Language Development across Childhood and Adolescence [1 ed.] 9789027295002, 9789027234735

Citation preview

Language Development across Childhood and Adolescence

Trends in Language Acquisition Research Official publication of the International Association for the Study of Child Language (IASCL). IASCL website: http://atila-www.uia.ac.be/IASCL

Series Editors Annick De Houwer University of Antwerp/UIA [email protected]

Steven Gillis University of Antwerp/UIA [email protected]

Advisory Board Jean Berko Gleason Boston University

Ruth Berman Tel Aviv University

Philip Dale University of Missouri-Columbia

Paul Fletcher Hong Kong University

Brian MacWhinney Carnegie Mellon University

Volume 3 Language Development across Childhood and Adolescence Edited by Ruth A. Berman

Language Development across Childhood and Adolescence Edited by

Ruth A. Berman Tel Aviv University

John Benjamins Publishing Company Amsterdam
/
Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Language development across childhood and adolescence / edited by Ruth A. Berman. p. cm. (Trends in Language Acquisition Research, issn 1569–0644 ; v. 3) Includes bibliographical references and index. 1. Language acquisition. I. Berman, Ruth Aronson. II. Series. P118.L262 2004 401’.93-dc22 isbn 90 272 3473 6 (Eur.) / 1 58811 582 8 (US) (Hb; alk. paper)

2004058010

© 2004 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents

Contributors

vii

Series editors’ preface

ix

Preface

xi

Research on later language development: International perspectives Marilyn A. Nippold Between emergence and mastery: The long developmental route of language acquisition Ruth A. Berman

1

9

Lexical acquisition in the early school years Julie E. Dockrell and David Messer

35

Derivational morphology revisited: Later lexical development in Hebrew Dorit Ravid

53

Discursive constraints on the lexical realization of arguments in Spanish Ekaterina Khorounjaia and Liliana Tolchinsky

83

Syntactic ability in children and adolescents with language and learning disabilities Cheryl M. Scott

111

Growing into academic French Harriet Jisa

135

Learning to spell in a deep orthography: The case of French Sébastien Pacton and Michel Fayol

163

Text–writing development viewed through on-line pausing in Swedish Åsa Wengelin and Sven Strömqvist

177

The role of peer interaction in later pragmatic development: The case of speech representation Shoshana Blum-Kulka

191



Table of contents

On reading poetry: Expert and novice knowledge Joan Peskin and David Olson

211

The nature and scope of later language development Liliana Tolchinsky

233

References

249

Author index

289

Subject index

297

Contributors

Berman, Ruth A. Department of Linguistics Tel Aviv University Ramat Aviv, Israel 69978 e-mail: [email protected]

Blum-Kulka, Shoshana Department of Communication and Journalism Hebrew University Jerusalem, Israel 91905 e-mail: [email protected]

Dockrell, Julie E. Psychology and Human Development Institute of Education 20 Bedford Way London WC1A 0AL, England e-mail: [email protected]

Fayol, Michel LAPSCO/CNRS UMR 6024 Université Blaise Pascal 34 avenue Carnot, 3000 Clermont-Ferrand, France e-mail: [email protected]

Jisa, Harriet Laboratoire Dynamique du Langage (UMR 5596) & Université Lyon 2 14 avenue Berthelot 69363 Lyon Cedex 07, France e-mail: [email protected]

Khorounjaia, Ekaterina Department of Linguistics University of Barcelona Gran Vma Cortes Catalanes 585 08007 Barcelona, Spain e-mail: [email protected] Messer, David Psychology Department London South Bank University 103 Borough Road London SE1 0AA, England e-mail: [email protected] Nippold, Marilyn A. Communication Disorders and Sciences University of Oregon Eugene, Oregon 97403, USA email: [email protected] Pacton, Sébastien Laboratoire Cognition et Diveloppement (CNRS, UMR-8605) Universiti Reni Descartes – Paris V 71 avenue Edouard Vaillant, F-92774 Boulogne-Billancourt Cedex, France e-mail: [email protected] Peskin, Joan Centre for Applied Cognitive Science, Department of Human Development and Applied Psychology, OISE/University of Toronto, 252 Bloor Street West, Toronto, Ontario, M5S1V6, Canada e-mail: [email protected]

 Contributors

Olson, David Centre for Applied Cognitive Science, Department of Human Development and Applied Psychology, OISE/University of Toronto, 252 Bloor Street West, Toronto, Ontario, M5S1V6, Canada e-mail: [email protected] Ravid, Dorit School of Education and Department of Communication Disorders Tel Aviv University Ramat Aviv, Israel 69978 e-mail: [email protected] Scott, Cheryl M. Rush University Medical Center 1653 West Congress Parkway, Suite 203 Senn, Chicago, IL 60612, USA e-mail: [email protected]

Strömqvist, Sven Department of Linguistics Helgonabacken 12 SE-223 62 Lund, Sweden e-mail: [email protected]

Tolchinsky, Liliana Department of Linguistics University of Barcelona Gran Vma Cortes Catalanes 585 08007 Barcelona, Spain e-mail: [email protected]

Wengelin, Asa Department of Linguistics Helgonabacken 12 SE-223 62 Lund, Sweden e-mail: [email protected]

Series editors’ preface

We are very proud to present the third volume in the series ‘Trends in Language Acquisition Research’. As an official publication of the International Association for the Study of Child Language (IASCL), the TiLAR Series aims to publish two volumes per three-year period between IASCL congresses. All volumes in the IASCL-TiLAR Series will be invited (but externally reviewed) edited volumes by IASCL members that are strongly thematic in nature and that present cutting edge work which is likely to stimulate further research to the fullest extent. Besides quality, diversity is also an important consideration in all the volumes: diversity of theoretical and methodological approaches, diversity in the languages studied, diversity in the geographical and academic backgrounds of the contributors. After all, like the IASCL itself, the IASCL-TiLAR Series is there for child language researchers from all over the world. Although it is IASCL policy to try and link one of the two tri-annual volumes in the Series to the main topic of the preceding IASCL congress, the IASCL-TiLAR series is emphatically not intended as congress or symposia proceedings. This means that in the volumes related to congress themes there can be contributions by IASCL members that were not presented at the congress. We are pleased to present the third volume of the TiLAR series, which is devoted to Language Development Across Childhood and Adolescence. The volume editor, Ruth A. Berman, brings together psycholinguistic studies of various aspects of language acquisition across childhood and adolescence, with a special emphasis on crosslinguistic research in this area. This volume intends to bring its readership up to date on the most recent developments in this area of research. The volume is introduced by an overview chapter written by Marilyn A. Nippold, and concludes with a discussion written by Liliana Tolchinsky. Finally we would like to thank Brian MacWhinney, the former president of the IASCL, and Michael Tomasello, the current president of the IASCL, for their enthusiastic support in setting up the IASCL-TiLAR Series, and Seline Benjamins and Kees Vaes of John Benjamins Publishing Company for their professional and creative input throughout the preparation of this volume as well as the series as a whole. We are also grateful to the external reviewers



Series editors’ preface

whose constructive criticisms and judgements contributed much to the quality of this book. ‘Trends in Language Acquisition Research’ is made for and by IASCL members. We hope it can become a source of information and inspiration which the community of child language researchers can continually turn to in their professional endeavors. General Editors Antwerp, June 2004

Preface

It is both a privilege and a pleasure to have edited this book, the third volume in the Trends in Language Acquisition Research (TiLAR) series of the International Association for the Study of Child Language. A key motivation for selecting as its topic the domain of later language development was to demonstrate that, although findings of language acquisition research since the 1960s are in many respects consistent with the Chomsky-inspired view of the process as rapid, effortless, and largely completed by age three to five years, in fact the route taken by children from the initial emergence of linguistic knowledge to mastery of both language structure and language use has “a long developmental history”. A second motivation was to supplement the educational and clinical concerns that have informed much prior work on language across the school years by research undertaken from psycholinguistic and crosslinguistic perspectives. To meet this aim, the chapters in this volume, both jointly and individually, examine the combined impact of cognitive and social maturation, on the one hand, and of increased experience with literacy-related activities, on the other, to demonstrate how schoolchildren and adolescents come to flexibly deploy an extensive repertoire of lexical elements in an augmented range of morphosyntactic constructions and discourse contexts in order to achieve maturely proficient language use. The nature of these questions, and the manner in which this book addresses them, are clearly articulated in the introduction to its contents by Marilyn Nippold, whose groundbreaking volume of 1988 set the stage for research on later language development. The bulk of the book is made up of ten chapters that provide overviews of current research conducted by their authors and by other scholars in different domains of linguistic structure and use. Ruth Berman’s chapter considers the paradox of the early emergence but late mastery of language knowledge in terms of a developmentally motivated phase-based model of language acquisition across adolescence, illustrated by three domains in acquisition of Hebrew as a first language: lexical semantics (different classes of adjectives), morpho-syntax (passive constructions), and syntactic structure (impersonals and nominalizations). The next two chap-

