of hand actions: an event-related fMRI study. Neuron, 42(2), 323-334 Bunge, S. A., Dudukovic, N. M., Thomason, M. E., Vaidya, C. J., & Gabrieli, J. D. E. (2002). Immature frontal lobe contributions to cognitive control in children: evidence from fMRI. Neuron, 33, 301-311. Burgess, P. W., Quayle, A., & Frith, C. D. (2001). Brain regions involved in prospective memory as determined by position emission tomography. Neuropsychologia, 39(6), 545555. Bush, G., Luu, P., & Postner, M. J. (2000). Cognitive and emotional influences in anterior cingulate cortex. Trends in Cognitive Science, 4(6), 215-222. Carter, R. (1998). Mapping the mind. Berkeley: University of California Press. Chakrabarti, B., & Baron-Cohen, S. (2006). Empathizing: neurocognitive developmental mechanisms and individual differences. Progress in Brain Research, 156, 403-417. Chaminade, T., & Decety, J. (2002). Leader or follower? Involvement of the inferior parietal lobule in agency. Neuroreport, 13(15), 1975-1978. Cheney, D.L., & Seyfarth, R.M. (1990). How monkeys see the world: Inside the mind of another species. Chicago: University of Chicago Press. Choi, S., & Bowerman, M. (1991). Learning to express motion events in English and Korean: The influence of language-specific lexicalization patterns. Cognition, 41, 83-121. Chomsky, N. (1980). Rules and representations. Oxford: Basil Blackwell. Chou, T. L., Booth, J. R., Burman, D. D., Bitan, T., Bigio, J. D., Lu, D., et al. (2006). Developmental change in the neural correlates of semantic processing. Neuroimage, 29, 1141-1149. Dapretto, M., & Bookheimer, S. (1999). Form and content: dissociating syntax and semantics in sentence comprehension. Neuron, 24, 427-432. Dapretto, M., Davies, M. S., Pfeifer, J. H., Scott, A. A., Sigman, M., Bookheimer, S. Y., & Iacoboni, M. (2006). Understanding emotions in others: mirror neuron dysfunction in children with autism spectrun disorders. Nature Neuroscience, 9(1), 28-30. Davis, M. H. (1994). Empathy: a social psychological approach. Boulder, CO: Westview Press. Decety, J., & Chaminade, J. (2004). When self represents the other: A new cognitive neuroscience view of psychological intention. Consciousness & Cognition, 12, 577-596. Decety, J., & Grézes, J. (2006). The power of simulation: imaging one’s own and other’s behavior. Brain Research, 1079(1), 4-14. Decety, J., Grézes, J., Costes, N., Perani, D., Jeannerod, M., Procyk, E. et al., (1997). Brain activity during observation of actions. Influence of action content and subject’s strategy. Brain, 120(10), 1763-1777.

Language and Thought

571

Dehaene, S., Spelke, E., Pinel, P., Stanescu, R., & Tsivkin, S. (1999). Source of mathematical thinking: behavioral and brain-imaging evidence. Science, 284, 970-974. De Villiers, J.G. (2000). Language and theory of mind: What are the developmental relationships? In S. Baron-Cohen, H. Tager-Flusberg, & D. Cohen (Eds.), Understanding other minds (2nd Edition, pp. 83-123). Oxford: Oxford University Press. De Villiers, J.G., & de Villiers, P.A. (2000). Linguistic determinism and the understanding of false beliefs. In P. Mitchell, and K.J. Riggs (Eds.) Children’s reasoning and the mind (pp. 191-228). East Sussex, UK: Psychology Press. De Waal, F. (2001). The ape and the sushi master: cultural reflections of a primatologist. New York: Basic Books. Feinberg, T. E. (2000). Altered egos: How the brain creates the self. New York: Oxford University Press. Ferstl, E.C., & von Cramon, D.Y. (2002). What does the frontomedian cortex contribute to language processing: coherence or theory of mind? Neuroimage, 17, 1599-1612. Fletcher, P.C., Happé, F., Frith, U., Baker, S. C., Dolan, R. J., Frackowiak, R. S. J., & Frith, C. D. (1995). Other minds in the brain: A functional imaging study of ‘theory of mind’ in story comprehension. Cognition, 57, 109-128. Fodor, J. A. (1975). The language of thought. New York: Crowell. Fodor, J. A. (1983). The modularity of mind. Cambridge, MA: MIT Press. Fogassi, L., Ferrari, P. F., Gesierich, B., Rozzi, S., Chersi, F., & Rizzolatti, G. (2005). Parietal lobe: From action organization to intention understanding. Science, 308, 662-667. Friederici, A. D., Rüschemeyer, S. A., Hahne, A., & Fiebach, C. J. (2003). The role of left inferior frontal and superior temporal cortex in sentence comprehension: Localizing syntactic and semantic process. Cerebral Cortex, 13, 170-177. Frith, U. (2003). Autism: Explaining the enigma. Malden, MA: Blackwell Publishing. Frith, U., & Frith, C. D. (2003). Development of neurophysiology of mentalizing. Philosophical Transactions of the Royal Society of London, B, Biological Sciences, 358(1431), 459-473. Gallagher, H.L., & Frith, C.D. (2003). Functional imaging of ‘theory of mind.’ Trends in Cognitive Sciences, 7(2), 77-83. Gallagher, H.L., Happé, F., Frunswick, N., Fletcher, P.C., Frith, U., & Frith, C.D. (2000). Reading the mind in cartoons and stories: an fMRI study of ‘theory of mind’ in verbal and nonverbal tasks. Neuropsychologia, 38, 11-21. Gallagher, H.L., Jack, A.I., Roepstorff, A., & Frith, C.D. (2002). Imaging the intentional stance in a competitive game. Neuroimage, 16, 814-821. Gallese, V. (2007). Before and below 'theory of mind': embodied simulation and the neural correlates of social cognition. Philosophical Transactions of the Royal Society of London - Series B: Biological Sciences, 362(1480), 659-669. Garavan, H., Ross, T. J., Murphy, K., Roche, R. A. P., & Steln, E. A. (2002). Dissociable executive functions in the dynamic control of behavior: inhibition, error detection, and correction. Neuroimage, 17, 1820-1829. German, T. P., Niehaus, J. L., Roarty, M. P., Giesbrecht, B., & Miller, M. B. (2004). Neural correlates of detecting pretense: Automatic engagement of the intentional stance under covert conditions. Journal of Cognitive Neuroscience, 16(10), 1805-1817.

572

Chiyoko Kobayashi

Giedd, J. N., Blumenthal, J., Jeffries, N. O., Castellanos, F. X., Liu, H., Zijdenbos, A., et al. (1999). Brain development during childhood and adolescence: a longitudinal MRI study. Nature Neuroscience, 2, 861-863. Goel, V., Grafman, J., Sadato, N., & Hallet, M. (1995). Modeling other minds. Neuroreport, 6(13), 1741-1746. Gogtay, N., Giedd, J. N., Lusk, L., Hayashi, K. M., Greenstein, D., Vaituzis, A. C., et al. (2004). Dynamic mapping of human cortical development during childhood through early adulthood. Proceedings of National Academy of Science, USA, 101, 8174-8179. Goldman, A. (1989). Interpretation psychologized. Mind & Language, 4, 161-185. Gopnik, A., & Wellman, H.M. (1992). Why the child’s theory of mind really is a theory. Mind & Language, 7(1-2), 145-171. Grèzes, J., Frith, C. D., & Passingham, R. E. (2004). Inferring false beliefs from the actions of oneself and others: an fMRI study. Neuroimage, 21, 744-750. Gusnard, D. A., Akbudak, E., Shulman, G. L., & Raichle, M. E. (2001). Medial prefrontal cortex and self-referential mental activity: relation to a default mode of brain function. Proceedings of the National Academy of Sciences, U. S. A. 98(7), 4259-4264. Happé, F.G. (1993). Communicative competence and theory of mind in autism: a test of relevance theory. Cognition, 48, 101-119. Happé, F., Ehler, S., Fletcher, P., Frith, U., Johansson, M., Gillberg, C., Dolan, R., Frackowiak, R., & Frith, C. (1996). ‘Theory of mind’ in the brain: evidence from a PET scan study of Asperger syndrome. Neuroreport, 8, 197-201. Harris, P. L. (1991). The work of the imagination. In A. Whitman (Ed.), Natural theories of mind: The evolution, development and simulation of everyday mindreading (pp. 283304). Oxford: Oxford University Press. Harris, P. L. (1992). From simulation to folk psychology: The case for development. Mind and Language, 7, 120-144. Harris, P. L. (2005). Conversation, pretense and theory of mind. In J. W. Astington, & J. A. Baird (Eds.), Why language matters for theory of mind (pp. 70-83). Oxford: Oxford University Press. Hayes, C.M. (1998). Theory of mind in nonhuman primates. Behavioral and Brain Sciences, 21, 101-134. Heiser, M., Iacoboni, M., Maeda, F., Marcus, J., & Mazziotta, J. C. (2003). The essential role of Broca's area in imitation. European Journal of Neuroscience, 17(5), 1123-1128 Hespos, S. J., & Spelke, E. S. (2004). Conceptual precursors to language. Nature, 430(6998), 453-456. Hobson, R. (1991). Against the theory of “Theory of Mind.” British Journal of Developmental Psychology, 9, 33-51. Hresko, W. P., Reid, D. K., & Hammill, D. D. (1981). The Test of Early Language Development. Austin, TX: Pro-Ed. Iacoboni, M. (2005). Neural mechanisms of imitation. Current Opinion in Neurobiology, 15, 632-637. Jackson, P. L., Brunet, E., Meltzoff, A. N., & Decety, J. (2006). Empathy examined through the neural mechanisms involved in imagining how I feel versus how you feel pain. Neuropsychologia 44(5), 752-761. Jackson, P. L., Meltzoff, A. N., & Decety, J. (2005). How do we perceive the pain of others? A window into the neural processes involved in empathy. Neuroimage 24(3), 771-779.

Language and Thought

573

Kain, W., & Perner, J. (2005). What fMRI can tell us about the ToM-EF connection: falsebeliefs, working memory, and inhibition. In W. Schneider, R. Schumann-Hengsteler, & B. Sodian (Eds.), Young children’s cognitive development (pp. 189-217). Mahwah, NJ: Lawrence Erlbaum. Kobayashi, C. (2007). Language and thought: Cultural and linguistic influence on developmental neural bases of theory of mind. Ph.D Thesis. Cornell University, NY. Kobayashi, C., Glover, G. H., & Temple, E. (2006). Cultural and linguistic influence on neural bases of ‘Theory of Mind’: An fMRI study with Japanese bilinguals. Bran and Language, 98(2), 210-220. Kobayashi, C., Glover, G. H., & Temple, E. (2007a). Children’s and adults’ neural bases of verbal and nonverbal ‘Theory of Mind’. Neuropsychologia, 45, 1522-1532. Kobayashi, C., Glover, G. H., & Temple, E. (2007b). Cultural and linguistic effects on neural bases of ‘Theory of Mind’ in American and Japanese children. Brain Research, 1164, 95107. Koelsch, S. (2005). Neural substrates of processing syntax and semantics in music. Current Opinion in Neurobiology, 15(2), 207-212. Koelsch, S., Fritz, T., Schulze, K., Alsop, D., & Schlaug, G. (2005). Adults and children processing music: an fMRI study. Neuroimage, 25(4), 1068-1076. Kotz, S. A., Cappa, S. F., von Cramon, D. Y., & Friederici, A. D. (2002). Modulation of the lexical-semantic network by auditory semantic priming: An event-related functional MRI study. Neuroimage, 17, 1761-1772. Lamm, C., Batson, C. D., & Decety, J. (2007). The neural substrate of human empathy: effects of perspective-taking and cognitive appraisal. Journal of Cognitive Neuroscience, 19(1), 42-58. Landa, R. (2000). Social language use in Asperger syndrome and high-functioning autism. In A. Klin, F.R. Volkmar, & S.S. Sparrow (Eds.), Asperger syndrome (pp. 125-155). New York: The Guilford Press, pp. 125-155. Lane, R. D., Fink, G. R., Chau, R. M., & Dolan, R. J. (1997). Neural activation during selective attention to subjective emotion responses. Neuroreport, 8, 3969-3972. Limb, C. J., Kemeny, S., Ortigoza, E. B., Rouhani, S., & Braun, A. R. (2006). Left hemispheric lateralization of brain activity during passive rhythm perception in musicians. Anatomical Record. Part A, Discoveries in Molecular, Cellular, and Evolutionary Biology, 288(4), 382-389. Liu, D. (2005). Neural correlates of children’s theory of mind development. Ph.D Thesis. University of Michigan, MI. Liu, D., Sabbagh, M. A., Gehring, W. J., & Wellman, H. M. (2005, April). An ERP study of 6-year-olds’ theory of mind. Poster presented at the 2005 annual meeting of the Cognitive Neuroscience Society, New York City, NY. Lucy, J. A. (1992). Grammatical categories and cognition: A case study of the linguistic relativity hypothesis. Cambridge: Cambridge University Press. Leekam, S. R., & Prior, M. (1994). Can autistic children distinguish lies from jokes? A second look at second-order belief attribution. Journal of Child Psychology & Psychiatry, 35: 901–15. Leslie, A.M. (1992). Autism and the ‘theory of mind’ module. Current Directions in Psychological Science, 1, 18-21.

574

Chiyoko Kobayashi

Leslie, A.M. (1994). Pretending and believing: issues in the theory of ToMM. Cognition, 50, 211-238. Leslie, A. M., Friedman, O., & German, T. P. (2004). Core mechanisms in 'theory of mind'. Trends in Cognitive Sciences, 8(12), 529-533. Leslie, K. R., Johnson-Frey, S. H., & Grafton, S. T. (2004). Functional imaging of face and hand imitation: Towards a motor theory of empathy. Neuroimage. 21(2), 601-607. Liu, D., Sabbagh, M. A., Gehring, W. J., & Wellman, H. M. (2005, April). An ERP study of 6-year-olds’ theory of mind. Poster presented at the 2005 annual meeting of the Cognitive Neuroscience Society, New York City, NY. Lohman, H., & Tomasello, M. (2003). The role of language in the development of false belief understanding: A training study. Child Development, 74, 1130-1144. Martin, R. C. (2003). Language processing: Functional organization and neuroanatomical basis. Annual Review of Psychology, 54, 55-89. McDonough, L., Choi, S., & Mandler, J. M. (2003). Understanding spatial relations: Flexible infants, lexical adults. Cognitive Psychology, 46(3), 229-259. Meltzoff, A. N., & Brooks, R. (2001). “Like Me” as a building block for understanding other minds: Bodily acts, attention, and intention. In B. F. Malle, L. J. Moses, & D. A. Baldwin (Eds.), Intentions and intentionality (pp. 171-191). Cambridge, MA: MIT Press. Mitchell, J. P., Banaji, M. R., & Macrae, C. N. (2005). The link between social cognition and self-referential thought in the medial prefrontal cortex. Journal of Cognitive Neuroscience, 17, 1306-1315. Morgan, G., & Kegl, J. (2006). Nicaraguan Sign Language and Theory of Mind: the issue of critical periods and abilities. Journal of Child Psychology and Psychiatry, 47(8), 811819. Moriguchi, Y., Ohnishi, T., Mori, T., Matsuda, H., & Komaki, G. (2007). Changes of brain activity in the neural substrates for theory of mind during childhood and adolescence. Psychiatry & Clinical Neurosciences, 61, 355-363. Moro, A., Tettamanti, M., Perani, D., Donati, C., Cappa, S. F., & Fazio, F. (2001). Syntax and the brain: Disentangling grammar by selective anomalies. Neuroimage, 13, 110-118. Müller, R. A. (2007). The study of autism as a distributed disorder. Mental Retardation and Developmental Disabilities Research Reviews, 13(1), 85-95. Nakamura, A., Maess, B., Knösche, T. R., Gunter, T. C., Bach, P, & Friederici, A. D. (2004). Cooperation of different neuronal systems during hand sign recognition. Neuoimage, 23, 25-34. Newman, A. J., Pancheva, R., Ozawa, K., Neville, H. J., & Ullman, M. T. (2001). An eventrelated fMRI study of syntactic and semantic violations. Journal of Psycholinguistic Research, 30(3), 339-364. Newton, A. M., & de Villiers, J. G. (2007). Thinking while talking. Psychological Science, 18(7), 574-579. Nichols, S., & Stich, S. P. (2003). Mind reading: an integrated account of pretence, selfwareness, and understanding other minds. Oxford: Oxford University Press. Oberman, L. M., & Ramachandran, V. S. (2007). The simulating social mind: The role of the mirror neuron system and simulation in the social and communicative deficits of Autism Spectrum Disorders. Psychological Bulletin, 133, 310-327. Ohnish, T., Moriguchi, Y., Matsuda, H., Mori, T., Hirakata, M., Imabayashi, E., Hirao, K., Nemoto, K., Kaga, M., Inagaki, M., Yamada, M., & Uno, A. (2004). The neural network

Language and Thought

575

for the mirror system and mentalizing in normally developed children: an fMRI study. Neuroreport, 15(9), 1483-1487. Onishi, K. H., & Baillargeon, R. (2005). Do 15-month-old infants understand false beliefs? Science, 308, 255-258. Paradis, M. (1998). The other side of language: Pragmatic competence. Journal of Neurolinguistics, 11, 1-10. Patel, R. S., Bowman, F. D., & Rilling, J. K. (2006). Determining hierarchical functional networks from auditory stimuli fMRI. Human Brain Mapping, 27(5), 462-470. Pekkola, J., Ojanen, V., Autti, T., Jaaskelainen, I. P., Mottonen, R., & Sams, M. (2006). Attention to visual speech gestures enhances hemodynamic activity in the left planum temporale. Human Brain Mapping, 27(6), 471-477. Perner, J., & Ruffman, T. (2005). Infants’ insight into the mind: How deep? Science, 308, 214-216. Piaget, J. (1962). Play, dreams, and imitation in childhood. New York: Norton. Poldrack, R. A., Wagner, A. D., Prull, M. W. Desmond, J. E., Glover, G. H., & Gabrieli, J. D. (1999). Functional specialization for semantic and phonological processing in the left inferior frontal cortex. Neuroimage, 10, 15-35. Premack, D., & Woodruff, G. (1978). Does the chimpanzee have a theory of mind? Behavioral and Brain Sciences, 1, 515-526. Price, C. J. (2000). The anatomy of language: contribution from functional neuroimaging. Journal of Anatomy, 197, 335-359. Price, J. L. (2005). Free will versus survival: Brain systems that underlie intrinsic constrains on behavior. Journal of comparative neurology, 493, 132-139. Rimol, L. M., Specht, K., & Hugdahl, K. (2006). Controlling for individual differences in fMRI brain activation to tones, syllables, and words. Neuroimage, 30(2), 554-562. Roskies, A. L., Fiez, J. A., Balota, D. A., Raichle, M. E., & Peterson, S. E. (2001). Taskdependent modulation of regions in the left inferior frontal cortex during semantic processing. Journal of Cognitive Neuroscience, 13, 829-843. Roth, D., & Leslie, A. M. (1998). Solving belief problems: towards a task analysis. Cognition, 66, 1-31. Sabbagh, M. A. (2004). Understanding orbitofrontal contributions to theory-of-mind reasoning: Implications for autism. Brain and Cognition, 55, 209-219. Sabbagh, M.A., & Taylor, M. (2000). Neural correlates of theory-of-mind reasoning: an event-related potential study. Psychological Sciences, 11(1), 46-50. Sanfey, A. G., Rilling, J. K., Aronson, J. A., Nystrom, L. E., & Cohen, J. D. (2003). The neural basis of economic decision-making in the ultimatum game. Science, 300, 17551758. Saxe, R. (2005). Against simulation: the argument from error. Trends in Cognitive Science, 9(4), 174-179. Saxe, R., Carey, S., & Kanwisher, N. (2004). Understanding other minds: linking developmental psychology and functional neuroimaging. Annual Review of Psychology, 55, 87-124. Saxe, R., and Kanwisher, N. (2003). People thinking about thinking people: the role of the temporo-parietal junction in “theory of mind.” Neuroimage, 19, 1835-1842. Saxe, R., & Wexler, A. (2005). Making sense of another mind: the role of the right temporoparietal junction. Neuropsychologia, 43, 1391-1399.

576

Chiyoko Kobayashi

Scholl, B. J., & Leslie, A. M. (1999). Modularity, development and ‘theory of mind.’ Mind & Language, 14(1), 131-153. Schultz, R.T., Romanski, L.M., & Tsatsanis, K.D. (2000). Neurofunctional models of autistic disorder and Asperger syndrome: Clues from neuroimaing. In A. Klin, F.R. Volkmar, & S.S. Sparrow. (Eds.), Asperger syndrome (pp. 172-209). New York: Guilford Press. Sevostianov, A., Horwitz, B., Nechaev, V., Williams, R., Fromm S., & Braun A. R. (2002). fMRI study comparing names versus pictures of objects. Human Brain Mapping, 16(3), 168-175. Siegal, M., Carrington, J., & Radel, M. (1996). Theory of mind and pragmatic understanding following right hemisphere damage. Brain and Language, 53, 40-50. Siegal, M., & Varley, R. (2002). Neural systems involved in ‘theory of mind’. Nature Reviews, 3, 463-471. Sinclair, M. (1970). The transition from sensory-motor behavior to symbolic activity. Interchange, 1, 119-126. Singer, T. (2006). The neuronal basis and ontogeny of empathy and mind reading: Review of literature and implications for future research. Neuroscience and Behavioral Reviews, 30, 855-863. Southgate, V., Senju, A., & Csibra, G. (2007). Action anticipation through attribution of false belief by 2-year-olds. Psychological Science, 18(7), 587-592. Sowell, E. R., Peterson, B. S., Thompson, P. M., Welcome, S. E., Henkenius, A. L., & Toga, A. W. (2003). Mapping cortical change across the human life span. Nature Neuroscience, 6: 309–315. Sperber, D., & Wilson, D. (2002). Pragmatics, modularity, and mind-reading. Mind & Language, 17(1-2), 3-23. Surian, L., Caldi, S, & Sperber, D. (2007). Attribution of beliefs by 13-month-old infants. Psychological Science, 18(7), 580-586. Tager-Flusberg, H. (2000). Language and understanding minds: Connection in autism. In S. Baron-Cohen, H. Tager-Flusberg, & D. Cohen (Eds.), Understanding other minds: Perspective from autism (2nd Edition, pp. 124-149). Oxford: Oxford University Press. Temple, E., Deutsch, G. K., Poldrack, R. A., Miller, S. L., Tallal, P., Merzenich, M. M., & Gabrieli, J. D. (2003). Neural deficits in children with dyslexia ameliorated by behavioral remediation: evidence from functional MRI. Proceedings of the National Academy of Sciences, USA, 100(5), 2860-2865. Tomasello, M. (2001). Perceiving intentions and learning words in the second year of life. In M. Bowerman, & S. C. Levinson (Eds.), Language acquisition and conceptual development (pp. 132-158). Cambridge, UK: Cambridge University Press. Tomasello, M. (2003). The key is social cognition. In D. Gentner, & S. Goldin-Meadow (Eds.), Language in mind: Advances in the study of language and thought (pp. 48-57). Cambridge, MA: MIT Press. Tracy, J., Flanders, A., Madi, S., Natale, P., Delvecchio, N., Pyrros, A., & Laskas, J. (2003). The brain's response to incidental intruded words during focal text processing. Neuroimage, 18(1), 117-126. Valsiner, J. (1989). Human development and culture: The social nature of personality and its study. Lexington, MA: Lexington Books. Vandenberghe, R., Nobre, A. C., & Price, C. J. (2002). The response of left temporal cortex to sentences. Journal of Cognitive Neuroscience, 14, 500-560.

Language and Thought

577

Varley, R., & Siegal, M. (2000). Evidence for cognition without grammar from causal reasoning and ‘theory of mind’ in an agrammatic aphasic patient. Current Biology, 10, 723-726. Venkatraman, V., Siong, S. C., Chee, M. W., & Ansari, D. (2006) Effect of language switching on arithmetic: a bilingual FMRI study. Journal of Cognitive Neuroscience, 18(1), 64-74. Völlm, B. A., Taylor, A. N. W., Richardson, P., Corcoran, R., Stirling, J., McKie, S., Deakin, J. F. W., & Elliot, R. (2006). Neural correlates of theory of mind and empathy: a functional magnetic resonance imaging study in a nonverbal task. Neuroimage, 29(1), 9098. Vogeley, K., Bussfeld, P., Newen, A., Herrmann, S., Happé, F., Falkai, P., Maier, W., Shah, N.J., Fink, G.R., & Zilles, K. (2001). Mind reading: Neural mechanisms of theory of mind and self-perspective. Neuroimage, 14, 170-181. Vygotsky, L. S. (1967). Thought and language. In E. Hanfmann, & G. Vakar (Eds. and Trans.). Cambridge, MA: MIT Press. Wang, A. T., Lee, S. S., Sigman, M., & Dapretto, M. (2006a). Developmental changes in the neural basis of interpreting communicative intent. Social, Cognitive and Affective Neuroscience, 1, 107-121. Wang, A. T., Lee, S. S., Sigman, M., & Dapretto, M. (2006b). Neural basis of irony comprehension in hcildren with autism: the role of prosody and context. Brain, 129(4), 932-943. Wellman, H.M., Cross, D.C., & J. Watson. (2001). Meta-analysis of theory of mind development: The truth about false belief. Child Development, 72(3), 655-684. Wernicke, C. (1874). Der aphasiche Symptomenkomplex. Bresiau, Poland: Cohen and Weigert. Wheelwright, S., Baron-Cohen, S., Goldenfeld, N., Delaney, J., Fine, D., Smith, R., Weil, L., & Wakabayashi, A. (2006). Predicting autism spectrum quotient (AQ) from the systemizing quotient-revised (SQ-R) and empathy quotient (EQ). Brain Reseach, 1079, 47-56. Whorf, B.L. (1956). Language, thought, and reality: Selected writing of Benjamin Lee Whorf. (J.B. Carroll, ed.). Cambridge, MA: MIT Press. Wimmer, H., & Perner, J. (1983). Belief about belief: representation and constraining function of wrong beliefs in young children’s understanding of deception. Cognition, 13, 103-128. Ziatas, K., Durkin, K., & Pratt, C. (1998). Belief term development in children with autism, Asperger syndrome, specific language impairment, and normal development: Links to theory of mind development. Journal of Child Psychology and Psychiatry, 39(5), 755763.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 25

IS A STORY THE SAME AS THE SUM OF ITS PARTS? THE DEFERENTIAL EFFECTS OF CONTEXT AND ISOLATED WORD TRAINING ON READING FLUENCY Sandra Lyn Martin-Chang* Concordia University Montreal, Quebec, Canada

Abstract The first priority for novice readers is to understand that the letters they see in written language correspond to the sounds they hear in speech. Once children grasp the letter-tosound pairings afforded to them by the alphabetic principle, they can begin to identify some of the written words that are contained in their spoken vocabularies. At this point, the question becomes whether reading practice should take place in stories (context) or in lists (isolation). Two camps have emerged with opposing views on this topic. On the one hand, investigators have shown increased long-term word recognition when children practice reading words in isolation. In contrast, it has also been suggested that reading words in context can act as a secondary self-teaching mechanism. Proponents of this view argue that the added value of context increases reading ability beyond that achieved by other methods of training. At present, then, the evidence on best training practices (context vs. isolation) is mixed. The studies discussed here begin to unify these two differing perspectives by elucidating the circumstances that mediate the utility of each type of training. In brief, during the five training studies reviewed, context training dominated isolated word training with regards to word acquisition. However, after words were learned, retention over an eight-day interval approached ceiling under both conditions. Finally, fluency transfer (reading words quickly and accurately in new texts or new lists) was maximized when the congruency between training and testing was high; words trained in isolation were read more fluently in novel lists, whereas words trained in context were read more fluently in novel stories. To conclude, the decision of which training method to use cannot be made independently from the task to be done after training has ceased. However, if the goal of *

Author Note: Correspondence should be addressed to Sandra Martin-Chang who is now at, 1455 De Maisonneuve Boulevard West, Montreal, Quebec, Canada, H3G 1M8. Tel: (514) 848-2424 x 8932, fax: (514) 848-4520, email: [email protected]

580

Sandra Lyn Martin-Chang instruction is to have students who excel during training, remember what they have learned, and who use their learning to read new texts fluently, the research plainly suggests that training in context is a more fruitful method of instruction than isolated word training.

Introduction The sight of an infant taking his or her first steps sends parents scrambling for video cameras – and with good reason. Those steps, however unsteady and tenuous, open up avenues for exploration beyond anything the infant could previously imagine. Along a similar vein, hearing a child read his or her first few words is an exciting and awe-inspiring event; although they may be cautious and tentative, those words mark the entrance into literacy. However, unlike motor development, which follows a natural progression from stepping to walking to running, without the aid of formal training reading often fails to progress from word-by-word decoding to fluent reading. Research has shown that even when interventions successfully improve reading accuracy, it is inordinately difficult to “close the fluency gap” once children have fallen behind their peers (Torgesen, Rashotte, & Alexander, 2001). Torgesen et al. attributed this difficulty to the loss of reading experience in the slow and plodding reader. When reading is fluent, it frees children to think about what they are reading without allotting finite resources attending to how they are reading it. Until skill development has reached this level, the benefits associated with reading are often too minimal for children to justify the effort it requires. It is for this reason that non-fluent readers tend to participate in other activities rather than spending their time engaging with print (Stanovich, 1986). It would seem that fluency is a natural precursor of reading enjoyment. As educators and parents, then, our goal is clear: to make the transition from non-reader to fluent reader as seamless as possible. But obvious questions remain: What methods of training increase the chances of achieving fluency? Should words be read in games, grocery lists, recipes, and flashcards (isolation)? Or should every effort be made to get children reading meaningful passages (context) right from the start? This chapter addresses these issues by describing a series of experiments illuminating the role of context in the development of reading fluency.

The Struggle for Fluency On June 22, 2004, the headline of the Toronto Star entitled, “New Help For Reading”, described a plan to quadruple the number of reading specialists in the public school system by 2006 (Boyle, 2004). Education Minister Kennedy explained: “If we don’t get basic literacy by 8 years old and advanced [fluent] literacy by 12, there will be immense struggle ahead for that particular student”. The $150 million-budget increase came in response to the finding that half of Ontario students were failing to meet provincial reading standards (Boyle, 2004). This is not a phenomenon unique to Canada – in 2000, it was reported that a similar number of children were failing to meet grade level standards in the United States (Smith, Borkowski, & Whitman, 2008). The long-term educational prognoses for children who are meeting reading standards are very different from those who are not. The “Matthew Effect” (Stanovich, 1986)

Is a Story the Same as the Sum of Its Parts?

581

plots two learning trajectories resulting from high and low levels of reading proficiency. On the one hand, children who easily grasp the correspondence between print and sound upon entering school are clearly advantaged. They are able to use these skills to set up early sight vocabularies (Ehri, 1999; Share, 1995), which encourage extra-curricular reading activities (Stanovich, 1986). Reading for enjoyment acts to further strengthen core phonological abilities, which in turn, expand sight word vocabularies (Perfetti, 1992). As reading becomes ever more facile, a greater allotment of cognitive resources can be dedicated to text integration and comprehension (laberge & Samuels, 1974; Perfetti & Roth, 1981; Stanovich, 1980). At this point, reading is said to be ‘fluent.’ The culmination of these events results in increasingly skillful readers, reading more advanced material, for greater lengths of time (Stanovich, 1986). This description does not, however, adequately summarize the experience for most poor readers. Children who struggle in the beginning stages of reading often become disenchanted with print, and as a result, receive substantially fewer self-initiated reading experiences than their skilled peers. This carries a double impact for poor readers; it denies them much needed reading practice and excludes them from a rich source of knowledge and vocabulary (Stanovich, 1986). While the discrepancies between good and poor readers at the initial stages of instruction may seem transient, empirical evidence suggests otherwise. As shown by a 10year longitudinal study, the ease of acquiring reading skills in Grade 1 accounted for a significant amount of variance in reading ability in Grade 11. Moreover, the children who learned to read more easily in Grade 1 continued to find greater enjoyment in reading a decade later (Cunningham & Stanovich, 1997). A recent study also found a positive relationship between print exposure and reading rate in adults (Martin-Chang & Gould, in press). University students who read more for pleasure also read faster and had higher comprehension scores than students who read less (as measured by an updated Author Recognition Test and an Activity Preference Questionnaire). One of the consequences of the effortful, halted reading, characteristic of poor readers is that meaningful phrases in texts are processed exceedingly slowly. It is estimated that skilled readers in Grade 4 process as many as 24 idea units per minute; this is contrasted with poor readers who average only eight (Jenkins, Fuchs, van den Broek, Espin, & Deno, 2003). It is perhaps for this reason that reading fluency (the number of words read correctly per minute) serves as a better predictor of reading comprehension than more standard measures, such as question answering skills, recall ability, or performance on cloze tasks (Fuchs, Fuchs, Hosp, & Jenkins, 2001; Jenkins et al., 2003). For instance, Spear-Swerling (2006) found that the best marker of fourth grade comprehension was a measure of oral reading fluency taken in third grade. Accordingly, fluency is recognized as one of the primary goals of reading instruction (Kuhn et al., 2006; Kuhn & Stahl, 2003). The Ontario government, then, is correct in allotting more funds to early reading instruction. However, there is a longstanding debate over which method of instruction carries the most benefit for the beginning reader. For example, researchers who encourage the direct instruction of phonics credit phonological decoding as being the single most influential skill in successful reading (for review see Mody, 2003; Shankweiler et al., 1999; Shankweiler & Fowler, 2004). They argue that children learning an alphabetic language, such as English, do not need to waste resources memorizing word forms or predicting words from context. Children who grasp the alphabetic principle can capitalize on the systematic relations from letters to sounds in order to “see” the way words are pronounced (Ehri, 1992, 1993, 1999,

582

Sandra Lyn Martin-Chang

2004). Moreover, as language continues to shift and develop, children can use letter-sound conversion skills in order to read words that are newly invented (e.g., names for children’s toys such as “Webkinz” or “Bakugan”), or words given new, unusual, spellings (e.g., slang terms such as “phat”). However, for the last several decades the instructional approach endorsed in Canada has focused on contextual reading, pushing the role of phonological skills to the periphery. In fact, according to Goodman (1992), it was Canadians who popularized the term “whole language” in reference to text-based reading programs. Text-based reading strategies orient children toward using cues from pictures, semantics and syntax to aid in word identification. As exemplified by the following statement from a language arts textbook, text-based reading programs are still highly regarded by some Canadian educators. Often it is assumed that phonemic awareness instruction is the most important component of the reading program for young children, put phonology is only one of the four language systems […] while it is important to help children understand how language works, such instruction should not lead to isolated skill instruction, separate from meaningful reading and writing experiences (Tompkins, Bright, Pollard, & Winsor, 2005, p. 93). Indeed, a brief survey of the materials used within the realm of education shows that the definition of ‘reading’ often entirely fails to include word reading accuracy as part of it’s description. In these documents the role of word reading is down played and the focus is placed instead on the “interactive process between the reader and the text” (Progress in International Reading Literacy Study, 2006). For example the International Association for the Evaluation of Educational Achievement (2004) described ‘reading’ as “the ability to […] construct meaning from a variety of texts” (International Association for the Evaluation of Educational Achievement, 2004). Many researchers would agree with this statement, however they would posit that accurate word reading is a fundamental component passage comprehension. For example, Compton, Appleton, and Hosp (2004) explored the text characteristics associated with fluent reading in second grade children. They measured texts in terms of passage readability (i.e., using the Flesch-Kincaid and Spache measures), decodability, high frequency words, multisyllabic words and average sentence length. In contrast to the view that word reading accuracy plays an inferior role in passage reading, Compton et al. Found that reading fluency increased as the number of highly frequent and easily decodable words increased; no such relationship was found with passage readability. Research examining the development of prosodic reading (fluent reading with appropriate intonation and timing) has also highlighted the importance of word reading accuracy. Schwanenflugel et al. (2004) found that the children who read with most advanced prosody while reading passages also had the fastest word decoding skills. They stated that “as children became skilled word decoders, reading prosody took on a culturally normative character” (p. 127). Importantly, these authors also concluded that prosody was less related to comprehension than to word decoding. Based on these findings, it seems misguided for educators to concentrate exclusively on the role of “meaning making” in reading, without acknowledging the important role of word decoding skills. Outside of the Education field, there has been mounting concern over the effectiveness of whole language instruction. Investigators who argue that the main focus of reading instruction should be letter names and sounds disagree with the premise that children should rely on nonalphabetic cues as a primary form of word identification (Ehri, 1992, 1993, 1999; Gough &

Is a Story the Same as the Sum of Its Parts?

583

Wren, 1999; Johnston, 2000; Perfetti, 1992, 1999; Share, 1995; Stanovich & Stanovich, 1999). Code-based strategies of reading encourage the use of letter-sound conversion rules to decode unknown words (Ehri, 1999; Perfetti, 1999; Share, 1995). The conflict over the direct instruction of decoding skills has become one of the main points of contention between proponents of whole language and supporters of a code-based approach to reading instruction. As indicated by its name, the whole language philosophy posits that language learning should not be broken down into smaller component skills. Instead, children should learn how to read in a similar fashion as they learn how to speak, via exposure. It is argued by Goodman (1973) that showing children the systematic correspondences between letters and sounds by using worksheets or by other direct approaches, focuses the child on the print and inadvertently detracts from the meaningfulness of the text as a whole. A second tenet of the whole language tradition is that reading lessons must involve “real” texts. Proponents of whole language propose that learning to read using children’s stories highlight the similarities between the written word and speech (Dalh, 1979; Goodman, 1965, 1967, 1992) which ultimately improves both the motivation and the ability to read.