 Preface

ters concern the lexicon, based on data from English and Hebrew respectively. Dockrell and Messer review studies of early vocabulary acquisition to demonstrate that most such accounts fail to take into consideration a rich enough constellation of factors to characterize what is involved in word knowledge in general, and in vocabulary growth in the early school years among normally developing and language impaired populations in particular. Ravid provides an overview of psycholinguistic and developmental approaches to the lexicon and lexical acquisition, followed by experimental data from a range of studies of later vocabulary acquisition among Hebrew-speaking schoolchildren and adults – with supporting evidence from language-impaired populations – to argue for the critical role of derivational morphology in development of the ‘literate lexicon’. The next three chapters focus on syntax. Khoroujaina and Tolchinsky consider the lexical realization of arguments in oral Spanish texts produced by schoolchildren, adolescents, and adults (elicited in the framework of a large scale crosslinguistic project on developing text construction abilities) to demonstrate that the linguistic notion of “Preferred Argument Structure” is indeed discursively well-motivated, that it is fully established by early school age, and that it finds distinct expression in different types of discourse, in their case, of personal experience narratives compared with expository discussions. Scott gives a detailed account of studies of her own and others on later developing syntax in contrast to the more widely studied domain of early grammatical development in language-impaired populations to consider the relevance of competence- versus performance-based accounts of these findings, while her comparison of children and adolescents with language and learning disabilities to their normally developing peers in such areas as textual connectivity sheds important light on advanced level syntactic acquisitions in general. Jisa’s chapter provides a carefully-documented characterization of research comparing more formal written usage and colloquial everyday French, as background to a survey of her own discourse-based studies on narratives and expository texts produced in speech and in writing by French-speaking schoolchildren, adolescents, and adults, demonstrating by her findings in such domains as subordination, nonfinite syntactic packaging, and the expression of discourse stance that ‘moving into academic French’ both fosters and is fostered by acquisition of later developing grammatical constructions. The next two chapters concern writing as a notational system from two different perspectives. The chapter by Pacton and Fayol on children’s acquisition of spelling in French, a language with a deep orthography and considerable surface opacity, surveys a large body of research on learning to spell in French, as

Preface 

a psycholinguistic text case of the role and effect of implicit compared with explicit learning in this as in other domains of school-age language development. From a very different perspective, Wengelin and Strömqvist analyze findings from text-writing of Swedish schoolchildren, adolescents, and adults (in the framework of the same crosslinguistic project as the chapter on Spanish and part of Jisa’s findings for French earlier in the book), derived by applying a specially devised key-logging computer program to compare pauses and transitions between words compared with sentences as a function of developing text construction abilities in the written modality. The next two chapters are also concerned with extended discourse but, again, from two very different perspectives. Blum-Kulka deals with conversational discourse, starting with a comprehensive review of research on peer-talk as a critical basis for developing ‘discursive literacy’, and proceeding to analysis of peer-interactions recorded between young school age Hebrew-speaking children, with a focus on speech representation and the relation between direct quotation and indirect means of representing talk as differently developing speaker perspectives. The chapter by Peskin and Olson, the only one in the book concerned with language comprehension rather than production, considers the special challenges that poetic discourse constitutes for later language development through comparison of how high school and college students interpret poems, construed as reflecting the cognitive distinction between what is said and what is meant, on the one hand, and the experiential distinction between novice and expert knowledge, on the other. Liliana Tolchinsky’s concluding chapter considers the nature and scope of later language development in light of different explanations (cognitive, experiential, and discursive) that have been proposed to account for late-emerging facets of language knowledge and then reviews a range of such abilities – including inter-sentential and discursive links between linguistic expressions, understanding and use of non-literal language, command of dialogue, and the role of multiple perspectives – in relation to theory of mind and the impact of literacy as sources of explanation. The book thus touches on a deliberately wide spectrum of domains of linguistic knowledge (morphology, lexicon, semantics, and syntax); contexts of language use (word spelling, text writing, interactive conversation, and poetry reading); research orientations (background disciplines of the authors include linguistics, psycholinguistics, developmental psychology, social psychology, and literacy); age groups and developmental stages (from kindergarten via the early school grades, and from high school into adulthood); and languages (English, French, Hebrew, Spanish, and Swedish). My hope is that while

 Preface

the subject-matter of the book involves a rich plethora of issues, the story that emerges forms a coherent and cohesive piece of discourse and a solid basis for future investigation. Many people were involved in the making of this book. I would like first and foremost to thank the editors of the series, Steven Gillis and Annick de Houwer, who spared no effort in providing me with timely, sensible, and helpful input and advice throughout the editing process. Thanks, too, to Kees Vaes of Benjamins Publishing House for his professionalism and patience. It is a pleasure to acknowledge our gratitude to the many scholars from all over the world who gave of their time and expertise to reviewing one or more of the chapters, as follows: Vibeke G. Aukrust, Shmuel Bolozky, Peter Bryant, Eve V. Clark, Denis Creissels, John Du Bois, Harriet Jisa, Brian MacWhinney, Jon Miller, Bracha Nir-Sagiv, Arnaud Rey, Hava Bat-Zeev Shyldkrot, Finn Egil Tønnessen, Liliana Tolchinsky, Iris Yaron, Penelope G. Vinden, and Rita Watson I would also like to thank my colleagues and the assistants in my lab at Tel Aviv University for their support and feedback, particularly Dorit Ravid, as well as Laly Bar-Ilan, Yael Fürst and Bracha Nir-Sagiv for help with references and the index. Last, but of course not least, I am grateful to all the contributors to the volume for their professionalism and goodwill in putting up with my constant reminders about deadlines and demands for rewritings. Ruth A. Berman Tel Aviv, June 2004

Research on later language development International perspectives Marilyn A. Nippold

This book, Language Development Across Childhood and Adolescence: Psycholinguistic and Crosslinguistic Perspectives, edited by Ruth A. Berman, makes a unique contribution to the study of later language development. With chapters written by internationally known researchers working in Canada, England, France, Israel, Spain, Sweden, and the United States, converging evidence is presented to demonstrate that language development beyond the preschool years is a gradual and protracted process, extending well into adulthood. While previous research on this topic focused primarily on native English speakers (Nippold 1998), recent studies reported in this book deal with additional languages such as French, Hebrew, Spanish, and Swedish, thereby generating new insights into this important domain of human development. From this burgeoning body of research, information has emerged that speaks to three critical questions in the domain of later language development: (1) What is the nature and substance of language growth beyond the preschool years? (2) How can this growth be revealed? and (3) What factors drive these later linguistic attainments? This chapter addresses these questions, based largely on the contents of this book.

.

What is the nature and substance of language growth beyond the preschool years?

. The literate lexicon and figurative language A central component of later language development is the attainment of literacy – the spoken and written language skills necessary for successful commu-



Marilyn A. Nippold

nication in formal settings such as school and the work place. As Ravid (this volume) explains in her chapter, a key contributor to literacy is the acquisition of a “literate lexicon” – a mental dictionary of thousands of complex and low frequency words, coexisting in an elaborate semantic network. “A rich and powerful system,” the literate lexicon steadily expands during the school years, enabling cognitive, linguistic, and academic growth to continue unabated. During early childhood, children are learning the meanings of concrete words (e.g., car, truck, tricycle) they hear spoken at home or at daycare. In contrast, school-age children and adolescents acquire knowledge of abstract vocabulary (e.g., navigation, proportionality, renaissance) they hear or read in literate environments such as the classroom where they listen to teachers’ lectures or read textbooks on topics such as geography, mathematics, and history (Berman, this volume). Words encountered in literate contexts often express multiple or figurative meanings that preschool children fail to understand (e.g., absorb, grounds, dark, sweet, sharp) (Dockrell & Messer, this volume). This includes multi-word expressions such as idioms and metaphors that occur in political and scientific discourse (e.g., the road to peace, capillaries are tiny pipe lines) (Ravid, this volume). In addition to building a literate lexicon of advanced vocabulary, other important aspects of later lexical development include the ability to retrieve words with speed and accuracy during conversational, narrative, expository, or persuasive discourse (Dockrell & Messer, this volume; Ravid, this volume), and the ability to define complex vocabulary for school assignments (Nippold et al. 1999). Indeed, growth in all aspects of the literate lexicon continues well into adulthood, and there appear to be no obvious limitations on lexical learning through the lifespan (Nippold 1998). In examining other aspects of literacy, additional evidence for the gradual and protracted nature of later language development can be observed. For example, when young people are asked to explain their interpretations of poetry, complex verbal reasoning processes are revealed. As Peskin and Olson (this volume) discuss in their chapter, poetry interpretation challenges the adolescent or young adult to look beyond the simple meanings of words to recognize their symbolic and multidimensional meanings – in essence, to recognize how the writer intends to express more than what is actually stated. These authors explain, for example, that Marvell’s poem, On a Drop of Dew, requires an understanding of how a dew drop, resting on a rose petal, “wishes to evaporate and return to the skies,” just as the soul of a human being “longs to return to heaven” (p. 214). Although analogical reasoning and metaphor comprehension emerge early in life (Nippold 1998), an advanced understanding of figurative

Research on later language development

language, such as that required by this poem, is attained only in adulthood. Similarly, the use of low-frequency metacognitive verbs (e.g., assume, construe, realize) is a late linguistic attainment, evident when students are challenged to interpret poems (Peskin & Olson, this volume) or to write persuasive essays at school (Nippold et al. 2003). . Syntactic attainments Later language development is also characterized by important syntactic attainments. For example, the ability to use low-frequency structures such as the passive voice, common in scientific writing (e.g., “Natural disturbances and the biological legacies produced by them are often poorly understood by policy-makers and natural-resource managers,” Lindenmayer et al. 2004: 1303), gradually increases in written language production throughout the school-age and adolescent years and into adulthood (as demonstrated in this volume by the chapters of Berman, of Jisa, of Ravid and of Scott). Other syntactic structures that evidence a gradual increase in use during these years include subordinate clauses such as center-embedded and object relative clauses (e.g., The cat that the dog chased pounced on the mouse), past perfect marking (e.g., The cat that had climbed the tree tried to catch a bird), modal auxiliaries (e.g., could, should, might), and low-frequency adverbial conjunctions (e.g., meanwhile, moreover, consequently), each of which results in the production of longer sentences in spoken and written discourse (Loban 1976; Nippold 1998, 2003; Scott, this volume). As Scott (this volume) emphasizes in her chapter, growth in the use of subordination (that is, inter-clausal density) makes an important contribution to later syntactic development along with the ability to deploy these advanced structures in a flexible manner (also see Ravid, this volume). Similarly, Khorounjaia & Tolchinsky’s (this volume) study of narrative and expository discourse in Spanish-speaking children, adolescents, and adults, reports a gradual increase in the production of noun phrase elaboration (e.g., “She gave her sister a sparkling silver bracelet for her birthday”), resulting in increasingly informative discourse during these years. Flexibility in knowing when to use formal versus informal aspects of language is an additional hallmark of later syntactic development. As Jisa (this volume) reports in her chapter, by age 10 years, French-speaking children are able to employ compact syntactic packaging such as pronoun ellipsis with nonfinite subordination (e.g., “Francois and Flipper took a little one and left while saying good-bye”) during formal written communication, structures that rarely occur informally. Similarly, Khorounjaia & Tolchinsky (this volume) report





Marilyn A. Nippold

that by age 9 years, Spanish-speaking children have learned to employ a more distant, impersonal style of speaking during expository compared to narrative discourse, and Scott (this volume) reports that English-speaking children learn to make use of compact, densely formulated sentences when producing formal written language, with the knowledge that this level of formality is not expected in casual conversation. . Spelling A relatively new topic of investigation is the development of spelling, a critical aspect of written language proficiency, and one that clearly taps a child’s metalinguistic awareness or “metamorphology”. Pacton & Fayol (this volume), in examining French-speaking children, report that spelling ability can be enhanced by knowledge and awareness of morphology. For example, a child who recognizes that fillette ‘little girl’ and vachette ‘little cow’ share the diminutive morpheme -ette ‘little’ will gain insight into the correct spelling of these two words. The importance of spelling to the production of fluent discourse is discussed by Wengelin & Strömqvist (this volume). Studying Swedish children, they report that once a young writer has learned to spell proficiently, by about age 13 years, the child is no longer hampered by the mechanical aspects of writing, and is therefore able to allocate greater cognitive and linguistic resources to the fluent expression of ideas.

. How can growth in later language development be revealed? Evident from the above discussion, growth in later language development reveals itself when young people are presented with challenging tasks of spoken and written communication. Tasks that involve the use of expository, narrative, or persuasive discourse are especially good candidates for tapping these abilities (Nippold 2003; Nippold et al. 2003; Ravid, this volume). With these sorts of tasks, the speaker or writer is called upon to engage in complex thought, a process that requires substantial cognitive and linguistic resources. As a result, the individual’s command of abstract vocabulary (e.g., metacognitive verbs) and advanced syntactic structures (e.g., non-finite subordination, center-embedded relative clauses) can become evident. For example, compared to conversation, expository discourse stimulates the use of longer utterances containing greater clausal density through adverbial, relative, and nominal clauses (Nippold 2003). Other tasks that can uncover advanced linguistic skills

Research on later language development

include having students interpret unfamiliar poems (Peskin & Olson, this volume) or write essays on controversial topics (Nippold et al. 2003). Because these tasks stimulate complex thought, complex language is called into play. Thus, to reveal growth in later language development, cognitively challenging tasks must be presented (Jisa, this volume; Scott, this volume).

. What factors drive these later linguistic attainments? . Education, literacy, and metalinguistic competence With its emphasis on literacy, schooling plays a critical role in later language development. As discussed above, education provides the school-age child or adolescent with the necessary exposure to complex and low frequency words, multilexemic expressions, and syntactic structures. This occurs when students read or listen to complex discourse. Circumstances that motivate students to actively analyze what they hear and read – to engage in metalinguistic behavior – also play a key role in the developmental process, along with the opportunity to use the new words and structures for their own communicative purposes. Dockrell and Messer (this volume), in discussing native-English speakers, report that young school-age children readily acquire new vocabulary during interactive book reading sessions where teachers provide explanations of unfamiliar words that occur in stories. Then, once children are able to read proficiently (by about age 10), they begin to acquire new words independently through the use of metalinguistic strategies. These include, for example, inferring the meanings of unfamiliar words from written context clues and performing a morphological analysis of the roots, suffixes, and prefixes of words. The ability to employ these sorts of metalinguistic strategies is itself a gradual developmental process, and one that is especially helpful in learning the meanings of words that occur in books on science (e.g., biology, geology, mineralogy, climatology) and mathematics (e.g., geometry, geometric, quadrilateral, quadratic, symmetry, asymmetry). In turn, this greater depth of word knowledge promotes the child’s reading comprehension, and provides further exposure to complex vocabulary. Similarly, Berman (this volume) and Ravid (this volume) report that Hebrew-speaking children in academic settings frequently are exposed to words that are morphologically complex, including denominal adjectives like personal, mountainous, industrial and derived nominals like delivery, testimony,





Marilyn A. Nippold

description. These types of words are characteristic of formal written language such as expository discourse but occur infrequently in casual spoken communication. Again, once students become proficient readers and are able to read a wide variety of books in academic subjects such as science, history, and mathematics, they gain the necessary exposure to increasingly difficult words (e.g., monozygotic, electromagnetic, gerrymandering, inalienable, equidistant, postulate) that often are embedded in long and complex sentences (e.g., “For the first time during a climatic transition of this nature and magnitude, the hominids dealing with global change were anatomically modern people who were living in Siberia, the Americas, and on all continents save Antarctica, sometimes at fairly high densities” (Straus 2004: 1300). In addition to vocabulary, complex syntax is acquired through exposure in literate contexts and subsequent metalinguistic analysis. Berman (this volume) reports that Hebrew-speaking children are exposed to academic discourse containing low-frequency structures such as the passive voice (e.g., “The forest was destroyed by the fire”) and nonfinite nominalizations (e.g., “Building a new library will be challenging”), both of which rarely occur during informal spoken communication. Similarly, Scott (this volume), citing Huttenlocher (2002), reports that English-speaking children whose teachers frequently employ multiclausal utterances acquire greater syntactic proficiency than children whose teachers speak more simply. Exposure to complex language in school provides the opportunity to map phonological, morphological, and lexical regularities of linguistic structure and content onto their orthographic conventions, a process that also promotes the development of spelling (Pacton & Fayol, this volume). Beyond exposure, the learner’s increased ability to actively analyze the structural aspects of language – including phonology, syntax, and morphology – comes to play a key role in the learning process. Thus, there is an ongoing, cyclical relationship between literacy and later language development, a process that is heavily supported by metalinguistic competence. Nevertheless, as mentioned above, exposure to and analysis of complex language is not sufficient for learning to occur. It is critical also that students have opportunities to actually use what they are learning. For example, by engaging in academic tasks such as formal speaking and writing assignments, students are called upon to employ the sophisticated lexical and syntactic elements they have been learning for their own communicative purposes. Writing, in particular, offers important advantages over speaking because it allows the learner more time to search the lexicon for the precise vocabulary and to organize the discourse more concisely. In contrast, speaking, because of its associated time pressures, often precludes this opportunity for thoughtful re-

Research on later language development

flection. Additionally, when writing, the learner is able to devote more thought to the use of formal aspects of language such as syntactic packaging via nonfinite verb phrases and passive voice (e.g., “People must be punished in order to stop the violence”, “The wildfire was ignited by lightning”) (Ravid, this volume). Similar points are made by Wengelin & Strömqvist (this volume), who argue that once the child has mastered the basic writing skills (e.g., spelling, punctuation, and keyboarding in the case of computers), writing becomes an opportunity for “planning and conscious reflection” (p. 189) and frees the child from online pressures that exist when speaking. Further, as Jisa (this volume) describes, the writing process supports the development of more sophisticated spoken language skills. Through opportunities to use the new words and syntactic structures in writing, students gain confidence in using them in formal speaking activities as well, such as oral presentations. Gradually, the child acquires advanced levels of spoken and written language competence (Scott, this volume). . Cognition In addition to education, later language development depends upon the child’s cognitive readiness to advance to higher levels of thought. Importantly, during late childhood (around age 11–12 years), Piaget’s stage of formal operational reasoning emerges. A hallmark of this developmental period is the ability to employ hypothetical-deductive reasoning and to consider multiple solutions to problems (Santrock 1996). These cognitive attainments often are reflected in the language used for school-related purposes. This can be seen, for example, in the appearance of metacognitive verbs during poetry interpretation (Peskin & Olson, this volume) and in persuasive essays for school assignments where young people increasingly are able to consider both sides of an argument (Nippold et al. 2003), and see, further, Tolchinsky (this volume). . Socialization A third factor promoting the development of language in older children and adolescents is the opportunity for socialization. As Blum-Kulka (this volume) describes in her chapter, opportunities to engage in peer talk through spontaneous role-playing activities and extended conversations can promote the development of linguistic and pragmatic competence. Citing examples from her research with Israeli children, she explains that when two 9-year-old girls played a game of “Saleslady and Customer,” they practiced using advanced lex-





Marilyn A. Nippold

ical and morphological forms characteristic of formal spoken Hebrew (e.g., “I’ll be with you shortly,” “This dress is very festive/gorgeous/amazing”), forms that typically would not occur in the girls’ conversational speech. Similarly, when two 9-year-old boys discussed a problem they were having with a mutual friend, they practiced using turn-taking skills, cohesive markers, logical sequencing, and collaborative comments. According to Blum-Kulka, these attainments were promoted, in part, by the shared culture of childhood – the knowledge that each child had of the situation, each other, and of other people. Other studies of peer talk and socialization indicate similar benefits to later language development (e.g., Dorval & Eckerman 1984).