Reading in Context The idea of learning to read words within connected text has gained some favor, even among investigators who have rejected all other aspects of whole language. For although there is now an impressive database underscoring the importance of phonological skills to reading development, it is still unclear whether contextual reading offers opportunities for learning beyond those experienced from reading words out of context (Nation & Snowling, 1998; Tunmer & Chapman, 1995, 2002). If contextual reading functions as a secondary selfteaching mechanism, it follows that children should be more successful learning to read words in stories compared to in isolation. Dahl (1979) asked a group of children to read passages to a criterion of 100 words per minute, at which point training on a new story began. A second group of children were trained to read 800 words presented individually to a criterion of 60 words per minute. A third group acted as controls and received only classroom instruction over the lengthy training interval. Despite of the extensive amount of single word exposure in Dahl’s study, she reported the greatest gains in accuracy, speed, and comprehension, for children who practiced reading in context. Reading words in isolation accrued no more benefits than regular classroom instruction. Dahl concluded that practicing words in passages was necessary to integrate the skills required for proficient reading. In a more recent investigation, Archer and Bryant (2001) compared learning words in isolation and in sentence contexts. They asked 6 and 7-year-old average readers to read from the Schonell reading list until 8 items were missed. The 8 unknown words for each participant were then divided into 2 groups of 4, half of which were taught within the context of meaningful sentences, while the others were taught via flash cards. Each target word was presented 3 times during training. Archer and Bryant found that children required fewer trials to learn the target words during context training compared to isolated word training. Nation and Snowling (1998) have also reported that reading in context may benefit children who experience atypical difficulty learning to read. For example, many dyslexics, despite having persisting phonological difficulties, achieve levels of functional literacy that approximate those of average readers. Nation and Snowling credit this achievement to the

584

Sandra Lyn Martin-Chang

ability to draw from context to facilitate learning. They observed that children with dyslexia gained the most from reading in context compared to both average readers, and readers with specific comprehension problems. Based on this finding they concluded that dyslexics might be using context as a secondary self-teaching mechanism to enhance overall reading ability. Strengthening the view that context facilitates learning in children with reading disabilities, Jenkins et al. (2003) compared children’s word reading accuracy while reading meaningful passages versus reading randomized word lists. Jenkins et al. reported that “about a third [of reading disabled students] demonstrated impairment in list reading but not in context reading. Apparently, some students with reading disability derived enough facilitation from context to escape the impairment classification for reading accuracy” (Jenkins et al., 2003, p.243). Further, Stanovich has demonstrated that the long-term effects of text reading are not limited to poor readers. Print exposure has been credited with improving several reading related skills in children of all ability, including vocabulary, reading speed, decoding ability, reading comprehension, and general world knowledge (Cunningham & Stanovich, 1997; Martin-Chang & Gould, in press; Stanovich, 1986; Stanovich, West, Cunningham, Cipielewski, & Siddiqui, 1996). This has led some investigators to ask whether the ability to use context could be a source of individual differences in the development of reading skill.

Context Use as a Source of Individual Differences Tunmer and Chapman (1995) examined the relationship between language prediction skill (as measured by performance on a cloze task) and reading ability in approximately 300 children. They found that language prediction skill made an independent contribution to reading ability even when phonological recoding skills were taken into consideration. Yet, Tunmer and Chapman continued to stress the importance of letter-to-sound correspondences, noting that children with emerging phonological skills showed the largest improvements from reading in context. In contrast, children who lacked a basic understanding of the alphabetic principle were unable to combine the graphophonemic cues from the print and the contextual cues from the text. The unique variance in reading ability accounted for by prediction skill led Tunmer and Chapman to conclude that the ability to use context is a source of individual differences in reading. Nation and Snowling (1998) came to a similar conclusion when examining the word reading skills of children with comprehension difficulties. These children have a range of deficits in drawing inferences, integrating sentences, and monitoring metacognitive aspects of comprehension. Children with comprehension difficulties tend to be quick and accurate readers, therefore these comprehension problems are observed in spite of adequate phonological skill. However, when Nation and Snowling divided a group of children matched on phonological skill into those with and without comprehension difficulties, they found differences in word reading ability despite both groups having similar levels of phonological awareness. Poor comprehenders had difficulty reading words that could not be sounded out by regular grapheme-phoneme conversion rules. Nation and Snowling proposed that children with comprehension difficulties are unable to capitalize on the contextual support offered from text. Consequently they show deficits reading words that benefit from being learned in context – such as those that are irregular or infrequent. Like Tunmer and Chapman (1995),

Is a Story the Same as the Sum of Its Parts?

585

Nation and Snowling concluded that the ability to benefit from semantic support is a source of individual differences in reading skill.

Context versus Lists While the concept of learning to read within connected text has gained popularity, very few empirical studies have been dedicated to examining the outcomes associated with contextual training. Before the results from our series of experiments are discussed, however, some important terms must first be defined. There is no question, of course, that context is crucial to many aspects of reading advancement. For example, context is necessary for both learning new word meanings (e.g., discovering that a “hansom” refers to a horse and buggy in the Narnia series), and for deciphering which word meaning (e.g., pig pen vs. writing pen) or pronunciation (e.g., read this, vs. she read) to use in a given sentence. However, there is far less consensus over whether context helps children reliably translate letter strings into spoken words. It is this specific facet of context that remains to be clarified, and thus forms the basis of the experiments described below. Even within this fairly narrow definition of reading, context training could benefit children in three different ways; context could enable children to read more words accurately at the time of reading (acquisition), to remember more words learned during training (retention), or to read those words more fluently when they are encountered in novel situations (transfer).

Experimental Overview The current chapter reviews the results of five training experiments that shared a common research design. When discussing the work below I purposely focus on the commonalities among the five experiments (i.e., the results observed during acquisition, retention, and transfer), rather than on the properties unique to each (i.e., number of participants, number of items in material sets, etc.) in order to highlight the consistent pattern of findings that emerge when the body of work is considered as a whole (more detailed descriptions can be found in Martin-Chang & Levy, 2005, 2006; and in Martin-Chang et al., 2007). The experiments reported here consisted of two phases: a training phase, and a test phase. During the training phase, each child was taught to read different sets of target words under experimentally distinctive conditions. Although the number of actual target words varied between experiments (ranging from a minimum of 10 to a maximum of 85 target words per condition) the number of training trials was held constant so that each target word was viewed 12 times throughout training. Each word appeared twice in the training materials, therefore accuracy was measured throughout phase 1 to gauge the amount of learning taking place over the 6 learning trials (2 words per story x 6 trials = 12 word repetitions in total). In phase 2, the children were tested to determine how many words from each condition could be read in different situations. Depending on the experiment, retention was tested by giving children the same materials used during training (e.g., stories or lists), or the amount of fluency transfer was tested by providing the children with novel materials containing trained

586

Sandra Lyn Martin-Chang

words (e.g., new stories or lists). The findings for accuracy, retention and transfer are discussed below.

Acquisition Traditionally, investigations into the role of context have focused on the number of additional items that are read when children are presented with stories compared to scrambled word lists. Although the exact values vary, it is now generally agreed upon that if word identification is difficult because of characteristics surrounding the reader (e.g., low ability, inexperience, second language learner, etc.), or the text (e.g., too advanced, visually degraded, etc.) performance improves when words are presented in context (e.g., Archer & Bryant, 2001; Biemiller, 1970; Landi, Perfetti, Bolger, Dunlap, & Foorman, 2006; Nicholson, 1991). My colleagues and I were interested in an issue related to this finding, namely we wondered if we could capitalize on the added value of context observed during single reading trials to create optimal situations for word acquisition during training. This question has been addressed by a series of experiments working with good, poor and average readers in Grades 2, 3, and 4 (Martin-Chang & Levy, 2005, 2006; Martin-Chang et al., 2007). As described above, a within-subjects design was employed in each of the experiments so that every child was taught to read two sets of different words; one set was taught in context, the second set was taught in isolation. In the context condition words were incorporated into meaningful stories. The stories we used were not predictable (i.e., Pilot studies using our materials have shown that when cloze procedures were employed, undergraduate students were not able to guess the identity of the target words), therefore children needed to attend to the print in order to arrive at correct word pronunciations. In the isolated word conditions, words were presented individually. In some cases, they were shown on a computer screen, in others they were presented on individual flashcards, but in both cases items in the isolated word condition were presented out of context. One of the methodological considerations involved when comparing context and isolated word training, is that the children who have the most to gain by reading words in context (less skilled readers), generally come across the most difficulty when faced with reading a whole passage. Levy (2001) encountered this problem during an experiment where she trained good and poor readers to read two sets of words, one set was incorporated into a story and the second set was shown within a computer game. Although Levy reported no differential effects of context training for fluency gains (accuracy, reading rate, and comprehension) in novel passages containing trained words, a limitation of the study was that the number of word exposures was not held constant between the two training conditions. Due to the fact that the children could only read a passage once or twice during the same amount of time in which they could read several lists, the items in the isolated word conditions received, on average, double the number of repetitions during training compared to those in the context condition. Building on the findings reported from Levy and her colleagues (Faulkner & Levy, 1999; Levy et al., 1997), Ring, Brown and Black (2008) compared reading improvements resulting from context and isolated word training in a group of 9-year-old students at risk for reading failure. Using a between-subjects design, Black et al. had children read either passages or lists containing target words adhering to specific vowel-consonant patterns (information regarding

Is a Story the Same as the Sum of Its Parts?

587

the target words can be obtained from the Texas Scottish Rite Hospital for Sick Children). Training occurred in small groups (app. 5 students), 3 to 4 days a week for 15-20 minutes a session. At the completion of training, greater gains on standardized tests of reading rate, reading accuracy, and word attack skills were reported in the children who received the isolated word training. These findings fit nicely with the corpus of work showing that training words in isolation improves subsequent reading of passages containing those same words. However, two key methodological points of this study bear consideration. First, in isolated word training the target words were presented to the children in a meaningful passage before they were pulled out of context and trained in isolation. It is possible that reading the words in a story at the onset of training set up a conceptual framework that influenced how the words were interpreted when they were later read in isolation. Second, because the experimenters controlled for training time (15-20 minutes) rather than word repetition, it is possible that the words read in isolation received more training trials than the words presented in context. In order to avoid similar complications in the series of experiments reported here, we controlled for word repetition between the two experimental training conditions rather than reading time. We also equated task difficulty by ensuring that children were reading single words in both conditions. In order to do this, we employed a shared reading paradigm in the context condition where the participants read the target words and the experimenter read the remainder of the story. Word acquisition was measured as the number of target words read correctly during each training trial. As depicted in Figure 1, the first trial of training replicated the well-established contextual facilitation effect in all five experiments (Alexander, 1998; Archer & Bryant, 2001; Kim & Goetz, 1994; Landi et al., 2006, Nation & Snowling, 1998; Nicholson, 1991; Nicholson,Bailey, & McArthur, 1991). Poor readers in third and fourth grade (Experiments 1 and 3) and average readers in second grade (Experiments 2, 4, and 5) were able to read substantially more items when the words were first encountered in context. But would the initial boost in word reading accuracy be carried over the remaining learning trials? Based on our data, the answer seems to be ‘yes’. The initial benefits of reading in context were maintained throughout an extended training program; at the end of 12 word repetitions, children with poor and average abilities were reading reliably more words in context compared to in isolation. This was a very robust finding, as evidenced by the consistent pattern of results observed across all five experiments. It is also especially noteworthy because every child received both methods of training. Regardless of how well the students were reading at the end of isolated word training they were reading statistically more accurately after context training. In sum, the data presented in this chapter suggest that the benefits provided by context – specifically, enabling children to repeatedly pair written words with correct pronunciations, accumulate over time to result in measurable increases in word reading accuracy. In addition to word reading accuracy, reading rate was measured throughout training in 3 of the 5 experiments. As would be expected, children of all ability (good, average, poor) became faster at reading the materials as training progressed. Because the units of measurement varied between the two conditions (individual words were measured in milliseconds in the isolated word condition, story readings were measured in minutes in the context condition), we were not able to comment on whether words were read faster in context or in isolation during training.

588

Sandra Lyn Martin-Chang Ex. 1. Poor readers in Grade 4

0.95 0.85 0.75 Ex 1 Context Ex 1 Isolation

0.65 0.55 0.45 0.35 Trial 1

Trial 2

Trial Trial 3 4

Trial 5

Trial 6

Ex. 2. Average readers in Grade 2

0.95 0.85 0.75 Ex 2 Context Ex 2 Isolation

0.65 0.55 0.45 0.35 Trial 1

Trial 2

Trial Trial 3 4

Trial 5

Trial 6

Ex. 3. Poor readers in Grade 3

0.95 0.85 0.75 Ex 3 Context Ex 3 Isolation

0.65 0.55 0.45 0.35 Trial 1

Figure 1. (Continued).

Trial 2

Trial Trial 4 Trial 5 Trial 6 3

Is a Story the Same as the Sum of Its Parts?

589

Ex. 4. Average readers in Grade 2

0.95 0.85 0.75 Ex 4 Context Ex 4 Isolation

0.65 0.55 0.45 0.35 Trial 1

Trial 2

Trial 3

Trial 4

Trial 5

Trial 6

Ex. 5. Average readers in Grade 2

0.95 0.85 0.75 Ex 5 Context Ex 5 Isolation

0.65 0.55 0.45 0.35 Trial 1

Trial 2

Trial Trial 3 4

Trial 5

Trial 6

Figure 1. Ratio of words acquired as a function of training condition and trial.

Retention Torgesen et al. (2001) outlined five major factors that have the potential to disrupt reading fluency: 1) how well context is used to aid in word identification, 2) how many words in the passage are recognized by sight, 3) how quickly those sight words are read, 4) how quickly unknown words are read, and 5) how quickly word meanings are identified. With regards to using context to aid in word identification, the findings reviewed above have shown that reading words in context increases the number of successful learning trials that occur during training. Poor and average readers in Grades 2, 3, and 4 could name more words presented in context at every trial during training compared to words presented in isolation. Nonetheless, as Torgesen identifies in his second point, the paramount goal of training is to permanently increase the number of words children can recognize by sight. This begs the question of whether words learned in context are remembered as well as words learned in isolation. One of the first studies to address this issue compared the effects of teaching words in isolation to teaching words that were accompanied by a picture. Samuels (1967) found that

590

Sandra Lyn Martin-Chang

children entering first grade were more likely to read items correctly during training if the words were illustrated. However, the inverse pattern of results was reported at test; fewer words from the picture conditions were accurately read once stripped of contextual cues. In a second experiment, Samuels tracked the rate of word acquisition during the reading of illustrated and unillustrated storybooks. Again he observed that poor readers in the isolated word (no-picture) condition were able to read significantly more words on a subsequent word identification task compared to children in the illustrated condition. Singer, Samuels and Spiroff (1973) examined the cumulative effects of pictures and sentences on the word acquisition of children in Grades 1 and 2. Words could be presented in one of four conditions: a) alone, b) in a sentence, c) with a picture, or d) in a sentence that was also accompanied by a picture. As the number of extraneous cues associated with target words increased (pictures and context) so did the number of repetitions required to reach the criterion of two successive error free readings. Once more, learning words in isolation resulted in the best long-term retention rates compared to learning words in the presence of sentences or illustrations. Nemko (1984) also examined whether reading instruction was more profitable in context or in isolation. Using a fully crossed, within subject design, Nemko trained and tested children in isolation and in context. During each condition the words were presented once with feedback (in either isolation or context) and then immediately again without feedback (in isolation or in context). This training/testing paradigm was repeated 6 times. Only 4 words were taught in each condition, therefore no training results were reported. The 6 testing trials were combined to result in a total accuracy score out of 24. Nemko reported the best performance when target words were presented in isolation during both training and test (isolation-isolation = 19.39). Words were learned with similar degrees of accuracy in the remaining three conditions (isolation-context = 15.22, context-isolation = 14.35, contextcontext = 14.75). Thus, once again, learning was maximized in isolation. Additional support comes from Johnston (2000) who trained children in Grade 1 to read words using 3 training methods: a) rereading, b) sentence context, and c) word banks. The amount of contextual support available to the reader systematically declined over the three training conditions. The rereading technique focused solely on the text as a whole and provided the greatest source of contextual constraint. In contrast, moderate contextual support was available in the sentence context treatment. Finally, the word bank technique used personalized sets of flashcards and consequently focused the children at the level of individual words. Johnston reported that all children, regardless of skill level, read more words correctly in the word bank condition, whereas the least amount of learning was recorded in the rereading condition. Thus, it seems that as reading becomes facilitated by context, the amount of attention allotted to the print decreases, resulting in compromised word learning. Consistent with these accounts, Landi and colleagues (Landi et al., 2006) identified a number of words that second graders were unable to read in isolation. The children then read half of the words once in isolation, and the other half once as the final word in a predicable sentence. During the single training trial, Landi et al. noted that children had higher accuracy rates for words presented in context. Yet, when the words were presented in isolation after a 7-day retention period, all of the items were read with similar accuracy. Perhaps more telling, when Landi et al. considered only those words that were read correctly during the practice trial, they found that the retention rates were superior for words that were initially read in

Is a Story the Same as the Sum of Its Parts?

591

isolation. These findings lend support to Samuels (1967) argument, suggesting that while context may result in more accurate on-line recognition, it might also reduce the need to attend to the print which inadvertently results in corresponding reductions in long-term word acquisition. The notion that words learned in context might be more easily forgotten was investigated in Experiments 4 and 5 reported in the present chapter (see Martin-Chang et al., 2007 for more details). During this set of experiments average readers in Grade 2 were first screened to identify two sets of words that could not be read in context. Once again, a within-subjects design was used so that each child learned half of the unknown words within a story while the remaining words were learned via flashcard training. The acquisition curves for these experiments are depicted in Figure 1 (Experiments 4 and 5). Eight days after training finished, the children were presented with the same materials used during training. The retention rates were highly encouraging. We found that once children learned a set of words, either in context or in isolation, very little forgetting took place over the 8-day interval. For example, in Experiments 4 and 5 participants learned to read 97% and 96% of the words presented during context training. After a week delay, they were able to read 98% and 97% of the words respectively. The non-significant increase from phase 1 to phase 2 seemed to suggest that the children were treating the retention task as an additional learning trial. The results were similar for words learned in isolation. At the end of isolated word training children were reading approximately 94% and 91% of the words correctly; once more, retention fell by only 1% and 3% after the 8-day delay (93% and 88% accuracy, respectively). There was no significant change in accuracy over the retention period for words trained in context or isolation in either Experiment 4 or 5. So why then, have others reported a drop in retention following context training and why we were unable to replicate it?

Transfer Fluency affords children the ability to read never-before-seen text in a fashion that is easy, automatic and effortless (Meyer & Felton, 1999). Therefore, it is not enough that children can read words in the same materials used during training, they must also be able to easily access word representations from memory so that unfamiliar texts containing familiar words can be read fluently. In addition to word acquisition and retention, our group has also explored how memorial representations formed under one set of circumstances (i.e., in context or in isolation), are transferred to aid in the reading of novel texts. Specifically, several of our studies have been aimed at determining whether one method of training (either context or isolation) offers advantages for optimal fluency transfer. Some pertinent work in this area has come from the work of Levy and her colleagues (Bourassa, Levy, Dowin, & Casey, 1998; Faulkner & Levy, 1994, 1999; Levy, Abello, & Lysynchuk, 1997; Levy, 2001). For instance, Bourassa et al. (1998) asked poor readers in fourth grade to read a training story 4 times in succession on day 1 of their experiment. On day 2, the training story was read one additional time after which the children were asked to read a transfer story and a control story. Both passages were novel, however the transfer story contained many of the words that had been practiced in the initial training passage, whereas the control story contained very few trained words. Bourassa et al. reported faster and more accurate reading of the transfer passages compared to the control passages. Importantly, they

592

Sandra Lyn Martin-Chang

also found comprehension improvements in the transfer stories, suggesting that training words in one context can lead to heightened understanding of those same words when they are encountered in a new context at a later time. Bourassa et al. (1998) also examined whether training words in context would transfer into the more fluent reading words presented isolation. Similar to the first experiment, poor readers in fourth grade read a training story 4 times in succession on the first day of the investigation. On the second day, the training story was read one additional time, however, at this point the children were asked to read two lists: one list containing trained words, and the second list containing untrained words. Bourassa et al. reported improvements in both reading rate and accuracy when words that were initially trained in context were later read in isolation. The work of Bourassa et al. (1998) substantiated the claim that learning to read words at one level (text or word) transfers to several other levels. However, Bourassa et al. did not directly compare the efficacy of two methods of training. Therefore, they were unable to comment about whether one method of training (either context or isolation) resulted in enhanced fluency transfer. Our work was aimed at addressing this interesting gap in the literature. Our basic research question, then, focused on how well children could read trained words when faced with new materials. As such, the transfer tasks were presented to the participants as new activities. No mention was made of the fact that the new stories or lists contained the same words that had been practiced during training. A methodological challenge associated with measuring fluency transfer in this way is creating materials that contain a high degree of word overlap between the training and testing passages that do not overlap in content. In the context condition, this was achieved by employing a shared reading paradigm during training that required the participants to read only individual target words. The remaining words, which were read by the experimenter, were used to create a semantic framework that was distinct from the transfer passage. For example, one of the training stories revolved around a Halloween party occurring in a third grade classroom. In this story, words such as only, branch, and breeze could be meaningfully incorporated into the story as finishing details describing the atmosphere of the classroom (e.g. “The only sound the kids heard was a branch in the breeze tapping on the window.”). In contrast, the transfer passages were much shorter because they contained only words that had been trained in phase 1 (during transfer the children read the entire story, therefore the passages did not contain any extraneous words read by the experimenter). For example, the transfer passage accompanying The Halloween Story was about two young boys who make discovery in the woods. In The Owl’s Nest the words only, branch and breeze were used to setup the main scene that takes place in the forest. In the isolated word condition, the transfer lists were made “novel” by adding a large number of untrained words at test. In both conditions, fluency transfer was monitored by tracking how accurately and quickly the trained words were read in the novel materials (passages and lists). Furthermore, reading comprehension was also measured in two of the experiments to ensure that the increase in reading speed did not come at the expense of reading for meaning. We found that when children were required to read passages at phase 2, fluency transfer was the highest when training occurred in context at phase 1 (Martin-Chang & Levy, 2005; Martin-Chang et al., 2007). We noted that children read new passages containing trained words more accurately after context training compared to isolated word training. This finding

Is a Story the Same as the Sum of Its Parts?

593

held after a 1-day delay (Martin-Chang & Levy, 2005) and an 8-day delay (Martin-Chang et al., 2007). In addition, both good and poor readers in fourth grade and average readers in second grade read the novel transfer passages faster following context training compared to isolated word training (Martin-Chang & Levy, 2005). Interestingly, we observed the exact opposite findings when participants were asked to read novel lists during phase 2. Namely, good and poor readers in Grade 3 (Martin-Chang & Levy, 2006) and average readers in Grade 2 (Martin-Chang et al., 2007) read more accurately and quickly in isolation, when they had first received training in isolation. Taking the transfer results from all five experiments into consideration, we are left to conclude that our original research question, asking “which method of training leads to universally better fluency transfer”, was overly simplistic. Instead, a more profitable question asks “what are the circumstances that mediate the best method of training for fluency transfer”. It would appear that the most advantageous method of training for optimal fluency transfer is directly dependent on the task employed at test.

Retention or Transfer? It will be recalled that: a) several researchers have reported that words learned in context are more susceptible to being forgotten than words learned in isolation, and that b) the data reported from Experiment 3 failed to replicate this finding. However, even a brief examination of the literature shows that many investigators have failed to differentiate between retention and transfer. For example, the ‘retention tasks’ reported in numerous studies involve reading words in lists at test (c.f. Archer & Bryant, 2001; Johnston, 2000; Landi et al., 2006; Samuels, 1967; Singer et al., 1973). Under these circumstances retention was measured when words had been both practiced and tested in isolation, whereas accuracy transfer was measured for words that had been practiced in context but tested in isolation. This may account for the retention advantage for words both learned and tested in isolation in the studies discussed above. Our findings fit nicely with this interpretation. In our experiments, retention was highest when the tasks presented during training and test were identical. Although retention was somewhat lower when learning had to be transferred to a novel task, advantages remained when the tasks during training and testing were congruent (both in context or both in isolation). In contrast, sharp decreases in retention were noted when the tasks presented during training and testing were incongruent (context to isolation, or isolation to context). Given this finding, it is interesting to note that many standardized reading tests involve reading words in isolation, even though it is widely acknowledged that children are rarely required to read words in isolation in day-to-day life.

594

Sandra Lyn Martin-Chang

Conclusion The importance of instilling young children with early reading success cannot be overestimated. The goal for educators is to make early reading instruction as valuable and rewarding as possible for all children, and to quickly minimize the gap between good and poor readers when difficulties arise (Cunningham & Stanovich, 1997; Stanovich, 1980, 1986; Stanovich et al., 1996). Research has clearly documented the effects of contextual facilitation during on-line word recognition in poor and inexperienced readers (Nation & Snowling, 1998; Nicholson, 1991; Perfetti & Roth, 1981; Stanovich, 1980). Consistent with these reports, all five experiments reviewed here found that poor and average readers were able to name substantially more words on trial 1 in context compared to in isolation. The selfteaching hypothesis (De Jong & Share, 2007; Share, 2004) argues that every time a word is read correctly the child gains the opportunity to acquire word specific orthographic information about that individual word. If words have a greater likelihood of being read correctly in context, it follows that contextual reading should result in greater word learning than isolated word training. Our data support this hypothesis. The benefits associated with contextual reading on trial 1 were maintained over training. Poor and average readers were naming reliably more words at the completion of context training compared to isolated word training in all five experiments. We went on to demonstrate that the benefits of learning words in context does not come at the cost of poor retention; words learned in context were retained just as accurately as words learned in isolation. These results are consistent with the premise that contextual reading offers opportunities for word acquisition beyond those experienced from reading words out of context (Nation & Snowling, 1998; Tunmer & Chapman, 1995, 2002). An examination of the transfer phases reveals a second general trend in the data. Namely, fluency transfer (as defined by reading speed and accuracy) is maximal when there is a high degree of congruency between the tasks employed at training and at test. For example, when the transfer task involves reading words in context, good and poor readers showed equivalent benefits in reading speed when practice takes place in context. In addition, these increases in speed occurred with no loss in reading comprehension. On the other hand, when the transfer task involves reading words in isolation, good and poor readers showed equivalent gains in reading speed after practicing words in isolation. Similarly, words are read more accurately in an unfamiliar passage when they were first trained in context, whereas words are read more accurately in isolation when they are first trained out of context.

Summary of Findings This chapter opened by suggesting that context training could affect the development of reading fluency in three distinct ways. Taken together the results reviewed here show that for acquisition, children reliably experience more success when learning words in context. Both children of poor and average reading skill learned more items, in fewer trials, when the words were incorporated into meaningful passages. With respect to retention, words learned in context and in isolation are remembered equally well after a week long delay, as long as the children were given the same materials at test as they were given during training. The situation becomes slightly more complicated when measuring the transfer of fluency; here,

Is a Story the Same as the Sum of Its Parts?

595

either form of training can offer measurable benefits above the other, depending on the kinds of materials the children will need to read after training has finished.

Educational Implications Reading, like walking, has the ability to grant access to new and unseen places – but only to those who posses adequate skill. Torgesen et al. (2001) suggested that fluency deficits remain resistant to remediation due to the lack of print exposure in the non-fluent reader. When reading is effortful and slow children simply choose to do other things. This notion finds support from the results reported by Martin-Chang and Gould (in press) showing that the slowest readers in university were also the least likely to read for enjoyment. In spite of the heated and discrepant views that are often brought to the table with regard to reading pedagogy, most researchers would agree that the goal of instruction is to provide children with the ability to read stories quickly, accurately, and with comprehension. Training is not an end unto itself; the ultimate goal is to give children the skills and motivation to be able to read for enjoyment. With this in mind, the data presented here suggest that, after children are well versed in the alphabetic principal, training words in context offers benefits over and above teaching words in isolation (Martin-Chang & Levy, 2005, 2006; MartinChang et al., 2007). However, the fact that isolated word training can sometimes result in superior performance (i.e, when children are asked to read out of context at test) serves as an important reminder that there is nothing magical about learning to read words in stories ‘per se’. Instead, it is simply the high degree of congruency between training (school) and test (extracurricular reading) that is expected to improve fluency transfer.

Acknowledgments I wish to thank Megan Ladd and Christopher Martin-Chang for their careful reading this chapter. I would also like to gratefully acknowledge financial support SSHRC for the experiments summarized herein.

References Alexander, J. C. (1998). Reading skill and context facilitation: A classic study revisited. The Journal of Educational Research, 9, 314-318. Archer, N., & Bryant, P. (2001). Investigating the role of content in learning to read: A direct test of Goodman’s model. British Journal of Psychology, 92, 579-591. Biemiller, A. (1970). The development of the use of graphic and contextual information as children learn to read. Reading Research Quarterly, 11, 75-96. Bourassa, D., Levy, B. A., Dowin, S., & Casey A. (1998). Transfer effects across contextual and linguistic boundaries: Evidence from poor readers. Journal of Experimental Child Psychology, 71, 45-61. Boyle, T. (2004, June 22). New help for reading. Toronto Star, pp. A1, A19.

596

Sandra Lyn Martin-Chang

Compton, D., Appleton, A., Hosp, M. (2004). Exploring the relationship between textleveling systems and reading accuracy and fluency in second-grade students who are average and poor decoders. Learning Disabilities Research, 19, 176-184. Cunningham, A. E., & Stanovich, K. E. (1997). Early reading acquisition and its relation to reading experience and ability 10 years later. Developmental Psychology, 33, 934-945. Dahl, P. R. (1979). An experimental program for teaching high speed word recognition and comprehension skills. In J.E. Button, T.C. Lovitt, & T. D. Rowland (Eds.), Communications research in learning disabilities and mental retardation (pp. 33-65). Baltimore, MD: University Park Press. De Jong, P., Share, D. (2007). Orthographic learning during oral and silent reading. Scientific Studies of Reading, 11, 55-71. Ehri, L. (1992). Reconceptualizing the development of sight word reading and its relationship to recoding. In P. Gough, L. Ehri, & R. Traiman (Eds.), Reading acquisition (pp. 107143). Hillsdale, NJ: Lawrence Erlbaum Association. Ehri, L. (1993). How English orthography influences phonological knowledge as children learn to read and spell. In R.J. Scholes (Ed.), Literacy and language analysis (pp. 21-43). Hillsdale, NJ: Lawrence Erlbaum Association. Ehri, L. (1999). Phases of development in learning to read words. In J. Oakhill & R. Beard (Eds.). Reading development and the teaching of reading (pp. 97-108). Oxford, England: Blackwell. Ehri, L. (2004, June). Learning to read words: Theory, findings and issues. Paper presented at the 11th Annual meeting for the Society of Scientific Study of Reading, Amsterdam, Netherlands. Faulkner, H. J., & Levy, B.A. (1994). How text difficulty and reader skill interact to produce differential reliance on word and content overlap in reading transfer. Journal of Experimental Child Psychology, 58, 1-24. Faulkner, H. J. & Levy, B.A. (1999). Fluent and nonfluent forms of transfer in reading: Words and their message. Psychonomic Bulletin and Review, 6, 111-116. Fuchs, L. S., Fuchs, D., Hosp, M. K., & Jenkins, J. R. (2001). Oral reading fluency as an indicator of reading competence: A theoretical, empirical, and historical analysis. Scientific Studies of Reading, 5, 239-256. Goodman, K. S. (1965). A linguistic study of cues and miscues in reading. Elementary English, 42, 639-643. Goodman, K. S. (1967). Reading: A psychological guessing game. Journal of the Reading Specialist, 6, 126-135. Goodman, K. S. (1973). The 13th easy way to make learning to read difficult: A reaction to Gleitman and Rozin. Reading Research Quarterly, 8, 484-493. Goodman, K.S. (1992). I didn’t found whole language. The Reading Teacher, 46, 188-199. Gough, P. B., & Wren, S. (1999). Constructing meaning: The role of decoding. In J. Oakhill & R. Beard (Eds.), Reading development and the teaching of reading (pp. 59-78). Oxford, England: Blackwell. International Association for the Evaluation of Educational Achievement. (2004). Retrieved June 16, 2008, from http://www.iea.nl/. Jenkins, J. R., Fuchs, L. S., van den Broek, P., Espin, C., & Deno, S.L. (2003). Sources of individual differences in reading comprehension and reading fluency. Journal of Educational Psychology, 95, 719-729.

Is a Story the Same as the Sum of Its Parts?

597

Johnston, F. R. (2000). Word learning in predictable text. Journal of Educational Psychology, 92, 248-255. Kim, Y.H., & Goetz, E.T. (1994). Context effects on word recognition and reading comprehension of poor and good readers: A test of the interaction-compensatory hypothesis. Reading Research Quarterly, 29, 178-188. Kuhn, M.R., & Stahl, S. A. (2003). Fluency: A review of developmental and remedial practices. Journal of Educational Psychology, 95, 3-21. Kuhn, M.R., Schwanenflugel, P.J., Morris, R.D., Mandel Morrow, L., Gee Woo, D., Meisinger, E.B., Sevcik, R.A., Bradley, B., Stahl, S.A. (2006). Teaching children to become fluent and automatic readers. Journal of Literacy Research, 38, 357-387. LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6, 293-323. Landi, N., Perfetti, C.A., Bolger, D., Dunlap, S., & Foorman, B. (2006). The role of discourse context in developing word form representations: A paradoxical relation between reading and learning. Journal of Experimental Child Psychology, 94, 114-133. Levy, B.A. (2001). Moving the Bottom. In M. Wolf (Ed.), Dyslexia, fluency, and the brain (pp. 357-379). Timonium, MD: York Press. Levy, B.A., Abello, B., & Lysynchuk, L. (1997). Transfer from word training to reading in context: Gains in reading fluency and comprehension. Learning Disability Quarterly, 20, 173–188. Martin-Chang, S. L., & Levy, B.A. (2005). Fluency transfer: Differential gains in reading speed and accuracy following isolated word and context training. Reading and Writing, 18, 343-376. Martin-Chang, S. L., & Levy, B.A. (2006). Word reading fluency: A transfer appropriate processing account of fluency transfer. Reading and Writing, 19, 517-542. Martin-Chang, S. L., Levy, B. A., & O’Neil, S. (2007) Word acquisition, retention, and transfer: Findings from contextual and isolated word training. Journal of Experimental Child Psychology, 96, 37-56. Martin-Chang, S. L., & Gould, O. N. (in press). Revisiting print exposure: Exploring differential links to vocabulary, comprehension and reading rate. Journal of Research in Reading. Meyer, M. S., & Felton, R. H. (1999). Repeated reading to enhance fluency: Old approaches and new directions. Annals of Dyslexia, 49, 283-306. Mody, M. (2003). Phonological basis in reading disability: A review and analysis of the evidence. Reading and Writing: An Interdisciplinary Journal, 16, 21-39. Nation, K., & Snowling, M. J. (1998). Individual differences in contextual facilitation: Evidence from dyslexia and poor reading comprehension. Child Development, 69, 9961101. Nemko, B. (1984). Context versus isolation: Another look at beginning readers. Reading Research Quarterly, 19, 461-467. Nicholson, T. (1991). Do children read words better in context or lists? A classic study revisited. Journal of Educational Psychology, 83, 444-450. Nicholson, T., Bailey, J., & McArthur, J. (1991). Context cues in reading: The gap between research and popular opinion. Reading, Writing, and Learning Disabilities, 7, 33-41.

598

Sandra Lyn Martin-Chang

Perfetti, C.A. (1992). The representation problem in reading acquisition. In P. Gough, L. Ehri, & R. Trieman (Eds.), Reading acquisition (pp. 147-174). Hillsdale, NJ: Lawrence Erlbaum Associates. Perfetti, C.A. (1999). Cognitive research and the misconceptions of reading education. In J. Oakhill & R. Beard (Eds.), Reading development and the teaching of reading (pp. 42-58). Oxford, England: Blackwell. Perfetti, C.A., & Roth, S. F. (1981). Some of the interactive processes in reading and their role in reading skill. In A.M. Lesgold, & C.A. Perfetti (Eds.), Interactive processing in reading (pp. 269-297). Hillsdale, NJ: Lawrence Erlbaum Associates. Progress in International Reading Literacy Study. (2006). An international perspective on fostering reading development. Retrieved June 16, 2008, from http://timss.bc.edu/ pirls2006/index.html. Ring, J., Brown, S., Black, J. (2008). Reading fluency instruction for students at risk for reading failure: Effects of two approaches to improving reading rate. Presentation at the Society for the Scientific Study of Reading, Asheville, NC. Samuels, S. J. (1967). Attentional progress in reading: The effect of pictures on the acquisition of reading responses. Journal of Educational Psychology, 58, 337342. Schwanenflugel, P. J., Hamilton, A. M., Kuhn, M. R., Wisenbaker, J. & Stahl, S. A. (2004). Becoming a fluent reader: Reading skill and prosodic features in the oral reading of young readers. Journal of Educational Psychology, 96, 119-129. Shankweiler, D., & Fowler, A. (2004). Questions people ask about the role of phonological processing in learning to read. Reading and Writing: An Interdisciplinary Journal, 17, 483-515. Shankweiler, D., Lundquist, E., Katz, L., Stuebing, K. K., Fletcher, J. M., Brady, S., et al. (1999). Comprehension and decoding: Patterns of association in children with reading difficulties. Scientific Studies of Reading, 3, 69-94. Share, D. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition, 55¸151-218. Share, D. (2004). Orthographic learning at a glance: On the time course and developmental onset of self-teaching. Journal of Experimental Child Psychology, 87, 267-298. Singer, H., Samuels, S. J., & Spiroff, J. (1973). The effect of pictures and contextual conditions on learning responses to printed words. Reading Research Quarterly, 9, 555-567. Smith, L.E., Borkowski, J.G., Whitman, T.L. (2008). From Reading Readiness to Reading Competence: The Role of Self-Regulation in At-Risk Children. Scientific Studies of Reading, 12, 131 – 152. Spear-Swerling, L. (2006). Children’s reading comprehension and oral reading fluency in easy text. Reading and Writing, 19, 199-220. Stanovich, K. E. (1980). Toward an interactive-compensatory model of individual differences in the development of reading fluency. Reading Research Quarterly, 16, 32-71. Stanovich, K. E. (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 21, 360-407. Stanovich, K. E., & Stanovich, P. J. (1999). How research might inform the debate about early reading acquisition. In J. Oakhill & R. Beard (Eds.), Reading development and the teaching of reading (pp. 12-41). Oxford, England: Blackwell.

Is a Story the Same as the Sum of Its Parts?

599

Stanovich, K.,E., West, R.F., Cunningham, A.E., Cipielewski, J., & Siddiqui, S. (1996). The role of inadequate print exposure as a determinant of reading comprehension problems. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties (pp.15-32). Lawrence Erlbaum Associates: New Jersey. Tomkins, G. E., Bright, R. M., Pollard, M. J., & Winsor, P. J. T. (2005). Language arts: Content and teaching strategies. Toronto, ON: Pearson Education Canada. Torgesen, J. K., Rashotte, C. A., & Alexander, A.W. (2001). Principles of fluency instruction in reading: Relationships with established empirical outcomes. In M. Wolf (Ed.), Dyslexia, fluency, and the brain (pp. 357-379). Timonium, MD: York Press. Tunmer, W.E., & Chapman, J.W. (1995). Context use in early reading development: Premature exclusion of a source of individual differences? Issues in Education, 1, 97100. Tunmer, W.E., & Chapman, J.W. (2002). The relation of beginning readers’ reported word identification strategies to reading achievement, reading-related skills, and academic selfperceptions. Reading and Writing: An Interdisciplinary Journal, 15, 341-358.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 26

CHILDREN’S LANGUAGE AND LITERACY LEARNING DURING A SCIENCE CONCEPT INQUIRY Ni Chang Indiana University, South Bend, Indiana, US

Introduction Using children’s literature to assist in science inquiry and in knowledge building in other subject areas has been on the rise due to the benefits of supporting children’s language and literacy learning. However, published research has reported a substantial variation in the methods of utilizing children’s literature to support learning and language development. A deficiency in the association between reading and writing, and in the amount of time children are given to write has also been reported. “Writing was not linked to reading, and students seemed to spend relatively little time composing” (Allington, Guice, Li, Michelson, & Baker, 1995). In some classrooms in the United States, science is separately taught and given relatively less time in comparison with English/Language Arts (Royce & Wiley, 2005). In the real world, however, when it comes to learning science concepts, children do not often separate science from other subject areas that assist them in the inquiry, because the human brain naturally makes meaningful connections when new information is presented (Bredekamp & Copple, 1997). “One of the most important constants throughout human development is that all domains of development are interrelated. Development in one dimension influences and is influenced by development in other realms” (Bredekam & Copple, 1997, p. 144). Furthermore, Allington, et al also found “. . . hardly any art or drama activities linked to the reading children were doing in any of the classrooms” (p. 10). Therefore, the purpose of this study was to examine what might provide a scaffolding to a child’s language and literacy competencies when reading was closely connected to writing and when art was linked to reading in the process of acquiring a science concept. The underlying research question is “What does an integrated curriculum provide to facilitate and how does an integrated curricular approach mediate children’s language and literacy competencies when they are engaged in acquiring a science concept?”