Conclusions The chapters in this book provide intriguing new information and insights into later language development. With contributions from international researchers, it represents an inspiring trend in the field of psycholinguistics, indicating that later language development is being investigated with intensity around the world. It is fascinating to consider the similarities in this process across languages and the manner in which this expanding body of knowledge contributes to an understanding of how the human capacity for language development remains active well beyond the preschool years. It is not the purpose of this book to explain how this body of knowledge can be applied for practical purposes. Nonetheless, it is important to note that it offers substantial implications for professionals working in a variety of disciplines, including linguistics, psychology, education, language pedagogy, and speech-language pathology. As more is learned about the nature of later language development, its relationship to literacy, and the factors underlying its growth, practitioners will gain insight into the difficulties experienced by school-age children and adolescents who demonstrate language-learning weaknesses, as some of the authors in this volume indicate (e.g., Dockrell & Messer; Scott). Importantly, this information will enable professionals to enhance the learning process in students who struggle to become literate language users. For these reasons, this book will surely speak to readers around the world that have very different interests.

Between emergence and mastery The long developmental route of language acquisition* Ruth A. Berman

.

Introduction

The chapter considers an apparent paradox in children’s acquisition of their native language: the incompatibility between early emergence versus late mastery of linguistic knowledge. The term ‘acquisition’ in the title is itself anomalous, since concern here is crucially with language development, in the sense of the path taken by children from initial emergence to mature mastery of linguistic knowledge. The study reflects a general thrust of my research in the domain, and its claims are supported by findings from other chapters in this volume: the need to take into account linguistic, cognitive, and social forces as shaping the path from becoming a ‘native speaker’ to being a ‘proficient speaker’ (and/or writer) of a given language. From this perspective, language proficiency involves a complex configuration of interrelated types of knowledge: (1) linguistic command of the full range of expressive options, both grammatical and lexical, available in the target language; (2) the cognitive ability to integrate forms from different systems of the grammar, and to deploy these options to meet different communicative goals and discourse functions; and (3) cultural recognition of what constitute the favored options of a given speech community, adapted to varied communicative contexts and to different norms of usage.1 As a starting point, I take a key finding of work with Dan Slobin on narrative development (Berman & Slobin 1994). Children of different ages and speaking different languages were asked to tell a story based on a wordless picture-book describing the adventures of a boy and his dog in search of a runaway frog. Even the youngest subjects, aged 3 to 4 years, produced texts that were syntactically constructed in accordance with the grammatical struc-



Ruth A. Berman

tures of their first language. This confirms the Chomskian view of language acquisition as a short-lived and efficient process, which “envisions a rapid and effortless transition from the ‘initial state’ to the ‘final state’ (Crain & McKee 1985: 94), and it supports the claim that “There is growing consensus that by the age of three, children have acquired the basic phonological, morphosyntactic, and semantic regularities of the target language irrespective of the language or languages to be learned” (Weissenborn & Höhle 2000: vii). On the other hand, our study showed that the oldest children in our sample, aged 9 to 10 years, produced narratives that differed markedly from those of the adults, not only in content but also in morpho-syntax and lexicon. From this we concluded that becoming a native speaker is a rapid and highly efficient process, but becoming a proficient speaker takes a long time. This chapter aims to provide further evidence that linguistic forms, many of which emerge at early preschool age, in fact have ‘a long developmental history’. Our earlier findings are supported by recent cross-linguistic research on the text production abilities of schoolchildren and adolescents compared with adults, as summarized in Berman & Verhoeven (2002a, b), and detailed by Berman (2004; in press b); Berman & Nir-Sagiv (2004); Ravid (in press); and the chapter by Jisa, this volume. This work supports findings of Englishlanguage researchers such as Nippold (1998) and Scott & Windsor (2000), which underscores the protracted nature of language development, into and even beyond adolescence. These and related studies also shed light on the cognitive and socio-cultural underpinnings of developing language use in constructing expository compared with narrative texts, and in writing compared with speech (Berman & Katzenberger 2004; Berman et al. 2002; Ravid & Berman 2003; in press; Reilly et al. 2002; Strömqvist et al. 2004; and by Wengelin & Strömqvist, this volume). In accounting for order of acquisition, researchers have tended to focus on linguistic complexity, of form – in morpho-syntactic structures and processes – or content, in semantic representations and lexical specificity. The approach taken here is that neither structural complexity nor transparency of form-meaning mappings suffice to account for what linguistic forms children acquire, and master, when. Rather, factors of cognitive demands, expressive options, and target language typology also play a role in the path from emergence to mastery of language. The cognitive demands involved in language acquisition have been freshly illuminated by recent cross-linguistic and cross-cultural research on the topic of ‘language and thought’ (Bowerman & Levinson 2001; Gumperz & Levinson 1996). These studies highlight the complexity of the task faced by children in

Between emergence and mastery

mastering language use along with linguistic structure. One reason is that language learners, like everybody else, face the problem of choice – both of what to say and how to say it. To this end, they need to learn how to tap into the full repertoire of expressive options available to them in ways that are appropriate to a particular context of discourse. This is not simply a question of the conceptual complexity of relevant notions, such as causality, attribution, or temporal sequence. The mental abilities needed to talk about such concepts emerge quite early in development. This is also true for other types of cognitive representations relevant to extended discourse, like knowledge of scripts, the ability to recall past events in constructing narratives, and to distinguish between different discourse genres such as descriptions versus stories. For example, in their initial attempts to express the discourse function of making reference to the characters in a story, young children can rely on their conceptual ability to distinguish between people and animals, between children and adults, between men and women. But these abilities still do not mean that preschoolers’ cognitive representations and processing abilities are adequate to the task of producing communicatively appropriate and well-organized pieces of discourse on a range of topics. On the other side of the ‘form-function’ coin, it is not enough to in some sense ‘know’ the relevant linguistic forms. Thus, 4-year-olds are familiar with many linguistic devices for making reference, and 5- and 6-year-olds know even more: naming and labeling by proper or common nouns; describing by adjectives, prepositional phrases, or relative clauses; and using pronouns, null subjects, and determiners. However, young children are not yet capable of coping with the cognitive load involved in the simultaneous execution of different tasks. In the case of narrative reference, they need to organize a text globally around the main protagonist(s), taking into account secondary protagonists, while at the same time recruiting the appropriate linguistic devices for encoding this by unambiguous means of expression (Hickmann 2003). In an earlier study (Berman & Katzenberger 1998), we interpreted preschoolers’ inability to cope with the cognitive demands involved in execution of these different tasks concurrently in light of the ideas of Shatz (1984) on storytelling and of Pascual-Leone (1987) in other cognitive domains – as deriving from the difficulty imposed by cognitive overload on young children’s online processing of linguistic output. Another factor affecting what children will learn relatively late is target language variability. The source of such variation may be dialectal, as in the standard language of school compared with the local dialect of the home; historical, as in the different strata of English vocabulary; and/or contextual, as in regis-





Ruth A. Berman

ter differences between everyday colloquial usage, standard intermediate-level usage of the media or of academic discourse, and the normative requirements of the official language establishment (Ravid 1995). Ready and flexible access to diverse linguistic registers, varied levels of usage, and different types of texts are prerequisites for ‘linguistic literacy’ (Ravid & Tolchinsky 2002), and require both protracted cognitive maturation and extensive experience with different communicative settings. These demands, as noted by several of the chapters in this volume, are met, if at all, only in adolescence or beyond. In what follows I try to show how structural and referential complexity interact with factors of target-language typology, register variation, and speaker preferences for particular expressive options in specifying development from emergence via acquisition to mastery. As background, the chapter outlines a phrase-based model of language development (Section 2), illustrated by research on three domains of Hebrew lexicon and grammar (Section 3), summed up by predictions based on these ideas and research findings (Section 4).

. A phase-based model of language development The discrepancy between early emergence and late mastery can best be explained by considering linguistic form in relation to language use across development. Within such a form-function approach, ‘linguistic form’ is a coverterm for the entire gamut of devices available in a language – grammatical inflections, derivational morphology, lexical expressions, syntactic constructions, and syntactic processes (Slobin 2001); and ‘discourse functions’ involve textual properties such as making reference to persons, objects, and ideas; expressing spatial and temporal relations; and interconnecting the parts of a discourse into thematically coherent units (Hickmann 2003). The basic idea underlying this approach is that, as a result of increased linguistic knowledge and greater experience with language use, speakers (and writers) come to deploy a wider, more flexible variety of linguistic forms to express a particular discourse function (Berman 1996, 1997a; Berman & Slobin 1994; Slobin 1996, 1997). A second facet of this conception is that language acquisition is not instantaneous but essentially developmental in nature. Moreover, the process follows a developmental route which is peculiar to linguistic development in some respects but which also shares key properties with development in other cognitive and social domains (as argued by Berman et al. 2002; Karmiloff-Smith 1986a, 1992).

Between emergence and mastery

In order to integrate a ‘form-function’ approach with the ‘developmental paradox’ noted at the outset, I propose a phase-based model of language acquisition and development, as summarized in (1). (1) Phases in language development: I Pre-grammatical: item-based, situation-bound II Grammaticized: structure-dependent, rule-bound III Conventionalized: context-oriented, discourse-motivated

This model was first articulated in relation to Hebrew inflectional and derivational morphology (Berman 1986a). Other domains to which it has since been applied are: morphological marking of transitivity and voice (Berman 1993b); syntactic constructions such as complex noun phrases, word classes, null subjects, and nominalizations (Berman 1987a, 1988a, 1990, 1993c); and narrative text construction (Berman 1988b, 1993a, 1995a). The focus of these studies, as well as of the analyses in the next section, is on the nature of these changes rather than on the ‘mechanisms of language acquisition’ (MacWhinney 1987). However, both the claims made in the preceding section and the evidence marshaled in Section 3 uphold the view that no single mechanism can account for the complex question of ‘what drives change in children’s grammar’ (Bowerman 1987). Several general properties characterize the changes from emergence to mastery, from juvenile to fully appropriate use of language. First, following Karmiloff-Smith (1986a), these changes occur – in fact recur – across developmental phases rather than in invariant, cross-domain Piagetian stages. For example, children may have acquired the grammar of number and gender agreement, yet still be at an ‘item-based’ phase of person marking (Tomasello 2001b); and children may have structure-dependent knowledge of inflections without rule-based command of form-meaning relations in derivational morphology (see Ravid, this volume). Second, this model takes into account ‘partial knowledge’ and reorganizations en route from ‘entry to exit’, from the initial to so-called end-state. In fact, it is intrinsically ‘open-ended’, to allow for individual differences, linguistic variation, and language change. Third, command of linguistic knowledge develops and is successively reintegrated with increased ability in the domain of language use. That is, there is a constant two-way interaction, rather than a dichotomous distinction, between competence and performance in this as in other domains (Berman 1995a). With age and increased cognitive and social maturation, the linguistic behavior of speaker-writers comes to have an increasing effect on their internal linguistic representations; concomitantly, growing knowledge in different linguistic





Ruth A. Berman

domains has an increasing impact on language use. Thus, later language development involves a more expert and flexible interplay between augmented linguistic knowledge, on the one hand, and greater experience with language use in varied communicative contexts, on the other. This general framework is elaborated in (2), which breaks down the three phases of (1) into a sequence of five steps. (2) Five developmental steps: a. Rote-knowledge – initial acquisition of individual items as unanalyzed amalgams, closely tied to the immediate situational context, with no generalization beyond form-meaning mappings of particular linguistic forms (words or set phrases). b. Initial alternations – a few highly familiar items are modified contrastively within or across paradigms and constructions, with no abstraction beyond certain limited groups of forms. c. Interim schemata – transitional, non-normative and idiosyncractic, although partly productive application of rules and tentative generalizations. d. Rule-knowledge – grammaticization, with strict adherence to rules plus some lacunae, including inadequate command of structural and lexical constraints expressed as over-regularizations and ‘creative errors’. e. Proficient integration of knowledge and use – in mature language usage, abstract rules are constrained by norms of usage, rhetorical conventions, and discourse appropriateness, yielding variation in style and register in accordance with particular communicative context (e.g., genre and modality) as well as individual predispositions, backgrounds, and levels of literacy.

Steps (a) and (b) are characteristic of ‘emergent’ phases of language knowledge and use; they are ‘pre-grammatical’ and ‘item-based’ in a highly specific sense: as non-rule-generated and not generalized to abstract categories of linguistic or discourse structure; they rely critically on the mechanisms of rote learning and imitation; yet at the same time they have recourse to universal semantic and conceptual distinctions that are shared across languages and hence do not need to be learned (see, further, Section 4 below). Steps (c) and (d) relate to what is generally termed ‘acquisition’, since they involve grammaticality, in the form of productive use of structure-dependent rules of morpho-syntax and the lexicon, along with internalization of relevant schemas of discourse structure. And step (e) represents proficient ‘mastery’ of both linguistic structure and language use.

Between emergence and mastery

The progression delineated in (1) and (2) is in a way ‘U-shaped’ (Strauss 1982), since the initial and final phases are both highly context sensitive, but in two different senses. Initially, children’s language knowledge and use is intimately tied to and dependent on the extra-linguistic contexts of language input and output. Maturely proficient language use is also peculiarly fine-tuned to context, but this is shaped by the interaction between the cognitive and cultural constraints of a given communicative situation (e.g., shared knowledge, social status of interlocutors), on the one hand, and discourse-internal constraints on the organization of linguistic information (e.g., global text construction, online processing of speech production), on the other. A related facet of this U-shaped progression is that there will be more individual variation and less stereotyped language use both in the initial and the more mature phases, but for different reasons. Early learning is tied to specific situations and individual items – words, constructions, conversational turns. With maturity, personalized styles of expression take shape in line with the particular socio-cultural background and life experience of individual speaker-writers. This view of the process is consistent with the idea of multiple mechanisms for bootstrapping children’s entry into initial acquisition (Shatz 1987) or consideration of perceptual, logical, and syntactic bootstrapping in accounting for how and why “grammar and the lexicon develop together in infancy” (Bates & Goodman 2001: 158). That is, moving from unanalyzed rote usage to rulebound command of form-meaning relations in early acquisition is driven by a ‘confluence of cues’ – perceptual, semantic, structural, and pragmatic (Berman 1993b, 1994). This can help explain the early and rapid emergence of abstract linguistic knowledge in the preschool years. But the idea of multiple bootstrapping needs to be extended to account for the manifold factors involved in later language development of the kind that occurs before and across adolescence (Hickmann 2003; Nippold 1998).

. Examples from acquisition of Hebrew Below, these ideas are examined in light of findings for three aspects of Hebrew grammar: types of adjectives (3.1), passive constructions (3.2), and nominalizations (3.3). These were selected to represent different linguistic domains: lexicon, clause-construction, and complex syntax respectively. And each focuses on distinct discourse functions: semantic modification in the case of adjectives, agent downgrading by passivization, and clause-linkage and connectivity by means of non-finite nominalization. Moreover, Hebrew typology





Ruth A. Berman

is such that each of these domains also interacts markedly with the domain of derivational morphology (see Ravid, this volume). The data discussed in this section derive from a range of sources: longitudinal and cross-sectional naturalistic language samples of young children in conversational interaction with adults; structured elicitations of morphological and syntactic constructions of preschool and schoolage children in comprehension and production; oral picture-based narratives elicited from children aged 3 to 9 years; and personal-experience narratives and expository texts produced in speech and writing by schoolchildren and adolescents.2 Findings and analyses are from Hebrew, but they are meant to illustrate quite general cross-language processes. . Classes of adjectives This section deals with vocabulary acquisition by comparing three classes of Hebrew adjectives that differ in the morphology-semantics interface that they represent – in contrast to the corresponding classes in English, for which relevant data are available. These three classes are illustrated in (3), listed in order of their acquisition, as ‘basic’, in (3a), ‘resultative’ in (3b), and ‘denominal’, in (3c). (3) Three classes of adjectives in Hebrew: a. Basic: gadol ‘big’ katan ‘small’ lavan ‘white’ shaxor ‘black’ tov, yafe ‘good, nice’ ra, garúa ‘bad, terrible’ b. Resultative ‘Endstate’ Passive Participles: Infinitives Present Past (Resultative) Participle li-sgor soger sagar CaCuC sagur3 ‘to close’ ‘(is) closed, shut’ li-xtov kotev katav katuv ‘to write’ ‘written’ le-taken metaken tiken meCuCaC metukan ‘to fix’ ‘fixed, in order’ le-saper mesaper siper mesupar ‘to-cut’ ‘cut, shorn’ le-hašxil mašxil hišxil muCCaC mušxal ‘to thread’ ‘threaden’ le-havin mevin hevin muvan ‘to understand’ ‘understood’

Between emergence and mastery

c.