602

Ni Chang

Theoretical Framework Science Inquiry through Integrated Curriculum Children advance science understanding through various means, such as exploration, discovery, cooperation, collaboration, experimentation, and reading. Literature affords children ample opportunity not only to become familiar with science concepts, but also to gain language and literacy competencies if books are meticulously and thoughtfully selected and if adult-child reading interactions are appropriately designed (Chang, 2007). Curricular integration is at the heart of linking science with children’s literature, which allows children to focus on scientific and literary elements at the same time (Cho & Kim, 1998). “Literature is not only an area of language development, but the center of integration for the entire curriculum through a whole language approach” (Cho & Kim, 1998, p. 3).

Literature-Based Science This literature-based science approach to teaching signifies an integrated curriculum that falls into what Huck (1992) classified a comprehensive literature program. In comprehensive literature program, books permeate every area of the curriculum. Teachers read aloud to children and encourage them to respond to books through discussion, drama, art, and writing (Machado, 2007). As children learn science, they use the language arts skills, such as listening, speaking, reading, writing, and viewing to comprehend science concepts under discussion. Gambrell and Mazzoni (1999) reasoned that this comprehensive approach enables teachers to interlock various activities in a logical fashion, making activities meaningful to young children. These activities provide children not only with book skills, but also with a sense of lifelong literacy learning. Literature-based science may help generate meaningful language learning activities that enable children not only to learn science through language, but also to learn language and to learn about language, which is supported by Halliday (1989 in the International Reading Association (IRA) & National Council of Teachers of English (NCTE), 2006-2007). Learning through language refers to the fact that children’s learning takes place through the use of language. Learning language occurs through listening, speaking, reading, writing, and viewing. Learning about language implies that children develop an understanding of how language works. When children are situated in a learning environment in which these three aspects operate together, the learning context is meaningful and supportive for literacy development and learning makes sense to young children. Sense-making learning also comes from children’s books. Children’s picture books afford children plenty of exciting and rich scientific experiences and encourage inquisitive minds. It is believed that linking science with children’s books helps improve children’s attitudes toward science, increase their interest in science, and promote their confidence in scientific abilities. In addition, “Connecting science with children’s books has a potential to provide children with a richer, more realistic, and broader view of the nature and scope of science” (Cho & Kim, 1998, p. 5). In such a meaningful context for learning science, children are able to identify science with their life experience in a familiar setting (Cho & Kim, 1998).

Children’s Language and Literacy Learning during a Science …

603

Learning becomes easy to children as book content is related and interesting to them and allows them to describe what they know orally and visually/pictorially.

Reggio Emilia Approach to Science Inquiry Art, such as visual/pictorial representations or symbolic expression in Reggio schools, is not at all a separate entity of curriculum, but an integrated part of the learning process, through which children are able to freely and comfortably express themselves. The Reggio Emilia approach to early childhood education advocates a curriculum that brings unity and continuity to teaching and learning. An analysis of science education research literature from an early childhood perspective enables Stegelin (2003) to conclude that the Reggio Emilia Approach can exceptionally and uniquely promote conceptual change within the constructivist paradigm. Reggio educators provide a scaffold for children’s learning of concepts by engaging them in meaningful dialogues and by offering them with occasions for discovery. Learning, from Reggio educators’ point of view, is a complex and upward spiral rather than a linear learning process (Malaguzzi, 1993). Children govern their own rate of learning as they develop a high level of curiosity with which they are likely to develop an ability to ask questions and a willingness to discover on their own. Learning moves back and forth, but in an advanced fashion. This notion is also construed through an understanding that children are intelligent, powerful, competent, capable, creative, curious, and elastic (Rinaldi, 1998). These attributes of young children are sufficient enough to make a persuasive conclusion: children are natural science researchers (Stegeline, 2003). The U.S. science education goal is to promote self-directed thinkers and to challenge students to master science concepts necessary for helping them to build a solid knowledge base so they can understand the world around them (Stegelin, 2003). Additionally, the National Science Education Standards (1996) specified what science concepts that students, at varying developmental and grade levels, should know and understand (National Research Council, 1996). Since the Reggio Emilia Approach is intended to encourage young children to explore, discuss, and reflect upon key concepts, the inclusion of Reggio Emilia into early childhood science education is likely to make learning science concepts a much more stimulating, encouraging, and unforgettable experience (Stegelin, 2003). Reggio educators affirm that visual documentation is critical to the learning and assessing processes (Gandini, 2002) and, hence, visually documenting children’s learning process and its outcomes in various formats is essential to teaching and learning (Edwards, Gandini, & Forman, 1998; Stegelin, 2003). Reggio educators place more emphasis on a learning process rather than on a product. Children’s learning processes can supply teachers with information about “how children arrive at specific conclusions, how they scaffold from one level of understanding to the next and how similar or different children’s cognitive processes are at different levels of development (Fu, Stremmel, & Hill, 2003 in Stegelin, 2003, p. 165). Documentation thus should focus more on children, on actual ongoing learning experiences, and on children’s thoughts, perceptions, and ideas (Katz & Chard, 1996) to reveal varying levels of children’s understanding of certain concepts (Chang, 2007). This practice also works as a qualitative tool for assessment (Chang, 2007), because it assists teachers in planning for “how to provoke and encourage children’s learning to the next and more complex level of understanding” (Stegelin, 2003, p. 165).

604

Ni Chang

Since its beginning, curriculum is designed based on children’s interests, prior knowledge, and/or life experiences, which work as a guiding torch throughout children’s learning experiences. Curriculum is not designed in advance, but emergent based on children’s interests and needs. Emergent curriculum usually is reflected through inquirybased projects, from which general educational goals are set forth. During the process of its implementation, possible outcomes as a result of pedagogical decisions may occur consistent with perceived needs and young children’s levels of learning, which result from their oral expressions as well as from expressions by the use of visual/pictorial language (Chang, 2007; Edwards, Gandini, & Forman, 1998).

From Pictorial Representation to Visual Literacy Drawing is something that naturally attracts most young children. According to Sulzby (1996), most kindergarteners draw and scribble as they write to express themselves (in Machado, 2007). Given paper and drawing tools, children leave marks to create symbols, representing their observations, perceptions, wishes, ideas, understandings, knowledge, impressions, and feelings (Chang, 1996; Machado, 2007). Durkin (1966) found that scribbling and drawing played a major role in forming a starting point of children’s curiosity about written language. The development of literacy should begin prior to children’s formal schooling, which can be partially achieved through activities that allow children to gain awareness of “associating spoken words with their graphic representations” (International Reading Association (IRA) & National Council of Teachers of English (NCTE), 2006-2007, n.d.). One such activity is talking about drawings. Some of the symbols created by young children may be readily recognizable to adults while some others require a young author’s interpretation or explanation to unveil the represented meaning (Chang, 2007), as these symbols are “unique and represent the word in the child’s own way” (Machado, 2007, p. 491). Therefore, talking about drawing is essential and important (Chang, 2007). When an adult pays attention to children’s graphic language and accept children’s comments and ideas, children learn their thoughts are valued (McCormick, 1988), which may propel youngsters’ continued effort to share their inner world (Machado, 2007). Dialogical interactions are also opportunities for children to recognize that created symbols can be read, spoken, or talked about and, at the same time, to see their thoughts can be written for expression and communication (Machado, 2007; McCormick, 1988). Young children’s creative writing ability has been strongly encouraged by many educators and researchers, such as Stauffer (1970), because it is basic to reading and a precursor to writing (Oken-Wright, 1998). Standards set forth by the IRA and NCTE (20062007) expect teachers to develop “curriculum and instruction that make productive use of the emerging literacy abilities.” Viewing or visual representations integrated into curriculum or inquiry-based projects can be powerful in assisting children in acquiring a science concept, as all these multi-faceted means, including reading aloud, speaking, thinking, and writing/drawing simultaneously come into play, aiding the inquiry. It is a meaningful learning process, providing ample room for children to “adjust their use of spoken, written, and visual language to communicate effectively with a variety of audiences and for different purposes”

Children’s Language and Literacy Learning during a Science …

605

(IRA & NCTE standard #4) and “. . .to accomplish their own purposes (e.g., for learning, . . . and the exchange of information” (IRA & NCTE standard #12). Visual literacy or viewing has been commonly recognized by the IRA as well as many researchers and experts in the early childhood literacy field. Through visual images, considered as a language, a child’s perception of the world is made known (Weismann, 1970). Visual literacy represents one’s ability to perceive, produce visual messages, and to think visually about images (Machado, 2007). Since “the primary literacy of the twenty-first century will be visual” (Machado, 2007, p. 165), children should have experiences that allow them to shift from word to illustration and vise versa. Visual literacy is a critical skill that can assist children in their learning across different educational disciplines, such as math, science, art, and language. It improves children’s cognitive, reading, writing, and creative skills. The needed skills also include attending behaviors, discrimination, identification, and analysis (Machado, 2007). Visual thinking, visual arts, visual intelligence, and visual awareness all depend on one’s ability to note various characteristics, including line, shape, color, texture, number, and various dimensions (Machado, 2007). Stieglitz (1972) believed that the sense of sight played a crucial role as a basic source of information about one’s immediate environment, because it not only involves the eye, but also the brain. When encountering something visual that either is of one’s own interest or relates to one’s past experiences, one has to make a note of it based on previous knowledge and ideas and to link it with feelings and emotions. Eye receptors note the perceived object or event with its unique characteristics relating to an individual’s purpose in noting this particularity. If children are encouraged to note down their learning through graphic representations, this act functions as recording and demonstrating their understandings. When communicating with others about what has been seen, they not only have a chance to use language and to advance its growth, but also are able to trace and revisit their discoveries and actions. “The exercise is an instrument for reflection . . . Reflection can lead to a refinement of ideas and further search and discovery” (Machado, 2007, p. 165). Morrow and Asbury (2003) highly recommended that visual literacy be integrated with speaking, listening, reading, and writing. It thus requires instruction to be explicit and well orchestrated, because when discussing children’s creative art, the discussion gives teachers insights into what children are thinking through their reading and talking about visual cues and symbols (Machado, 2007). In this process, the adult and child are deliberately seeking meaning from children’s expressed views and experiences (Barry, 1999 in Machado, 2007). Visual images and conversations about visual representations are conducive to advancing children’s understanding of abstract concepts (Burmark, 2002) and concurrently promote language competencies.

Reading Aloud The benefits of reading aloud have long been acknowledged by teachers and researchers alike in the early childhood field. Reading aloud enables children to hear and see picture books, which is advantageous to an increased interest of children in print. Reading aloud promotes children’s concentration, develops their auditory perception, builds vocabulary, enlarges grammar, enhances thinking, and develops brain power. In addition, it widens the imagination, which, in turn, advances the creativity that is conducive to problem solving.

606

Ni Chang

Reading aloud teaches children skills necessary to “differentiate the significant ideas in the content from the less important and to place the important ideas in a logical sequence” (Homeschool-curriculum-for-life.com, n.d.). According to Strickland (1990), reading-aloud activities are positive and beneficial to children’s development because interactive behaviors exhibited by an adult and a child include “questioning, scaffolding (modeling dialogue and responses), praising, offering information, directing discussion, sharing personal reactions, and relating concepts to life experiences” (in Machado, 2007, p. 591). These practices are fundamental to conceptual development and the advancement of the language arts skills, including listening, speaking, reading, writing, and visual literacy. In addition, children’s oral vocabulary increases as a result of active and oral participation in these activities. Oral vocabulary refers to words that are spoken and that are heard (Machado, 2007). Children’s vocabulary, in turn, plays a crucial role in their comprehension and ease of learning to read (the National Reading Panel, 2000) just as “Oral vocabulary is a key in making the transition from oral to written forms” (Machado, 2007, p. 543). Therefore, teachers are expected to involve children in discussions about books to arrive at these recognized and worthwhile ends. When talking about what has been read, children develop the ability to think and express their ideas (Homeschool-curriculum-for-life.com, n.d.). To acquire language, children need not only to engage with language for a certain purpose, but also to be situated in a social context. Two people interact with each other, in part, through the medium of language. While using language to communicate, a dyad of two individuals pays attention to one another, constructing meanings and influencing thoughts as well (Fries, n.d.). Discussions guided by sensitive adults help construct meanings and clarify comprehension that otherwise would distance children from the learning process. In dialogical communication, children’s responses can offer teachers valuable clues, allowing them to know how to effectively work out the next step of instruction to further promote learning (Chang, 1996; Chang, 2007). Effective ways, aiming to encourage children to use language authentically, apparently need to be seriously sought out by concerned teachers in an effort to enhance children’s learning and their language competencies (Machado, 2007). Purposeful teacher dialogues with children, working as a stimulant, are intended to promote the mental and language growth of young children. In these intellect-building encounters, teachers explain and talk with children about objects and interesting concepts in order to “advance further discovery or to help put the discovery or experience into words” (Machado, 2007, p. 113). Dialogues enable teachers to be exposed to aspects of children’s inner world, their concept learning, and feelings, which are deemed valuable in a continuum of teaching and dyadic interactions (Chang, 2007).

Methodology Subjects and Site The participants of the study included 71 children, ages 4-8, and 71 pre-service teachers across four semesters (spring and fall, 2005, spring, 06, and spring 07) from a Midwest

Children’s Language and Literacy Learning during a Science …

607

comprehensive university. Most of the children (38%) were five years old, who were either in preschool or kindergarten (see Table 1). Table 1. Demographics of Young Participants Ages 4-8 Age 4 5 6 7 8 Total

Gender Boy 3 14 8 4 2 31 (44%)

Girl 8 13 11 5 3 40 (56%)

Total n 11 (15%) 27 (38%) 19 (27%) 9 (13%) 5 (7%) 71 (100%)

The majority of adult and young participants were Caucasians. The adult participants in this institution were all Block One students (most were juniors). Block One is the first block for which the students must enroll after entering into a teacher education program to gain their baccalaureate degree and teaching licensure. These participants were identified through an early childhood course, “Inquiry in Math and Science.” The college students were intentionally sampled because they had received rigorous training prior to actually working with young children on a one-on-one basis either in a school setting or in an alternative setting comfortable for both the adult and the child. The intentional sampling was also employed due to the participants’ first-hand experiences of working with a child on a science concept.

Data Collection Under the guidance of the researcher/ trainer (hereafter referring to as the researcher), each adult participant must exert an effort to tap into the interests of a child who would be the interactive partner of the adult. A child might be a cousin, a relative’s child, or a child from a school where the participant was engaged in field experience. With an established intimate knowledge of the child, a science concept was determined. The participant planned and refined an integrated curriculum, consisting of English/Language Arts (book sharing and conversations), Science (a scientific concept) and Art (children's drawings). What an adult participant was required to do included (1) searching for a children’s picture book conveying an accurate science concept and (2) designing a hands-on-activity that was plainly and simply related to the chosen concept. Based on the lesson plan that had been approved by the researcher, the interaction was implemented and was either tape recorded or videotaped before being transcribed. Each pre-service teacher must submit a written report, documenting his or her learning reflection besides the transcription of the conversation. The information provided by these students allowed the researcher to explore what opportunities an integrated curriculum provided to facilitate and how an integrated curricular approach mediated children’s language and literacy competencies when they were engaged in acquiring a science concept.

608

Ni Chang

Data Analysis The data were analyzed qualitatively and quantitatively. The analysis of the conversation on the transcripts were based on the following: A) Halliday's Model of Language Functions (1973) was used to examine young children's utterances (see Table 2) and B) the language and literacy development information provided by Otto (2006) (see Table 3) coupled with Halliday's Language Functions was adopted to explore adult utterances in terms of opportunities provided to young children for oral language and literacy acquisition. The data were coded in line with Halliday’s Model of Language Functions as well as the language and literacy development information provided by Otto. The coded data then were tallied and their individual frequencies were calculated to partially answer the research question concerning opportunities that the children had in the development of language and literacy competencies when interacting with adults in the process of learning science. Table 2. Halliday’s theory transformed for the purpose of data analysis

Code

Meaning

Personal (P)

to express opinions or emotions

Regulatory (R)

to influence the behaviors of others

Informative/ Representational (I) Instrumental (IS) Interactional (IN) Imaginative (IM) Heuristic (H) Yes or no answer

to give information facts and information

Example

to express their needs

-I’ll make something that you really like. - Doesn’t make any. -You do that other thing. - Mommy, smell this one. -I made a Daddy snowman. -They help him, they help him walk like this. -I need help making a robot.

to form relationships

Mommy.

to express creative language

-Why do you have all those ears?

to seek information and ask questions

-How do you make purple? -Yeah. -No.

Even though the codes were predetermined in light of Halliday’s theory and information congregated by Otto, the researcher also employed the constant comparative method suggested by Creswell (2002) and Lincoln & Guba (1985). This analysis was employed to examine the data to identify significant and meaningful patterns in relation to the research question, focusing on opportunities for offering scaffolding for young children’s language and literacy uses. This exercise was undertaken continuously throughout the analysis process for data refinement.

Children’s Language and Literacy Learning during a Science …

609

Table 3. Codes of the language and literacy development information provided by otto

Code

Name

LL

Low level questions

HL

High-level question

CQ

Clarification question

RQ

Recitation question

IQ

Meaning

Example

Requiring a little thinking or a listener having an answer already Stimulating a listener’s thinking

What color is this marker? Why did you draw this picture?

Intending to understand better the message of what was just said. Knowing the answer, but intending to check for understanding

Child: They are fireworks. Adult: You drew fireworks?

Information questioning

Intending to know information

Adult: What are those giggly lines on your pictures?

TATQ

There-and-then question

Focusing on things or events not presented in the immediate context

*LES

Expansion or recasting

*LER

Repetition

Completing what a child was just said into a full sentence to demonstrate more complex syntax, morphology, semantics, and phonology. Repeating what was just said

Adult: We went to that apple orchard last year. Did you remember? Child: I do not remember. But I like apples. Child: A bird. Adult: Yes. I see the bird is flying to the South because Fall is approaching.

Why is not a caterpillar’s color the same as a butterfly’s? How is rain formed?

Adult: What is the first letter of your name? Child: My name starts with J.

Child: This is a dinosaur. Adult:

*Linguistic scaffolding (LS)

Knowing the data thoroughly, the researcher coded and interpreted the data with different symbols and words representing language and literacy development. For example, READ/BO-IM denotes the impact that a book had on a child. LIT/VO-WR refers to a child’s initial desire to write (see Table 4). The purpose of coding was to form categories emerging from the data that had the same content based on the researcher’s carefully considered judgments. In the constant comparative method process, the data were constantly compared internally within a context and externally across categories. That is, the analysis was done by comparing incidents applicable to each category and then by classifying the coded data based on their individual categories. Even though the data were categorized according to their individual characteristics, the information within each category was further compared, being focused on similarities and differences.

Table 4. Codes used for categorizing language development Code LIS/FO-DI

Interpretation Listening through following direction

LIS/MADE SP/BOWA SP/WO-RE

Listening through mathematical concept development Speaking through book walk Speaking through word recognition

SP/TU-TA

Speaking through turn taking and vocabulary learning

SP/IN-IN

Through an investigating process of a child’s interest

Example S: Now, you are going to get to do an activity to help you remember what the order of the colors in the rainbow look like. S: These are arches that are representing each of the colors in the rainbow. S: I would like you to try and order them in the order they go in according to what you have learned. S: Do you think you can do that for me? G: Yeah. Y H: Can you count out the eight legs for me? K: One, two, three, four, five, six, seven, eight. [pointing to picture as he counts] S: In this picture what colors do you see in the rainbow? G: [Points at the colors] Red, orange, yellow, green, blue, and purple. N: Do you remember what they are called? A: Um…. N: (Made the sound of the letter “t.”) A: Tadpole. (She smiles) M: . . . The pumpkin seed grew into a pumpkin . . . T.J.: Ummm, that thing (pointing to the sequence card with a sprout on it). M: That’s right, it’s a sprout. T.J.: Sprout T: What are you interested in nature? T: And what do you know about [it]? K: Frogs—some are poisonous and some are not. K: They are different colors green, blue. K: Green are not poisonous. ...

Table 4. (Continued) Code READ/THINK

Interpretation Ability to think while being read

READ/THINK

Ability to connect with one’s own prior knowledge Repeating what was being read voluntarily

READ/VO-RE

READ/REPE

Adult’s requesting a child to repeat words

READ/BO-IM

Empowerment of the book on a child’s learning Building a concept of a library’s functionality

READ/LIBRARY

LIT/PH-EN

Phonological enhancement

Example Child: “I made sure to get the tongue right this time. I saw in the book that I didn’t draw it right the first time. Child: “Do snakes really do that, or do snakes really need that?” while listening to a book. While talking about the book content to enhance the child’s understanding of the physical characteristics of an ant, MyKaila commented, “Don’t touch the eyes” when the location of the ant’s eyes were being discussed. An adult sounded out each word carefully while pointing at each of the words. Adult: One….Tiny….T..u..r..t..l..e. T-U-R-T-L-E. Child: Oh! Turtle! T-u-r-t-l-e. Turtle, that sounds like purple to me. P-u-r-p-l-e! Expect instead of p’s there are t’s! Adult: Can you repeat those words after me? … Adult: Thorax (pointing to the thorax). Child: Thorax. Adult: Abdomen. Child: Abdomen. Adult: So the egg turns into a tadpole? Child: Yeah, didn’t you read the story? Child: Can I take the book too? Adult: How about I write down the title and who wrote it, then she can get it at your school library? Child: Yeah and I am going to check the book out when we go to the library R: Alright, I’ll write and you tell me what to say. A: (points to the largest elephant) R: You want to tell me about this elephant first? A: Happy. Happy Elephant. Happy Elephant R: You want it to say Happy…

Table 4. (Continued) Code LIT/LA-PI

Interpretation Asked to label pictures

LIT/VOWR LIT/ENWR

Voluntary labeling and writing

LIT/ASWR LIT/VORE VIS/DRA

VIS/TADR

Encouraged to write

Asked to write a name New vocabulary development Mental ability to recognize differences visually and symbolically Talking about drawing

Example M: If you could please label what you are drawing, that would be of great help. A: Okay, here is the picture, and it is labeled too. E: But let me write something down here (opens the book that Shannon just has read to him to a page with the names of different butterflies and starts to write down a specific type). A: This is a baby frog. (She pointed to the smallest frog) J: Do you think that you could write “baby frog?” A: Yep (She writes ‘baby.’) J: Can you put your name on it? A Is that a good N? A child was exposed to and learned three new big words when acquiring a science concept of the water cycle:, evaporation, condensation, and precipitation Tyler quickly pulled his drawing close to him and attempted to modify it soon after viewing the picture of a spider on the book cover. H: Can you tell me . . . about what you have drawn this time? K: Sure! I drew the sun in the corner, because snakes need heat. Then, I drew some rain, lots and lots of rain, because snakes need a lot of water. . . .

Children’s Language and Literacy Learning during a Science …

613

Categories that were, in some way, connected with one another and that were needlessly overlapping were combined together. For example, some questions that were categorized as low-level questions (LLQ) according to language and literacy information by Otto (2006). Examples are “Do you know what I want you to do for me now?” “Do you think you can do that for me?”, “Can you tell me about your drawing?” “What is the first stage of a butterfly’s life cycle?” A continuous comparison within the category show that although these questions all fell into the category of LLQ (low-level questions), the last two questions also were labeled as IQ adult. IQ denotes information questions according to Halliday’s Language Function Model (1973). Collected drawings and the researcher's notes were referenced to verify and refine the data, as were the communicative notes between the researcher and the participants. Reliability was sustained by the researcher’s effort to code and re-code the data and to classify categories consistently across the four semesters.

Results and Discussion The study was intended to explore what opportunities an integrated curriculum provided to facilitate and how an integrated curricular approach mediated children’s language and literacy competencies when they were engaged in acquiring a science concept. While a child was acquiring a science concept, several different subject areas, such as language arts, math, and arts were involved and assisted the learning process. During the dialogical interaction, adult participants mostly asked the children low-level questions (LLQ) (34.45%) or shared ideas with the children (P adult) (34.08%). Young participants, in contrast, primarily offered information (I child) (60.23%) or stated their own ideas and thoughts (P child) (26.55%) (see Table 5). The following will first present the analysis of questioning that occurred throughout a lesson in order to demonstrate what opportunities an integrated curriculum provided to facilitate and how an integrated curriculum mediated children’s language and literacy acquisition in the process of learning a science concept. To further respond to the research question, the findings will also be presented specifically related to five areas of language arts development, namely, listening, speaking, reading, writing, and visual literacy/viewing.

Questioning Questioning technique was vastly utilized throughout a lesson to maintain the verbal interactions. The analysis shows that most of the low-level questions asked by the adults served either to seek information (RQ adult) (32.69%) or to solicit answers that they already knew (recitation questions--IQ adult) (31.26%). In interacting with the children, the adults infrequently utilized clarification questions and closed-ended questions that elicited either yes or no answers (see Table 6). The finding is consistent with the Otto’s (2006) notion that questioning is essential to interactions, as it is one of the principal strategies to elicit responses from listeners.

Table 5. Children’s Utterances In Light of Halliday’s Model of Language Functions Semester

LLQ

HLQ

IQ

CQ

P

I

R

H

IS

IN

Total

Y/N

Spring 05

57

1

3

6

110

200

15

6

4

0

402

74

Fall 05

29

2

0

0

161

351

1

6

2

0

552

67

Spring 06

22

3

0

0

87

246

13

10

0

0

381

103

Spring 07

19

4

0

0

105

253

3

17

5

2

408

82

Total

127

10

3

6

463

1050

32

39

11

2

1743

326

7.28%

.56%

.34%

26.55%

60.23%

1.83%

2.31%

.62%

.11%

100%

.17 %

Table 6. Adults’ utterances in light of both the language and literacy development information provided by Otto (2006) and Halliday's language functions Semester

LLQ

HLQ

RQ

CQ

IQ

TATQ

P adult

I adult

R adult

LSE

LSR

Total

Y/N

Spring 05

229

8

53

2

63

1

271

67

83

1

0

659

4

Fall 05

307

2

73

3

97

0

276

105

58

2

2

752

5

Spring 06

318

15

109

4

124

0

282

164

100

3

2

884

18

Spring 07

283

28

154

13

88

2

234

160

78

31

10

824

6

Total

1137

53

389

22

372

3

1063

496

319

37

14

3119

33

34.45%

1.70%

32.69%

1.85%

31.26%

.25%

34.08%

15.90%

10.23%

1.19%

.45%

100%

Children’s Language and Literacy Learning during a Science …

615

There were various purposes for an adult to ask questions in the learning of a science concept. Questions, according to Machado (2007), encouraged the children to ponder and to express their thoughts verbally, which allow the adults to check children’s understanding of concepts. During the interactive process, the adult participants modified questions to fit specific language levels of the children in order for them to respond with ease. In this process, the adults helped the children experience “turn-taking” (Otto, 2006; Pine, 1994), active listening, speaking, thinking, and following directions to further his or her language development (Otto, 2006). Exposing children to questioning also impacts their five aspects of language knowledge, including phonetics, semantics, syntax, morpheme, and pragmatics (Otto, 2006). According to Otto (2006), phonetic knowledge involves children’s receptive awareness and production of language-related sounds (phonemes). Semantic knowledge refers to children’s receptive and expressive vocabulary. Syntactic knowledge means children’s ability to use utterances comprised of more than three words. Morphemic knowledge indicates young children’s linguistic competencies in communicating intended meaning more precisely with an ability to change word structures. Pragmatic knowledge is about children’s ability to use language for a wider range of purposes, such as requesting permission, expressing emotions, or making judgments.

Attracting the attention Seianna utilized questioning at the very beginning of a lesson to attract the attention of Gabby, a five-year-old child (see below). To respond to the question, Gabby had a chance to listen and to be exposed to the structure of a complete sentence as the adult expanded the words she just had said. S: What do you think this is a picture of? [RQ adult, LLQ adult] G: [Smiles and her eyes get really big] A rainbow! [I child] S: You are right! [P adult] S: It is a rainbow. [P adult] [LSE adult] The structure of the question is short and well formed with no false start. The question asked is contextual, which was closely associated with the immediate context (the picture held in the teacher, Seianna’s hand). It promotes the act of communication to enhance the child’s language acquisition (Otto, 2006). The excitement observed in the child’s eyes served as a vehicle to prepare the child for the subsequent dialogue. A conversational interchange between a five-year-old boy, Josiah, and an adult, Sarah, in this same fashion can also be illustrated as below: S: What do you think these are [in my hand]? [RQ adult LLQ adult] J: [giggling] Seeds. [I child] S: Yes, they are seeds. [LSE adult] The first recitation question asked by both of the adults served the purpose of attracting the child’s attention to a lesson that was about to start. The adults utilized a linguistic

616

Ni Chang

scaffolding technique to fill in words to make complete sentences out of the children’s utterances: from “A rainbow” to “It is a rainbow” or from “Seeds” to “[T]hey are seeds.” This behavior of expanding or filling in words to a child’s verbal expression provides a means to enhance the child’s acquisition of language competencies and is widely recognized by researchers (Machado, 2007, Otto, 2006). In addition, the adult’s repetition of these key words, such as “rainbow” and “seeds” is intended to “reinforce pronunciation and to confirm the meaning of what the child said” (Otto, 2006, p. 51). Scaffolding technique is characterized by Bruner (1978) and Wells (1986) as a supportive teaching strategy with challenge, constituting responsive communication, open-ended questioning technique, and the facilitation of the child’s initiatives (Machado, 2007).

Setting the stage The questioning technique continued as an adult made an effort to prepare a child for the participation in the lesson. Seianna posed the following questions in order for Gabby, five years old, to concentrate on the scope of this lesson. S: Have you ever seen an actual rainbow? [IQ adult, LLQ adult] G: [Smiles] Yeah, it’s in the sky and on the road like water. [I child] S: What did the rainbow look like? [RQ adult, LLQ adult] G: Like colors. [I child] S: What colors did you see? [IQ adult, LLQ adult] G: Green, red, yellow, blue, and uh [rolls her eyes up to think] pink. [I child] The information question, “Have you eve seen an actual rainbow?” (IQ) led the child to share with the adult what she knew about a rainbow. With the adult’s continual inquiry: “What colors did you see [on the rainbow]?” the child verbally described her prior knowledge of the science topic of the sequential colors of a rainbow, which, from the child’s point of view, was comprised of five different colors. Not only did the adult get to know the child’s knowledge level of the colors of the rainbow, but also encouraged the child to use her language skills to describe her mental picture about a rainbow (productive-expressive speech). In the second short episode, the adult asked two information questions with one recitation question, and the child was able to understand them all (receptive speech). The number of words constituting the questions yielded about the same number of words from the two responses from the child (I child) with 18 words each. The questions were well formed and fairly simple in their structure. Following the adult’s questions did not seem to be difficult for Gabby. This conversational interaction with accurately modeled utterances allowed the child to think in order to provide information and impacted on the child’s comprehension and language growth (Otto, 2006). In a similar way, Sarah also captivated Josiah’s attention (a five-year-old child) with a number of questions:

Children’s Language and Literacy Learning during a Science …

617

S: Do you know where they [seeds] are from? [RQ adult LLQ adult] J: Maybe an apple. [I child] S: You are right. [P adult] S: These are apple seeds that will eventually turn into apples. [LES adult] [Sarah brings out a whole apple to show.] S: . . . Do you know how an apple grows? [RQ adult LLQ adult] J: Uh-huh. [Y child] In the dialogic interaction between Sarah and Josiah, with the concrete materials, seeds, displayed in front of the child, the child’s attention was quickly captured. This paved the way for the child’s concentration on the recitation question (RQ) posed by the adult. The child’s comprehension of the question elicited a short response (I child). This, in turn, allowed Sarah to employ a linguistic scaffolding strategy (Otto, 2006) as she extended the child’s short responses, “Maybe an apple” by saying, “These are apple seeds that will eventually turn into apples.” It is a way to fill out what a child has said (Otto, 2006; Reich, 1986; Wells, 1986). Expansion or recasting (Camarata, 1995) used as a linguistic scaffolding strategy helps advance the child’s acquisition of language skills (Machado, 2007; Otto, 2006). Expansion models more complex syntax, morphology, semantics, and correct pronunciation (phonology) (Machado, 2007; Otto, 2006). It is contingent on the experience of a child encountered at the very moment. The responsive communication of the adult expands the syntactic composition of the child’s utterance (Machado, 2007). In addition, Sarah’s verbal statement helped facilitate Josiah’s language skills because they are closely tied to the jointly focused objects (seeds and an apple). “Referents of utterances found in the immediate context give the young child a visible connection between the oral symbol and its actual referent” (Otto, 2006, p. 48).

Requesting a Drawing After the expansion of the child’s utterance, Sarah brought out a real apple. This additional concrete material further excited Josiah, age 6, and helped activate his prior knowledge. At this moment, Sarah requested a drawing from Josiah in order to receive detailed information of what a child knew about the growth cycle of an apple: S: Can you show me with a picture how an apple grows? [LLQ adult] [Josiah picks up the red crayon. He draws a tree with apples on it. He also adds his name to the picture.] (see Figure 1)

618

Ni Chang

Figure 1. Josiah’s first attempt to describe the growth cycle of an apple

A similar situation took place in the interaction between Seianna and Gabby, age 5: S: Can you draw me a picture of how the colors appear in a rainbow? LLQ adult] G: [Giggles] Yeah. [Gabby picks up the green crayon first and draws the top of the rainbow. She then grabs the purple crayon and draws the next arch in the rainbow. She then continues with the red crayon, followed by the yellow crayon, the orange crayon, and finally ending the rainbow with the blue crayon.]. (see Figure 2) The verbal exchange between the adult and child enabled the child to focus on the science concept. The children’s active listening seemed to allow them to understand the requests. While the questions facilitated the children’s listening, the children had an opportunity to learn that a mental image can be expressed through a graphic representation (Paquette, Fello, & Jalongo, 2007).

Figure 2. Gabby’s prior understanding of the rainbow colors.

Children’s Language and Literacy Learning during a Science …

619

Sometimes, when a request is made, the child’s action does not directly reflect what had been requested. An adult then flexibly takes advantage of this opportunity to engage the child in a scaffolded conversation to enhance his or her acquisition of language competencies (Otto, 2006). In addition, the adult’s intimate listening facilitated the child’s language skills (Machado, 2007). The following segment of conversation illustrates the supportive scaffolding that the adult provided to the child’s listening, speaking, turn-taking, and direction following skills. J: Yeah, can you draw that [tornado] for me? [IQ adult LLQ adult] G: I like this new crayon. It’s pointy. [P child] J: I like new crayons too. [LSR adult] J: This box has lots of new crayons if you want to use those. [LSR adult LSE adult] G: This looks like my tornado friend. [P child] J: Your mom told me you have a tornado friend. [LSR adult] She says it makes big messes just like a tornado. [P adult] Is that what you’re drawing? [IQ adult LLQ adult] G: Yeah. [Y child] This conversation, unlike the two that were presented earlier, did not take in the pattern of the child’s quick compliance with the adult’s request. Unexpectedly, the conversation focus shifted from tornados to crayons. Yet, Jenni let six-year-old Gena assume control over the conversation by consciously following the lead of the child. After her information question (IQ adult), Jenni realized that the child’s focus at that time was on the new crayons provided. With the understanding that a young child usually is fascinated by new writing tools, Jenni patiently waited for the child’s enthusiastic exploration of the drawing tool. Yet, to facilitate the child’s linguistic skills, Jenni followed suit by repeating the child’s utterance about the new crayons: “I like new crayons too.” The repetition changed the sentence as structured by the child to show her more complex sentence structure achieved by switching from the singular (this new crayon) to the plural (new crayons). In the meantime, she extended the sentence to make it more complex in its structure: “This box has lots of new crayons if you want to use those.” The child then quickly changed a topic from new crayons to that of tornados: “This looks like my tornado friend.” By this lead, Jenni made an attempt to shift the child’s focus to the science topic by emphasizing the key words, “a tornado friend,” in the sentence, “Your mom told me you have a tornado friend.” The repetition serves to confirm the child’s utterance. The next sentence was further intended to arouse the child’s interest in the topic. The question that Jenni asked afterwards purposefully lured the child back to the topic of the tornado formation. In this process, Jenni facilitated the child’s turntaking, listening, speaking, and direction following, because the child had to listen to Jenni carefully in the entire interchanging communication in order to follow and respond to the questions. This finding is consistent with the notion made by Otto (2006). Jenni was able to maintain the conversation even though the topic appeared to be derailed at a time, because she employed contingent questioning to pose a series of questions that were built on Gena’s response to engage the child’s thinking and oral responses (Otto, 2006; Trousdale, 1990). Moreover Jenni’s ability to verbally map (Otto, 2006) what Gena said shows that she was able to take into account the developmental level of the child’s receptive language skills. The action of the child’s drawing of tornados and the science topic itself promote the child’s

620

Ni Chang

conceptual development. “Verbal mapping is not just idle chatter on the part of an adult. Instead, it involves a conscious focus by the adult on the concepts and vocabulary relevant to the ongoing learning activity” (Otto, 2006, p. 49). In a communication loop, a child does initiate a change of a topic. An adult sensitive to the changed topic quickly becomes responsive by providing supportive scaffolding to sustain the communicational flow. As the topic of tornados was being discussed, the child was exposed to syntactic, morphemic, and phonemic knowledge related to the concept, extending and expanding the child’s receptive and expressive language (Otto, 2006).

Eliciting Prior Knowledge In soliciting the content of a drawing, an adult’s information question was an opportunity for a child to use language to describe what appeared on a piece of paper (see the dialogue below-S denotes Sarah and J represents Josiah, age 6): S: Okay, can you tell me what is happening in your picture? [IQ adult LLQ adult] J: They are growing, a tree and the apples are growing. [I child] The child described his picture in roughly two complete sentences: “They are growing” and “the apples are growing.” The child was able to comprehend the adult’s complex question that is composed of a compound structure with one main clause and one subordinate clause. In addition, it also involves the aspects of semantics, syntax, and morpheme. It is apparent that the child had no difficulty understanding the question and responded with correct morphology: “They are growing. . . the apples are growing” besides selecting an appropriate verb and forming grammatically correct sentences. The information also enabled Sarah to gain some insight about Josiah’s understanding of the growth cycle of an apple. The verbal exchanges taking place between Seianna and Gabby, a five-year-old child, further show language enhancement as a result of the information question posed by the adult: S: . . . Can you describe your picture to me? [LLQ adult] G: It’s a pretty rainbow with lots of colors. [I child] G: It has purple in it. [I child] G: That’s my favorite color! [Smiles] [P child] S: It is a pretty rainbow. [P adult] S: You did a very good job drawing the arches and filling them each in with different colors. [P adult] The adult’s low-level question (LLQ) elicited three complete sentences with the first sentence structure being more complex than the other two: “It’s a pretty rainbow with lots of colors.” Not only did the child describe the drawing, but also expressed her own feeling about a certain color (P child). The pictorial explanation by the child was then acknowledged by the adult. The acknowledgement made the child see that her utterances were understood by the adult and that the rainbow was shaped in arches and did have different colors: “It is a pretty rainbow. You did a very good job drawing the arches and filling them each in with different colors.” All the utterances about the picture that Gabby produced coupled with the verbal

Children’s Language and Literacy Learning during a Science …

621

expression about the colors of a rainbow prior to the drawing fairly well informed Seianna of how much the child knew about the rainbow colors (see Figure 2).

Gauging the Comprehension After book reading and follow-up discussions, communication continued with questions and answers. Seianna utilized a series of questions to gauge where four-year-old Gabby was conceptually. Concurrently, the child was assisted in solidifying or constructing the science concept by speaking and thinking. S: Okay, now I am going to ask you some questions that relate to what we have just read from the book and we will not use the book to find the answers this time. [R adult] G: Uh…okay. [Y child] S: What colors are in the rainbow? [RQ adult LLQ adult] G: [Takes her time recalling the colors and thinks carefully to herself] Red, orange, mmm… yellow, uh….green, uh…blue, and purple! [I child] S: Great job! [P adult] S: You named all of them and in the correct order too. [P adult] S: I am so proud of you. [P adult] S: What color is always on top? [RQ adult LLQ adult] G: Red. [I child] S: Good job. [P adult] S: What color is always on the bottom? [RQ adult LLQ adult] G: Purple! [I child] G: That’s my favorite color! [Smiles]. [P child] The initial directional statement was relatively long for a young child. However, the child seemed to be able to follow it: “Uh . . . okay.” The recitation question made the child think hard before sharing the answer verbally. The targeted and purposeful praise further inspired the child to cooperate with the adult. Soon after the adult posed another recitation question (RQ adult LLQ adult), the child’s quick response consistently showed her close attention to the adult’s question: “Red.” The next response given by the child clearly demonstrated her listening comprehension, because instead of asking the second color of the rainbow, the adult gave an unexpected twist to solicit the answer of the last color of the rainbow. Accurately, the child was able to answer: “Purple.” Not only did she answer the question, but also expressed her comment on the color: “That is my favorite color.” This interactive segment strengthened the child’s listening ability, but offered the opportunity for the child to communicate orally with the adult. Literal or low-level questions (Genishi & Dyson, 1984; Otto, 2006) serve to have children recall facts or learned concepts and help establish a basic level of comprehension and understanding. In the meantime, questions like these enable adults to perceive whether children are on the right track in the learning process (Otto, 2006). Jaime engaged McKenzie (a five-year-old child) with questions and answers in order to informally gauge the child’s level of understanding of the growth cycle of a pumpkin. This interactive process not only was an opportunity to reinforce the child’s acquisition of the science concept, but also exposed the child to more formal language structure.

622

Ni Chang J: How does a pumpkin start out? [RQ adult LLQ adult] M: By seed. [I child] J: Good! A pumpkin starts out as a seed. [I adult] [LSE adult] J: Then what does that seed grow into? [RQ adult LLQ adult] M: Um……..a stem and leaf. [I child] J: Right! The seed grows from the ground into a little green stem you can see above the ground. [I adult] [LSE adult] J: In the book they call that little stem a sprout. [I adult] M: Oh yeah. I forgot. [P child]

Jaime’s two recitation questions resulted in three utterances from the child. The incomplete sentences by the child as the information, e.g., “By seed” and “a stem and leaf,” then were expanded by the adult: “A pumpkin starts out as a seed” and “ . . the seed grows from the ground into a little green stem you can see above the ground.” In this verbal interaction, the adult was modeling for the child by expanding the sentences to show the child how to structure a formal sentence with the correct tense (starts), a correct corresponding verb phrase (starts out) and a correct use of a preposition (as). This effort can further promote the child’s understanding of semantics, syntax, morpheme, and phonetics (Otto, 2006). Otto recognizes the expansion of a child’s utterance as linguistic scaffolding; it is the adult’s attempt to use language in a supportive manner for children to communicate and to express their insights. The second sentence is more complex than the first one, because it has a main clause, “The seed grows from the ground into a little green stem” with an adjective sub-clause, “you can see above the ground” to modify “a little green stem.” Exposed to this sentence, the child again was reinforced with the correct selection of a verb and a correct verb tense use. This learning opportunity provided the child with more to know, including how to use spatial words, such as “above,” another verb phrase, “grow into,” a prepositional phrase, “from the ground,” an adverb phrase, “above the ground.” Also in this learning process, the child was further informed of the usefulness of the book and learned a new vocabulary word, “sprout.”

Hands-on Activity To help Kevin (a five-year-old child) further construct a concept of physical characteristics of a spider, Hannah provided him with Styrofoam balls, pipe cleaners, glue, black buttons, etc to create a spider. The questions that Hannah asked passed the dialogical turn to the child in order to bring out Kevin’s understanding of the physical characteristics of a spider: H: Which part of your spider is the front? [RQ adult LLQ adult] [Kevin points to the front of his spider.] H: How do you know? [RQ adult LLQ adult] K: It has legs. [I child] H: So, you remembered that the legs are on the front body part? [CQ adult LLQ adult LSE adult] K: Yes. [Y child]

Children’s Language and Literacy Learning during a Science …

623

H: Good job! [P adult] H: That is an awesome spider! [P adult] K: I put two eyes on the top of the head so that it can see if someone is trying to kill it. [I child] H: Wow, the book did say that some spiders have eyes on the top of their head. [LSE adult] Questioning in this discourse consists of one clarification and two recitation questions. These two types of questions informed the adult about Kevin’s level of understanding of the science concept. Linguistically, Kevin had several attempts to use language to communicate while solidifying his newly acquired knowledge by Hannah’s repetition of “legs” in the sentence: “ . . you remembered that the legs are on the front body part?” (CQ adult LLQ adult LSE adult) following the child’s utterance of “It has legs” (I child). Because of the appropriate questions raised by Hannah, even the adult’s personal expression: “This is an awesome spider!” could inspire Kevin to add more information by speaking a complex sentence: “I put two eyes on the top of the head so that it can see if someone is trying to kill it” (I child). This oral expression, on the other hand, reflected the focused attention that the child had on the book content while it was being read to him. The adult then provided linguistic scaffolding by adding plural nouns to the sentence and by changing the sentence structure. Her response was, “Wow, the book did say that some spiders have eyes on the top of their head.” Such a response facilitated the child’s language acquisition (Otto, 2006).

Gauging the Learning Outcome In the end of the learning process, another drawing was solicited in order for an adult to “assess” a learning outcome. This was an opportunity for Josiah, six years old, to use language to share what had been learned with Sarah. S: Can you tell me about your picture? [IQ adult LLQ adult] J: First the seed came, then the tree, then the bud, then the flower, and then the apple. [I child] This information-seeking provided the child with an opportunity to express what he confidently knew about the life cycle of an apple. Although the question only had 7 words, the child responded with 17 words. In a similar case, Gabby also acquired not only the new science concept, but also new vocabulary. Moreover, she was able to freely express herself. S: . . can you tell me about your picture? [IQ adult LLQ adult] G: Yes, I colored a rainbow with colors in order. [I child] S: What is the order of the colors you drew? [RQ adult LLQ adult] G: Red, orange, yellow, green, blue, and purple. [I child] The child demonstrated to the adult, as a result of the question, that she learned new vocabulary, “in order.” It also shows her capability to make a relatively more complex

624

Ni Chang

sentence with two prepositional phrases: “with colors” and “in order.” The second recitation question posed by Seianna was a complex sentence with a main clause and a subordinate clause. However, the child was able to follow with a response that was conceptually correct even though Gabby had just acquired it (see Figure 3). Otto (2006) theorized that the way a question is worded by an adult would influence how a child would structure his or her response.

Figure 3. Gabby’s post-learning drawing.

The general finding shown above was also in line with another notion made by Otto, (2006) and by Trousdale (1990). The sustainability of the conversations demonstrates the adults’ uses of contingent questioning, which composes a series of questions building on each other to engage children’s thinking and elicit oral responses. Taking into consideration the child’s responses, follow-up questions were thereby formulated (Otto, 2006; Trousdale, 1990).

Five Languages Arts Areas To further respond to the research question, the following text will set forth opportunities the young participants had specifically related to five areas of language arts development, namely, listening, speaking, reading, writing, and visual literacy/viewing in the process of learning a science concept through an integrated curriculum.

Listening While engaging the children in the learning of a science concept, the adults facilitated the children’s listening acquisition through questioning. Yet, the study also found that a child’s listening skills were facilitated through the following:

Children’s Language and Literacy Learning during a Science …

625

Following directions For the child to construct a learned science concept, a hands-on activity was provided after book reading and follow-up discussion. Although ittle verbal exchange happened during this interaction, prior to a child’s engagement in the activity, he or she had to listen in order to know how this activity worked (see the example given below). S denotes Seianna, an adult, while G denotes Gabby, a five-year-old child: S: Now, you are going to get to do an activity to help you remember what the order of the colors in the rainbow look like. [R adult] [Seianna shows her the colored arch pieces.] S: These are arches that are representing each of the colors in the rainbow. [P adult] S: I would like you to try and order them in the order they go in according to what you have learned. [R adult] S: Do you think you can do that for me? [LLQ adult] G: Yeah. [Y child] [The child grabs the pieces and begins to lay them out on the table.] G: [Talks out loud as she places each piece in its order] Red, orange, yellow, blue, green, purple. [P child] Seianna’s instruction was quite long. It consists of two regulatory speeches and one personal speech. Although the adult’s comment in between those two long regulatory speeches was relatively short, it was a complex sentence with a main clause and an adjective sub-clause, describing the arches. To this five-year-old child, Gabby, it provided a change in following the direction. While Gabby was listening to the instruction, she had to actively process the adult’s utterance in order to understand what she was hearing. The child’s response, in return, shows to the adult that Gabby was successful in receiving the information. Machado (2007) supported the practice as she believed that listening required practice since it is a learned behavior. Sometimes, children hear, but may not understand what they have heard. “Language development depends on the auditory process” by sustaining attention span and following directions (Machado, 2007, p. 244).

Learning Mathematics Listening took place through the child’s engagement in mathematics learning. After reading aloud, Hannah and her five-year-old child, Kevin, were engaged in a book walk about “the physical characteristics of a spider.” The book walk not only granted the child an enhanced understanding of the number of legs that a spider has, but also reviewed and practiced counting, solidified cardinal number (four), and reinforced his spatial sense (on this side) (see the conversation below): H: Now, let’s look at this book again. [R adult] H: And we are going to point to the parts as we read them. [R adult] Okay? H: Can we point to the parts of the spider as we read them? [LLQ adult] K: Uh-huh. [Y child]

626

Ni Chang H: Can you count out the eight legs for me? [LLQ adult] K: One, two, three, four, five, six, seven, eight. [pointing to picture as he counts] [I child] H: How many legs are on this side? [pointing at book] [RQ adult LLQ adult] K: Four. [I child] H: How many legs are on this side? [pointing at book] [RQ adult LLQ adult] K: Four. [I child] H: Yes. [Y adult]

This episode of book walk promoted the child’s listening by having the child’s attention drawn on the adult’s utterances. In this practice, the child’s attentive listening made him respond to every step that the adult expected him to do. His answers confirmed that the child was able to decipher the adult’s spoken language as he understood what a cardinal number is. Learning the physical characteristics of a spider involved learning or reviewing the mathematical concept. At the same time, it encouraged the child’s careful listening. Discussion following a read-aloud helps develop young children’s language skills (Short, Kauffman, & Kahn, 2000; Soundy, Guha, & Qin, 2007). The adult’s affirmative answer “Yes” to the child’s responses became a rewarding experience that continued to sustain the child’s attention to learning (Machado, 2007).

Recognizing Words N: Okay, now we’ll go back through the book. [R adult] N: You were right that in the first stage of a frog’s life cycle they are eggs. [I adult] N: They are wrapped in jelly. [I adult] N: The second stage they have gills. [I adult] N: Do you remember what they are called? [LLQ adult] A: Um…. N: (Made the sound of the letter “t.”) A: Tadpole. (She smiles) [I child] Although the child did not say a lot, the combination of the regulatory speech (R adult) and information speech to provide information, either to verify what a child knew or to reinforce the new concept (I adult), encouraged the child to listen. Because of her attentive and active listening, the child knew when it was time for her to respond: “Um . . .” Quickly reacting on the child’s cue, the adult retaught the child with an elicitation method. This episode shows that interaction between an adult and a child works as an opportunity for promoting a letter sound (phoneme) and vocabulary construction (Otto, 20006).

Reading Aloud Reading aloud evoked a child’s thinking when a child had a great interest in the topic covered by the book. This is consistent with Machado’s (2007) notion that listening is a learned behavior. Although it seemingly appears to be passive and receptive, listening

Children’s Language and Literacy Learning during a Science …

627

promotes a child’s thinking and encourages the child to interpret what has been read to him or her, which, in turn, draws more attention from the child to the book being read. After Kevin, a five years old child, drew his first picture of the physical characteristics of a spider, Hannah read a children’s book, Spiders by Monica Hughes. Kevin was looking intently at the pictures while the book was being read aloud. His apparent active thinking about the book content led him to make a heuristic speech to query the picture in the book: “What are these sticking out?” [pointing to the hairs on the spiders legs]. The answer given by the adult encouraged his further thinking by closely listening to the book: “Let’s keep reading the book and see if it tells us.” This encouragement, in turn, piqued the child’s interest in the book. His desire to seek the answer from the book grew. Duke and Pearson (2002) and Johnson (2009) supported the use of authentic books that may be enjoyable to children and that may increase the children’s desire to read. Johnson further posited, “A child who is engaged in reading employs both his mind and his heart. . .” (p. 48). Kevin’s displayed behavior completely demonstrates this notion. It was because of careful listening and his deep engagement of his heart and mind that he asked a higher-level question. Table 7. Adult and child utterances across the four semesters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Total Average

Spring 05 Adult Child 43 28 36 18 28 6 36 30 27 26 31 30 70 61 31 15 21 8 36 26 33 12 29 21 46 15 34 28 52 37 46 7 26 17 34 17

659 36.6

402 22.3

Fall 05 Adult Child 20 13 25 9 33 23 33 22 39 27 18 8 22 25 60 45 37 18 41 20 46 60 70 48 60 46 23 14 64 31 16 16 25 30 35 23 60 35 25 39 752 552 37.6 27.6

Spring 06 Adult Child 93 28 39 12 46 22 83 23 32 18 68 25 44 20 46 33 40 26 82 28 42 17 56 19 27 27 120 59 66 24

884 58.9

381 25.4

Spring 07 Adult Child 56 22 74 37 73 36 60 36 51 13 96 34 46 12 72 40 24 30 54 19 64 25 30 37 72 25 25 21 33 21

830 55.3

408 27.2

628

Ni Chang

Speaking Through the science concept learning experiences, the total utterances of the adults were 3125. The average utterances for each adult were 45.96 either in the form of a statement or that of a question. In contrast, the young participants spoke 1743 times with 25.63 times as the average. The results excluded utterances made by adults during reading aloud (see Table 7). Oral language is a key area of literacy development in the early childhood field (Strasser & Seplocha, 2008).

Discussing Book Content (Book Walk) After reading a carefully selected book, an adult often would engage a child in an activity to discuss the book content. Machado (2007) posited that dynamic classroom discussion lent itself to the development of literacy competency. Discussion of the book by means of a book walk not only develops the child’s oral language acquisition, but also is conducive to the child’s listening comprehension as reading a book once may not effectively help young children construct the meaning of a story (Machado, 2007). The discussion in the form of book walk, if enjoyed by both adult and child, may further the child’s listening, thinking, and speaking. In this process, the child has to pay attention to what is being said, use language to think, and then verbalize his or her thoughts. The acquisition of language competency is promoted when a reciprocal interaction between an adult and a child is built on a communication that is authentic and substantive (Machado, 2007). The following dialogue taking place between Seianna and Gabby (a five-year-old child) about “the sequential colors of a rainbow” illustrates these aspects. S: In this picture what colors do you see in the rainbow? [RQ adult LLQ adult] G: [Points at the colors] Red, orange, yellow, green, blue, and purple. [I child] S: What color appears on top in the picture? [RQ adult LLQ adult] G: Red! [I child] S: Do you think it is always on top? [IQ adult LLQ adult] G: Mmm…yes! [Y child] S: Let’s check it by re-reading the page that describes the order. [R adult] The book walk specifically highlighted this particular science concept. Centering on it, both of the parties’ attention was focused on the page under discussion. As the adult raised the question: “In this picture what colors do you see in the rainbow?” the child’s thinking was evoked, “Yes, what exactly am I seeing in this picture?” The child must take a quick action to examine the page in order to find proper words to answer this question. The answer to this recitation question (RQ adult LLQ adult) represented the child’s undivided attention to the page: “Red, orange, yellow, green, blue, and purple” (I child). Although there was no apparent confirmation from the adult to the child’s answer, the adult’s continual questioning, “What color appears on top in the picture?” signaled to the child that the naming of each color was correct. In the meantime, the question elicited one-word response from the child, based on which the adult framed another question, “Do you think it is always on top?” In this question-and-answer process, these two recitation questions (RQ adult LLQ adult) exposed

Children’s Language and Literacy Learning during a Science …

629

the child to syntactic, pragmatic, morphemic, and semantic knowledge about the science concept. The first question was a question with “what” as a starting word, which was followed by a noun, a verb, an adverb phrase, and a prepositional phrase. Whereas the second question was structured with a main clause “Do you think” and an objective sub-clause: “it is always on top.” The child was able to hear the pronunciation of each word spoken by the adult and to process the way these questions were formed. The hesitated response from the child: “Mmm . . . yes” quickly allowed the adult to determine the next-step action: “Let’s check it by re-reading the page that describes the order”(R adult). The verbal interaction during the book walk inspired the child to want to know about colors of a rainbow. Apparently, the child gained the understanding of the concept by the adult’s discussing and analyzing the text. In learning a science concept of the life cycle of a flower, Sara, a five-year-old child, demonstrated her increased capability in speaking (see the episode below). L denotes Lena and S denotes Sara. L: How many stages did this seed go through before it was an adult flower? [IQ adult LLQ adult] S: Um? L: Let’s go through them together and we’ll count using our fingers, ok? [R adult] S: Yeah. [Y child] L: First is the…? [RQ adult LLQ adult] S: Seeds! [I child] L: Yes, then the root grows…? [RQ adult LLQ adult] S: Down to the ground. (Sarah is pointing her finger down toward the floor). [I child] L: Wow, you are doing great. Next the…? [RQ adult LLQ adult] S: Green thing grows up to the sky. (Sarah is using her finger to indicate the growth of the shoot). [I child] L: The green thing is called a shoot and it grows up to the sky just like you said. [I adult] L: What is the last stage in the flower’s lifecycle? [RQ adult LLQ adult] S: The flower and the leaves come. [I child] L: How many stages are there? [RQ adult LLQ adult] S: One, two, three, four. [I child] L: Yes. There are four stages. [I adult] The child’s brain was moving as she was listening to the questions posed by the adult. She had to pay very close attention in order to fill in the blanks that the adult deliberately made. In this verbal interaction, while sharing what she knew about each stage of the life cycle of a flower, the child was able to use the adverb phrase “down to the ground” to describe the second stage after the seeds are planted in dirt. The next sentence shows the child’s ability to describe the third stage. Instead of using one word to represent it, the child used several words to express her knowledge in this regard: “Green thing grows up to the sky.” This expression also reflected her awareness of morpheme (grows up) and her ability to use the preposition phrase: “to the sky.” The adult, who was sensitive and responsive to the child’s reply, provided linguistic scaffolding (Machado, 2007; Otto, 2006), expansion or recasting (Camarata, 1995) for the child’s utterance with the presentation of a new vocabulary

630

Ni Chang

word, “shoot.” The adult’s scaffolding had an effect on the child’s thinking and speech; the language scaffolding helped the child use inner speech to clarify her thinking. “Interactive teachers provide verbal assistance and nudge discovery based on their individual child’s degree of sophistication” (Machado, 2007, p. 222). The child’s four responses as information giving (I child) presented her evolved language abilities from one word, “seeds” to one whole sentence: “The flower and the leaves come.” It is consistent with the report by Snow and Tabors (1993) that a discussion led by an adult following book reading is powerful in developing a child’s complex oral language, vocabulary, and story understanding via the child’s use of language. These are critically essential in young children’s later literacy tasks (Machado, 2007). Additionally, Lena was assisted in acquiring a concept of cardinal number when she was asked to sum up the stages that she and the adult were just jointly counted together. Instead of reporting “four” as a cardinal number, indicating a number of stages in this set, the child counted it again. Lena’s sensitivity, based on Sara’s utterances, quickly intervened the process by directing the child’s attention to the cardinal numeral “four:” “. . .There are four stages.” This short learning experience enabled Sara not only to learn a new word, “shoot,” but also had a chance to solidify counting and a cardinal number that further enhanced the understanding of the science concept being studied. The learning experienced by Sara is also consistent with the notion by Otto (2006) and Tizard (1981). Both the adult and the child were situated in a context that both adult and child were cognizant of, forming an immediate learning context, which is advantageous of learning. Furthermore, Sara, due to the one-on-one interaction with the adult, obtained more opportunities to respond and to speak than children in a group setting (Otto, 2006).

Turn-Taking and Vocabulary Learning T.J. (a four-year-old child) had fun learning a concept of the pumpkin’s growth cycle. The child’s interest and enjoyment propelled his thinking of and active participation in the activity. Interacting with the adult afforded this child an opportunity not only to enhance listening, speaking, thinking, and following directions, but also to enlarge his vocabulary. M denotes Michelle, an adult while T.J. denotes a four-year-old boy. M: Okay, let’s read these and fill in the missing words. [R adult] M: T.J. planted a pumpkin . . . [P adult] T.J.: SEED! [I child] M: Good job! The pumpkin seed grew into a pumpkin . . . [P adult] T.J.: Ummm, that thing (pointing to the sequence card with a sprout on it). [I child] M: That’s right, it’s a sprout. [P adult] T.J.: Sprout [P child] M: The pumpkin sprout grew into a pumpkin . . . [P adult] T.J.: A plant. [I child] M: You are just so smart! [P adult] M: The pumpkin plant grew into a pumpkin . . . [IQ adult] T.J.: Ummm, a flower? [I child] M: Good! The pumpkin flower grew into a . . . [P adult]

Children’s Language and Literacy Learning during a Science …

631

T.J.: PUMPKIN! [I child] M: The pumpkin (grew) and (grew) and (grew) until… [P adult] T.J.: grew and grew and grew until I TOOK IT HOME! [P child] The turn-taking exercise drew the child’s close attention to the activity. He had to listen well in order to partake in the activity. The child’s verbal expression demonstrates that this four-year-old had fun interacting with the adult by filling in the blank left by Michelle. He started to become more and more comfortable while this exercise was unfolding. Toward the end, the child spoke more than one word: “grew and grew and grew until I TOOK IT HOME!” In this interaction, T.J. had to comprehend words spoken by Michelle, to comply with the rule of turn-taking, and to search for proper words deriving from his understanding of a science concept. This illustrated dialogue is in line with the perspectives of Machado (2007) and Otto (2006) that a child is able to acquire an understanding of changing roles in a dialogic conversation. A speaker can become listener and vice versa. Completing simple sentences or filling in missing words is a teaching technique promoting language skills (Machado, 2007; Otto, 2006).

Investigating the Child’s Interests: A Process Prior to teaching six-year-old Kolton, Trista was in search for an appropriate topic. In this interview process, the child was given an opportunity to share enthusiastically and impressively with Trista what he knew. The respect shown by the adult’s attentive listening to Kolton encouraged him to develop more linguistically (Machado, 2007) by unveiling his knowledge: T: What interests you in nature? [IQ adult LLQ adult] T: And what do you know about [it]? [IQ adult LLQ adult] K: Frogs—some are poisonous and some are not. [I child] K: They are different colors green, blue. [I child] K: Green are not poisonous. [I child] K: All frogs croak at dark. [I child] K: They eat flies and dragon flies. [I child] K: Butterflies—caterpillars turn into butterflies. [I child] K: Their colors are white, orange, and black. [I child] K: Crickets—Austin chases crickets, croak out in the dark, hide in dark places. [I child] K: Snakes—there are different kinds rattle snakes, king snake, some are poisonous. [I child] K: Uncle Andrew has snakes. [I child] K: Big, big enough to eat people, but maybe not eat mice, and different creatures. [I child] K: Spiders—hairy spiders, little spiders, daddy long legs protect people, stick to stuff, bite people, big hairy ones bite also. [I child] With “What interests you in nature?” as a starting question, the child seemed to receive a signal that the adult was here with great intention to listen to what he knew about nature. The

632

Ni Chang

right question elicited the flood of answers from the child, which consisted of 12 sentences (I child). Some of them have more than two verb phrases, e.g. “Austin chases crickets, croak out in the dark, hide in dark places,” “Big, big enough to eat people, but maybe not eat mice, and different creatures.” These 12 sentences also demonstrate the child’s sophisticated knowledge of different insects, amphibians, and reptiles that the child could fluently and eloquently describe to the adult listener. This interaction is also supported by Machado (2007) and OkenWright (1998) because open-ended questions result in more than one fixed and short response. As this particular question starting with a word “what” was posed to the child, the child’s mind was quickly working and his active thinking then enabled the child to use language to inform the adult of much that he knew about nature.

Reading The young participants acquired a science concept initially through a book reading. While interacting with the young participants, the adults read either an entire book or selective pages closely related to the science concept to assist the science concept inquiry. All the children’s books used in this study were informational picture books. In light to Machado (2007), non-fiction books can be of great assistance in offering science concepts or precise vocabulary. According to Machado, McViker (2007), Otto (2006), and Strasser and Seplocha (2008), reading enhances language and literacy competencies as children are exposed to the progression of texts from left to right and book reading enables them to understand that a book is read from the front to the back. Children may also become aware that it is the print, rather than pictures in the book, that carries real information about the story. Reading aloud supports children’s observation of how written language works and how one goes about reading (Otto, 2006). As children become involved in reading or begin reading aloud wholeheartedly, they are well on their way to become committed readers (Johnson, 2009).

Thinking while reading While listening to an adult reading aloud, children commented on things that related to their life experiences or background knowledge to aid their comprehension. When talking about the physical features of an ant, 4-year-old MyKaila shared with Heather what she knew about the care of her eyes: “Eyes. Don’t touch the eyes.” The book was appropriate to this young child’s level of understanding. She was able to decipher the adult’s oral language that assisted her in understanding the book content. The appropriateness also was reflected through the child’s ability to attempt to understand the reading content by linking her lifeexperience. This is encouraging to the child, who may want to learn more from books in her future endeavors: “When children can make a connection between what they hear when adults read aloud and their personal lives, they are more likely to be involved, interested, and constantly learning from their books” (Johnson, 2009, p. 48). In Silina’s reading only two pages of the book, “Baby Whales Drink Milk” by Barbara Juster Esbensen, a six-year-old Geron, could say five or six sentences by commenting on things that were familiar to him. S denotes the teacher and G represents Geron.

Children’s Language and Literacy Learning during a Science …

633

S: [S reads] Humans are mammals. Cats, dogs, and horses, and pigs are mammals too. G: I got two dogs. [P child] G: Every time we take the little one for a walk, she gets hyper. [P child] G: Then we put her back into the house. [P child] G: She tries to bite my ankle off! [P child] S: Oh no! That is not good! [P adult] ... S: [S continues to read] They all feed their babies milk that comes from their bodies. Whales are mammals. Whale babies drink milk too. G: My little sister used to drink milk, but she doesn’t now. [P child] Geron tried to make sense of what he was hearing by relating it to his own experiences and observations. The child’s inner-world was clearly manifested through his self-talk. A carefully selected book, suited to the child’s individuality and cognitive capacity, elicited so many comments from the child (P child) concerning his background knowledge. Experiences of this type are supported by Johnson (2009), Machado (2007), and Ruddell (2004). The interaction between adult and child provides a scaffold for the child to construct meaning, which helps enhance their comprehension if the book offers content relates to the child’s daily experience (Machado, 2007). In studying what snakes need to survive, Katie, a seven-year-old child, read a book, “All About Snakes,” by Martha E. H. Rustad.” Her interest and a high degree of concentration made her stop the reading several times. The book engaged her in active thinking: “Do snakes really do this?” Or “do snakes really need that?” Even though her second drawing fully reflected her mastery of the science concept, Katie was also able to self-correct herself on the shape of a snake’s tongue: “I made sure to get the tongue right this time (see Figure 4).

Figure 4. Katie’s awareness of the appearance of a snake’s tongue.

634

Ni Chang

I saw in the book that I didn’t draw it right the first time” (see Figure 5). This book meticulously selected by the adult based on the child’s interest and her prior knowledge enabled the child to build new knowledge on what she already knew. It is consistent with the strategies supported by Johnson (2009) to improve comprehension through making connections by combing new with old after the child synthesizes recently encountered information. Communicating the science concept, verbally describing it, graphically writing it down, and talking about the drawing were a sufficient base for the child to grasp the newly acquired concept (Paquett et al., 2007).

Figure 5. Katie’s understanding of the snake’s tongue prior to learning.

Voluntary Repeating Silina, an adult, read the title: “Baby Whales Drink Milk” Geron, a six-year-old child, voluntarily repeated, “Baby whales drink milk.” In this way, Geron was able to begin to establish a science concept about one characteristic of a mammal. In a different case, Hilina reported that MyKaila, four years old, was very much interested in what was being read to her. Listening to a book, “Ants” by Coughlan, Mykaila kept repeating the key words off of each page. For example, page 7 says, “Ants are black and brown.” MyKaila repeated, “Black and brown.” These two young children were engrossed in the books. The title or the content of the book captivated their attention. Their willingness to repeat after the adults provided them with practice involving semantic, syntax, phonetic, and phonological knowledge (Machado, 2007; Otto, 2006). In Geron’s case, the book title of “Baby whales drink milk” includes a plural noun and a complete sentence structure with a subject, verb, and an object.” In MyKaila’s case, the child was exposed to four aspects of language as the sentence, “Ants are black and brown,” contains a complete sentence with a subject and predicate adjectives. In addition, with the aid of picture, the child might have enhanced her concept of colors, black and brown.

Children’s Language and Literacy Learning during a Science …

635

The verbal imitation helps enhance the child’s phonologic knowledge and provided a way for a child to increase language acquisition. Repeating new words is an attempt for the child to become familiar with them (Machado, 2007). In this sense, book reading opens the door to literacy and is an opportunity for the child to gain new vocabulary and information. Clay (1991a) and Machado (2007) believed that, most likely, children would use the language heard or acquired from book reading in their futures book talks.

Requested Repeating Sometimes, an adult asked a child to repeat words during reading aloud. The following is an example showing how Ashley asked the child, Tyler, a five-year-old child, to read after her. A: That’s right, there are three body parts. They are called the head, thorax, and abdomen. Can you repeat those words after me? A: Head (pointing to the head) T: Head A: Thorax (pointing to the thorax) T: Thorax A: Abdomen (pointing to the abdomen). T: Abdomen. ... Repeating after an adult enabled the child to practice pronouncing new words and to sustain the child’s listening. The child had to pay attention to the reading aloud in order to follow the direction. The words, such as “thorax” and abdomen,” are relatively hard to say and are rarely used in a child’s daily life. Reading after the adult served as a signal to the child that these words were important. While saying these words, the child would intentionally look at the corresponding pictures to try to comprehend the meaning of these words. While the child was able to enhance her phonological skills, she also had a chance to increase vocabulary and pragmatic knowledge (Otto, 2006). In addition, this is a technique for an adult to encourage the child to learn new words. This strategy was also supported by Machado (2007) in that repetition is necessary in teaching a new word to a child.

Book Impact on Learning Anna intentionally drew the attention of Olivia, a six-year-old child, to the book reading with the following words after listening to the child’s explanation of her first drawing (see Figure 6): A: Do you think dolphins eat other things besides seaweed and fish? O: [Thinking] I don’t know A: Well why don’t we find out. I brought a book to read to you today that talks about dolphins and what they eat.

636

Ni Chang

Figure 6. Book impact.

With intrigued interest in what a dolphin eats, the child paid undivided attention to the book reading. Concept or informational picture books tell a story about a specific concept through pictures or illustrations, helping children think about ideas (Johnson, 2009). The effect of the book reading was reflected evidently in a hands-on activity. Olivia was asked to choose among several cut-outs, such as apples, carrots, doughnuts, cakes, fish, shrimp, squid, seaweed, turkey, and chicken, to glue on a piece of blue construction paper that represented the sea to demonstrate and construct the science concept of the diet of a dolphin. When asked to rationalize why apples, carrots, doughnuts, cakes, seaweed, turkey, and chicken were not the candidates for the construction paper, Olivia responded very loudly, “Because that’s not what the book said!” The empowerment of the book on a child’s learning can also be demonstrated through the following instance. After Mindi (a six-year-old child) answered all the questions about the growth cycle of a plant, Nina asked her how she was so knowledgeable of the topic. Mindi attributed her knowledge to the book: N: . .[C]ould you tell the stages of a seed growing, starting with the seed? [IQ adult LLQ adult] M: Yep! First, there is the seed that is in the ground under the dirts and soils, then it gets fat and the roots and hair roots grow, it comes out of ground and grows leaves. [I child] N: That’s right, and how did you know all of that? [IQ adult LLQ adult] M: Cuz I learned it in that book. [I child]

Children’s Language and Literacy Learning during a Science …

637

These two examples demonstrate the children’s mastery of the concepts. Additionally, reading aloud not only was a remarkable source of listening activity to a child, but also an important first step for a child to move to literacy (Machado, 2007). These two children benefited from the book reading by having an opportune time for building vocabulary and extending phonological awareness. Book reading also enabled them to be exposed to morpheme, syntax, semantics, and a literate form (Machado, 2007; Snow & Tabors, 1993). The child’s comprehension was consistent with the belief held by Cullinan (1992) and Machado (2006) that picture books could have a long lasting effect, helping children remember events by the meaningful framework the books offer.

Learning to Value Books and the Library When Brenda was ready to read a book to Jasmine, the six-year-old girl requested, “Can I read it?” B: Sure. As you read it, pull on your string to see the different stages the seed goes through to become a pumpkin. [P adult] J: How about you read it and I’ll pull the string. [P child] Brenda knew that Jasmine could read and comprehend the poem. With that knowledge kept in mind, Brenda suggested they would both read it. B: How about we take turns reading? [P adult] B: You start with the first line, and then I’ll read the next. [P adult] J: OK. [Y child] Jasmine read first. After reading the first line, she decided to keep going. As she arrived at the parts where the seed changed into a plant, she tugged on the string to make the leaves appear. She read some more until the orange pumpkin appeared. J: This is cool! I am going to show this to my teacher. [P child] B: That’s a good idea and you can tell her what you have learned about the lifecycle of pumpkins. [P adult] J: Can I take the book too? [LLQ child] B: How about I write down the title and who wrote it, then she can get it at your school library? [P adult] J: Yeah and I am going to check the book out when we go to the library. [P child] In this interactive communication, negotiation led the child to have an opportunity to read the book while an adult was listening. The dialogue modeled for the child the social skills in making a request and a suggestion, if viewed necessary: “Can I read it?” “How about you read it and I’ll pull the string.” In this interactive dialogue, the child did not only express her own thoughts, but also asked a question. In addition, she demonstrated her capability to use somewhat complex sentence structure in unveiling her thinking: “How about you read it and I’ll pull the string.” This idea was confirmed by the adult with a similar sentence structure: “How about we take turns reading?” It further taught the child about turn-taking. Toward the

638

Ni Chang

end, the child’s utterance is more advanced, showing a complex sentence structure: “I am going to check the book out when we go to the library.” The enjoyable discovery by the child heightened her interest in books and fed her with the knowledge of how interesting information can be found and how books can be used in searching for answers. The joy of book reading frivolously paves the way for establishing a formal literal form as books contain complete sentences with correct semantics and syntax (Machado, 2007). In addition, the interaction reinforced the idea of the usefulness of a library: “Yeah and I am going to check the book out when we go to the library.” This learning opportunity not only enabled the child to develop the science concept, but also exposed her to the value of the book and the usefulness of a library.

Writing Literacy involves a series of strategies young children use in their attempts to comprehend and produce oral and written language (Machado, 2007; Teale, 1995). It is imperative that adults help young children increase their awareness that both written or pictorial marks and print words carry meaning (Machado, 2007; Wells, 1981).

Phonological Learning Ruby emphasized phonological awareness while helping Austin (a five-year-old child) acquire a science concept by soliciting information and by legibly writing down what the child told her. R: Oh. I see. [P adult] R: Alright, I’ll write and you tell me what to say. [P adult] A: (points to the largest elephant) R: You want to tell me about this elephant first? [LLQ adult] A: Happy. Happy Elephant. Happy Elephant [I child] R: You want it to say Happy… [CQ adult] A: This, is, a, ha-ppy, el-e-phant. [I child] R: (writes) Okay. Anything else? [LLQ adult IQ adult] A: (points to the smallest elephant) on this one…um, mad. This, is, a, mad, el-e-phant.[I child] R: (writes) A: (points to the middle sized elephant) Okay, on this one it’s a sad el-e- phant. Sad. [I child] R: (writes) This, is, a…. [P adult] A: (joins in) sad, el-e-phant. I child R: Elephant. Okay. Any other words you want on your paper? [LLQ adult IQ adult] R: No. [I child] Ruby was dictating the child’s utterances, facilitating the child’s phonological awareness by working as a scribe (Machado, 2007; Oken-Wright; 1998; Otto, 2006). While writing down the child’s utterances, Ruby deliberately sounded out each of the syllables to facilitate the child’s phonemic awareness (Machado, 2007).