Denominated Adjectives: klal / klali ish / ishi ‘rule’ ‘regular’ ‘person’ ‘personal’ tsava / tsva’i sifrut / sifruti ‘army’ ‘miltary’ ‘literature’ ’literary’ midbar / midbari har / harari ‘desert’ ‘desertlike’ ‘mountain’ ‘mountainous’

‘Basic adjectives’ like those in (3a) have the mono- or bisyllabic form CVC(VC) and they are morphologically underived, hence ‘basic’. They also generally have morphologically related verbs – transitive causatives and intransitive inchoatives – analogous to English enlarge, whiten, or prettify, which share the same consonantal root as the adjectives, e.g., gadol ‘big’ / li-gdol ‘to grow (larger)’ / le-hagdil ‘to make larger, enlarge; yafe ‘pretty’ / le-yapot ‘to make pretty’ / lehityapot ‘to become pretty’.4 (See Ravid, this volume). ‘Resultative adjectives’ are passive participles, corresponding to Germanic or Romance past participles with an end-state or completive sense, like English written, spoken, torn. In Hebrew, these are marked by an internal vowel -u, in one of three forms; each such form relates to one of the three transitive binyan verb-patterns, labeled here as ‘Pn’. Thus, in (3b), the P1-pattern verb li-xtov ‘to write’ yields katuv ‘written’, the P3 pi’el verb le-taken ‘to fix’ yields metukan ‘fixed, unbroken’, and the P5 hif’il form le-hašxil ‘to thread, string’ yields mušxal ‘threaded, strung’. Denominal adjectives as in (3c) are formed by a (bound or free) noun stem plus the adjectivizing suffix, stressed -i, e.g., klal ‘rule’ > klal-i ‘regular, general’, sifrut ‘literature’ > sifrut-i ‘literary’, or ir‘town, city’ > iron-i ‘urban, municipal’. The order of the examples in (3) reflects the general sequence in which children add adjectives to their productive lexicon: they start with basic adjectives, then add end-state resultatives, and only subsequently denominals. The question is what determines this difference in the acquisitional schedule. The first group – ‘basic’ adjectives as in (3a) – are acquired much like their counterparts in English, from around age two years (Ravid & Nir 2000). That is, they are learned like other words in children’s early vocabulary, as ‘pregrammatical’, ‘item-based’, morphologically unanalyzed amalgams unrelated to their associated causative or inchoative verbs – hence representing Phase I of the model outlined in Section 2. Moreover, the form-meaning mappings entailed by such lexical items develop, like in other languages, by relative semantic complexity in relation to other items in the same lexical category, for example, dimensional terms like big (acquired before small as its negative polarity converse and before tall or wide as semantically more specific terms); color terms





Ruth A. Berman

(e.g., red before orange or crimson); or evaluative attributives (e.g., good before excellent, bad before awful). As for the second class of adjectivals – u-marked resultatives as in (3b) – Hebrew-speaking children show productive use of passive participles forms to express resultative endstates, corresponding to English broken, torn, lost, from age 3 to 4 years. That is, children ‘acquire’ these forms at the same time as numerous other word-formation processes of derivational morphology (Berman 1995b; 1999; 2003a). Prior to this, children’s knowledge of these forms is similar to their ‘pre-grammatical’, non-analytic use of other, basic adjectives, e.g., sagur ‘shut, closed’, meluxlax ‘dirty’ (literally ‘dirtied’, cf. basic naki ‘clean’), mesudar ‘neat, orderly’ (literally ‘tidied’, cf. basic yafe ‘nice, pretty’). Structuredependent acquisition of these forms at around age 3 to 4 is attested by both structured elicitations and innovative use of u-marked forms in children’s spontaneous speech. In a structured elicitation task (Berman 1994), children were shown pictures of an initial state and given an appropriate description (e.g., ‘Here is a man with a beard’), a means for incurring a change-of-state using the source verb (e.g., ‘and here is a razor to shave [source form = le-galeax] his beard’), and then asked to describe the resultant endstate (e.g., ‘and here the man is . . ., his face is . . . [target form: megulax ‘shaven’]). Children used u-marked, potentially passive participial forms of verbs to describe these end-states only rarely at age 2, nearly half the time by age 3 to 4 years, and close to 80% of the time from age 5 years. But the forms they produced, while ‘grammatical’ since they accorded with the structure of possible passive participles in the language, were often not those in the conventional lexicon (less than a quarter of the u- marked forms produced by the 3-year-olds, rising to 40% by age 4 years and 60% among 5-year-olds). Support for such ‘rule-based’ but non-conventionalized knowledge of how to mark resultative end-state by passive participles is provided by children’s ‘creative errors’ in their naturalistic speech output from as young as age 2 to 3 years (Berman 2000). The examples in (4) are of possible, but non-established forms of resultative participles – CaCuC, meCuCaC, and muCCaC respectively. They are thus examples of step (c), signaling entry into the Phase II grammaticized command of the system. (4) Spontaneous Coinages of u-Based Passive Participles: a. ha-mecax sheli kasuy be-se’ar [Keren, 2;10] CaCuC: cf. mexuse ‘covered’ ‘My forehead (is) covered with hair’

Between emergence and mastery

b. ha’or kan (ba-sandal sheli) kamut [Nir, 3;7] CaCuC: cf. mekumat crumpled’ ‘The leather here (=on my sandal) (is) crumplened’ c. ima, ha’orez kvar bashul [Erez, 4;8] CaCuC: cf. mevushal ‘cooked’ ‘Mom, the rice (is) already cookered’ d. tir’i, ima, ha-sefer merutav [Smadar, 2;1] meCuCaC: cf. ratuv‘wet’ ‘Look, Mom, the book (is) wettied’ e. axšav ze mešutaf (al matbea še šatfa) [Hagar, 2;10] meCuCaC: cf. šatuf ‘rinsed’ ‘Now it (is) rinsened’ (about a coin she had put in water) f. ani me’od me’ulevet mimex [Yael, 4;6] meCuCaC: cf. ne’elav(ti)‘got-insultered’ ‘I (am) very insultered from = by you’ g. ani kvar muxlecet [Rona, 2;9] muCCaC: cf. xalcu li ‘removed (shoes)’ ‘I (am) already unshoed’ h. ha-raglayim sheli yihyu mukfot mikor [Shay, 4;10] muCCaC: cf. kfu’ot ‘frozen+Fem,PL’ ‘My feet will be freezen from-cold’

These examples from young preschoolers reflect both general conceptual and language-particular knowledge of form-meaning mappings: (1) the idea of a resultant endstate deriving from a change-in-state activity; (2) the fact that this is typically encoded in Hebrew by means of one of three possible morphological patterns – CaCuc as in (4a) to (4c), meCuCaC (4d–4f), or muCCaC (4g, 4h); and (3) how to alternate these forms correctly by inflectional marking of agreement for gender and number (e.g., by the feminine singular ending -et or the feminine plural ending -ot in (4g, 4h). What these children still need to acquire is the morpho-lexical convention that stipulates which of the three passive participial forms matches the particular active verb pattern associated with it in the established lexicon, knowledge which typically consolidates with increased vocabulary in the early school years. The third class of adjectives – denominal adjectives like those illustrated in (3c) and examined in detail by Ravid (this volume) – correspond to words like English industrial, military, mountainous. In Hebrew, these are formed by addition of a single invariant suffix, the stressed syllable -i, a highly productive device for forming adjectives.5 These are derived not from verbs, as in the case of passive participles, but from nouns, and they are a much later, typi-





Ruth A. Berman

cally school-age acquisition. A large corpus of unconventional lexical usages of preschool children (Berman 2000) contained very few such forms, mostly from older children. For example, Ran, aged 5;4, describes his father’s metal gun as barzel-i ‘iron-y’, and Tal, aged 6;1, refers to a glass bottle as zxuxit-i ‘glass-y’ = ‘made of glass, transparent’. And in the structured elicitation tasks described by Ravid (this volume), 5- to 6-year-old preschoolers produced such forms only two-thirds of the time, and even first-graders made structural errors in around two-thirds of their responses, while denominal adjective formation was found to be difficult for even older normally-developing gradeschoolers and quite beyond the abilities of their SLI peers. This was supported by findings for extended discourse; in informative texts produced orally and in writing by 6th graders, 11th graders, and adults, denominal adjectives were rarely used before high school age (Ravid & Zilberbuch 2003a). Both spontaneous usage and structured elicitations thus show that denominal -iy suffixed adjectives are a very late acquisition, they emerge in Hebrew child language mainly with the onset of literacy, around age 6, and they are mastered as a productive lexical option even later, in marked contrast to the u- marked participial adjectives which, as noted, emerge much earlier. This contrast between the early and robust mastery of resultative adjective formation with the later, only occasional, use of denominal adjectives is not immediately obvious in terms of structural complexity or the principle of ‘formal simplicity’ (Clark 1993). Morphologically, resultative participles are marked by a stem-internal vowel (-u) compared with the linear formation by a stressed suffixal vowel (-iy) added to a stem in denominated adjectives. Then, too, the form of a resultative participle depends on which active conjugation pattern it is derived from, in contrast to the invariant adjective forming suffix -iy. Besides, in terms of derivational processes, the surface form of resultative participles diverges from the verbs or nouns that they derive from far more than denominal adjectives differ from their source nouns. This contrast between the internal u-marked resultative adjectives and the superficially more transparent denominals is shown in (5a) and (5b) for items from the established lexicon, while the unconventional forms in (5c) through (5e) include examples of -u marked resultatives that were coined by preschoolers (iced from Hila, aged 2;9, flowered from Sivan, aged 4;0, and lemoned from Vered, 5;2).6 (In the examples (5), the stem-internal resultative marker -u- and the stem-external denominal suffix -i are bolded).