Children’s Language and Literacy Learning during a Science …

639

Requested Matching of Printed Labels to Pictures Engaging Abbi (an eight-year-old child) in the learning of the ladybug’s life cycle, Melissa invited the child to not only draw, but also label her drawing. M: Abbi will you please draw me another picture of what the life cycle of a ladybug is? [LLQ adult] A: You mean will I draw you a picture of what I KNOW the life cycle of a ladybug is? [CQ child LLQ child] M: Yes, that’s what I mean. [P adult] M: If you could please label what you are drawing, that would be of great help. [P adult] [five minutes later] A: Okay, here is the picture, and it is labeled too. [P child] Melissa creatively incorporated literacy skill building into this content learning opportunity. The learning of the science concept apparently was a suitable opportunity for the child to use both types of written languages (drawing and conventional writing) to express what she knew. This palatable technique is consistent with Machado’s (2007) suggestion to intentionally incorporate skill building into daily conversations and daily activities. When engaging the child in learning, Jaime invited McKenzie (a six-year-old child) not only to place cards with visual images, but also to place, in order, those cards with printed words only: J: Now let’s try an activity I brought along for you to try. [R adult] J: On these cards are [pictures of] the different stages of the life cycle of a pumpkin. [P adult] J: Can you put them in the correct order of how a pumpkin grows? [LLQ adult] M: O.K. (put cards in order after 18 seconds). [Y child] J: Good job! [P adult] J: On these cards is the name for every stage. [I adult] J: Do you think you can put them above the right picture? [LLQ adult] M: [In 21 seconds, she puts the cards above the correct stage as Jaime reads her the different words] This was a meaning-centered activity (Puckett & Diffily, 2004), in which the child not only was exposed to listening and was encouraged to think, but also had an opportunity to increase the awareness of print and phonetic awareness. Jaime’s explanation of this activity was fairly long. The three whole sentences were comprised of two statements and one question. Without paying attention and having the ability to comprehend the requests, this young child would not have been able to respond when it was her turn to speak: “O.K,” nor would she have manipulated the cards according to the instructions. Visual representations are a semi-abstract means (Charlesworth & Lind, 2006) for a child to acquire knowledge conceptually, a step closer to abstract thinking and learning. The child’s capability of ordering these cards in a sequential order represented her complete grasp of concepts as a result of corresponding mental pictures that she had established in the process of learning. It also shows that she was at the threshold of being able to think semi-abstractly. Her way of thinking further was promoted by the cards printed with words only. At this point, the child

640

Ni Chang

must listen carefully to each of the words and must then quickly relate to a just acquired concept, in order, to make a decision of where this particular card should be placed. This activity matched Machado’s (2007) recommendation concerning the integration of language arts into daily activity and conversation. Again, this is consistent with the notion made by Machado and Otto (2007) that literacy development is enhanced when children are provided with opportunities to connect images and words.

Requested Labeling of Pictures When Shannon was asking Erik, age 7, if she could take his drawing with her to show her teacher, Erik’s desire to write became evident (see the ensuing dialogue shown below): S: You have made a beautiful picture Erik and you have gotten all the stages of the lifecycle correct. [P adult] S: Would you mind if I took this to school and showed my teacher? [LLQ adult IQ adult] E: Nope. You can. [P child] E: But let me write something down here (opens the book that Shannon just has read to him to a page with the names of different butterflies and starts to write down a specific type). [P child] E: What does this say? (He points to a dragonfly). [IQ child LLQ child] S: That is an island paradise dragonfly and [it lives] in Madagascar. [I adult] E: Okay, well! That’s the kind that I drew so I want to put its name on the paper. [I child] E: Can I do that? (Erik adds colors to his adult dragonfly that resembles a dragonfly shown in the book). [IQ child LLQ child] (see Figure 7) His immense interest in the dragonfly urged him to examine closely dragonflies shown on every page while listening to the book being read aloud. The visual image of the Island Paradise dragonfly captivated the child’s attention. The uniqueness of this dragonfly, from the child’s perspective, inspired him to want to write. Even though the dragon’s name is fairly complex, it did not seem to deter this seven-year-old child. This is consistent with the position of Caswell and Duke (1998) and Johnson (2009) that a child’s desire to read and write enlarges, if a non-fiction book caters to a child’s interest. In a similar case, Tracy’s child partner, Kolton, at the age of 5, labeled each figure he just drew, such as “egg, tadpole, frog leg, and adult frog” (see Figure 8). When interviewed by the adult about why he went a few “extra miles” to do the writing, Kotlon responded confidently, “Because the cards had some labeling.” The cards that Tracy designed were used to enable Kolton to construct the science concept of the dragonfly’s life cycle. Impacted by the words in addition to pictures printed on these cards, the child was motivated to write. This demonstrates that if a strategy employed by an adult is well suited to a child’s background knowledge and daily life experience, the eagerness of learning becomes a catalyst in inspiring the child to move beyond what had been expected by the adult. The “extra miles” strode as it were by the child benefited his literacy inquiry. It was due to the appropriateness and meaningfulness of this activity that this child became an active learner who delighted in the activity, which further intensified an early literacy foundation (Gundling, 2002, Machado, 2007).

Children’s Language and Literacy Learning during a Science …

641

Figure 7. Voluntary writing.

Figure 8. Voluntary labeling.

Megan’s child partner, Rylee (a four-year-old child), voluntarily wrote, “To Megan From Rylee” on her second bug picture. The act of a drawing promoted this young child to practice her writing skills (see Figure 9). The child’s liking for her teacher encouraged her to write, an attempt to engage in early literacy (Machado, 2007; Oken-Wright, 1998).

642

Ni Chang

Figure 9. Voluntary Writing.

Encouraged Writing Alie, a six-year-old child, was asked to sound out and to write a few new words with Jennifer’s help while explaining her first drawing on a science concept of the life cycle of a frog: J: Can you tell me about your picture? [LLQ adult] A: This is a baby frog. (She pointed to the smallest frog) [I child] J: Do you think that you could write “baby frog?” [IQ adult LLQ adult] A: Yep (She writes ‘baby.’) [Y child] J: Great! [P adult] A: This is a frog growing up. (She points to the medium frog. [I child] J: I bet you could write ‘growing up’ with my help. [P adult] A: Yep (She writes ‘gr’ and Jennifer helps her with the rest.) [Y child] A: This is the mommy frog. (She points to the largest frog.) [I child] J: Let’s write it. (She writes ‘mom.’ ) [R adult] (see Figure 10)

Children’s Language and Literacy Learning during a Science …

643

Figure 10. Encouraged Writing.

The visual images played an important role in the child’s inquiry of a science concept that kept the child’s attention and promoted her conceptual learning in science. Simultaneously, Jennifer capitalized on the child’s deep interest in the topic to advance her literacy competency by encouraging her to label her pictorial language. This first drawing, used as an assessment tool to gauge the child’s prior knowledge (Chang, 2007), not only showed how much the child knew of this science concept, but also enabled the child to be aware of quite a number of new written words. This opportunity helped increase Alie’s semantic and phonemic knowledge (Otto, 2006).

Requested Name Writing When her child partner completed a drawing, Jessica encouraged Anthony, a 5-year-old child, to write his name and a number: J: Can you put your name on it? [LLQ adult] A: Is that a good N? [I child LLQ child] J: That’s great. [P child] J: Can you put a one on the picture so we know that it’s your first drawing? [LLQ adult] (see Figure 11).

644

Ni Chang

Figure 11. Requested name writing.

Printing a child’s own name on artwork is one of the beneficial activities that are fully endorsed by researchers and factually practiced by many early childhood teachers as this activity helps lead young children to the path of early literacy (Machado, 2007; Otto, 2006)

Recognizing and Enlarging Vocabulary The learning process not only afforded an opportunity for a child to talk about the concept just learned, but to listen to the instructions in order to know how to engage in the activity provided. In the meantime, this was also an opportunity for a young child to check his level of understanding about what he knew with an adult’s presence while acquiring new words and increasing the awareness of phonology. Sabrina put out a series of cards for Josiah (a six-year-old child) to place in a sequential order to represent the grow cycle of a tree. Soon after the cards were shown, Josiah immediately pronounced what he knew:

Children’s Language and Literacy Learning during a Science …

645

S: Now, you are going to put the labels where they go. [R adult] J: Okay. I think I know what this spells: seeds. [P child] S: You are right. [P adult] S: Can you put it with its picture? [LLQ adult] [Josiah glues the word SEEDS under the picture of the seeds. Sabrina helps Josiah read the rest of the labels. Josiah glues them with the corresponding pictures without assistance.] [see Figure 12]

Figure 12. Recognizing vocabulary.

In the process of learning the science concept of the water cycle, Dominic, a six-year-old child, learned how to say three big words: evaporation, condensation, and precipitation with Bridges, a caring adult: B: . . . Will you explain what’s happening in your picture? [IQ adult LLQ adult] D: Yup! Water, cloud, rain. (Points to each part) It goes like this see (Traces the order with his finger) [I child] (see Figure 13) B: What are those stages called again? [RQ adult LLQ adult] D: Vaprtization, condentration, and precipitation. [I child] Although Bridges was satisfied with the child’s mastery level of the concept, as Dominic was able to describe it with the words that he felt comfortable using, the adult utilized the moment to talk about the child’s drawing to reinforce the three newly acquired words she did this despite the fact that the adult knew it was very challenging for the child to master these words within this lesson. Assisting a child in learning new words is one of commonly identified critical behaviors that a teacher has in facilitating the child’s language acquisition (Machado, 2007).

646

Figure 13. Enlarging vocabulary.

Ni Chang

Children’s Language and Literacy Learning during a Science …

647

Visual Literacy/Viewing The young participants were requested to produce two drawings throughout a lesson with one at the very beginning as a pre-assessment tool and with the other at very end as a postassessment tool (Chang, 2007). Drawing is viewed as part of the visual arts, which is also termed, visual literacy. Visual literacy or viewing is the fifth language arts area. It concerns a child’s perception of the world and how he or she sees and interprets viewed images. “. . .art offers a rich and easily accessible way to express meaning” (Soundy, Guha, & Qin, 2007, p. 82). The ability to understand and produce visual messages involves visual literacy (Machado, 2007).

Drawings The first drawing made by Kevin (a five years old child) increased the child’s desire to know more about physical characteristics of a spider. Following his explanation of the drawing, Hannah, the adult, was ready to read aloud a relevant book to him. While she was reading the book title and its author and illustrator, the spider on the cover of the book immediately caught Kevin’s attention and spurred him into action. He immediately grabbed his picture and drew another circle for the body (see Figure 14) Kevin’s reaction toward the visual image presented on the book cover demonstrates the child’s ability to think visually. Visual thinking is useful in “improving children’s cognitive, reading, writing, and creative skills” (Machado, 2007, p. 164). Jennie worked with Samantha, a four year old child, on the life cycle of a pumpkin. Samantha volunteered to turn every page while the book was being read. Jennie observed that the child was able to visually think of the picture on each page and to interpret its content (see the conversation below): J: Do you know what you will find when you cut the top part off of the pumpkin? S: Duh! Seeds! [Laughing] J: How did you know that? S: From the picture in the book. The act of page turning assists a child in his or her literacy development (Machado, 2007). In this interaction, turning pages focused the child on listening to the book reading, which, in turn, was advantageous to her knowledge building about the relationship between pictures and print. Visual images in the book are a language (Machado, 2006), helping to formulate mental pictures in a child’s mind. Incorporating visual literacy into other language arts areas, such as listening, speaking, reading, and writing, has been advocated by Machado (2007) and Morrow and Asbury (2003).

648

Ni Chang

Figure 14. Drawing.

Talking About Drawing Kate (an eight years old child) was very articulate about what a snake needs to survive. Her explanation of the drawing also explicitly displayed that the child had a clear idea of what color a snake was portrayed by the book. Even though the color purple dominated this drawing, the adult understood that the noticeable dominance of purple did not represent the child’s misunderstanding of the color of a snake, because purple was one of the child’s favorite colors. H denotes the adult and K denotes Kate, the child. H: Can you tell me . . . about what you have drawn this time? [LLQ adult] K: Sure! I drew the sun in the corner, because snakes need heat. [I child] K: Then, I drew some rain, lots and lots of rain, because snakes need a lot of water. [I child]

Children’s Language and Literacy Learning during a Science …

649

K: This time I made my snake purple, because in the book it showed an orange snake, and well, I like the color purple. [I child] In comparison with her first drawing (see Figure 5), Kate’s second drawing (see Figure 4) shows the child’s increased listening and reading comprehension. An analysis of the child’s verbal representation showed that Kate had mastered the science concept. The finding was consistent with Paquett et al.’s (2007) research conclusion: “Listening and reading comprehension can be assessed by analyzing children’s visual, verbal, and written representations of their understandings” (p. 65). Talking about drawings is a strategy for assisting learners in combining their prior knowledge with the new information, stemming from the use of a non-fiction book and being presented by symbol systems, such as verbal discussion and drawing. Responding through art develops young children’s language skills (Short, Kauffman, & Kahn, 2000; Soundy, Guha, & Qin, 2007). The finding is consistent with Snow, Burns, and Griffin’s (1998) notion that activity, such as drawing, could provoke and direct children’s attention to the relationship between print and speech. A kind of exploratory play, drawings and scribbling are used by young children to communicate with others, as they convey their underlying understandings and represent their awareness about print (Machado, 2007, Sulzby, 1996). It paves the way to early printing and reading attempts, as children are helped to be aware that print symbolizes oral language and written marks hold meaning (Machado, 2007). Once “it is written, it can be read” (Kalmar, 2008, p. 89). Through their ability to decipher the created symbols verbally, their acquisition of speaking and visual literacy competencies are facilitated (Machado, 2007).

Drawing to Know Abbi (an eight-year-old child) was asked to draw the life cycle of a ladybug at the beginning of a lesson. It took the child about five minutes to complete the drawing. Melissa then invited the child to share the drawing verbally with her: “Okay, so that is what you think the life cycle of a ladybug is?” The child replied with great confidence: “Yes, I know what the life cycle of a ladybug is.” As the adult encouraged the child to say more about her drawing, the child interrupted, “Okay, wait a minute. Let me label them for you.” She wrote, “egg,” “with out spots,” “with spots,” and “steeped on” either above or beneath a symbol she created (see Figure 15). Yet, the full confidence about this particular science concept soon was questioned by the adult’s statement, urging the child to pay close attention to the book that was about to be read. M: Okay, so you think that ladybugs start as an egg, then it turns into a ladybug without spots, then it is a ladybug with spots, and then it gets killed by being stepped on? [I child] A: Yes, that is what the ladybug’s life cycle is. [P adult] M: Okay. How about I read you a book on the life cycle of a ladybug? [LLQ adult] A: Alright. Wait does that mean? [IQ child LLQ child] A: You are telling me I’m wrong? [IQ child LLQ child] M: I’m not saying you are wrong, but I don’t think you are completely correct. [P adult] A: Okay, let’s read the book, so I can learn about the life cycle and get it right. [P child]

650

Ni Chang

Figure 15. Learning to write.

The book reading was followed by discussion, questions and answers, and a hands-on activity focused on this science concept. The child was then invited to produce another drawing: M: Abbi will you please draw me another picture of what the life cycle of a ladybug is? [LLQ adult] A: You mean will I draw you a picture of what I KNOW the life cycle of a ladybug is? [CQ child LLQ child] M: Yes, that’s what I mean. [P adult] A: Okay, here is the picture, P child and it is labeled too. [P child] ... M: Thanks, Abbi. [P adult] A: Alright. Thanks for asking me. P child A: I really enjoyed doing that, and learning about the life cycle of a ladybug. [P child] The analysis of the pre- and the post-drawings reveals the mastery of the science concept, which was also evident in the findings of Paquett et al. (2007). The researchers upheld the strategy of Talking Drawing (McConnell, 1993) as it increased reading and listening comprehension. The strategy is comprised of several steps from having children unveil their prior knowledge of a topic that has been selected for instruction. The created graphic representations are verbally shared by the child artists. Instruction on the specific topic ensues that is followed by the production of a second drawing and by the comparison of two drawings made by the children during the process. It is believed that this is an effective framework for organizing, remembering, and constructing meaning from text, as one makes visual images within the mind while reading or being read to (Paquett et al., 2007; Reutzel,

Children’s Language and Literacy Learning during a Science …

651

2003). The functionality of the first drawing works as an avenue to exhibit children’s created mental pictures. The visual representations then are orally explained by the children, so the listener will know the real meaning of the graphic language. It is an essential step for a viewer to undertake, which is fully supported by Church (2005) and Paquett et al. (2007). The conceptual development demonstrated through the post-drawing primarily is attributable to book reading and discussion, which enhance student comprehension skills and broaden knowledge bases. Verbal exchanges about the content scaffolds children’s topical understanding (Paquett et al., 2007), thinking, language uses, and writing (Machado, 2007; Otto, 2006). In Abbi’s case, the child’s disequilibrium, derived from her pre-drawing, oral elaboration of the drawing, and the way Melissa invited the child to the book reading, stimulated the child’s thinking and encouraged the child to pay closer attention to the book. Resulting discussion about the book content furthered the topic understanding, extending her confidence to express what she knew about the science concept pictorially. She also had a chance to develop her writing and vocabulary; she was able to add “s” to “egg” and to correctly spell “larva” and “pupa” in her second drawing. The strategy of Talking Drawings is beneficial to enhancing emergent literacy as it involves talking and drawing. They both also may provide an opportunity to those who have difficulty reading to taste the fruit of success (Paquett et al., 2007).

Conclusion This study was intended to explore the opportunities that an integrated curriculum offered to facilitate language learning and how an integrated curricular approach mediated children’s language and literacy competencies while children were engaged in learning a science concept. The findings show that the children’s listening and speaking primarily was facilitated by answering questions posed by the adult participants. Questions were largely recitation questions (RQ) and information questions (IQ), which provided a basis for maintaining an ongoing conversation (Pine, 1994; Otto, 2006) and encouraged language development (Otto, 2006). Adults’ personal speech (P adult) also worked as another source of language to enhance children’s listening and speaking acquisition. In their acquisition of a science concept via a lesson that incorporated language arts and the visual arts, the young participants were offered many opportunities that could increase their language and literacy skills. The lesson included introduction, a request for a first drawing as an assessment means to determine prior knowledge, follow-up discussion after reading aloud a book, hands-on activity, and a request for a second drawing as an assessment means to formally gauge the learning outcome. The entire process of interaction is in line with Otto’s (2006) notion that language development activities need to be integrated into the curriculum. The integration was also in correspondence with the principles of developmentally appropriate practices in that teachers need to provide many opportunities for children to hear and to use language for different purposes (Bredekamp & Copple, 1997). This study also examined the facilitation of language and literacy competencies through discretely exploring each of the five language arts areas, namely, listening, speaking, reading, writing, and visual literacy. It was found that there were various opportunities available to the young children, assisting their acquisition of language and literacy skills. The young children’s listening acquisition was facilitated through following directions and learning

652

Ni Chang

mathematics, such as counting and ordinal numbering, recognizing words, and reading aloud in addition to listening to questions posed by adults. Even though speaking was a result of responses to the adults’ recitation questions (RQ), information questions (IQ), and the adults’ personal speech (P adult), it was also facilitated through discussing book content (a book walk), turn-taking and vocabulary learning, and through the investigation of a topic of interest to the particular child. During the time of book reading, the young participants displayed various behaviors which were conducive to literacy development, including thinking while reading, voluntary repeating, requested repeating, book impact on learning, and learning to value books and the library. The act of reading aloud nonfiction texts increases conceptual development and vocabulary (Elster, 1994; Otto, 2006) as picture and/or story books expose children to more precise and varied vocabulary than other sources do. Children’s written language was facilitated over the course of the integrated lesson through the following aspects: phonological learning, requested matching of printed labels to pictures, requested labeling of pictures, encouraged writing, requested name writing, and recognizing and enlarging vocabulary. The children’s visual literacy acquisition was facilitated through drawings, talking about drawing, and drawing to know. The findings suggest that teachers should read pictures to children in a children’s book, provide opportunities for children to arrange pictures in a sequence that tells a story (Machado, 2007), pique children’s interest in libraries, use new words with children, extend children’s comments through questioning, focus children’s attention on an analysis of books read to them, engage children in intellectual challenging conversations (Dickinson & Tabors, 2001), obtain and maintain children’s attention (Dickinson & Tabors, 2001), support children’s writing attempts (Dickinson, 2001), give individual children adequate time to speak (Schickendanz, 2003), engage children in extensive conversations (Schickendanz, 2003). A body of literature recognizes the importance of infusing the visual arts into the curriculum, because it assists young children to visually express themselves and to communicate with others. Yet, when communicating with young children, the focus often seems to be placed on concepts, such as shapes, colors, and texture. Language goals for art activities focus on encouraging children to develop conceptual knowledge related to art and to be able to verbally describe the colors, textures, or shapes with which they are working. . . . Children should also learn the concept name of the particular technique they are using. Rather than labeling the technique as simply “painting,” a teacher should explain the type of painting involved, such as sponge painting, string painting, finger painting, block painting, or brush painting. Otto, 2006, p. 203 This study points to a different direction when it comes to the use of art center or visual arts. In enhancement of language and literacy skills, “talking about drawing” is a form of visual arts conversation taking place at the art center, which enables the child to express her thoughts and mental images verbally. Considering young children share their inner world in a symbolic manner, children’s paintings can represent meaning, which, in turn, can be talked about. This behavior is more appropriate than simply placing its focus on experimenting with different colors or novice art tools when children are engaged in painting. This study shows it is possible to create an art center and visual art experiences to serve more than one purpose. Through this study, the young participants were able to put words to their experiences via visual art/drawings, the symbolic language, because young children express and learn through

Children’s Language and Literacy Learning during a Science …

653

symbols (Edwards, Gandini, & Forman, 1998; Otto, 2006). If texture, shapes, and colors are part of the curriculum goal, creative adults or teachers can flexibly integrate the concepts into the children’s learning process. At times, the exploration of various mediums of artistic expression can be artfully linked with subject areas-related inquiry. If children are encouraged to use writing throughout the curriculum for various purposes in different learning centers in an early childhood classroom setting, integration of visual arts then makes a lot of sense. Drawing should take place not only in a writing center, where the exploration of communication through symbols takes place (see Otto, 2006), but also in other locations and occasions as well. Dialogic interactions then widen learning scope and spectrum, making learning meaningful, purposeful, interesting, and educational.

Implications for Education This study offers several educational implications for teachers and adults working with young children. Children’s learning experiences should be meticulously and thoughtfully planned in light of their prior knowledge (Kalmar, 2008), interest (William, Hedrick, & Tuschinski, 2008), and present conceptual understandings in order to motivate them to learn (William et al., 2008). Inquiry of science concepts can be integrated into curriculum with the premises that children’s interests are clearly underscored and understood and that the topic of study is interesting to the majority of the children involved in the learning process. The lesson should be orchestrated and organized such that it is coherent and logical, and considers all possible subject areas that will support and buttress the curriculum. “To promote literacy, teachers think about how each theme activity involves listening, speech, reading, and writing and how to logically connect these areas during ongoing activities” (Machado, 2007, p. 191). The national Reading Panel Report (2000) and Morrow and Asbury (2003) urged teachers in early literacy to organize instruction in a systematic way. Planning also involves the consideration of a meticulously and thoughtfully selected book that is accurate in science content. The decision of book selection should be in line with the child’s prior experience (Otto, 2006), age, interests, and background knowledge (Chang, 2007). Professional literature stresses the importance of the impact that an integrated approach to language arts has on outcomes for learning and recommends a literature-based curriculum for both elementary and preschool settings (see Machado, 2007). Informational or non-fiction books, employed in the literacy/literature-based approach, are supportive of thematic teaching in the early childhood field, and are of great value to literacy development (Machado, 2007). Early childhood teachers can use this teaching strategy to start discussion, drama, art, and other language arts activities to strengthen various concepts inquiries, which can, in turn, spark children’s desire to explore what interests them, to seek out answers, to ponder questions, or to provide a response (Machado, 2007). Engaging children in reading is equally important in the child’s acquisition of a science concept and language and literacy competencies. Motivation is the heart and soul of engagement: “[a]s motivation increases, students want to spend more time reading” (Johnson, 2009, p. 48). It is an adult’s responsibility to create and seek out strategies suitable for reaching this end. An excellent book selection is the key to motivating young children to participate in learning actively.

654

Ni Chang

Considering that a child is a novice communicator, an adult needs to play a crucial role in supporting language acquisition in a dialogic dyad (Brunner, 1990; Vygotsky, 1978; Otto, 2006) to make a scafffolded conversation successful. After a read-aloud, a discussion is useful to enhance the child’s listening, speaking, thinking, comprehension, phonetic, and phonemic acquisition. It would be helpful for the adult to ask question about the child’s learning, including what the child is thinking, how further learning experience be built on as a result of this interaction. Listening to children throughout a lesson, emphasizing what is said rather than how it is said, is essential and valuable to the continuation of discourse between adult and child. With intimate listening, adults are able to be on the same wavelength with children. The absence of the adult who monopolizes verbal interactions not only encourages children to talk about what they know, but also assists them in putting ideas into words. Conversations are likely to be sustained when adults try to talk and pose questions pertaining to children’s lines of thought. To maintain children’s attention to an ongoing conversation and an inquiry regarding a science concept, shared reference between an adult and a child should be one of the key elements (Otto, 2006; Tronick, Als, & Adamson, 1979). A jointly interesting object or event, such as drawing and a science concept, can gain the attention of both the adult and the child (Otto, 2006), making conversation meaningful and purposeful. Having children use drawings to express what they know and what they have learned plays an important role in an integrated learning process. Arts “are visually expressive and communicative by nature, they also provide opportunities to enhance language development” (Otto, 2006, p. 203). Children are able to experiment with and explore how written language is used in specific contexts (Otto, 2006). Displaying drawings or art products in the classroom or storing them in the students’ journals for later reference would allow teachers to gain very little information about children’s insights and their learning. Asking students to demonstrate their inner thoughts through a unique drawing is much more meaningful and beneficial than having them produce a paragraph in writing about a particular topic determined by the teacher (Paquett, Fellow, & Jalongo, 2006/2007). Talking about drawing is useful and necessary as the pictures may represent very different meaning to children, compared to an adult’s interpretation of the same drawing. While talking with the child about his or her drawing, the adult provides a scaffold for the child’s language and literacy acquisition and helps those who struggle with reading to experience success (Paquett et al., 2006/2007) while, at the same time, obtaining some knowledge of the child’s mental picture.

Recommendation for Future Research Effort Research on visual literacy as a means of support to children’s learning in language arts and other subject areas should continue in its scope and depth as more and more children are attracted by TV, videos, and computer games and many early childhood teachers pay more attention to oral language and written text than to visual thinking and learning (Karchmer, Mallette, & Leu, 2003). Research is also needed to examine the effect of visual literacy on later elementary reading achievement (Roskos, Christie, & Richgels, 2003) and on the relationship between non-fiction books and children’s learning (Casergue & Plauche, 2003). The effect resulting from the well-selected book on children’s science concept inquiry and the

Children’s Language and Literacy Learning during a Science …

655

development of language and literacy acquisition should also be an area of further investigation. The replication of this study is useful and necessary. Future effort may also be exerted to examine the effect of the same approach employed by this study on a small group of children.

Significance of the Study While the public eye is focusing on children’s reading and writing achievement and classroom instruction in reading and writing, the focus on instructional time to enhance children’s reading and writing skills is inappropriate, as it is not the remedial solution for children’s reading deficiency in US schools (Hirch, 2004). Researchers (Machado, 2007; Morrow & Asbury, 2003; Otto, 2006) and the National Reading Panel Report (2000) mutually emphasized the essence of systematic and well-orchestrated instruction concerning early literacy. Equally recognized by the professional literature is the important role that an integrated approach to language arts plays in children’s learning achievement, because children benefit from the exposure to background information in history, science, literature, and arts. Literacy instruction should be seamlessly blended into subject matter learning with reading and writing working as a tool for students’ content learning (Fisher & Ivey, 2005; Lacina & Watson, 2008). It is, therefore, recommended that a literature-based curriculum be employed in both elementary and preschool settings, as students make sense of content knowledge with the use of listening, speaking, reading, writing, and viewing (Fisher & Ivey, 2005) and as the acquisition of language and literacy skills derives from content-rich-readingselection curricula (Hirch, 2004; Machado, 2007) as well as from situating children in a meaningful learning context (Machado, 2007). Moreover, Kalmar (2008) and Pappano (2008) called on early childhood educators to give children something to talk in order to develop their literacy. On the basis of aforementioned reasons, the results of this study answer all these calls. In this learning experience, the young participants acquired a science concept with the support of a combination of five areas of language arts, including listening, speaking, reading, writing, and visual literacy. The children had ample opportunities to express thoughts verbally. In addition, this study makes a contribution to learning a science concept in an enjoyable and somewhat lighthearted manner. This study can be helpful to elicit a more promising discussion about ways to facilitate young children’s acquisition language and literacy competencies while assisting them in learning a science concept through an integrated curriculum.

Limitation A limitation about this means of data collection might be that the researcher was unable to engage in follow-up interviews with the adults who were teaching the children. An additional limitation relates to the fact that the transcription might contain errors due to typos or careless mistakes. This limitation was minimized, to a degree, by the prolonged data collection and by additional means used in the data collection, such as drawings and communication notes between the researcher and the participants. This study also is limited in its generalizability to all children due to its small size and because the sample lacks diversity.

656

Ni Chang

The study, however, did make a contribution to the children’s learning in that they were exposed to five language arts and had opportunities to use language and to gain literacy competencies through an integrated curriculum that was systematically planned and organized and that was designed to help young children develop a science concept.

References Allington, R., Guice, S., Li, S., Michelson, N., & Baker, K. (1995). Patterns of implementation of literature-based curriculum (Report Series 1.14). National Research Center on Literature Teaching and Learning, Albany, NY. (ERIC Document Reproduction Service No. ED ED386731) Beals, D. E. (1993). Explanatory talk in low-income families’ mealtime conversations. Applied Psycholinguistics, 14, 485-513. Bredekamp, S. & Copple, C. (Eds.). (1997). Developmentally appropriate practice in early childhood programs serving children from birth through age 8. Washington D.C.: National Association for the Education of Young Children. Bruner, J. (1978). The role of dialogue in language acquisition. In A. Sinclair, R. Jarvella, & W. Levelt (Eds.), The child’s conception of language (pp. 241-256). New York: Springer-Verlag. Burmark, L. (2002). Visual literacy: Learn to see, see to learn. Alexandria, VA: Association for Supervision and Curriculum Development. Camarata, S. (1995). A rationale for naturalistic speech intelligibility intervention. In M. Fey, J. Windsor, & S. Warren (Eds.), Language intervention: Preschool through the elementary years (pp. 63-84). Baltimore, MA: Paul H. Brookes. Cabergue, R. M., & Pauche, M. B. (2003). Immersing children in nonfiction: Fostering emergent research and writing. In D. M. Barone & L. M. Morrow (Eds.), Literacy and young children: Research-based practices. (pp. 243-260). New York: The Guilford Press. Caswell, L., & Duke, N. K. (1998). Non-narrative as a catalyst for literacy development. Language Arts, 75(2), 108-117. Chang, N. (2007, October). A Medium for nurturing the development of language and literacy skills of young children at home. Paper presented at the meeting of the National Association of the Education of Young Children, Chicago, IL. Chang, N. (1996). The role of the teacher in children's acquisition of concepts based on Reggio principles and related theories. Unpublished doctoral dissertation, Vanderbilt University of Nashville, TN. Charlesworth, R., & Lind, D. (2006). Science and math for young children. Cincinnati, OH: Delmar. Cho, B. K., & Kim, J. J. (1998, April). Literature based science activities in kindergarten through children’s picture book. Paper presented at the Annual Conference of Association for Childhood Education International, Tampa, FL. Church, E. B. (2005). Learning from children’s art and writing. Early Childhood Today, 19(5), 35-42. Clay, M. (1991a). Becoming literate: The construction of inner control. Portsmouth, NH: Heinemann.

Children’s Language and Literacy Learning during a Science …

657

Creswell, J. W. (2002). Educational research: Planning, conducting, and evaluating quantitative and qualitative research. Columbus, OH: Upper Saddle River. Cullinan, B. E. (1992). Whole language and children’s literature. Language Arts, 69(6), 426430. Duke, N., & Pearson, P. D. (2002). Effective practices for developing teaching comprehension. In A Farstrup & S. J. Samuels (Eds.), What research has to say about reading instruction (3rd ed.). Newark, NJ: International Reading Association. Edwards, C., Gandini, L., & Forman, G. (1998). The hundred languages of children: The Reggio Emilia Approach-advanced reproach (Eds.). Greenwich, CT: Ablex. Fisher, D., & Ivey, G. (2005). Literacy and language as learning in content-area classes: A departure from “every teacher a teacher of reading.” Action n Teacher Education, 27(2), 3-11. Fries, P. (n.d.). Systemic approaches to child language. Retrieved August 1, 2007, from Developmenthttp://www.isfla.org/Systemics/Definition/ChildDev.html Fu, V. R., Stremmel, A. J., & Hill, L. T. (2002). Teaching and learning: Collaborative explorations of the Reggio Emilia Approach. Columbus, OH: Merrill/Prentice-Hall. Gambrell, L. B., & Mazzoni, S. A. (1999). Principles of best practice: Finding the common ground. In L. B. Gambrell, L. M. Morrow, S. B. Neuman, & M. Pressley (Eds.), Best practice in literacy instruction. New York: The Guilford Press. Gandini, L. (2002). The story and foundations of the Reggio Emilia approach. In V. Fu, Stremmel, & L. Hill (Eds.), Teaching and learning: Collaborative exploration of the Reggio Emilia approach (pp. 13–21). Columbus, OH: Merrill/Prentice-Hall. Genishi, C., & Dyson, A. (1984). Language assessment in the early years. Norwood, NJ: Ablex. Gundling, R. (2002). Promoting early literacy in early childhood program. Child Care Bulletin, 27, 7-8. Halliday, M. A. K. (1973). Explorations in the functions of language. London: Edward Arnold.Homeschool-curriculum-for-life.com (n.d.) Why read aloud to your child. Retrieved January 28, 2008, from http://www.homeschool-curriculum-for-life.com/whyread-aloud.html Huck, C.S. (1992). Literacy and literature. Language Arts, 69(7), 520-526. International Reading Association and National Council of Teachers of English (2006-2007). Literacy engagements. Retrieved November 14, 2007, from http://www.readwritethink.org/ literacy/index.html International Reading Association and National Council of Teachers of English (2006-2007). IRA/NCTE standards for the English Language Art Retrieved November 14, 2007, from http://www.readwritethink.org/standards/index.html Johnson, D. (2009). The joy of children’s literature. New York: Houghton Mifflin Harcourt Publishing Company. Kalmar, K. (2008). Let’s give children something to talk: Oral language and preschool literacy. Young Children, 63(1), 88-92. Karchmer, R. A., Mallette, M. H., & Leu, D. J. Jr. (2003). Early literacy in a digital age: Moving from a singular book literacy to the multiple literacties of networked information and communication technologies. In D. M. Barone & L. M. Morrow (Eds.), Literacy and young children: Research-based practices. (pp. 175-194). New York: Guilford. Katz, L. G., & Chard, S. C. (1996). The contribution of documentation to the quality of early childhood education. (ERIC Documentation Reproduction Services No. EDO-PS-96–2).

658

Ni Chang

Lacina, J., & Watson, P. A., (2008). Focus on literacy: Effective content teachers for the middle grades. Childhood Education, 84(3), 159-162. Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Newbury Part, CA: Sage. Machado, J. M. (2007). Early childhood experiences in language arts: Early Literacy. (8th Ed.). Clifton Part, NY: Thomson Delmar Learning. Malaguzzi, L. (1993). History, ideas, and basic philosophy. In C. Edwards, L. Gandini, & G. Forman (Eds.), The hundred languages of children: The Reggio Emilia Approach to early childhood education, (pp. 41–89). Norwood, NJ: Ablex. McCormick, T. W. (1988). Theories of reading in dialogue: An interdisciplinary study. New York: University Press of America. McConnell, S. (1993). Talking drawings: A strategy for assisting learners. Journal of Reading, 36(4), 260-269). McVicker, C. (2007). Young readers respond: The importance of child participation in emerging literacy. Young Children, 62(3), 18-22. Morrow, L. M., & Asbury, E. (2003). Current practices in early childhood literacy development. In L. M. Morrow, L. B. Gambrell, & M. Pressley (Eds.), Best practices in literacy instruction (pp. 43-63). New York: The Guilford Press. National Reading Panel (2000). Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction (NIH Publication No. 00-4769). Washington, DC: National Institute of Child Health and Human Development. National Research Council (1996). National Science Education Standards. Washington, D.C.: The National Academy Press. Oken-Wright, P. (1998). Transition to Writing: Drawing as a Scaffold for Emergent Writers. Young Children, 53(4), 76-81. Otto, B. V. (2006) Language development in early childhood. 2nd Edition. New Jersey: Upper Saddle River. Pappano, L. (2008). The power of family conversation: School and community programs help parents build children’s literacy from birth. Harvard Education Letter. Retrieved May 22, 2008, from http://www.edletter.org/insights/familyconversation.shtml Paquette, K. R., Fello, S. E., & Jalongo, M. R. (2006/2007). Talking drawings: Improving intermediate sutdents’ comprehension of expository science text. Childhood Education, 83(2), 80-86. Pine, J. (1994). The language of primary caregivers. In C. Gallaway & B. Richards (Eds.), Input and interaction in language acquisition (pp. 15-37). Cambridge, London: Cambridge University Press. Puckett, M., & Diffily, D. (2004). Teaching young children: An introduction to the early childhood profession (2nd Ed.). New York: Harcourt Brace. Reich, P. (1986). Language development. Upper Saddle River, NJ: Prentice Hall. Reutzesl, R. D. (2003). Teaching children to read: Putting the pieces together. New York: Prentice Hall. Rinaldi, C. (1998). Projected curriculum constructed through documentation Progettazione: An interview with Lella Gandini. In C. Edwards, L. Gandini, & G. Forman (Eds.), The hundred languages of children: The Reggio Emilia Approach-advanced reproach, (pp. 113–126). Greenwich, CT: Ablex.

Children’s Language and Literacy Learning during a Science …

659

Roskos, K. A., Christie, J. F., & Richgels, D. J. (2003). The essentials of early literacy instruction. Young Children, 58(2), 52-60. Royce, A., & Wiley, D. (2005). The common ground. Science and Children, 42(5), 40-42. Ruddell, R. (2004). Researching the influential literacy teacher: Characteristics, beliefs, strategies, and new research directions. In R. Ruddell & N. Unrau (Eds.), Theoretical models and processes of reading (5th ed.) (pp. 979-997). Newark, DE: International Reading Association. Short, K. G., Kauffman, G., & Kahn, L. H. (2000). “I just need to draw:” Responding to literature across multiple sign system. The Reading Teacher, (54), 160-171. Snow, C. E., & Tabors, P. O. (1993). Language skills that relate to literacy development. In B. Spodek & O. Sracho (Eds.), Yearbook in early childhood education (pp. 222-246). New York: Teachers College Press. Snow, C., Bruns, S., & Griffin, P. (Eds). (1998). Preventing reading difficulties in young children. Washington, DC: National Academy Press. Soundy, C. S., Guha, S., & Qin, Y. (2007). Picture power: Placing artistry and literacy on the same page. Young Children, 62(3), 82-88. Stauffer, R. C. (1970). The language experience approach to the teaching of reading. New York: Harper and Row. Stegelin, D. A. (2003). Application of the Reggio Emilia approach to early childhood science curriculum. Early Childhood Education Journal, 30(3), 163-169. Stieglitz, M. G. (1972). The visual differential: An experimental study of the relation of varied experiences with visuals to shape discrimination. Unpublished doctoral dissertation. University of Wisconsin. Strasser, J., & Seplocha, H. (2008). Using picture books to support young children’s literacy. Childhood Education, 83(4), 219-224. Strickland, D. S. (1990). Family literacy: Sharing good books. Young Children, 43(7), 518519. Sulzby, E. (1996). Roles of oral and written language as children approach conventional literacy. In C. Pontevecorvo, M. Orsolini, B. Burge, & L. B. Resnick (Eds.), Children’s early text construction (pp. 25-46). Mahwah, NJ: Lawrence Erlbaum. Teale, W. H. (1995). Introduction. In T. Harris & R. Hodges (Eds.), The literacy dictionary. Newark, DE: International Reading Association. Tizard, B. (1981). Language at home and at school. In C. Cazden (Ed.). Language in early childhood education (pp. 17-27). Washington, DC: National Association for the Education of Young Children. Trousdale, A. (1990). Interactive storytelling: Scaffolding children’s early narratives. Language Arts, 67(2), 164-173. Weismann, D. (1970). The visual arts as human experience. Englewood Cliffs, NJ: Prentice Hall. Wells, G. (1981). Learning through interaction: The study of language development. Cambridge, MA: Cambridge University Press. Wells, G. (1986). The meaning makers: Children learning language and using language to learn. Portsmouth, NH: Heinemann. William, L. M., Hedrick, W. B., & Tuschinski, L. (2008). Motivation: Going beyond testing to a lifetime of reading. Childhood Education, 84(3), 135-141.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 27

READING IMPAIRMENT IN CHILDHOOD: OVERVIEW OF THE ELECTROPHYSIOLOGICAL CORRELATES OF DEVELOPMENTAL DYSLEXIA Barbara Penolazzi1, Chiara Spironelli1and Alessandro Angrilli1, 2 1

Department of General Psychology, University of Padova, Via Venezia 8, 35131 Padova, Italy 2 CNR Institute of Neuroscience, Padova section, Italy

Abstract Electrophysiological measurements enabled us to greatly increase our knowledge on the most spread learning disability in childhood represented by the Developmental Dyslexia (DD). The present chapter reviews the most relevant studies, which used either Event Related Potentials (ERPs) or Electroencephalographic (EEG) bands to investigate reading disabilities in developmental age. Several studies are here described, which succeeded in showing processing abnormalities not only in dyslexics, but also in children genetically at risk of dyslexia, through the analysis of both the "classical" electrophysiological components (i.e., MMN, P300, N400) and earlier evoked potentials. The electrophysiological markers of neuronal dysfunctions found in these children, helped scientists to uncover the psychophysiological mechanisms chiefly involved in this language disorder. These are: deficits in speech sound processing and impairment in manipulating the phonological features of grapheme strings. A considerable advance in the field has been recently reached by using EEG bands, traditionally used for detecting group differences in resting state, but currently extended to the measure of cognitive activation obtained through different experimental tasks. Such studies revealed functional differences in both fast and slow EEG rhythms between dyslexics and controls. The use of two functionally distinct EEG rhythms, theta and beta bands, in line with several ERP results, supports the view that dyslexics' reading difficulties are related to a linguistic impairment which is represented essentially at phonological level. Similar conclusions were further supported by the analysis of delta rhythm - a functional marker of cortical inhibition - during the performance of different linguistic tasks. Group differences in delta activation points to a delay in dyslexics' brain maturation. The review of recent literature on electrophysiological correlates of DD, on one hand supports the view that

662

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli ERPs, with their excellent time resolution, can represent an optimal tool for investigating language disorders mainly along time domain. On the other hand, EEG bands are able to show both functional and maturational aspects of dyslexics' brains, thus representing a distinct measurement with respect to ERPs. These two electrophysiological methods provide complementary information for a better understanding of the neural and cognitive mechanisms involved in DD. In conclusion, electrophysiological measures could represent an effective tool for making early diagnoses of dyslexia or for predictions of future reading problems in infants at risk, but they could also be valuable for testing the efficacy of rehabilitative trainings.

1. Developmental Dyslexia 1.1 Main Features of Developmental Dyslexia Developmental Dyslexia, or Specific Reading Disability (hereafter often referred to as DD, or simply dyslexia) is a clinic syndrome classified among developmental learning disorders. It consists of a pronounced and persistent difficulty in learning to read, despite normal intelligence and motivation, spared sensory acuity, and adequate educational and socioeconomic opportunities [World Health Organization, 1993]. Both incorrect and non-fluent decoding of written language mark dyslexics’ reading performance, which, in order to diagnose the disability, must be significantly inferior (two standard deviations or more, in standardized reading tests) compared with the performance of normal readers matched for age or reading age. In dyslexia (as well as in the other developmental learning disorders) the main dysfunction is characterised by a relative “specificity”, that is, despite the presence of the reading disability, the general intelligence is rather preserved (IQ, assessed with standardized intelligence tests, is in these children typically less than two standard deviations from the average IQ). In regard to this issue, it is noteworthy that, although dyslexics can sometimes show comprehension and reasoning problems during reading, these must be considered as secondary deficits, due to their lowlevel decoding disability [Shaywitz, 1998]. Further, individuals suffering from DD must be free from other severe sensory or neurological impairments, which can englobe reading disability as secondary symptom (e.g., even if individuals suffering from deafness or from mental retardation can show reading deficits, these have to be interpreted as a mere consequence of their primary sensory or neurological disturb, and therefore, in these cases, a diagnosis of DD would result inadequate). Finally, as last diagnostic criterion, reading disability must interfere not only with scholastic career, but also with every daily activity involving reading skills, and the interference must be persistent. Indeed, dyslexia does not represent a transient developmental lag, rather it is a chronic condition [Shaywitz, 1998]: accuracy, speed, and automation of dyslexics’ reading will always show a gap with respect to those of matched normal readers, although early and continuative rehabilitative trainings can allow them to achieve high levels of education. A large amount of evidence, especially coming from twin studies [DeFries, Fulker, & LaBuda, 1987; Stevenson, Graham, Fredman, & McLoughlin, 1987] and surveys showing familial clustering of dyslexia [Smith, Gilger, & Pennington, 1996] has confirmed that DD is a phenotypically heterogeneous neurological syndrome with a genetic aetiology [Schulte-

Reading Impairment in Childhood

663

Körne, Deimel, Müller, Gutenbrunner, & Remschmidt, 1996; Pennington, 1997; Smith, Kelley, & Brower, 1998; Castles, Datta, Gayan, & Olson, 1999; Flint, 1999; Francks, MacPhie, & Monaco, 2002]. Although dyslexia does not commonly segregate in families in a simple mendelian way, it shows a strong genetic involvement, and familiarity for reading impairment is considered one of the most reliable predictor for the following development of the disorder. Further, the high comorbidity of dyslexia with oral or written language developmental disabilities, which is another main risk factor of future reading deficits, supports a particular vulnerability of many linguistic brain areas to hereditary factors [Geschwind & Galaburda, 1985]. Genetic linkage analyses have identified loci on several chromosomes (e.g. 1, 2, 3, 13, 15 and, above all, chromosomes 6 and 18), which have shown strong and replicable effects on reading skills. The neural anomalies, determined by these genetic alterations (and producing, in turn, decoding dysfunctions at cognitive and behavioural level), are still under investigation. However, although opinions are not unanimous on this issue, many data (especially from animal studies, see paragraph 1.3) suggest that genetic factors would cause reading impairment by producing focal anomalies in the cerebral regions specifically involved in language processing, that are left perisylvian areas [Ramus, 2004]. It is necessary to underline that, even assuming genetic bases of dyslexia, people suffering from this disorder differ in their individual profiles for many features (at neurobiological, cognitive and behavioural levels), also produced by a complex set of environmental factors. For instance, the presence of certain hormonal environments during the prenatal period (i.e., high levels of foetal testosterone) can be crucial in producing a wide range of sensorimotor deficits often associated to the reading impairment [Rosen, Herman, & Galaburda, 1999], and contributes in this way to enormously vary the phenotypic manifestations of dyslexia (see, paragraph 1.3). At the same time, environmental factors external to the specific individual histories of dyslexics have been shown to add variations in the occurrence and severity of the reading disability. Among these factors, the intrinsic structure of the language to which the individual is exposed is decisive for the prevalence of the disability. Indeed, it has been demonstrated that DD occurrence is much higher in language with an irregular (or deep) orthography (i.e., English, French, etc.), than in language with a regular (or shallow) orthography (i.e., Italian, German, etc.), the latter showing more transparent correspondence between graphemes and phonemes, and thus being easier to learn [Lindgren, De Renzi, & Richman, 1985]. Therefore, considering the complex relationships between inherited and environmental factors at the origin of the different phenotypic subtypes of DD, we should consider this disorder as an extremely heterogeneous and multifarious syndrome. Epidemiological data [for a review see Shaywitz & Shaywitz, 2005], indicate that the specific reading disability, which can be seen as the extreme inferior limit of a normal distribution of reading skills, is probably the most frequent neurobehavioral disorder in developmental age and its prevalence rates range from ~ 5 to ~17.5%, depending on several factors (i.e., diagnostic criteria, assessment tests, regularity of the language orthography, etc.). Although someone reported that DD affects males and females equally [Wadsworth, DeFries, Stevenson, Gilger, & Pennington, 1992; Flynn & Rahbar, 1994; Shaywitz, 1998], by ascribing the classical sexual differences in its occurrence to a sampling bias (following this hypothesis, dyslexic males should be easily identified with respect to females, and therefore included in the sample, because of their more disruptive behaviour), DD is most typically

664

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

reported in males, with a variable sex ratio depending on different factors (i.e., IQ, phenotypic sub-type of dyslexia, impairment severity, etc. [Olson, 2002; Ramus, 2004]). Children affected by this disorder show often other deficits in many linguistic and nonlinguistic domains: oral and/or written language (at the basis of specific language impairment, dysphasia and/or dysgraphia), mathematical skills (dyscalculia), motor coordination (dyspraxia), temporal orientation (dyschronia), visuospatial skills (developmental right hemisphere syndrome), attentional capabilities (attentional deficit disorder with hyperactivity) [Dewey, 1995; Gross-Tsur, Shalev, Manor, & Amir, 1995; Fawcett, Nicolson, & Dean, 1996; Rapin & Allen, 1998]. This comorbidity of dyslexia with many other developmental disorders proves their partially common origins, involving both genetic factors and environmental influences, and contributes to produce a complex syndrome in which a combination of several other deficits can complicate or worsen the main disadvantageous condition. Just because of this phenotypical variability of the syndrome, many efforts have been made in the attempt to classify its different subtypes. The distinction which mirrors the classification of acquired dyslexia (i.e., reading impairment with a lesional origin), based on dyslexics’ error patterns, distinguishes a phonological type of dyslexia (alteration of the reading mechanisms which use the grapheme-phoneme conversion rules typical of each language), a surface type (impairment of the reading mechanisms which retrieve the phonology starting from a visual identification of the global lexical unity), and a mixed type (impairment of both the previous reading mechanisms) [Castles & Coltheart, 1993]. A different classification of dyslexia subtypes, made on the basis of aetiological assumptions, differentiates, instead, between a dysphonetic type and a dyseidetic type of reading disability, involving phonological/speech discrimination deficits or visual perceptual impairments, respectively [Boder, 1973]. Despite these not much fruitful classificatory attempts, at present, it is more prudent and useful a characterization, rather than a categorization, of dyslexic syndrome. In this view, only an early and complete assessment of cognitive, sensory and motor functions, going over taxonomic purposes, may allow to identify the severity of decoding problems and the presence of other kinds of dysfunctions. This characterization of the deficit, targeted on individual’s symptoms and features will permit to create the basis for its treatment, expected to be as much effective as much early and specific.

1.2 Causal Hypotheses of Developmental Dyslexia In spite of the general consensus on the genetic component of DD, after many years of research, theorists are still in disagreement about the neurological and cognitive mechanisms which, defective and dysfunctional not only on genetic basis, can have a causal role in explaining the reading disability. Thus, researchers’ efforts are directed to find the common factors which account for most symptoms, and to characterize this disorder by uncovering causal relationships between different levels of explanation (i.e., genetic, neurobiological and cognitive). The phonological deficit theory of DD supports its purely linguistic origin, referring to a specific deficit in representation (or in the access of representation) and processing of the smallest speech sounds: the phonemes [Bradley & Bryant, 1983; Shaywitz, 1998; Snowling, 2001; Shaywitz, Shaywitz, Pugh, Mencl, Fulbright, Skudlarski et al., 2002; Ramus, Rosen, Dakin, Day, Castellote, White et al., 2003; Shaywitz, Shaywitz, Fulbright, Skudlarski, Mencl,

Reading Impairment in Childhood

665

Constable et al., 2003; Ramus, 2004; Shaywitz et al., 2005]. This deficit is usually characterized by at least three main components [Wagner & Torgesen, 1987]: the first is low phonological awareness or metaphonology (difficulty in understanding the relationships between the phonemes of spoken language and the corresponding graphemes of written language, which causes further difficulties to consciously manipulate these sublexical units); the second component is limited verbal short-term memory (both considering the span of the memory store, and the functioning of the working memory); the third component is slow lexical retrieval (particularly clear-cut for rapid naming tasks). As direct consequence of the above deficits, dyslexics show a general impairment in decoding processes (especially when based on grapheme-phoneme conversion rules), which produces very inadequate reading performance and other common linguistic deficits as well. In this perspective, the sensorimotor symptoms, often associated to decoding impairments, are thought as simple comorbid factors, without causal relationships with the reading disability. Starting from evidence showing the presence of several auditory deficits in the dyslexic population, an alternative theoretical approach, the temporal (or rapid) processing deficit theory of DD [Tallal & Piercy, 1973; Tallal, 1980; Merzenich, Schreiner, Jenkins, & Wang, 1993; Stein & Walsh, 1997], assumes a basic impairment in the perceptual processing of stimulus temporal features as the main cause of reading disability. According to this theory, dyslexics are unable to process stimuli, especially in acoustic modality, when these are presented briefly or in rapid temporal succession (as in the case of normal speech). In this view, dyslexics’ phonological deficits are considered secondary to the main deficit in the temporal processing of auditory stimuli. Based on a reversed causal scenario, the visual processing deficit theory of DD, even admitting a common defective phonology in dyslexics, assumes that this component is accessory in decoding impairment and that a dysfunction in brain visual mechanisms of reading is the core deficit [Stein, 1993; Stein & Talcott, 1999]. The hypothesis that reading disorder is pre-linguistic in nature arose from a number of studies showing additional visual abnormalities in many dyslexics: slow processing of visual information, longer visual persistence for stimuli with low spatial frequencies, impaired contrast sensitivity, binocular fixation instability, reduced sensitivity to visual motion [Lovegrove, Bowling, Badcock, & Blackwood, 1980a; Lovegrove, Heddle, & Slaghuis, 1980b; Martin & Lovegrove, 1987; Eden, VanMeter, Rumsey, & Zeffiro, 1996; Cornelissen, Hansen, Hutton, Evangelinou, & Stein, 1998; Stein, 2001]. Since all these symptoms specifically pertain to the magnocellular (or transient) visual system, this hypothesis, presuming a dysfunction of visual information processing by this sensory pathway (supported by neurobiological evidence, see paragraph 1.3), is often called magnocellular (or magnosystem) visual theory of DD [Anderson, Essick, & Siegel, 1985; Livingstone & Hubel, 1988]. One attractive and parsimonious interpretation of available evidence, in which all quoted theories could converge, points to dyslexia as a general sensorimotor syndrome. The unification of the different approaches has been made by extending the hypothesis of a magnocellular visual deficit to a general multisensory deficit in the whole magnocellular sensory system. In this view, the phonological deficit, considered as a part of a more general sensorimotor syndrome, has been ascribed to a more general auditory impairment and is considered to have the same neurobiological origin as the visual impairment: a sensory dysfunction of the magnocellular system, which is also evident in the tactile domain [Galaburda, 1999; Grant, Zangaladze, Thiagarajah, & Sathian 1999]. Since the magnosystem

666

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

reaches the cerebellum, via the posterior parietal cortex, this extended magnocellular theory of DD provides a possible explanation also for the motor deficits frequently associated with the reading disability, thus including the cerebellar/motor dysfunction hypothesis of DD as well [Nicolson & Fawcett, 1990; Nicolson, Fawcett, & Dean, 2001]. Further, as the magnocellular system is also important for directing visual attention and for visual search processes, in virtue of its wide connections with the posterior parietal cortex, some authors have argued, as a variant, that a visuo-spatial attentional deficit, indirectly mediated by the magnocellular deficit, may cause reading disorders [e.g., Valdois, Gèrard, Vanault, & Dugas, 1995]. Currently, most theories on the causes of DD coexist, and have not been falsified as each one is supported by a whole body of empirical evidence. Nevertheless, the amount and strength of experimental data in favour of either the one or the other approach varies, leading to theories with different degree of scientific value. However, at present, the strongest evidence converges in identifying the phonological deficit as the core dysfunction of the reading disability [Bradley et al., 1983; Paulesu Frith, Snowling, Gallagher, Morton, Frackoviak et al., 1996; Shaywitz, 1998; Snowling, 2001; Ramus, 2003, 2004; Shaywitz et al., 2005; Ramus & Szenkovits, 2008]. Indeed, after the initial great interest and enthusiasm generated by the extended magnocellular deficit hypothesis of dyslexia, we are currently observing a reversed trend in favour of the phonological hypothesis, due to the inherent problems that the magnosystem theory presents. The first big limit of this theory is that dyslexia has been defined as a complex sensorimotor syndrome, but many recent empirical data have shown that, even thought the prevalence and the significance of the sensorimotor deficits remain uncertain, these non-linguistic deficits involve only one third of dyslexic population [Ramus, 2003]. Indeed, whereas phonological impairment characterized all dyslexics, the other frequently reported deficits (in visual, auditory, attentional and motor domains), considered relevant for theories of the magnosystem, are actually present only in a small subsample of impaired readers, and therefore they can not be considered the primary cause of reading difficulties. The second important trouble of magnocellular hypothesis is that, even in the case in which one or more sensorimotor deficits are co-occurring with the reading disability, they do not always meet the requirements made on the basis of the theory. Indeed, dyslexics can present sensory deficits also for stimuli in frequency domains which not exclusively pertain to the magnosystem pathway; similarly, their auditory impairment, when present, can also be extended, in addition to rapid frequency changes, to slow frequency changes, or can even regard only these latter [Amitay, Ben-Yehudah, Banai, & Ahissar, 2002; Chiappe, Stringer, Siegel, & Stanovich, 2002; Ramus et al., 2003]. Finally, another critical issue for the magnocellular theory of DD is that, whenever present, simple sensory deficits can hardly be connected by causal links to reading disability, and this is confirmed by the fact that visual or auditory deficits are not reliable predictors of reading disturbance as much as phonological impairments [Kronbichler, Hutzler, & Wimmer, 2002; Share, Jorm, MacLean, & Matthews, 2002]. Taken together, these lines of evidence suggest a clear weakness of the magnocellular dysfunction theory of DD in favour of the phonological deficit theory. Someone could argue that ascribing a linguistic disability, like reading impairment, to a purely linguistic cause, like phonological deficit, can sound tautological, rather than explanatory. This because phonological awareness has been demonstrated to be positively correlated with reading skills, but the causal direction between the two factors is not so clear. However, Ramus [2003] suggested some arguments in favour of this issue: first, since

Reading Impairment in Childhood

667

phonological awareness has been proved to be a good pre-school predictor of later reading skills [Pennington & Lefly, 2001; Lyytinen, Guttorm, Huttunen, Hämäläinen, Leppänen, & Vesterinen, 2005], the causal direction is far from being ambiguous (i.e., phonological awareness enhances reading skills, even though these can produce a positive feedback on the phonological awareness itself); second, as already mentioned, the phonological impairment can not be ascribed to the phonological awareness deficit only, as other components (i.e., poor verbal short-term memory and lexical retrieval) have been demonstrated to be, although related, quite independent from phonological awareness [Compton, DeFries, & Olson, 2001; Wolf, Goldberg O’Rourke, Gidney, Lovett, Cirino, & Morris, 2002]. Therefore, although impaired phonological components in dyslexics deserve further investigations, the phonological alteration remains the most consistent factor capable of explaining the core of this reading disability. The sensorimotor syndrome, instead, is considered optional with respect to reading dysfunction, and it has been hypothesized [Ramus, 2004, 2006] that sensory and motor deficits may be present only in the less pure cases of dyslexia, when the written language decoding impairment is associated with other developmental disorders (i.e. specific language impairment, attentional deficit, autism, etc.). In this view, these deficits should be interpreted as consequent and secondary to other comorbid disorders, with limited causal links with reading skills, but able to enormously worsen the primary disability, by adding new dysfunctions. For these reasons, DD should be considered a multiform syndrome, characterized by different sub-types, in which, side by side with the phonological deficit (the main and sole symptom in the “pure dyslexia”), sensorimotor and attentional dysfunctions can additionally contribute, in different ways, depending on individual differences, to make the basic disability more severe and complex.

1.3 Neurobiological Abnormalities in Developmental Dyslexia A considerable amount of data confirmed the presence of structural and functional abnormalities in dyslexics’ brains [for reviews see Habib, 2000; Démonet, Taylor, & Chaix, 2004; Eckert, 2004]. The first neuropathological examinations and the following neuroimaging studies run on dyslexics showed several microscopic cerebral structural anomalies (i.e., ectopias, microgyri, neurons with small cell bodies or scarce dendritic ramification), which for the first time led to hypothesize a possible abnormality in specific stages of their prenatal brain maturation [Galaburda & Kemper, 1979; Galaburda, Sherman, Rosen, Aboitiz, & Geschwind, 1985; Kaufmann & Galaburda, 1989; Humphreys, Kaufmann, & Galaburda, 1990; Livingstone, Rosen, Drislane, & Galaburda, 1991; Galaburda, Menard, & Rosen, 1994; Robichon & Habib, 1998]. These neural anomalies were identified in the left perisylvian cortex, in the white matter, in the magnocellular layers of thalamus (both in the lateral, and in the medial geniculate nuclei, in the corpus callosum and in the cerebellum. Structural hemispheric anomalies due to incomplete cerebral lateralization were found in dyslexics at a more macroscopic cerebral level as well, and these consisted in absent or reversed asymmetry of the planum temporale [Geschwind & Levitsky, 1968], decreased asymmetry of the parietal operculum [Habib & Robichon, 1996], and bilateral or reversed asymmetry of inferior frontal gyrus [Jernigan, Hesselink, Sowell, & Tallal, 1991; Robichon, Lévrier, Farnarier, & Habib, 2000].

668

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

Functional brain imaging studies supported many of the above structural findings, showing dysfunctions in a broad cortical network of dyslexics’ brains, mainly distributed in the left hemisphere, which includes perisylvian and temporo-parieto-occipital brain areas (i.e., planum temporale, inferior frontal gyrus, superior, middle and inferior temporal gyri, fusiform gyrus, angular gyrus and supramarginal gyrus). In particular, evidence suggested an impairment of the left posterior brain system for reading, that is, the disruption of the connections between dorsal (parietal) and ventral (occipito-temporal) circuits, related to grapheme-phoneme mapping and to global lexical entry recognition [Helenius, Tarkiainen, Cornelissen, Hansen, & Salmelin, 1999; Simos, Breier, Fletcher, Bergman, & Papanicolaou, 2000; Paulesu, Démonet, Fazio, McCrory, Chanoine, Brunswick et al., 2001; Temple, Poldrack, Salidis, Deutsch, Tallal, Merzenich et al., 2001; Démonet et al., 2004]. As a consequence, to overcome this failure of left posterior reading system, in dyslexics the neural activity has been found shifted towards more anterior left regions (which are sometimes overactivated compared with normal readers), or right perisylvian areas (typically not involved in low-level decoding processes) [Shaywitz et al., 1998; Brunswick, McCrory, Price, Frith, & Frith, 1999; Simos et al., 2000; Georgiewa, Rzanny, Gaser, Gerhard, Vieweg, Freesmeyer et al., 2002; Shaywitz et al., 2002; Démonet et al., 2004]. All these neurobiological abnormalities are open to many interpretations, especially because, depending on the study (with its peculiar variables: i.e., dyslexic samples, experimental tasks, neurobiological methods for measuring neural structure or activity, etc.), the set of detected anomalies is always partial with respect to the whole set of anomalies above reported. Therefore, on the basis of the theoretical approach, different subsets of abnormalities have been used to support each explicative hypothesis of DD. More precisely, structural and functional anomalies in left perisylvian areas have been typically used in favour of linguistic/phonological deficit theory of DD, as these regions have a main role in linguistic and phonological processing [Price, 1998]. On the contrary, the structural alteration of thalamus have been used to claim the magnosystem theory of DD, in such a direction that thalamic anomalies in the magnocellular layers of lateral geniculate nucleus are considered to cause dyslexics’ visual deficits, and anomalies in the medial geniculate nucleus are expected to cause auditory deficits. The crucial question is related to the discovery of the causal direction between these two kinds of anomalies (i.e. cortical versus thalamic). A top-down explanation, consistent with phonological deficit theory of DD, supposes that genetically driven cortical abnormalities in the left perisylvian cortex (responsible for phonological impairment) are able to start a retrograde degeneration till the sensorial thalamic nuclei (which are responsible for sensory deficits), which in turn can extend to the posterior parietal cortex and to the cerebellum (responsible for motor and attentional deficits, respectively). Conversely, a bottom-up explanation, consistent with magnosystem dysfunction theory of DD, supposes an opposite scenario: genetically driven thalamic anomalies are able to propagate in all the areas connected to the thalamus (i.e. perisilvian cortex, and possibly also posterior parietal cortex and cerebellum). Data on this issue are provided by animal studies and suggest, in agreement with behavioural data attesting the phonological origin of DD, a top-down explanation. This top-down theory is based on experimental evidence that mice and rats with spontaneous or experimental induced cortical abnormalities subsequently develop thalamic anomalies [Herman, Galaburda, Fitch, Carter, & Rosen, 1997; Peiffer, Rosen, & Fitch, 2002]; in addition, these animal exhibit a variety of learning deficits, whereas sensory dysfunctions are present only depending on whether specific hormonal condition are met

Reading Impairment in Childhood

669

[Rosen et al., 1999]. Based on these finding, it has been proposed a neurobiological model which, although originally thought to explain causal factors of DD, can be applied to other developmental disorders as well (starting from the assumption that the kind of disorder is strictly related to the cortical abnormality location) [Ramus, 2004, 2006]. According to this model (see Figure 1), which currently offers the most plausible framework with a considerable amount of explanatory arguments, genetically driven focal anomalies in the left perisilvian cortex cause the phonological impairment which characterized all individuals with DD. Depending on environmental influences, especially the prenatal exposure to certain hormonal factors (e.g., sex hormones like androgens), cortical anomalies can drive thalamic anomalies, which in turn are responsible for the sensorimotor syndrome sometimes associated to reading disability.

Figure 1. Neurobiological model proposed by Ramus to explain the aetiology of Developmental Dyslexia. Solid lines represent core traits of developmental dyslexia, dashed lines represent associated traits, not necessarily present in each affected individual. LGN: lateral geniculate nucleus, MGN: medial geniculate nucleus (redrawn from Ramus, 2004).

670

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

2. Electrophysiological Correlates of Developmental Dyslexia 2.1 Event Related Potentials in the Study of Developmental Dyslexia Event-related potentials (ERPs) are voltage fluctuations evoked from specific synchronously active neuron assemblies, which are temporally associated to some kind of external (i.e., physical stimulation) or internal (i.e., cognitive processing, motor preparation) events. They are usually recorded from the human scalp and extracted from the electroencephalographic spontaneous rhythmic oscillations (typically much larger in amplitude) through an averaging procedure time-locked to the specific eliciting event [Fabiani, Gratton, & Federmeier, 2007]. ERP components span a continuum between the exogenous, or early, potentials (i.e., obligatory responses evoked by the physical features of the eliciting event) and the endogenous, or late, potentials (i.e., responses evoked by cognitive processes associated to external events, but without a necessary direct link between the former and the physical features of the latter). Despite a limited spatial resolution, ERPs, with their ability to determine the sequence of events closely occurring in time (on the order of milliseconds), can be used in paradigms specifically designed either to understand how the brain process the various cognitive functions (referring in particular to their time-course and constitutive subcomponents) or to make specific diagnoses of physical and mental dysfunctions. For its excellent time resolution, ERP technique turns out to be particularly suitable to investigate both reading (a complex cognitive function consisting of many subprocesses, hard to be investigated with behavioural methods) and reading disability, especially when the latter manifests itself with a time-based symptom, like decoding dysfluency. In recent years, the use of evoked potentials for studying reading disability, allowed the collection of many results (briefly reported in the next sub-paragraphs, following the order of component latencies), which suggested a wide range of ERP anomalies in dyslexics. Although the most frequent anomalies are related to the auditory evoked components (namely to the MMN), which point to a general weakness of auditory processing at the basis of reading disability, it is still difficult to establish, on the bases of electrophysiological evidence alone, whether this can support the temporal processing deficit theory rather than phonological impairment theory of DD. A correct interpretation of the available findings should take into account many variables (especially related to subjects’ sample selection or to the choice of experimental paradigm). Thus, given the difficulty to support with a reasonable certainty a specific causal theory of DD, it is more useful to use these ERP markers for achieving more effective diagnoses and detailed phenotypical profiles of the examined readers.

Early Evoked Components Event related positive or negative deflections peaking around the first 100 ms following an acoustic or visual stimulus are usually called “early” evoked components. They are fast and obligatory (i.e., not dependent on attentional resources) responses of the physiological system to stimulus physical features. Considering the use of these early components to investigate DD in the auditory domain (which was more investigated than the visual one),

Reading Impairment in Childhood

671

several studies reported results mainly related to the hemispherical asymmetries. A group of dyslexic children characterized as dyseidetics showed left lateralization in early evoked components to musical and speech stimuli, whereas a group of dysphonetics failed to show any asymmetry [Fried, Tanguay, Boder, Doubleday, & Greensite, 1981]. Similar results were found in a sample of adult dyslexics, classified into similar sub-groups (i.e., dyseidetics and dysphonetics) [Rosenthal, Boder, & Callaway, 1982]. In line with this evidence, Brunswick and Rippon [1994] observed larger N1 amplitudes over the left hemisphere in normal readers during dichotic presentation of syllables, and the reversed asymmetry in children with reading disability. To our knowledge, only one study [Yingling, Galin, Fein, Peltzman, & Davenport, 1986] was not able to find differences between dyslexics and controls in response to simple acoustic stimuli. Many other studies reported smaller amplitudes, longer latencies, reduced or inverse hemispherical lateralization of early obligatory components in children with developmental disorders, but these findings will be not reported in the present review as participant selection did not meet the specific criteria for a clear diagnosis of DD (and very often the samples investigated were constituted by children with reading impairments mixed to children with more general learning or language disabilities).

MMN The mismatch negativity (MMN) is a middle-latency negative component, with a frontocentral distribution, peaking between 100 and 250 ms after an acoustic stimulus, which is evoked by infrequent and physically deviant stimuli (in one of several features: i.e., pitch, duration, intensity, rise time, etc.) with respect to the sequence of homogeneous stimuli in which it randomly occurs. Since it is generally evoked by unattended stimuli, the MMN is considered to reflect automatic change detection [Kraus, McGee, Carrell, & Sharma, 1995], and therefore can be used to test the pre-attentive discriminatory skills within the auditory system. There are many studies which employed MMN to investigate reading impairments, however, with a few exceptions, findings are quite consistent on the whole and attest alterations in amplitude, latency and hemispherical lateralization of this evoked potential in dyslexic readers. In particular, depending on the study, reduced, delayed or inversely lateralized MMNs have been shown to mark acoustic discrimination impairments in dyslexics. On one hand, several studies reported this discriminatory deficit in response to all kinds of deviant auditory stimuli, comprising simple pure tones [Baldeweg, Richardson, Watkins, Foale, & Gruzelier, 1999; Maurer, Brem, Bucher, & Brandeis, 2003a; Maurer, Bucher, Brem, & Brandeis, 2003b]. On the other hand, other studies found discrimination impairments for deviant speech-related sounds, thus supporting a pre-attentive, linguistic explanation of reading disabilities [Schulte-Körne, Deimel, Bartling, & Remschmidt, 2001]. As Lyytinen at al. [2005] have suggested, MMN differences between normal and impaired readers appear more related to complexity and processing demands, rather than just mere speech vs. non-speech distinction of the used stimuli. This viewpoint was supported by the findings of Kujala and colleagues [Kujala, Alho, & Näätänen, 2000], who found MMN differences between adult dyslexics and controls only for deviant stimuli occurring within complex (not speech-related) sound patterns, whereas deviant sounds failed to differentiate groups when occurring within simple sound patterns. Overall, results have suggested that impairments in implicit (or pre-attentive) auditory processing (often so small that escape behavioural measurements) could give rise to dyslexics’ deficits. In details, altered MMN can be explained by deficits both at the lowest

672

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

level of auditory discrimination processing and at the highest level of auditory (and in particular phonological) short term memory processing. Therefore MMN anomalies have been considered to support, depending on the theoretical orientation, either the auditory (temporal processing or magnocellular) deficit hypothesis of DD, or the phonological deficit hypothesis. Taking out the theoretical implications, it should be reminded that a very important applicative aspect of MMN is represented by its predictive value for the future occurrence of reading disabilities in infants with familial risk for DD [for a review see Lyytinen et al., 2005]. Indeed, it was largely proved that, with respect to other ERP markers, a reversed hemispherical asymmetry of this component (which in normal controls is wider in the left hemisphere) to acoustic and speech sounds, represents a reliable predictor for later development of dyslexia in children at-risk, even before the first year of age (see the last paragraph for a more detailed report). In addition, Kujala e colleagues [Kujala, Karma, Ceponiené, Belitz, Turkkila, Tervaniemi et al., 2001] demonstrated that audio–visual training (not specifically based on speech sounds) improves the reading skills of the treated dyslexics and that this behavioural advance was correlated with greater MMN amplitude. Therefore, despite the relatively few conclusive findings concerning the significance of auditory processing in dyslexia, important advances can be expected in future from the use of ERP markers, like MMNs, as basis for the diagnosis, prognosis and evaluation of training efficacy in reading disorders.

N200 The N200 is a middle-latency negative deflection whose scalp distribution and functional significance vary according to modality and experimental manipulations. It can be detected both in auditory (with maximum amplitude at fronto-central sites), and in visual modality (with maximum at occipital locations). In many experimental paradigms, the N200 amplitude appears to reflect the detection of some kind of mismatch between the entering stimulus and the representations of the prior stimuli, actively generated and stored in short term memory. This component diverges from the MMN since, differently from the latter, it is usually elicited when subjects orient attention and must respond in active tasks of mismatch detection [Fabiani et al., 2007]. Differently from the above reported findings on MMN, an interesting research of Kujala et al. [Kujala, Halmetoja, Näätänen, Alku, Lyytinen, & Sussman, 2006] did not find MMN differences between impaired and normal readers, during a passive condition in which deviant speech sounds (of different duration compared with standard ones) must be ignored. On the contrary, in the active target condition (requiring a response to deviant stimuli) the N200 amplitude was considerably lower in the group of poor readers. For authors, this would demonstrate an association between reading difficulty and deficits in a conscious (i.e., attentive, rather than pre-attentive) stage of stimulus processing. By using a completely different paradigm, in which written words and letter strings have to be processed by normal and impaired readers, a result in the opposite direction has been found [Bergmann, Hutzler, Klimesch, & Wimmer, 1995]. In detail, increased N200 (which authors named N220) were found in impaired readers specifically for word processing. This was interpreted as reflecting dyslexics’ compensatory mechanisms aimed to face a reduced anticipatory activation of cortical regions involved in word processing. This was indexed by a decreased Contingent Negative Components (CNV), measured in an interval preceding the N200. It is evident, from the few examples of N200 here reported, that the modulation of this component

Reading Impairment in Childhood

673

is strictly related to the experimental variables and can be functionally interpreted only by a careful examination of each specific paradigm.