Between emergence and mastery

(5) Verb-based (resultative) vs Noun-based (denominal) adjectives: a. le-ta ’es ‘to-industrialize’ >metu’as ‘industralized’ vs. ta’asiya ‘industry’ >ta’asiyat-i ‘industrial’ b.

le-cayer ‘to-draw’ vs. ciyur ‘picture’

> mecuyar > ciyuri-

‘drawn, painted’ ‘picturesque’

c.

kérax vs. kérax

‘ice’ ‘ice’

> krux-ot > kirxi-

‘iced, icy, Fem-Pl’ ‘ice-like’

d.

pérax vs. pérax

‘flower’ ‘flower’

> mefurax > pirxoni-

‘flowered’ ‘florial’

e.

limon vs. limon

‘lemon’ ‘lemon’

> meluman ‘lemoned, with lemon’ > limoni- ‘lemony’

Internal u-form resultative participles are clearly structurally complex compared with denominal -i ending adjectives. Besides, it is not obvious that in semantic terms attributing resultant end-states to an object is more accessible to children than attributing associated properties to nouns. The notion of ‘resultant end-state’ implies mastery of a conceptually complex causal chain leading from an object in an initial state (say, whole, complete) to an activity applied to that object (e.g., someone breaks, cuts, or tears it) to a change-of-state which results in a different end-state (not whole, incomplete, in pieces, in parts or broken, cut, torn). As evidence, note that English-speaking children produce denominal adjectives with the (unstressed) suffix -y freely from as young as age two, both conventional forms like dirty, rainy, sandy and innovative forms like cracky, jammy, nighty (Clark 1993). How account for this contrast between early versus late acquisitions in the two languages? Two apparently distinct factors play a role – the one structural and the other pragmatic: linguistic typology and language use. As regards the first, typologically, Hebrew and English both distinguish adjectives from the two other major lexical categories, nouns and verbs. As a result, acquisition of the basic, morphologically nonderived type (3a) adjectives follows a similar developmental pattern in the two languages, governed by semantic complexity common to both languages and lexical conventions specific to each. Acquisition of such items is determined by factors like what kinds of attributive meanings young children are able to encode, how these become more specialized as a function of general vocabulary expansion, and what form-meaning mappings happen to be encoded by lexical adjectives in the target language. For example, in the class of dimensional adjectives, Hebrew gavóa is equivalent to both English ‘high’ and ‘tall’, in contrast to the antonymous terms namux





Ruth A. Berman

‘short in height’ and katsar ‘short in length’, for which English has the single adjective ‘short’. But Hebrew and English differ considerably in their derived adjectives: from verbs as in (3b) and from nouns as in (3c). Why would Hebrew-speaking children acquire the complex alternations involved in forming -u marked verbal adjectives at the same time that English-speaking children gain command of the more transparent, linear addition of final -ed in past participial resultatives like closed, washed, fixed or -en in words like broken, eaten, written? The answer lies partly in the impact of typological structure: Hebrew as a Semitic language typically creates new words synthetically, by assigning affixal patterns and alternating vowels to shared consonantal skeletons (Ravid, this volume; Shimron 2003). For children, this means that deriving words by means of verb-pattern alternations is a highly productive and hence accessible option, one that they command by age 3 to 4 years (Berman 1993b). Besides, their everyday wordstock includes many items with the same surface forms as resultative participles (e.g., ratuv ‘wet’, mesudar ‘tidy’, mufra ‘crazy’). These words illustrate the initial alternations of a few familiar items defined as step (b) in the developmental progression delineated in (2) above. That is, knowledge of the form-meaning mapping of such adjectives is still pre-grammatically ‘itembased’ rather than productively associated with a lexico-grammatical category (specifically, resultative end-state attributes / u- marked passive participles). These ‘initial exemplar’ items form the basis for subsequent abstracting out of the categorical form-meaning relation. In contrast, children’s early vocabulary, at the phase when they are first adding type (3a) ‘basic’ adjectives to their repertoire, is lacking in denominal adjectives that end in -i. This suggests that the contrast between early versus late acquisition of denominal adjectives in Hebrew and English cannot be attributed to structural complexity per se – whether semantic or morphological. Rather, the explanation lies in two interconnected socially embedded factors of language use: lexical productivity and linguistic register. Here, ‘productivity’ refers to speaker preferences for encoding novel form-meaning mappings in the lexicon (and see, further, endnote 5); and ‘register’ refers to the accepted ways for expressing established form-meaning mappings in different contexts of usage, in different types of discourse, and at various levels of style. In English, many denominal adjectives are formed regularly from basic, monosyllabic, everyday words of Germanic origin, for example, windy, dirty, soapy (similar to familiar nonderived adjectives like pretty, ugly, silly). Children use these early on, as a preferred, highly transparent method for attributing properties to nouns. In contrast, English denominal adjectives with Latin-based stems, e.g., military,

Between emergence and mastery

characteristic, or industrial are associated with later, school-age vocabulary. In Hebrew, denominal adjective-formation in general is typical of academic and journalistic discourse, types of language usage that are neither relevant nor accessible to preschool children. Adjectives with the surface form of passive participles like those in (3b), e.g., shavur ‘broken’, patuax ‘open’, mesudar ‘tidy’, or meluxlax ‘dirty’ are common items for Hebrew preschoolers. But denominal adjectives like those illustrated in (3c) are not at all typical of either the input or output language of young Hebrew-speakers. As a result, they have little recourse to the device of adding suffixal -i in order to innovate or use denominal type adjectives, even though the device itself is structurally productive and semantically transparent. In sum, factors of typologically motivated structural preferences (for word-internal vowel alternation in Hebrew compared with linear suffixation to a stem in English) interact across development with socially determined, language-specific patterns of lexical productivity and level of linguistic register, on the one hand, and the more universal factor of conceptual complexity, on the other. Together, these factors may outweigh more ‘obvious’ perceptual or processing factors of relative structural complexity in determining the order in which children acquire and use various classes of adjectives in different languages. . Adjectival versus syntactic passives A second example of the multiple factors that both drive and delay acquisition concerns syntactic alternations of voice and valence in Hebrew. These are illustrated in (6) for the two verb roots š-b-r ‘break’ and t-k-n ‘fix, repair’.7 (6) a.

Active Tensed Verb with Accusative Marker et: Ron šavar ve Boaz tiken et ha-bérez ‘Ron broke and Boaz fixed ACC the faucet’ b. Resultative Participle = Adjectival (Perfective) Passive: ha- bérez lo haya šavur, hu haya metukan ‘the faucet was not broken, it was fixed’ c. Syntactic Passive: ha- bérez nišbar (al ydey Ron) ve tukan (al ydey Boaz) ‘the faucet was-broken (by Ron) and fixed (by Boaz)’

Consider the different developmental history of these three related constructions. First, in simple-clause active sentences like (6a), children master use of the prepositional et as a unique marker of accusative case by the young age of





Ruth A. Berman

2;6, even though this involves complex, abstract knowledge which is not always semantically motivated since, first, the accusative marker et is used with verbs that refer to both states and activities; second, it is confined to definite objects (compare Ron ohev baxurot ‘Ron likes girls’, Ron ohev et ha-baxurot / Rina ‘Ron likes et the girls / Rina’); and third, many transitive activity verbs govern prepositions other than et: Compare Ron šavar et ha- bérez ‘Ron broke et the-faucet’ with Ron ba’at ba-bérez ‘Ron kicked in [=at] the-tap’, Ron hika et ha-kelev ‘Ron beat et the-dog’ with Ron hirbits la-kelev ‘Ron hit to-the-dog’. Use of accusative et is thus a morpho-syntactic phenomenon that is both early to emerge and rapidly mastered, in a way that cannot be accounted for by one-to-one form/meaning mapping. Rather, its early acquisition is explainable by a combination of universal patterns of acquisition and linguistic structure, on the one hand, and the language-particular factor of ‘typological imperatives’, on the other (Berman 1986b). Across languages, children gain command of simple clause case-marking relations between ages 2 to 3 (Perdue & Bowerman 1990) and transitivity is most typically associated with definite object NPs (Hopper & Thompson 1980). In language-specific terms, Hebrew non-subject NPs generally take a case-marking preposition, so that the typical surface form of a Hebrew simple clause construction is {N V Prep N}. The early mastery of the accusative marker et can thus be attributed to its centrality in Hebrew syntax and the regular marking of direct objects in simple active clauses as in (6a). Next, as noted in the preceding section, children show productive command of u- marked passive participle forms encoding resultative end-states, akin to English broken, mended, written – as illustrated in (3b) and again in (6b) – by age 3 to 4 years. In marked contrast, syntactic passives like those in (6c) are largely avoided by children even at early school-age. This was shown by a range of structured elicitations with different designs and from different populations. Thus Ravid (this volume) reports on studies showing that acquisition of passive morphology is not fully mastered as late as age nine years. Further, in a test administered to schoolchildren at different ages, six-year-olds consistently gave non-passive responses when presented with obligatory contexts for passive-formation, although they performed morphological alternations to give causative responses (e.g., changing li-cxok ‘laugh’ to le-hacxik ‘make laugh, amuse’) on other items in the same test (Berman 1993b; 1997c). In this, they differed from 11- to 12-year-olds who regularly provided passive constructions where required on the same test. A third source of evidence is the paucity of innovative verbal passive forms recorded in children’s spontaneous speech output compared with other coinages relying on manipulation of verb-pattern