P300 The P300 is a middle-latency positive component, peaking around 300 ms after an infrequent and deviant task-relevant stimulus, interspersed within a series of more frequent and standard stimuli (the classical “odd-ball” paradigm). This component is considered related to conscious processing and evaluation of stimuli in the working memory, as subjects must attend and actively respond to the odd-ball stimuli [Fitzgerald & Picton, 1983]. In impaired readers, it has been generally found P300 with reduced amplitude [Lovrich & Stamm, 1983; Holcomb, Ackerman, & Dykman, 1986; Erez & Pratt, 1992], delayed latency [Mazzotta & Gallai, 1992] or with an altered lateralization (i.e. earlier latencies to speech stimuli in the right hemisphere with respect to controls’ left dominance [Brezniz & Misra, 2003]) in many studies, all results suggesting an inefficient processing of task-relevant information [Taylor & Keenan, 1990]. This observation has generally been used by the theorists who support an attentional deficit at the basis of reading disability, as a proof in favour of their explicative theory. However, more controlled studies revealed altered P300s only in specific sub-samples of dyslexics, who in addiction suffered for attentional disorders [Duncan, Rumsey, Wilkniss, Denckla, Hamburger, & Odou-Potkin, 1994]. This finding undermines attentional deficit theory of DD and raises a crucial point in the employment of this component to study reading disorder. Since attentional deficits are often associated to reading disability (as reported in the first section of this chapter), anomalies in P300 can be considered to index attentional impairment as one of the possible causes of reading deficits (and not as a comorbidity factor) only when the samples of dyslexics under investigation are clearly diagnosed as free from specific attentional disorders, like in the case of the well controlled research of Holcomb and coworkers [Holcomb, Ackerman, & Dykman,1985]. However, even in the mentioned study, the relationships between attentional and reading deficits, based on dyslexics’ reduced P300, can be open to an alternative explanation: dyslexics’ repeated experiences of scholastic failures could produce more general and diffuse difficulties in attention engaging tasks, therefore attentional problem would be a consequence more than a cause of reading disorders. N400 The N400 is a negative endogenous component, peaking around 400 ms after the eliciting stimulus, with a posterior distribution, and a tendency to right lateralization. Since N400 amplitude has proved to be inversely correlated with many variables of a stimulus, in relation to its context (i.e., stimulus probability of occurrence, stimulus repetition, stimulus congruency with the context, subjective expectation of stimulus occurrence based on individual’s semantic memory, etc.), this component has been interpreted to reflect high level processing, such as meaning extraction and semantic integration of a stimulus within the context [Kutas & Federmeier, 2000]. Anomalous N400s during sentence comprehension have been reported in many studies on developmental reading disabilities, unfortunately findings are not univocal, and this makes their interpretation quite difficult. Some authors reported decreased N400 amplitude in dyslexics children [Stelmack, Saxe, Noldy-Cullum, Campbell, & Armitage, 1988; Helenius, Salmelin, Service, & Connolly, 1999], which was interpreted as a failure to engage semantic

674

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

memory during linguistic comprehension. On the contrary, other studies reported increased N400 amplitude in impaired readers [Neville, Coffey, Holcomb, & Tallal, 1993; Robichon, Besson, & Habib, 2002], which was considered to reflect their compensatory effort to extract word meaning and integrate it within the sentence. As reported in the first part of the present chapter (paragraph 1.1), many behavioural data demonstrated that impaired readers can sometimes show comprehension and reasoning deficits during text reading [Shaywitz, 1998]. However, it is noteworthy that such high–level cognitive problems must be regarded as secondary deficits, due to dyslexics’ main impairment in low-level decoding. Therefore, text comprehension dysfunctions, as indexed by altered N400s, should be considered as a common natural consequence of more basic decoding impairments, and, as such, not very informative with regard to the causal mechanisms possibly involved. In other words, since high-order impairment in reading can be expected with elevated probability in dyslexics, as a consequence of their basic low-order deficits, the investigation of earlier ERP components appears more effective in our opinion, being these indexes more strictly related to the primary impaired mechanisms of DD. Conversely, the so-called N420 detected by Penolazzi and colleagues [Penolazzi, Spironelli, Vio, & Angrilli, 2006] has to be considered in a completely different way. In fact, although this component has been found within a time interval roughly corresponding to that of the traditional N400, its different spatial distribution suggested to authors a completely different functional meaning. Within a linguistic paradigm in which the same set of words were processed on the basis of various linguistic criteria, the N420 was considered as related to a low-level processing phase, common to all the different experimental tasks: i.e., the grapheme-phoneme conversion during reading. In addition to spatial distribution and the paradigm used, this interpretation was supported by the mean reading times of the sample investigated, which showed that, during the N420 interval, both normal and impaired readers were still engaged in the basic processes of word decoding, a condition which makes the execution of a more high–level linguistic processing (like semantic elaboration) very unlikely. Indeed, whereas in controls this component was left lateralized, in disabled readers it was more distributed across hemispheres. Dyslexics’ lack of lateralization, found also in a later slow negative wave (700-1500 ms time interval), points to an altered cerebral asymmetry of the basic mechanisms necessary for reading the most important of which is the orthographyphonology mapping. In this view, these findings would therefore support the hypothesis of a phonological deficit underlying developmental reading disabilities.

2.2 Electroencephalographic Bands in the Study of Developmental Dyslexia Electroencephalography (EEG) is the measurement of electrical activity (post-synaptic currents) generated by neuronal assemblies, as recorded from electrodes placed on the scalp. Brain electrical oscillations cover a large range of values in amplitude (~ 5-200 µV) and frequency (~ 0.5- 100 Hz), and the constituent waves of EEG are usually characterized by specific ranges of such frequencies, giving rise to the well-known electroencephalographic (EEG) bands. These rhythms are classically linked to particular states of cortical arousal. As a general rule, waves with low frequency and large amplitude, typical of slow bands (i.e., delta and theta), have been traditionally associated to absent or decreased cognitive processing, whereas high frequency and small amplitude waves, characteristic of fast bands (e.g., beta

Reading Impairment in Childhood

675

and gamma) have been related to active cognitive processing [Pizzagalli, 2007]. However the more recent employment of EEG in experimental paradigms appropriately created to test task- and stimulus-related linguistic processing, has contributed to enrich (and sometime partially change) the functional significance of many bands, which therefore can be used nowadays as markers of more specific processes. Like ERPs and despite a low spatial resolution, EEG band analysis offers the possibility to study brain functions with a time resolution in the middle between fMRI and ERPs. In the study of dyslexia, EEG bands have been traditionally investigated, under resting condition, in order to detect basic brain activity differences between impaired and normal readers. Although data are not univocal, most of such studies found an overall increase in slower rhythms (i.e., delta and theta bands) and a decrease in mid range frequencies (especially alpha) in children with reading disabilities associated or not to other learning disorders [Sklar, Hanley, & Simmons, 1972; Colon, Notermans, DeWeerd, & Kap, 1979; Ahn, Prichep, John, Baird, Trepetin, & Kaye, 1980; John, Prichep, Ahn, Easton, Friedman, & Kaye, 1983; Harmony, Hinojosa, Marosi, Becker, Rodríguez, Reyes et al., 1990a; Fonseca, Tedrus, Chiodi, Cerqueira, & Tonelotto, 2006]. Only a few studies were not able to find any electrophysiological differences between impaired and normal readers in resting periods [Fein, Galin, Yingling, Johnstone, Davenport, & Herron, 1986; Yingling et al., 1986]. However these findings are the minority and many reasons probably account for these inconsistent results, above all confounded variables related to sample selection (e.g., few participants, phenotypic variability, comorbidity) and experimental setting (e.g., different parameters of EEG recording and analysis). Focusing on to the majority of the studies which found differences between groups, it is critical to note that they showed in dyslexics abnormal EEG patterns (i.e., relative preponderance of slower rhythms over faster ones), analogous to those which typically mark infancy and early childhood with respect to adulthood [Matoušek, M., & Petersén, 1973; John, Ahn, Prichep, Trepetin, Brown, & Kaye, 1980; Gasser, Verleger, Bacher, & Sroka, 1988; Harmony, Marosi, Díaz de León, Becker, & Fernández, 1990b]. For this reason, the main interpretation of the EEG abnormalities observed in impaired readers, especially in the case of less severe disabilities, has been often limited to a general, more or less persistent, maturational lag in brain development. Although EEG studies under resting state provided important information about basic differences between normal and impaired readers, a significant advance in the research on electrophysiological correlates of DD has been reached only recently, by using experimental paradigms aimed at stressing (through specific stimuli or tasks) the functional cognitive processes assumed to be potentially impaired in dyslexics (i.e., linguistic, perceptual, attentional). Within this new perspective of event related EEG studies, which will be briefly reviewed in the present paragraph, a further progress has been reached by shifting the focus from a more coarse band or spectral analysis, performed on the entire task period [Ackerman, McPherson, Oglesby, & Dykman, 1998; Rippon & Brunswick, 1998, 2000], to a more finegrained analyses based on time intervals corresponding to different processing phases required by the task [Klimesch, Doppelmayr, Wimmer, Gruber, Röhm, Schwaiger et al., 2001a; Klimesch, Doppelmayr, Wimmer, Schwaiger, Röhm, Gruber et al., 2001b; Spironelli, Penolazzi, Vio, & Angrilli, 2006; Spironelli, Penolazzi, & Angrilli, 2008; Penolazzi, Spironelli, & Angrilli, in press]. With such strategy, differences between dyslexics and controls, actively engaged in different experimental tasks, have been found particularly in beta, alpha and theta bands [Klimesch et al., 2001a, 2001b; Milne, Hamm, Kirk, & Corballis,

676

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

2003; Spironelli et al., 2006, 2008], and, more recently, in delta rhythm as well [Penolazzi et al., in press]. This new approach to the issue allowed to greatly enlarge the restrictive maturational lag hypothesis of reading disabilities, by detecting precise impaired mechanisms in specific brain regions at the basis of dyslexics’ behavioural deficits. In agreement with the fact that the functional significance of alpha rhythm (8-12 Hz) is heavily debated (particularly in studies investigating evoked EEG activity), and it is still difficult to clearly define its physiological role [Pizzagalli, 2007] the many studies which used it to investigate DD have led to different and inconsistent results [Duffy et al.1980; Harmony et al., 1990a; Ortiz, Exposito, Miguel, Martin-Loeches, & Rubia, 1992; Rippon & Brunswick, 2000] Depending on the study, dyslexics showed increased or diminished alpha levels compared with controls, possibly because different alpha sub-bands may be functionally dissociated [Klimesch, 1999]. Therefore, given that studies which used alpha band did not allow to deepen the dynamics involved in reading disorders, they will be not treated in the present review on event related EEG correlates of dyslexia. On the contrary, the research on reading disability has recently received a great impulse by the employment of theta band in several experimental paradigms. This EEG rhythm, characterized by wide and slow waves (4–8 Hz), prevails (together with the delta rhythm) in infancy also during waking, and gradually decreases over developmental age. Whereas traditional EEG studies reported theta activity in adults prominently during sleep, or drowsiness state, more recent findings have associate the increase of this rhythm with enhanced effective cognitive processing, improved ability to focus and sustain attention, and increased mental effort or task difficulty [Gevins & Smith, 2000]. In addition, increased theta activity has been interpreted as reflecting the extensive use of specific mechanisms, like information retention or scanning in working memory [Gevins, Smith, McEvoy, & Yu, 1997; Klimesh, 1999; Gevins et al., 2000; Jensen & Tesche, 2002; Sauseng, Klimesch, Dopplemayr, Hanslmayr, Schabus, & Gruber, 2004; Sauseng, Klimesch, Schabus, & Dopplemayr, 2005]. Since the most robust explicative hypothesis of DD narrows the reading disorder mainly to deficits in phonological working memory (see paragraph 1.2), the use of theta band as proved functional marker of verbal working memory and linguistic executive functions involvement turned out to be natural. Ackerman et al. [Ackerman, Dykman, Oglesby, & Newton, 1995] found increased theta (and decreased beta) power in poor readers with respect to controls during letter and word reading. In line with this result, Rippon and Brunswick [1998, 2000] reported an increase of frontal theta activity in impaired compared with normal readers, selectively during a phonological processing, contrasted with a visual task. This finding, which confirmed the discriminative power of theta band in relation to different cognitive tasks (whereas, in the same investigation, alpha and beta band analyses failed to reveal taskdependent differences in dyslexics), was interpreted as reflecting dyslexics’ decreased task automaticity and increased effort to compensate the basic phonological deficit. Klimesh et al. [2001a] investigated tonic and basic levels of theta band in a sample of dyslexic children, engaged in number, word and pseudo-word reading tasks. The authors found, in the patient sample, a bilateral increase in theta activity specifically during word reading, conversely controls showed a highly selective left lateralized activation for words. Interestingly, results on dyslexics’ anomalous language-related theta lateralization have been confirmed by a series of studies, which, extending the investigation to other bands, succeeded in correlating specific altered EEG patterns to particular phases of dyslexics’ word reading [Spironelli et al., 2006, 2008; Penolazzi et al., in press]. These studies, using a well-tested linguistic CNV paradigm,

Reading Impairment in Childhood

677

based on word-pair visual presentation [Angrilli, Dobel, Rockstroh, Stegagno, & Elbert, 2000; Angrilli, Elbert, Cusumano, Stegagno, & Rockstroh, B., 2003; Penolazzi et al., 2006; Spironelli & Angrilli, 2006; Spironelli et al., 2006, 2008], engaged dyslexics and agematched controls in different kinds of linguistic tasks. In detail, children were asked to judge word pairs on the basis of different criteria (each corresponding to an experimental task), appositely employed to test different mechanisms of written word processing: from the low level visuo-perceptual processing (in the control Orthographic task), to the higher level rhyming and semantic association processes (in the Phonological and Semantic tasks, respectively). Besides the possibility to test several subprocesses potentially impaired in dyslexics, an innovative feature of this paradigm consisted in the use, in all tasks, of the same set of words as first stimulus of each pair, to exclude interfering effects related to the features of the specific selected words (e.g., length, frequency, lexical category, etc. [Penolazzi, Hauk, & Pulvermuller, 2007]). Thus, the adopted paradigm allows to attribute all significant effects only to the specific processes engaged by the task and not to possible confounding factors. This allowed authors to find in dyslexics an altered pattern of theta activation (4-8 Hz), both in the temporal dimension and in the cortical space [Spironelli et al., 2006]. Analyses performed on three different phases of word processing (i.e., “W1”: word encoding interval; “iISI” and “tISI”: respectively initial and terminal parts of the inter-stimulus-interval between words of each pair), compared with a baseline interval, showed in controls a time-dependent and task-related cortical activation: a peak of theta activity during word encoding interval, and, always during this processing phase, a greater theta amplitude on the left hemisphere for the Phonological task, compared with the more bilateral activity of the Semantic task. On the contrary, dyslexics showed a peak of theta activity which was delayed to the first inter stimulus interval, and shifted to the right hemisphere during all phases of the Phonological task, and in some intervals of the Semantic task. Therefore dyslexics’ dysfunctional, unspecific and anomalous (in both time course and hemispherical asymmetry) theta activation to language has been interpreted as due to a defective recruitment of left hemisphere structures during the verbal working memory phase, especially in the Phonological task. Similar results were found, using the same linguistic paradigm on a different sample of impaired readers, in a following study [Spironelli et al., 2008] which added to previous findings with the novel contribution of beta band (13-30 Hz), a more classical index of cortical activation. Evidence of dysfunctional beta band alteration in dyslexic children has recently been provided by several studies. Klimesch et al. (2001b), by comparing alpha (812.5 Hz) and beta (12.5-16 Hz) band activations in the same experimental paradigm above described, found a selective deficit in dyslexics’ word processing. Controls showed a taskdependent beta activation: i.e., higher beta percentages during word reading in left hemisphere locations roughly corresponding, for authors, to Broca’s area and angular gyrus, and increased right/midline beta amplitude during number processing. Instead, dyslexics showed a complete lack of task selectivity. Therefore, beta band was interpreted, by Klimesh and colleagues, as a cortical index able to measure the capability to process specific stimuli (i.e., words, numbers, etc.), and, in particular in linguistic paradigms, to reflect graphemephoneme processing only in normal readers. Further evidence of dysfunctional beta band alteration in dyslexic children was also provided by Milne et al. (2003). These authors compared two subsamples of compensated dyslexics, i.e., dysphonetics and dyseidetics children, with a control group, during a lexical decision task on visually presented words and pseudo-words, and found no overall differences in mean beta power between normal and

678

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

impaired readers. However, a significant interaction group by anterior-posterior axis was found: whereas dysphonetic children showed increased beta activity over anterior sites, and dyseidetics over posterior ones, controls revealed a different pattern without antero-posterior differences. In agreement with most neuroimaging studies (see paragraph 1.3), Milne et al. interpreted these findings as a demonstration of the segregation of different word subprocesses in anterior and posterior regions of linguistic areas, thus confirming the capability of beta as a marker able to locate specific cognitive functions. In their study, Spironelli and colleagues (2008) introduced EEG beta band with the aim to collect more detailed topographical information on reading impairment. Indeed, whereas theta band is controlled by the deep subcortical structures of the temporal lobe, involved in working memory [Leung, 1998; Tesche & Karhu, 2000; Vertes, 2005], beta band is essentially produced by highly confined superficial cortical activity, depending by many independent generators typically recruited by high-level cognitive processing [Pantev, Makeig, Hoke, Galambos, Hampson, & Gallen, 1991; Tallon-Baudry & Bertrand, 1999]. Results revealed that the two EEG bands had very similar patterns of activation (see Figure 2), and were therefore, to some extent, functionally correlated, as previously demonstrated for working memory processes [Leung, 1992; Sarnthein, Petsche, Rappelsberger Shaw, & von Stein, 1998; Slotnick, Moo, Kraut, Lesser, & Hart, 2002; Sarnthein, Morel, von Stein, & Jeanmonod, 2003]. Thus, beta band helped to interpret theta band as an effective index of complex cognitive processes implemented on the cortical surface. Indeed, whereas controls showed greater theta and beta activation at left frontal sites during the Phonological task, dyslexics showed at anterior locations a dysfunctional right lateralization in all tasks. On the contrary, at posterior sites, impaired readers showed greater left lateralization during both Phonological and Orthographic tasks, possibly indicating an altered and difficult phonological transcoding process during verbal working memory phases of word processing. Therefore, also results of the last study, by linking dyslexics’ defective verbal working memory to the altered patterns of EEG activity, supports phonological deficit hypothesis of DD. Further, it promotes the use of theta and beta bands as useful instruments for assessing task-related language lateralization and its impairment in developmental age, being enough sensitive to task manipulation and to different phases of word processing. In the last study, the use of the same CNV linguistic paradigm [Angrilli et al., 2000] for measuring delta band (0-4 Hz) has not only strengthened previous results, but also represented a positive evolution for the investigation of the developmental disorder by means of EEG rhythms [Penolazzi et al., in press]. Since delta waves predominantly characterise EEG in developmental age, it marks cerebral immaturity, but, at the same time, delta waves are present in adulthood during the deepest stages of sleep, or in severe pathological conditions [De Jongh, de Munck, Baayen , Jonkman, Heethaar, & van Dijk, 2001; Wienbruch, Moratti, Elbert, Vogel, Fehr, Kissler et al., 2003; Babiloni, Frisoni, Steriade, Bresciani, Binetti, Del Percio et al., 2006]. Thus, this rhythm is generally considered a clearcut index of cortical inhibition. The study of Penolazzi et al. [in press] aimed at investigating DD by using delta band both for detecting group differences in brain maturation and for measuring, with an innovative approach, functional differences in task-related linguistic brain asymmetry.. Results of this study are in line with past findings of the same research group [Spironelli et al., 2006, 2008]: indeed delta activity, a clear marker of cortical inhibition, showed a specular pattern with respect to the previously investigated theta and beta bands, used as markers of functional activation (see Figure 3).

Reading Impairment in Childhood

679

Figure 2. Percentages of normalized Theta (a) and Beta (b) bands in dyslexics and controls (dashed and full lines, respectively) during W1 and iISI intervals throughout three linguistic tasks in the four cortical quadrants. Asterisks indicate significant post-hoc test results (redrawn from Spironelli, et al., 2008).

680

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

Figure 3. Percentages of Delta band in dyslexics and controls (dark and light bars, respectively) throughout three linguistic tasks in the four cortical quadrants (from Penolazzi et al., in press).

A first result concerning dyslexics’ greater overall delta amplitude represented a clearcut evidence of their greater brain electrical immaturity compared with controls. Further, unlike controls’ ability to differentiate tasks and linguistic processes, dyslexics showed essentially the same pattern of delta activity in all tasks (although the highest delta values were reached during the phonological processing): that is, greater delta percentages on anterior left and posterior right areas (corresponding to the disinhibition of anterior right and posterior left cortical regions). Therefore, these data confirmed both a greater brain electrical immaturity of impaired readers compared with controls, and a specific deficit in phonological processing mediated by anterior left linguistic networks (more severely impaired in dyslexics’ with respect to other brain regions). With this approach, the maturational lag hypothesis of DD has been integrated with that of phonological deficit, thus providing a more detailed picture of impaired readers’ main features. Turning the attention to delta band, its classical employment in measuring basic differences between groups appears at least restrictive, given its proven ability to discriminate different kinds and phases of linguistic processes in normal readers. Thus, side by side to more popular and traditional EEG rhythms sensitive to cognitive processing, a more extensive use of delta band is encouraged, being this last one of the few tools suited to investigate cognitive functioning and brain maturational lags, at the same time.

Reading Impairment in Childhood

681

3. Why to Use Electrophysiological Measures in The Study of Developmental Disabilities Electrophysiological methods are relatively non-invasive and less expensive compared with other brain functional imaging techniques, and, although they do not stand out for spatial resolution, provide multidimensional measures (i.e., which can be characterized for amplitude, latency/ frequency, scalp distribution) of neural activity, with an excellent temporal resolution of the processes induced by specific stimuli or tasks. This makes electrophysiological measures particularly suitable to investigate cognitive processes, which generally consist of several substages, hardly detectable with behavioural measures. The possibility of disentangle the subcomponents of a process, offered by ERPs and EEG measures, becomes even more important for studying cognitive disorders, especially when different explicative theories on the underlying impaired mechanisms still coexist (like in the case of DD). Further, electrophysiological methods can become unrivalled instruments to identify, through specific markers or patterns of electrical activity, all cases in which deficits are present, although not completely (or not yet) manifest at the behavioural/clinical level, and could worsen without an immediate treatment (i.e., light disorders; infants or child genetically at risk for some disability, but still not able to be identified by behavioural assessments). Thanks to all the above described features, electrophysiological methods represent the elective tool for a more comprehensive understanding of the cognitive and neural mechanisms underlying DD. Focusing on a more clinical/practical perspective, the first prominent benefit, which electrophysiological methods can provide to the study of DD, is the facilitation and acceleration of the diagnostic process. The improved detection of impaired readers can be favoured by electrophysiological measures at least in three ways. First, although behavioural testing will probably remain the simplest and less expensive way of screening preschool and school children, electrophysiological measures can contribute to anticipate and strengthened the diagnosis of reading disability delineated with the behavioural assessment. In particular, the power of neurobiological data, when supported by behavioural ones, could be very important for reaching a correct diagnosis, especially in transparent orthographies, where reading impairment is more difficult to detect in comparison with deep orthographies [Landerl, Wimmer, & Frith, 1997; Zoccolotti, De Luca, Di Pace, Judica, Orlandi, & Spinelli, 1999]. Indeed, whereas in the latter the reading impairment is identified without difficulty for deficits in both accuracy and speed (because of the high percentage of irregular words), in the former a slower reading speed is often the only symptom easily detectable (since more regular grapheme-phoneme correspondences hide a part of the disorder). For this reason, reading disabilities in regular orthographies are frequently underestimated or exchanged for normal developmental delays in reading skill acquisition, and therefore their diagnosis would greatly gain from additional electrophysiological evidence. Second, whereas in many cases behavioural testing fails to identify marks or precursors of reading impairments, electrophysiological measures can be particularly suited to isolate predictive cues of future reading problems in infants or children, genetically at risk for dyslexia [for a review see Lyytinen et al., 2005]. Many prospective studies indeed have found that, at a very early age (6 months or even earlier) several ERP patterns (overall, pointing to an anomalous right hemisphere dominance of children at-risk, during low level processing of auditory and

682

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

speech-related stimuli, instead of the left typical dominance of controls) can differentiate children with and without familial risk of dyslexia [Molfese & Molfese, 1997; Molfese, 2000; Molfese, Molfese, & Modglin, 2001]. Further, it has been shown that these brain responses have a reliable predictable value, since they correlate with the delayed interval of acquisition and development of language and reading skills [Leppänen, Pihko, Eklund, & Lyytinen, 1999; Pihko, Leppänen, Eklund, Cheour, Guttorm, & Lyytinen, 1999; Guttorm, Leppänen, Richardson, & Lyytinen, 2001; Guttorm, Leppänen, Tolvanen, & Lyytinen, 2003]. Third, electrophysiological methods, being able to distinguish the subcomponents of a process which can be selectively impaired, can contribute, more than behavioural measurements, to make differential diagnoses among the various sub-types of dyslexia. As stressed in the first paragraph, DD must be considered an extremely heterogeneous and multifactorial syndrome, in which side by side to the decoding impairment, a combination of several other deficits can complicate or worsen the disadvantageous condition. In this view, an electrophysiological paradigm, specifically created to assess the various cognitive components which could be impaired (i.e., linguistic, perceptual, attentional), can provide a detailed picture of the individual’s symptoms and features, favouring, in turn, more personalized (and, thus, more effective) treatment procedures. The improvement in the diagnostic process, which can be boosted by electrophysiological data, is aimed, first of all, at allowing the immediate implementation of a supportive training, since treatment efficacy is successful as much it occurs early and is individually targeted. In this view, besides the diagnosis, the second main applicative contribution of electrophysiological methods to the study of DD concerns its ability to test and compare the efficacy of different rehabilitative therapies. Indeed, by repeatedly measuring electrophysiological activity (for instance to coincide with the beginning and the end of the treatment, or with regular follow-up examinations), it can be possible to test the efficacy of specific remediation trainings (or eventually of coping strategies naturally developed with experience, when specific trainings are missing). Starting from the assumption that a wider neural reorganization after the training is associated, with high probability, to a more extensive and long-lasting behavioural improvement in reading performance, the comparison of electrophysiological recordings in succession may allow to evaluate behavioural progresses by the extent and the localization of cortical reorganization. From a theoretical perspective this will increase our current knowledge on cerebral plasticity phenomena related to linguistic processes, and from a clinical viewpoint it will drive the selection of the most effective treatments suited for a variety if individuals affected by dyslexia. In conclusion, ERPs, with their excellent time resolution, represent an ideal tool to study and detect language disorders, especially when the reading disability manifests itself mainly along time domain. EEG bands, being able to show not only functional, but also maturational features of dyslexics' brains, represent a distinct complementary method with respect to ERPs. Therefore, from a theoretical point of view, these two electrophysiological methods, less invasive and expensive compared with other brain imaging techniques, provide complementary information for a better understanding of the neural and cognitive mechanisms involved in DD. In addition, from a clinical/practical point of view, electrophysiological measures promise to be an effective instrument for making early diagnoses of dyslexia. Used together with behavioural testing, they can help to achieve an early detection of reading disorder, and, more crucial, they promise to offers, in future, reliable markers able to identify the risk of developing future reading disabilities in infants

Reading Impairment in Childhood

683

genetically at-risk (sample on whom, for obvious reasons, behavioural tests are still inapplicable). Finally, electrophysiological measures, indirectly favouring early and personalized linguistic trainings (naturally resulting from early diagnoses), can provide a valuable support to test their efficacy, by means of periodically repeated examinations of neural reorganizations underlying behavioural progresses. Therefore, addressing the initial question of this paragraph, there are good reasons to encourage, in the next future, a more extensive use of electrophysiological methods in the study of dyslexia (and the other developmental disorders as well): besides the possibilities they give to increase our theoretical knowledge on reading disabilities, they promise to become an effective tool to make early and differentiated diagnoses/prognoses, to predict future problems in infants with familial risk of DD, to test the efficacy of the therapy, by correlating neural reorganization phenomena to behavioural advances promoted by the therapy itself.

References Ackerman, P.T., Dykman, R.A., Oglesby, D.M., & Newton, J.E.O. (1995). EEG power spectra of dysphonetic and nondysphonetic poor readers. Brain and Language 49, 140152. Ackerman, P.T., McPherson, W.B., Oglesby, D.M., & Dykman R.A. (1998). EEG Power spectra of adolescent poor readers. Journal of Learning Disabilities, 31, 83-90. Ahn, H., Prichep, L., John, E.R., Baird, H., Trepetin, M., & Kaye, H. (1980). Developmental equations reflect brain dysfunctions. Science, 210, 1259-1262. Amitay, S., Ben-Yehudah, G., Banai, K., & Ahissar, M. (2002). Disabled readers suffer from visual and auditory impairments but not from a specific magnocellular deficit. Brain, 125, 2272-2285. Anderson, R.A., Essick, G.K., & Siegel, R.M. (1985). Encoding of spatial location by posterior parietal neurons. Nature, 301, 419-421. Angrilli, A., Dobel, C., Rockstroh, B., Stegagno, L., & Elbert, T. (2000). EEG brain mapping of phonological and semantic tasks in Italian and German languages. Clinical Neurophysiology, 111, 706-716. Angrilli, A., Elbert, T., Cusumano, S., Stegagno, S., & Rockstroh, B. (2003). Temporal dynamics of linguistic processes are reorganized in aphasics’ cortex: An EEG mapping study. Neuroimage, 20, 657-666. Babiloni, C., Frisoni, G., Steriade, M., Bresciani, L., Binetti, G., Del Percio, C., Geroldi, C., Miniussi, C., Nobili, F., Rodriguez, G., Zappasodi, F., Carfagna, T., & Rossini, P.M. (2006). Frontal white matter volume and delta EEG sources negatively correlate in awake subjects with mild cognitive impairment and Alzheimer’s disease. Clinical Neurophysiology, 117, 1113–29. Baldeweg, T., Richardson, A., Watkins, S., Foale, C., & Gruzelier, J. (1999). Impaired auditory frequency discrimination in dyslexia detected with mismatch evoked potentials. Annals of Neurology, 45, 495–503. Bergmann, J., Hutzler, F., Klimesch, W., & Wimmer, H. (2005). How is dysfluent reading reflected in the ERP? Journal of Neurolinguistics, 18, 153-165. Boder, E. (1973). Developmental dyslexia: a diagnostic approach based on three atypical reading-spelling pattern. Developmental Medicine and Child Neurology, 15, 663-687.

684

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

Bradley, L., & Bryant, P.E. (1983). Categorizing sounds and learning to read: a causal connection. Nature, 301, 419-421. Brezniz, Z., & Misra, M. (2003). Speed of processing of the visual-orthographic and auditoryphonological systems in adult dyslexics: The contribution of asynchrony to word recognition deficits. Brain and Language, 85, 486–502. Brunswick, N., McCrory, E., Price, C.J., Frith, C.D., & Frith, U. (1999). Explicit and implicit processing of words and pseudowords by adult developmental dyslexics: A search for Wernicke’s Wortschatz? Brain, 122, 1901-1917. Brunswick, N., & Rippon, G. (1994). Auditory event-related potentials, dichotic listening performance and handedness as indices of lateralisation in dyslexic and normal readers. The International Journal of Psychophysiology, 18, 265–275. Castles, A., & Coltheart, M. (1993). Varieties of developmental dyslexia. Cognition, 47, 149180. Castles, A., Datta, H., Gayan, J., & Olson, R.K. (1999). Varieties of developmental reading disorders: genetic and environmental influences. Journal of Experimental Child Psychology, 72, 73-94. Chiappe, P., Stringer, R., Siegel, L.S., & Stanovich, K.E. (2002).Why the timing deficit hypothesis does not explain reading disability in adults. Reading and Writing, 15, 73-107. Colon, E.J., Notermans, S.L., DeWeerd, J.P.C., & Kap, J. (1979). The discriminatory role of EEG power in dyslexic children. Journal of Neurology, 221, 257-262. Compton, D.L., DeFries, J.C., & Olson, R.K. (2001). Are RAN- and phonological awarenessdeficits additive in children with reading disabilities? Dyslexia, 7, 125-149. Cornelissen, P.L., Hansen, P.C., Hutton, J.L., Evangelinou, V., & Stein, J.F. (1998). Magnocellular visual function and children’s single word reading. Vision Research, 38, 471-482. De Jongh, A., de Munck, J., Baayen, J., Jonkman, E., Heethaar, R., & van Dijk, B. (2001). The localization of spontaneous brain activity: First results in patients with cerebral tumors. Clinical Neurophysiology, 112, 378–385. DeFries, J.C., Fulker, D.W., & LaBuda, M.C. (1987). Evidence for a genetic aetiology in reading disability of twins. Nature, 329, 537–39. Démonet, J.F., Taylor, M.J., & Chaix, Y. (2004). Developmental dyslexia. The Lancet, 363, 1451-1460. Dewey, D. (1995). What is developmental dyspraxia? Brain and Cognition, 29, 254-274. Duffy, F.H., Denckla, M.B., Bartels, P.H. & Sandini, G. (1980). Dyslexia: regional differences in brain electrical activity by topographic mapping. Annals of Neurology, 7, 412-420. Duncan, C.C, Rumsey, J.M., Wilkniss, S.M., Denckla, M.B., Hamburger, S.D., & OdouPotkin, M. (1994). Developmental dyslexia and attentional dysfunction in adults: brain potential indices of information processing. Psychophysiology, 31, 386-401. Eckert, M. (2004). Neuroanatomical markers for dyslexia: a review of dyslexia structural imaging studies. The Neuroscientist 10, 362–371. Eden, G.F., VanMeter, J.W., Rumsey, J.M., & Zeffiro, T.A. (1996). The visual deficit theory of developmental dyslexia. Neuroimage, 4, 108–117. Erez, A., & Pratt, H. (1992). Auditory event-related potentials among dyslexic and normalreading children: 3CLT and midline comparisons. International Journal of Neuroscience, 63, 247–264.

Reading Impairment in Childhood

685

Fabiani, M., Gratton, G., & Federmeier, K.D. (2007). Event-Related Brain Potentials: Methods, Theory, and Applications. In J. Cacioppo, L.G. Tassinary, G.G. Bernstom (Eds.), The handbook of psychophysiology (pp.85-119). Cambridge: Cambridge University Press. Fawcett, A.J., Nicolson, R.I., & Dean, P. (1996). Impaired performance of children with dyslexia on a range of cerebellar tasks. Annual Dyslexia, 46, 259-283. Fein, G., Galin, D., Yingling, C.D., Johnstone, J., Davenport, L., & Herron, H. (1986). EEG spectra in dyslexic and control boys during resting conditions. Electroencephalography and Clinical Neurophysiology, 63, 87-97. Fitzgerald, P.G., & Picton, T.W. (1983). Event-related potentials recorded during the discrimination of improbable stimuli. Biological Psychology, 17, 241–276. Flint, J. (1999). The genetic basis of cognition. Brain, 122, 2015-2032. Flynn, J.M., & Rahbar, M.H. (1994). Prevalence of reading failure in boys compared with girls. Psychology in the Schools, 31, 66-71. Fonseca, L.C., Tedrus, G.M.A.S., Chiodi, M.G., Cerqueira, J.N., & Tonelotto, J.M.F. (2006). Quantitative EEG in children with learning disability. Arquivos de Neuropsiquiatria, 64, 376-381. Francks, C., MacPhie, I.L., & Monaco, A.P. (2002). The genetic basis of dyslexia. The Lancet neurology, 1, 483-490. Fried, I., Tanguay, P. E., Boder, E., Doubleday, C., & Greensite, M. (1981). Developmental dyslexia: Electrophysiological evidence of clinical subgroups. Brain and Language, 12, 14–22. Francks, C., MacPhie, I.L., & Monaco, A.P. (2002). The genetic basis of dyslexia. The Lancet neurology, 1, 483-490. Galaburda, A.M. (1999). Developmental dyslexia: A multilevel syndrome. Dyslexia 5, 183– 191 Galaburda, A.M., & Kemper, T.L. (1979). Cytoarchitectonic abnormalities in developmental dyslexia: a case study. Annals of Neurology, 6, 94-100. Galaburda, A. M., Menard, M.T., & Rosen, G.D. (1994). Evidence for aberrant auditory anatomy in developmental dyslexia. The Proceedings of the National Academy of Sciences of the U.S.A., 91, 8010-8013. Galaburda, A. M., Sherman, G.F., Rosen, G.D., Aboitiz, F., & Geschwind, N. (1985). Developmental dyslexia: four consecutive patients with cortical anomalies. Annals of Neurology, 18, 222- 233. Gasser, T., Verleger, R., Bacher, P., & Sroka, L. (1988). Development of the EEG of schoolage children and adolescents. I. Analysis of band power. Electroencephalography and Clinical Neurophysiology, 69, 91-99. Georgiewa, P., Rzanny, R., Gaser, C., Gerhard, U.J., Vieweg, U., Freesmeyer, D., Mentzel, H.J., Kaiser, W.A., & Blanz, B. (2002). Phonological processing in dyslexic children: a study combining functional imaging and event related potentials. Neuroscience Letters, 318, 5–8. Gevins, A., & Smith, M.E. (2000). Neurophysiological measures of working memory and individual differences in cognitive ability and cognitive style. Cerebral Cortex 10, 829– 839.

686

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

Gevins, A., Smith, M.E., McEvoy, L., & Yu, D. (1997). High-resolution EEG mapping of cortical activation related to working memory: effects of task difficulty, type of processing and practice. Cerebral Cortex, 7, 374– 385. Geschwind, N., & Galaburda, A.M., (1985). Cerebral lateralization. Biological mechanisms, associations, and pathology: II. A hypothesis and a program for research. Archives of Neurology, 42, 521-552. Geschwind, N., & Levitsky, W. (1968). Human brain: left-right asymmetries in temporal speech region. Science, 161, 186-187. Grant, A.C., Zangaladze, A., Thiagarajah, M.C., & Sathian, K. (1999). Tactile perception in developmental dyslexia: a psychophysical study using gratings. Neuropsychologia 37, 1201–11. Gross-Tsur, V., Shalev, R.S., Manor, O., & Amir, N. (1995). Developmental right-hemiphere syndrome: clinical spectrum of the nonverbal learning disability. Journal of Learning Disabilities, 28, 80-86. Guttorm, T. K., Leppänen, P.H.T., Richardson, U., & Lyytinen, H. (2001). Event-related potentials and consonant differentation in newborns with familial risk for dyslexia. Journal of Learning Disabilities, 34, 534–544. Guttorm, T. K., Leppänen, P.H.T., Tolvanen, A., & Lyytinen, H. (2003). Event-related potentials in newborns with and without familial risk for dyslexia: Principal component analysis reveals differences between the groups. Journal of Neural Transmission, 110, 1059–1074. Habib, M. (2000). The neurological basis of developmental dyslexia: an overview and working hypothesis. Brain, 123, 2373-2399. Habib, M., & Robichon, F.(1996). Parietal lobe morphology predicts phonological skills in developmental dyslexia. Brain and Cognition, 32, 139-142. Harmony, T., Hinojosa, G., Marosi, E., Becker, J., Rodríguez, M., Reyes, A., & Rocha, C. (1990a). Correlation between EEG spectral parameters and an educational evaluation. International Journal of Neuroscience, 54, 147-155. Harmony, T., Marosi, E., Díaz de León, A.E., Becker, J. & Fernández, T. (1990b). Effect of sex, psychosoeial disadvantages and biological risk factors on EEG maturation. Electroencephalography and Clinical Neurophysiology, 75, 482-491. Helenius, P., Salmelin, R., Service, E., & Connolly, J.F. (1999). Semantic cortical activation in dyslexic readers. The Journal of Cognitive Neuroscience, 11, 535-550. Helenius, P., Tarkiainen, A., Cornelissen, P., Hansen, P.C., & Salmelin, R. (1999). Dissociation of normal feature analysis and deficient processing of letter-strings in dyslexic adults. Cerebral Cortex 9, 476–83. Herman, A.E., Galaburda, A.M., Fitch R.H., Carter, A.R., & Rosen, G.D. (1997). Cerebral microgyria, thalamic cell size and auditory temporal processing in male and female rats. Cerebral Cortex, 7, 53–464 Holcomb, P. J., Ackerman, P. T., & Dykman, R. A. (1985). Cognitive event-related brain potentials in children with attention and reading deficits. Psychophysiology, 22, 656–667. Holcomb, P. J., Ackerman, P. T., & Dykman, R. A. (1986). Auditory event-related potentials in attention and reading disabled boys. International Journal of Psychophysiology, 3, 263–273. Humphreys, P., Kaufmann, W.E., & Galaburda, A.M. (1990). Developmental dyslexia in women: neuropathological findings in three patients. Annals of Neurology, 28, 727-738.

Reading Impairment in Childhood

687

Jensen, O., & Tesche, C.D. (2002). Frontal theta activity in humans increases with memory load in a working memory task. European Journal of Neuroscience 15, 1395– 1399. Jernigan, T.L., Hesselink, J.R., Sowell, E., & Tallal, P.A. (1991). Cerebral structure on magnetic resonance imaging in language- and learning-impaired children. Archives of Neurology, 48, 539-545. John, E.R., Ahn, H., Prichep, L., Trepetin, M., Brown, D., & Kaye, H. (1980). Developmental equations for the EEG. Science, 210, 1255- 1258. John, E.R., Prichep, L., Ahn, H., Easton, P., Friedman, J., & Kaye, H. (1983). Neurometric evaluation of cognitive dysfunctions and neurological disorders in children. Progress in Neurobiology, 21, 239-290 Kaufmann, W.E., & Galaburda, A.M. (1989). Cerebrocortical microdysgenesis in neurologically normal subjects: a histopathologic study. Neurology, 39, 238-244. Klimesch, W. (1999). EEG alpha and theta oscillations reflect cognitive and memory performance: A review and analysis. Brain Research Reviews, 29, 169-195. Klimesch, W., Doppelmayr, M., Wimmer, H., Gruber, W., Röhm, D., Schwaiger, J., & Hutzler, F. (2001a). Alpha and beta band power changes in normal and dyslexic children. Clinical Neurophysiology, 112, 1186-1195. Klimesch, W., Doppelmayr, M., Wimmer, H., Schwaiger, J., Röhm, D., Gruber, W., & Hutzler, F. (2001b). Theta band power change in normal and dyslexic children. Clinical Neurophysiology, 112, 1174-1185. Kraus, N., McGee, T., Carrell, T.D., & Sharma, A. (1995). Neurophysiologic basis of speech discrimination. Ear and Hearing, 16, 19-37. Kronbichler, M., Hutzler, F., & Wimmer, H. (2002). Dyslexia: verbal impairments in the absence of magnocellular impairments. Neuroreport, 13, 617-620. Kujala, T., Alho, K., & Näätänen, R. (2000). Cross-modal reorganization of human cortical functions. Trends in Neurosciences, 23, 115–120. Kujala, T., Halmetoja, M. A., Näätänen, R., Alku, P., Lyytinen, H., & Sussman, E. (2006). Speech- and sound-segmentation in dyslexia: evidence for a multiple-level cortical impairment. European Journal of Neuroscience, 24 , 2420–2427 Kujala, T., Karma, J., Ceponiené, R., Belitz, S., Turkkila, P., Tervaniemi, M., & Näätänen, R., (2001). Plastic neural changes and reading improvement caused by audiovisual training in reading-impaired children. Proceedings of the National Academy of Sciences, 98, 10509–10514. Kutas, M., & Federmeier, K.D. (2000). Electrophysiology reveals semantic memory use in language comprehension. Trends in Cognitive Sciences, 4, 463-470. Landerl, K., Wimmer, H., & Frith, U. (1997). The impact of orthographic consistency on dyslexia: A German-English comparison. Cognition, 63, 315–334. Leppänen, P. H. T., Pihko, E., Eklund, K. M., & Lyytinen, H. (1999). Cortical responses of infants with and without a genetic risk for dyslexia: II. Group effects. Neuroreport, 10, 901–905. Leung, L.S. (1992). Fast (beta) rhythms in the hippocampus: A review. Hippocampus 2, 9398. Leung, L.S. (1998). Generation of theta and gamma rhythms in the hippocampus. Neuroscience and Behavioral Reviews 22, 275-290.

688

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

Lindgren S. D., De Renzi E., & Richman L. C., (1985). Cross national comparisons of developmental dyslexia in Italy and the United States. Child Development, 56, 14041417. Livingstone, M.S., & Hubel, D.H. (1988). Segregation of form, color, movement and depth: anatomy, physiology and perception. Science, 240, 740-749. Livingstone, M.S., Rosen, G.D., Drislane, F.W., &, Galaburda, A.M. (1991). Physiological and anatomical evidence for a magnocellular defect in developmental dyslexia. The Proceedings of the National Academy of Sciences of the U.S.A., 88, 7943–7947. Lovegrove, W.J., Bowling, A., Badcock, D., & Blackwood, M. (1980a) Specific reading disability: differences in contrast sensitivity as a function of spatial frequency. Science, 210, 439–440. Lovegrove, W.J., Heddle, M., & Slaghuis, W. (1980b). Reading disability: spatial frequency specific deficits in visual information store. Neuropsychologia, 18, 111-115. Lovrich, D., & Stamm, J. S. (1983). Event-related potential and behavioral correlates of attention in reading retardation. Journal of Clinical Neuropsychology, 5, 13–37. Lyytinen, H., Guttorm, T.K., Huttunen, T., Hämäläinen, J., Leppänen, P.H.T., & Vesterinen, M., (2005). Psychophysiology of developmental dyslexia: a review of findings including studies of children at risk of dyslexia. Journal of Neurolinguistic, 18, 167-195. Martin F., & Lovegrove, W. (1987). Flicker contrast sensitivity in normal and specifically disabled readers. Perception, 16, 215-221. Matoušek, M., & Petersén, I. (1973). Frequency analysis of the EEG in normal children and in normal adolescents. In P. Kellaway & I. Petersén (Eds.), Automation of Clinical Electroencephalography (pp. 75-102). New York: Raven Press. Maurer, U., Brem, S., Bucher, K., & Brandeis, D. (2003a). Altered tone and phoneme mismatch negativity in children at risk for dyslexia. Neuroreport, 14, 2245–2250. Maurer, U., Bucher, K., Brem, S., & Brandeis, D. (2003b). Development of the automatic mismatch response: From frontal positivity in kindergarten children to the mismatch negativity. Clinical Neurophysiology, 114, 808–817. Mazzotta, G., & Gallai, V. (1992). Study of the P300 event-related potential through brain mapping in phonological dyslexics. Acta Neurologica, 14, 173–186. Merzenich, M.M., Schreiner, C.E., Jenkins, W.M., & Wang, X. (1993). Neural mechanisms underlying temporal integration, segmentation, and input sequence representation: some implication for the origin of learning disabilities. In P. Tallal, A.M. Galaburda, R.R. Llinas, & C. von Euler (Eds.), Temporal information processing in the nervous system: special reference to dyslexia and disgraphia (Annals of the New York Academy of Sciences, 682, pp.1-22). New York: New York Academy of Sciences. Milne, D.R., Hamm, J.P, Kirk, I.J., & Corballis, M.C. (2003). Anterior-posterior beta asymmetries in dyslexia during lexical decisions. Brain and Language, 84, 309-317. Molfese, D. L. (2000). Predicting dyslexia at 8 years of age using neonatal brain responses. Brain and Language, 72, 238–245. Molfese, D. L., & Molfese, V. J. (1997). Discrimination of language skills at five years of age using event-related potentials recorded at birth. Developmental Neuropsychology, 13, 135–156. Molfese, V. J., Molfese, D. L., & Modglin, A. A. (2001). Newborn and preschool predictors of second-grade reading scores: An evaluation of categorical and continuous scores. Journal of Learning Disabilities, 34, 545–554.

Reading Impairment in Childhood

689

Neville, H.J., Coffey, S.A., Holcomb, P.J., & Tallal, P. (1993). The neurobiology of sensory and language processing in language-impaired children. The Journal of Cognitive Neuroscience 5, 235–253. Nicolson, R.I., & Fawcett, A.J. (1990). Automaticity: a framework for dyslexia research? Cognition, 35, 159-182. Nicolson, R.I., Fawcett, A.J., & Dean, P. (2001). Developmental dyslexia: the cerebellar deficit hypothesis. Trends in Neurosciences, 24, 508–11. Olson, R.K., (2002). Nature and nurture. Dyslexia, 8, 143-159. Ortiz, T., Exposito, F.J., Miguel, F., Martin-Loeches, M., & Rubia, F.J. (1992). Brain mapping in dysphonemic dyslexia: in resting and phonemic discrimination conditions. Brain & Language, 42, 270-285. Pantev, C., Makeig, S., Hoke, M., Galambos, R., Hampson, S., & Gallen, C. (1991). Human auditory evoked gamma-band magnetic fields. of the National Academy of Science of the United States of America 88, 8996-9000. Paulesu, E., Démonet, J.F., Fazio, F., McCrory, E., Chanoine, V., Brunswick, N., Cappa, S.F., Cossu, G., Habib, M., Frith, C.D., & Frith, U.(2001). Dyslexia: cultural diversity and biological unity. Science, 291, 2165–67. Paulesu E., Frith, U., Snowling, M., Gallagher, A., Morton, J., Frackoviak, R.S., & Frith, C.D. (1996). Is developmental dyslexia a disconnession syndrome? Evidence from PER scanning. Brain, 119, 143-157. Pennington, B.F. (1997). Using genetics to dissect cognition. The American Journal of Human Genetics 60, 13. Pennington, B.F., & Lefly, D.L. (2001). Early reading development in children at family risk for dyslexia. Child Development, 72, 816-833. Penolazzi, B., Hauk, O., & Pulvermüller, F. (2007). Early semantic context integration and lexical access as revealed by event-related potentials. Biological Psychology, 74, 374388. Penolazzi, B., Spironelli, C., & Angrilli, A. (in press). EEG delta activity as index of linguistic impairment in dyslexic children (Psychophysiology). Penolazzi, B., Spironelli, C., Vio C., & Angrilli, A. (2006). Altered hemispheric asymmetry during word processing in dyslexic children: an event related potential study. Neuroreport, 17, 429-433. Peiffer, A.M., Rosen, G.D., & Fitch, R.H. (2002). Rapid auditory processing and MGN morphology inmicrogyric rats reared in varied acoustic environments. Developmental Brain Research, 138, 187–193. Pihko, E., Leppänen, P.H.T., Eklund, K. M., Cheour, M., Guttorm, T. K., & Lyytinen, H. (1999). Cortical responses of infants with and without a genetic risk for dyslexia: I. Age effects. Neuroreport, 10, 901–905. Pizzagalli, D.A. (2007). Electroencephalography and High-Density Electrophysiological Source Localization. In J. Cacioppo, L.G. Tassinary, G.G. Bernstom (Eds.), The handbook of psychophysiology (pp.56-84). Cambridge: Cambridge University Press. Price, C.J., (1998). The functional anatomy of word comprehension and production. Trends in Cognitive Sciences, 2, 281-288. Ramus, F. (2003). Developmental dyslexia: Specific phonological deficit or general sensorimotor dysfunction? Current Opinion in Neurobiology 13, 1-7.

690

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

Ramus, F. (2004). Neurobiology of dyslexia: A reinterpretation of the data. Trends in Neurosciences, 27, 720-726. Ramus, F. (2006). A neurological model of dyslexia and other domain-specific developmental disorders with an associated sensorimotor syndrome. In G.D. Rosen (Ed.), The dyslexic brain: new pathways in neuroscience (pgg.75-101). Mahwah: Lawrence Erlbaum Associates. Ramus, F., & Szenkovits, G. (2008). What phonological deficit? The Quarterly Journal of Experimental Psychology, 61, 129-141. Ramus, F., Rosen, S., Dakin, S.C., Day, B.L., Castellote, J.M., White, S., & Frith, U., (2003). Theories of developmental dyslexia: Insights from a multiple case study of dyslexic adults. Brain 126, 841-865. Rapin, I., & Allen, D.A. (1988). Syndromes in developmental dysphasia and adult aphasia. In F. Plum (Ed.), Language, comunication and brain (pgg. 57-75). New York: Raven Press. Rippon, G., Brunswick, N. (1998). EEG correlates of phonological processing in dyslexic children. Journal of Psychophysiology 12, 261– 274. Rippon, G., Brunswick, N. (2000). Trait and state EEG indices of information processing in developmental dyslexia. International Journal of Psychophysiology, 36, 251– 265. Robichon, F., Besson, M., & Habib, M. (2002). An electrophysiological study of dyslexic and control adults in a sentence reading task. Biological Psychology, 59, 29-53. Robichon, F., & Habib, M. (1998). Abnormal callosal morphology in male adult dyslexic: relationships to handedness and phonological abilities. Brain and Language, 62, 127-146. Robichon, F., Lévrier, O., Farnarier, P., & Habib, M. (2000). Developmental dyslexia: atypical asymmetry of language areas and its functional significance. The European Journal of Neurology, 7, 35-46. Rosen, G.D., Herman, A.E., & Galaburda, A.M. (1999). Sex differences in the effects of early neocortical injury on neuronal size distribution of the medial geniculate nucleus in the rat are mediated by perinatal gonadal steroids. Cerebral Cortex 9, 27–34. Rosenthal, J. H., Boder, E., & Callaway, E. (1982). Typology of developmental dyslexia: Evidence for its construct validity. In R. N. Malastesha (Ed.), Reading disorders: Varieties and treatments (pp. 93–117). New York: Academic Press. Sarnthein, J., Morel, A., von Stein, A., & Jeanmonod, D. (2003). Thalamic theta field potentials and EEG: High thalamocortical coherence in patients with neurogenic pain, epilepsy and movement disorders. Thalamus and Related Systems, 2, 231-238. Sarnthein, J., Petsche, H., Rappelsberger P., Shaw, G.L., & von Stein, A. (1998). Synchronization between prefrontal and posterior association cortex during human working memory. Proceedings of the National Academy of Science of the United States of America, 95, 7092-7096. Sauseng, P., Klimesch, W., Dopplemayr, M., Hanslmayr, S., Schabus, M., & Gruber, W.R. (2004). Theta coupling in the human electroencephalogram during a working memory task. Neuroscience Letters, 354, 123-126. Sauseng, P., Klimesch, W., Schabus, M., & Dopplemayr, M. (2005). Fronto-parietal EEG coherence in theta and upper alpha reflect central executive functions of working memory. International Journal of Psychophysiology, 57, 97-103. Schulte-Körne, G., Deimel, W., Bartling, J., & Remschmidt, H. (2001). Speech perception deficit in dyslexic adults as measured by mismatch negativity (MMN). International Journal of Psychophysiology, 40, 77–87.

Reading Impairment in Childhood

691

Schulte-Körne, G., Deimel, W, Müller, K., Gutenbrunner, C., & Remschmidt, H. (1996). Familial aggregation of spelling disabilities. The Journal of Child Psychological Psychiatry, 37, 817-822. Share, D.L., Jorm, A.F., MacLean, R., & Matthews. R. (2002). Temporal processing and reading disability. Reading and Writing, 15,151-178. Shaywitz, S.E., (1998). Dyslexia. The New England Journal of Medicine, 338, 307-312. Shaywitz, B.A., Shaywitz, S.E., Pugh, K.R., Mencl, W.E., Fulbright, R.K., Skudlarski, P., Constable, R.T., Marchione, K.E., Fletcher, J.M., Lyon, G.R., & Gore, J.C., (2002). Disruption of posterior brain systems for reading in children with developmental dyslexia. Biological Psychiatry, 52, 101-110. Shaywitz, S.E., & Shaywitz, B.A., (2005). Dyslexia (Specific Reading Disability). Biological Psychiatry, 57, 1301-1309. Shaywitz, S.E., Shaywitz, B.A., Fulbright, R.K., Skudlarski, P., Mencl, W.E., Constable, R.T., Pugh, K.R., Holahan, J.M., Marchione, K.E., Fletcher, J.M., Lyon, G.R., & Gore, J.C. (2003). Neural systems for compesation and persistence: Young adult outcome of childhood reading disability. Biological Psychiatry, 4, 25-33. Simos, P.G., Breier, J.L., Fletcher, J.M., Bergman, E., & Papanicolaou, A.C., (2000). Cerebral mechanisms involved in word reading in dyslex children: A magnetic source imaging approach. Cerebral Cortex, 10, 809-816. Snowling, M.J. (2001). From language to reading and dyslexia. Dyslexia, 7, 37-46. Spironelli, C. & Angrilli, A. (2006). Language lateralization in Phonological, Semantic and Orthographic tasks: A slow evoked potential study. Behavioral Brain Research, 175, 296304. Spironelli, C., Penolazzi, B., & Angrilli, A. (2008). Dysfunctional hemispheric asymmetry of theta and beta EEG activity during linguistic tasks in developmental dyslexia. Biological Psychology, 77, 123-131. Spironelli, C., Penolazzi, B., Vio, C., & Angrilli, A. (2006). Inverted EEG theta lateralization in dyslexic children during phonological processing. Neuropsychologia, 44, 2814-2821. Sklar, B., Hanley, J., & Simmons, W.W. (1972). An EEG experiment aimed toward identifying dyslexic children. Nature, 240, 414-416. Slotnick, S.D., Moo, L.R., Kraut, M.A., Lesser, R.P., & Hart, J.Jr. (2002). Interaction between thalamic and cortical rhythms during semantic memory recall in human. Proceedings of the National Academy of Science of the United States of America, 99, 6440-6443. Smith, S.D., Gilger, J.W., & Pennington, B.F., (1996). Dyslexia and other specific learning disorders. In D.L. Rimoin, J.M. Connor, & R.E. Pyeritz, (Eds.), Principles and practice of medical genetics (pgg. 1767–89). New York: Churchill Livingstone. Smith, S.D., Kelley, P.M., & Brower, A.M. (1998). Molecular approaches to the genetic analysis of specific reading disability. Human Biology, 70, 239-256. Stein, J.F. (1993).Visuospatial perception in disabled readers. In D.M. Willows, & R.S. Kruk (Eds.), Visual processes in reading and reading disability (pp.331-346). Hillsdale: Lawrence Erlbaum Associates. Stein, J. F. (2001). The magnocellular theory of developmental dyslexia. Dyslexia, 7,12-36. Stein, J.F., & Talcott, J.B. (1999). Impaired neuronal timing in developmental dyslexia-The magnocellular hypothesis. Dyslexia, 5, 59-77.

692

Barbara Penolazzi, Chiara Spironelli and Alessandro Angrilli

Stein, J.F., & Walsh, V. (1997). To see but not to read: the magnocellular theory of dyslexia. Trends in Neurosciences, 20, 147-152. Stelmack, R.M., Saxe, B.J., Noldy-Cullum, N., Campbell, K.B., & Armitage, R. (1988). Recognition memory for words and event-related potentials: A comparison of normal and disabled readers. The Journal of Clinical and Experimental Neuropsychology, 10, 185– 200. Stevenson, J., Graham, P., Fredman, G., & McLoughlin, V.A., (1987). Twin study of genetic influences on reading and spelling disability. The Journal of Child Psychology and Psychiatry, 229–47. Tallal, P. (1980). Auditory temporal perception, phonics and reading disabilities in children. Brain and Language, 9, 182-198. Tallal, P., & Piercy, M. (1973). Defects of non-verbal auditory perception in children with developmental aphasia. Nature, 241, 468-469. Tallon-Baudry, C., & Bertrand, O. (1999). Oscillatory gamma activity in humans and its role in object representation. Trends in Cognitive Sciences 3, 151-162. Taylor, M.J., & Keenan, N.K. (1990). Event –related potentials to visual and language stimuli in normal and dyslexic children. Psychophysiology, 27, 318-327. Temple, E., Poldrack, R.A., Salidis, J., Deutsch, G.K., Tallal, P., Merzenich, M.M., & Gabrieli, J.D.E., (2001). Disrupted neural responses to phonological and orthographic processing in dyslex children: An fMRI study. Neuroreport 12, 299-307. Tesche, C.D., & Karhu, J. (2000). Theta oscillations index human hippocampal activation during a working memory task. Proceedings of the National Academy of Science of the United States of America, 97, 919-924. Valdois, S., Gèrard C., Vanault, P., & Dugas, M. (1995). Pheripheral developmental dyslexia: a visual attentional account? Cognitive Neuropsychology, 12, 31-67. Vertes, R.P. (2005). Hippocampal theta rhythm: A tag for short-term memory. Hippocampus 15, 923-935. Wadsworth, S.J., DeFries, J.C., Stevenson, J., Gilger, J.W., & Pennington, B.F. (1992). Gender ratios among reading-disabled children and their siblings as a function of parental impairment. The Journal of Child Psychology and Psychiatry, 33, 1229-39. Wagner, R.K., & Torgesen, J.K., (1987). The nature of phonological processing and its causal role in the acquisition of reading skills. Psychological Bulletin, 101, 192-212. Wienbruch, C., Moratti, S., Elbert, T., Vogel, U., Fehr, T., Kissler, J., Schiller, A., & Rockstroh, B. (2003). Source Distribution of Neuromagnetic Slow Wave Activity in Schizophrenic and Depressive Patients. Clinical Neurophysiology, 114, 2052-2060. Wolf, M., Goldberg O’Rourke, A., Gidney, C., Lovett, M., Cirino, P., & Morris, R. (2002). The second deficit: an investigation of the independence of phonological and namingspeed deficits in developmental dyslexia. Reading and Writing, 15, 43-72. World Health Organization (1993). The International Classification of Diseases, vol.10: Classification of Mental and Behavioural Disorders. Geneva: World Health Organization. Yingling, C.D., Galin, D., Fein, G., Peltzman, D., & Davenport, L. (1986). Neurometrics does not detect "pure dyslexics". Electroencephalography and Clinical Neurophysiology, 63, 426-430. Zoccolotti, P., De Luca, M., Di Pace, E., Judica, A., Orlandi, M., & Spinelli, D. (1999). Markers of developmental surface dyslexia in a language (Italian) with high grapheme– phoneme correspondence. Applied Psycholingusitics, 20, 191–216.

In: Encyclopedia of Language and Linguistics (2 Volume Set) ISBN: 978-1-61324-223-0 Editor: Kimberly L. Meinken ©2012 Nova Science Publishers, Inc.

Chapter 28

INTERACTION BETWEEN FLEXIBLE COGNITION AND LANGUAGE COMPREHENSION IN CHILDREN WITH AND WITHOUT LANGUAGE IMPAIRMENT Klara Marton1 Brooklyn College, City University of New York, US & Barczi Gusztav Faculty of Special Education, Eotvos Lorand University, Budapest, Hungry

Abstract The focus of this chapter is on the impact of various cognitive functions on language acquisition and language processing in different groups of children. Empirical data are presented on individual variations of language and cognition in typically developing children, children with specific language impairment, and in young adults with dyslexia. The interaction between flexible cognition (attention switching, working memory, and inhibition control) and language comprehension was examined in English-speaking and Hungarian participants. The findings suggest that working memory performance and language comprehension are strongly influenced by task complexity, age, and language structure. Working memory storage and processing are affected differently by stimulus complexity. Children with specific language impairment and young adults with dyslexia showed deficits in flexible cognition and language comprehension (oral and written). Children with SLI performed more poorly than their age-matched and language-matched peers in tasks that measured working memory capacity and inhibition control. In addition to the group effects, participants with language impairment showed different errors and performance patterns than their peers.

1

Correspondence: Klara Marton, Address: Brooklyn College, 2900 Bedford Ave., Brooklyn, NY 11210, Phone: (718) 951-4402, Fax: (718) 951-4167, Email: [email protected]

694

Klara Marton

Introduction: Interaction Between Language and Non-Linguistic Cognitive Skills Language development and processing are entwined with non-linguistic cognitive skills in children and adults. The focus of this chapter is on the interaction between language acquisition and disorders and the development of higher level cognitive skills. Empirical results from typically developing school-age children, from children with specific language impairment (SLI), and from adolescents and young adults with dyslexia will be used to demonstrate the impact that working memory, inhibition control and other executive functions have on written and oral language performance. The theoretical framework for this discussion is the neuroconstructivist account (Karmiloff-Smith, 1998). This theory suggests that cognitive development is closely linked to the changes in the underlying neurological structures. According to the neuroconstructivist approach, the adult modular information processing system is the result of development. Children are born with a less differentiated system that has domain-relevant capacities instead of domain-specific ones (Thomas, 2005). The infant’s system shows more flexibility and different strengths and weaknesses than the adult structure (Paterson, Brown, Gsödl, Johnson, & Karmiloff-Smith, 1999). In accordance with the neuroconstructivist theory, there are constraints at each level from the genes to the social environment. Over the course of development, these structures interact with each other. Cognitive development cannot be understood without comprehending the constraints on neural development (Westermann, Mareschal, Johnson, Sirois, Spratling, & Thomas, 2007). There is continuous competition and collaboration among the different structures. The level of development will depend on the specific context in which the given structures compete and collaborate with each other. The neuroconstructivist account has further implications for developmental disorders. In response to multiple constraints, the system goes through an adaptation process, similar to the one observed in typical development, but the constraints differ in atypical cases. Thus, this approach promotes a different view of atypical development from the more modular idea that assumes failures in isolated functions. Variations in the genes may change the course of development and their effects will differ across outcomes. They will have stronger impacts on certain functions and weaker influences on others. Because of the interactions across various levels in a dynamic context, one cannot separate intact and impaired modules. According to this view, there is a shift from dissociations to cross- syndrome associations (KarmiloffSmith, 1998). Thus, the neuroconstructivist approach provides a more dynamic framework of developmental disorders than the traditional neuropsychological models. Researchers studying neurocognitive disorders have argued that children diagnosed with different impairments show many overlapping characteristics in their cognitive and linguistic profiles. Although the groups of children with different neurocognitive disorders may vary in the rate of development (e.g., children with Down syndrome, with Williams syndrome, with specific language impairment, or with autistic spectrum disorder), they show many similarities in their performance profiles and errors (Bates, 2004). Thus, these impairments may lie on a continuum (e.g., Bishop, 2000). During the first year of life, infants acquire numerous non-linguistic skills that promote the development of language. More complex language structures are established through associations between speech sounds and their meanings (Chiat, 2001). This mapping process

Interaction between Flexible Cognition and Language Comprehension …

695

is influenced by a number of non-linguistic cognitive skills, such as cross-modal perception and sensori-motor precision (Bates, 2004). Joint attention facilitates the mapping process during which children link the words and sentences with their corresponding objects and events (Baldwin, 1995). Infants are born with a higher level memory system that enables them to imitate actions even after a significant delay. The first 18 months are characterized by a shift from experience-based mental representations to more hypothetical ones (Meltzoff, 1990). During this time, cross-modal integration develops and supports the emergence of various forms of imitation, which plays a central role in the development of motor control, speech-language-communication and social life (Tomasello, Kruger, & Ratner, 19993). The combination of joint reference and imitation facilitates the development of both the theory of mind and observational learning. Based on the assumption that language and cognition develop in interaction, if there is a deficit in any of the above mentioned cognitive functions, language learning will be altered. The rate of language development is also determined by more general cognitive factors, such as IQ and auditory memory (Tager-Flusberg, 2005). During the preschool years, children’s language level is highly correlated with their nonlinguistic skills. For example, Marchman (1997) suggests that young children’s production of suffixes is related to their rate of language acquisition. Children with more advanced phonological systems produce fewer errors in English past tense morphology than their peers. Flexible cognition is an umbrella term for the cognitive system’s ability to adapt to contextual and task-related changes by shifting attention, selecting relevant information, forming plans and monitoring performance (Deak, 2003). Flexible cognition has been found to influence language development to a great extent. In a conversation, children must remember the topic (topic maintenance), shift attention according to the progress of the dialogue, inhibit irrelevant thoughts, and consider their partner's perspective. To comprehend complex or potentially ambiguous sentences, children must recognize the words, selectively activate relevant meanings, inhibit irrelevant ideas, remember the words, and parse the sentence, keeping track of element order as well as long-distance grammatical relations (e.g., subject-verb agreement, tense, elements that have undergone movement and their traces, etc.). Sentences that can have more than one meaning (i.e., are initially ambiguous) also require suppression of one potential meaning. At a lexical level, in tasks like picture naming, children must selectively activate the correct word in a cohort of phonologically and semantically related words and inhibit related non-target items. They must selectively attend to segments and features that distinguish the target word from phonological neighbors. The relationship between flexible cognition and language is bidirectional, but we do not know how much language development contributes to the advancement of cognition. Behavioral studies’ results show that preschool-age children have problems in tasks that test flexible cognition. These children experience difficulty in inhibiting thoughts and responses that are not relevant anymore (Diamond, 1998), in flexible switching (Zelazo, Muller, Frye, & Marcovitch, 2003), and in selecting task cues (Deak, 2000). In the following parts of this chapter we will present data on the interaction between language and verbal working memory, inhibition control, and other executive functions.

696

Klara Marton

Interaction Between Language and Verbal Working Memory: A Developmental Perspective There is a clear relationship between working memory and language comprehension and production (e.g., Gathercole & Baddeley, 1993; Just & Carpenter, 2002; Marton, Schwartz, Farkas, & Katsnelson, 2006). Phonological working memory, which is a measure of shortterm storage, is closely linked to vocabulary development, particularly during the first years of language acquisition. Young children with more limited phonological working memory capacity, as measured by nonword repetition, show slower vocabulary development than children with better nonword repetition skills. Nonwords do not activate the long-term lexical representations, therefore nonword repetition reflects the person’s phonological storage capacity. As children become more advanced language learners, they rely more on their lexicon. They access their conceptual, semantic and phonological knowledge during vocabulary acquisition (Gathercole, 2006). Language comprehension is also linked to verbal working memory. Various linguistic span tasks (reading span, listening span) have been shown to be great predictors of language comprehension (Daneman & Philip, 1996) because language processing has high computational and storage demands. Participants’ comprehension performance reflects individual differences in working memory capacity (e.g., Caplan & Waters, 1999; Just & Carpenter, 2002; King & Just, 1991). These differences, however, are not about the individual’s phonological storage capacity, but imply more general executive functions (e.g., Conway & Engle, 1994; Engle & Cane, 2004). Working memory span is constrained by controlled attention (Engle, Kane, & Tuholski, 1999), by task-switching ability (Towse, Hitch, & Hutton, 1998), and by the inhibition of irrelevant information (Hasher & Zacks, 1988). According to Baddeley's working memory model (Baddeley, 2000), participants may demonstrate individual variations in processing and storage independently from a general executive capacity in complex working memory span tasks (Bayliss, Jarrold, Gunn, & Baddeley, 2003). The extent to which performance on complex working memory span tasks depends on processing and storage functions is still subject to debate (see more detailed conceptual reviews of working memory e.g., Baddeley, 2000; Engle & Kane, 2004; Miyake & Shah, 1999). Developmental studies on working memory performance evidenced that executive functions develop gradually until adolescence (DeLuca, Wood, Anderson, Bucanan, Proffitt, Mahony, & Pantelis, 2003). Changes in working memory span reflect a combination of efficiency and speed of processing. Older children perform faster on the same tasks as younger children, therefore they need fewer resources to perform a particular activity (e.g., counting) and have more resources available for memory operations (Case, Kurland, & Goldberg, 1982). A different interpretation of age-related changes is provided by the taskswitching model (Towse & Hitsch, 1995). In linguistic span tasks, participants may use the sentence context to help reconstruct the list of words that have to be remembered. If this reconstitution process is too difficult for the individual’s system, then it may take so much time that some information could be lost. Younger children may make more errors because of difficulty in attention switching. In complex working memory tasks, span measures may reflect the ability to control/switch attention, whereas response times depend on retrieval speed (Hitsch, Towse, & Hutton, 2001; Kail & Salthouse, 1994). Increases in attentional

Interaction between Flexible Cognition and Language Comprehension …

697

capacity and advance in attention switching result in working memory span enhancement (Barrouillet & Camos, 2001). In our own research, we examined the effects of age and sentence complexity on verbal working memory performance (Marton, Schwartz, & Braun, 2005). Previous studies that have investigated age-related changes in working memory structures used short and/or simple sentences in linguistic span tasks (Cain, Oakhill, & Bryant, 2004; Gathercole, Pickering, Ambridge, & Wearing, 2004). Although the findings of adult studies showed an effect of linguistic complexity on working memory performance, the relations between linguistic complexity and age in children's working memory production is an uncharted area. Our study extended previous work by exploring the effect of sentence length and complexity on working memory performance in groups of participants that differed in age. We examined whether stimulus length or complexity has a larger effect on working memory performance accuracy and language comprehension, and whether this relationship changes with age during childhood and adolescence. Further, this study tested the theory of working memory capacity and executive attention (Engle & Kane, 2004) from a developmental perspective. Three groups of children and young adults participated in this experiment (n=75). There were 25 participants in each group; the groups differed in age (Group1: 7;0-8;6 years; Group 2: 9;6-11:00 years; Group 3: college students, 19-22 years). The rationale for choosing these age groups was based on previous findings that indicated immature executive functions, such as the maintenance of task goals, sustained attention, set shifting, and inhibition below 8-8;6 years of age. Although memory span improves between age 10 and adulthood, resistance to interference matures around 10 years of age (DeLuca, Wood, Anderson, Buchanan, Proffitt, Mahony, & Pantelis, 2003; Hale, Bronik, & Fry, 1997). To measure the relations between working memory and language comprehension, we developed a Modified Listening Span task (MLS). The task included 90 sentences (30 syntactically simple short sentences, 30 syntactically complex short sentences, and 30 syntactically complex long sentences) with a question for each sentence. Sentence-length was determined by the number of syllables (M = 7.91, SD = 1.64 for the short sentences; M = 17.64, SD = 3.47 for the long sentences). The complex sentences included relative clauses. The last word in each sentence was replaced with a nonword (2-, 3-, 4-syllable). The nonwords were part of the syntactic structure of the sentences. The task required that participants listen to a sentence, repeat the sentence-final nonword and answer a question about the content of the sentence. Thus, the task required simultaneous processing and attention switching. The traditional linguistic span task (TLS) consisted of 9 sets of sentences with 5 sentences in each set (3 sets with syntactically simple short sentences, 3 sets with syntactically complex short sentences, and 3 sets with syntactically complex long sentences). Participants were asked to listen to the sentences and repeat the 5 sentence-final words in the order of presentation. The words were real words with similar phonotactic patterns, syllable length, and frequency of occurrence. Following sentence presentation, yes/no questions were randomly asked. In addition to the listening span tasks, a single nonword repetition task was used as a baseline measure of phonological working memory. The task consisted of 24 nonwords (2-3-4 syllables). The results in each working memory task showed an age effect; younger children made more errors than the older participants. Performance accuracy also differed between Groups 2 and 3, particularly with more complex items.

698

Klara Marton

Age-related changes were further observed in different error types and performance patterns, especially with an increase in stimulus complexity. In addition to an age effect, the results of the linguistic span tasks showed an effect of sentence complexity on working memory performance. Participants in each group showed higher working memory performance accuracy following the simple sentences than the complex sentences. Sentence length differences did not impact performance accuracy in this study. This finding is similar to the results of Willis & Gathercole (2001). The authors reported that increased length of stimulus items did not affect sentence comprehension. In contrast to sentence length, syntactic complexity did have a significant effect on working memory performance. The two dependent measures of the MLS task reflected the contribution of storage and processing efficiency separately. Sentence complexity affected the answers to the questions, but not nonword repetition. Nonword repetition accuracy was influenced by word length, but not by syntactic complexity. The idea of separate contributions of storage and processing efficiency in the MLS task was further supported by the high correlation between single nonword repetition and nonword repetition in the MLS task, and by the high correlation between the answers to the questions in the MLS and the TLS tasks. These findings are in line with the results of Bayliss and colleagues (2003). The authors suggested that processing efficiency and storage capacity constrain complex span performance independently. The results of the linguistic span tasks (MLS, TLS) in our study were also in agreement with the findings of Gathercole, Pickering, Ambridge, and Wearing (2004), and showed the effect of age on executive functions. In linguistic span tasks, executive functions are used to integrate phonological, lexical, syntactic, and semantic representations. Younger children rely more on their storage resources and show more difficulty in attention switching and flexible adaptation to changing task requirements. In our study, this was evidenced not only by these children’s lower scores, but also by the age-related differences in performance pattern. In contrast to the results of the older participants, younger children (Group 1) produced more interference errors than omissions with each sentence type. They showed more difficulty in blocking the effects of prior sentence information. The children in Group 2 produced a higher number of interference errors only with increased sentence complexity. Most of their errors were omissions with the simple sentences. Performance accuracy in young adults (Group 3) was also influenced by sentence complexity, but they made significantly fewer interference errors than the children in Groups 1 and 2, even with the complex sentences. Young adults showed better monitoring of previously recalled items. These data are similar to the results of McCormack, Brown, & Vousden (2000); the authors reported an age-related decrease in interference errors in short-term memory tasks. A similar age-related pattern emerged when performance on the TLS task was scored with 2 different methods: free recall versus recall with order. There was no difference between the scores using the 2 scoring methods for the participants in Group 3. Children in Group 2, received similar scores with the two methods for the simple sentences, but not for the complex sentences. In complex sentences, children’s recall scores were higher with the free scoring method than with the one that included order scoring. The youngest children (Group 1) showed score differences with each sentence type, even with the simple sentences. The finding that children in Group 1 received higher scores with the free scoring method

Interaction between Flexible Cognition and Language Comprehension …

699

following each sentence type shows that they were able to remember the words but not the order of presentation. This finding supports the idea that younger children may rely more on storage resources than on general processing. Remembering both the items and their order of presentation requires the ability to maintain goal-relevant information and continuous attention switching. Although processing efficiency was influenced by sentence complexity, but not by sentence length, nonword repetition performance -that reflects the storage function rather than processing- showed a word-length effect. Performance on both tasks, single nonword repetition and nonword repetition with sentence context, decreased with the increase of syllable length. In the MLS task, there was an interaction between word length and age. The younger children performed more poorly than the older participants at each syllable length; performance accuracy between Groups 2 and 3 did not differ for the 2-syllable nonwords, only with the longer nonwords. Working memory span continues to improve between 1011 years of age and adulthood. Taken these results together with the findings on the linguistic span measures, the two functions of working memory, storage and processing efficiency were differently affected by the structure and complexity of the stimuli. Nonword repetition, the measure of the storage function, was influenced by stimulus length, but not by the sentence type, whereas the answers to the questions and listening span, that reflect processing efficiency, were affected by sentence complexity, but not sentence length. The results of the two listening span tasks support the theory of working memory capacity and executive attention (Engle & Kane, 2004). Performance accuracy increased and the proportion of interference errors decreased with age. These age-related differences reflect better attention switching and executive control. Participants needed to switch attention continuously during the processing of complex linguistic structures in the listening span tasks. There was a gradual increase in performance accuracy across groups: the listening span tasks were most demanding on working memory for the youngest children (Group 1), who performed more poorly than groups 2 and 3 with each sentence type. Group 2 differed from Group 3 only with the complex sentences. The development of executive functions and their effect on working memory performance were also reflected in children's errors. Younger children showed more difficulty in blocking previously presented information. This problem resulted in a high number of interference errors. Monitoring previously recalled items requires executive functions that develop with age. In conclusion, both age and language structure influence working memory performance. There is an interaction between these two factors that has an additional impact on individual performance variations in working memory tasks. The findings indicate that different working memory tasks do not involve the same processes and that both factors, language and age, may influence them differently.

700

Klara Marton

Interaction Between Language and Verbal Working Memory In Children With SLI Phonological Storage and Processing Children with SLI have significant limitations in their language abilities in the absence of hearing impairment and frank neurological damage. These children show typical interaction with people and objects and they show no signs of pervasive developmental disorders, such as autism. Their language scores on standardized tests are at least 1.25 SD lower than the average (