Between emergence and mastery

morphology (Berman 2000). Finally, a cross-linguistic study of passive usage in texts written by 9- to 10-year-old schoolchildren compared with adolescents and adults showed that Hebrew speaker-writers rely on passive constructions significantly less than their English, Dutch, and French-speaking counterparts (Jisa et al. 2002): The two canonic passive verb-patterns do not occur at all in the texts of Hebrew grade school and junior high students, and they occur only rarely, although significantly more, in the high school (1% of all verb forms) and the adult texts (2% altogether), indicating that this is a ‘late developing’ form of expression in the language. This difference between resultative participles (adjectival passives) compared with syntactic or verbal passives is noteworthy for several reasons. First, both sets of forms derive from alternations in verb-pattern morphology, a system that children apply productively by age 4 years. Second, both the resultative participles like those in (3b) and passive-voice verbs like tukan ‘was-fixed’ in (3c) are formed with the same passive-marking vowel u. Third, passive constructions are relatively structurally unconstrained in Hebrew, although more so than in English, in the following ways: (1) Hebrew passives are typically confined to verbs that govern the accusative marker et but not other prepositions (Hebrew has no construction parallel to, say, English he was laughed at); (2) nonfinite verbs lack a passive form, so that a construction like English may be fixed needs to be paraphrased; and (3) passives tend to be avoided with stative verbs like hate, smell. Apart from these constraints, passivization is a syntactically productive process in Hebrew. Besides, Hebrew-acquiring children should be able to cope easily with syntactic passives, since the language has numerous constructions in which a non-agentive noun phrase functions as grammatical subject. The well-recorded delay in use of passives into school-age thus cannot be attributed to semantic opacity, morphological difficulty, or syntactic complexity in the expression of verb-argument relations. As noted, young preschool children show early command of form/function relations in direct object marking as in (3a) and in use of resultative participles as in (3b). Nor can the delay be due to formal constraints in the target-language syntax, since structurally, syntactic passives are as productive as resultative participles. Rather, it can be attributed to the fact that Hebrew affords speakers a range of readily accessible alternative means of expressing the discourse functions associated with passive voice: downgrading of the agent and presenting an ‘undergoer perspective’ on events (Keenan 1985; Tolchinsky & Rosado 2004). Several of these options are illustrated in (7), to show that, in addition to the passive construction illustrated in (7a), Hebrew speaker-writers can downgrade agency by





Ruth A. Berman

using middle-voice verb-morphology to provide a patient rather than an agent perspective as in (7b), an option that shows up increasingly with age (Berman & Slobin 1994: 515–538); they can focus on the predicate by using subjectless impersonal constructions as in (7c), common in the naturalistic speech output of children as young as 2 years of age, and even earlier in child-directed speech input (Berman 1980, 1990; Berman & Nir-Sagiv 2004); or they can topicalize nonagentive nominals by fronting to pre-subject position or by left-dislocation, as in (7d). (7) Alternatives for Agent Downgrading or Patient Perspectives: a. Syntactic passive: ha-balon pucac (al ydey Ron). ‘the-ballon was-burst (by Ron)’ b. Middle voice: ha-balon hitpocec. ‘the-balloon burst’ c. Impersonal: pocecu et ha-balon. ‘(they) burst-pl acc the-balloon’ d. Topicalization: et ha-balon lo ani pocacti. ‘acc the-balloon I didn’t burst’

Findings from spontaneous speech output, oral narratives, written narrative and expository texts, and from structured elicitations all show that Hebrewspeaking children well on in school-age prefer means other than syntactic passives to meet the discursive functions of passive voice in a language like English. This can be explained by a combination of three converging, usage-based factors. One is that speakers of Hebrew in general, not only children, tend to avoid passive constructions in their colloquial usage and even in more literary style, showing that usage preferences or ‘speaker productivity’ (Berman 1987c; Clark & Berman 1987) outweigh considerations of structural availability in children’s as in adult Hebrew. Relatedly, as noted for denominative adjectives, linguistic register plays a role in syntax, too: Passive voice is typical of the academic discourse of learned journals and lecture podia and of journalistic reporting in the media, contexts where activities are said to be ‘conducted’ and events are ‘caused’ or ‘expected’ rather than described in more personalized terms. These contexts are accessible to educated, literate speakers, but they are inappropriate and so irrelevant to the everyday colloquial discourse that governs the speech input and output of young children. Besides, in terms of what Jisa et al. (2002) define as ‘competition’,8 Hebrew speaker-writers have a rich range of alternative means for expressing the discourse functions of passive voice – downgrading of agency and focusing on the undergoer as in (7b) or on the event as in (7c).

Between emergence and mastery

. Nonfinite forms and nominalizations Two interrelated ‘operating principles’ (MacWhinney 1985; Slobin 1985) will be invoked to explain the final example of early emergence yet late mastery of linguistic knowledge. These are the factors of structural complexity, in the sense of degree of deformation of simple clause-structure, and the fact that later language development builds on earlier acquired, more basic materials as prerequisites, which enable children to use what they know ‘to learn more’ (Shatz 1987). The domain selected to illustrate these principles is nominalized constructions, which combine features of both the lexicon (the focus of Section 3.1) and of syntax (Section 3.2). Consider the examples in (8). (8) a.

Simple Clauses with Finite Verbs [Hebrew root b-n-y ‘build’]: He’ll build / construct the castle. It’ll be easy for him to-do so. hu yivne / yakim (et) ha-migdal. yihye lo kal la’asot zot. b. Infinitival Complement / Extrapositioned Subject: It will be easy (for him) to build / construct the castle. yihye (lo) kal li-vnot / le-hakim (et) ha-migdal. (For him) to build / construct the castle will be easy. (bishvilo) li-vnot / le-hakim (et) ha-migdal yihye kal. c. Gerundive Complement / Adverbial: He proved himself in-building / constructing the castle. hu hoxiax (et) acmo bi-vnot-o / be-hakim-o (et) ha-migdal. In building / constructing the castle, he proved himself. bi-vnot-o / be-hakimo (et) ha-migdal, hu hoxiax (et) acmo. d. Nominalized Objects / Subjects: He’ll succeed with (his) building / constructing / construction of the castle. hu yacliax bi-vniyat (-o) / hakamat-(-o) ( et) ha-migdal. (His) building / constructing / construction of the castle will be easy. bniyat (-o) / hakamat-(-o) (et) ha-migdal tihye kala.

Space does not permit detailed analysis of these constructions. Note, only, that the examples in (8) are ranged in Hebrew as in English on a cline of descending transparency, in the sense of increased distance from the simple-clause structure illustrated in (6a) and (8a). Moreover, this scale largely reflects the order in which these constructions are acquired. In Hebrew as in other languages, simple-clause structure with a tensed verb and direct object is acquired between age 2 to 3 (See Section 3.2). Infinitival complements with modal or aspectual verbs like want to build, start to build are mastered slightly later, but they are also an early acquisition. In these contexts, development takes the form





Ruth A. Berman

mainly of increased semantic and lexical variation of the range of tensed verbs that take infinitival complements and the extent these will be concatenated (e.g., want to be able to start to build / building). In contrast, gerundive and nominalized forms corresponding to those in (8c) and (8d) are relatively late, school-age acquisitions. In our cross-linguistic narrative study (Berman & Slobin 1994), we defined as ‘late acquisitions’ forms which were absent from the narrations of the preschool 3- to 6-year-olds in our sample, which appeared only occasionally in those of 9- to 10-year-old schoolchildren, and which were well represented in texts of the adults. The children’s texts were quite typically lacking in nonfinite and nominalized verbforms in four of the five languages we studied: English, Hebrew, German, and Spanish (the exception being Turkish, which requires nonfinite, nominalized subordination). In English, for example, participial -ing, used widely for progressive aspect by the youngest children, and in complement constructions by the older children, appeared in an adverbial or other modifying function only in mature narrations (e.g., And the deer carried him off to the edge of a cliff, with his dog chasing after). In Hebrew, high-register gerundive forms like be-hagio liktse ha-matsok ‘in-reaching-his to-the-end the cliff = on reaching the edge of the cliff ’ did not appear in a single text, as expected in the colloquial style suited to a children’s adventure story. And only a few adult narrators made use of the highly productive, morphological mechanism of Hebrew for deriving abstract nominals from activity verbs (e.g., ha-yeled haya asuk be-xipus axarey ha-tsfardea ‘The-boy was busy in-search after the-frog = in searching for the frog’; hem mamshixim bi-meruca-tam ‘They continue in-flight-their = with their flight, fleeing’). Subsequent research on Hebrew narrative and expository texts of schoolchildren, adolescents, and adults, reveals that nonfinite gerundive constructions (e.g., be-vo’enu la-dun ba-nose ‘in-coming-us to-consider the-topic = when we come to consider’) are confined to the expository texts produced by high-school and university students in writing, and not before then (Berman & Nir-Sagiv 2004). This more current research reveals that use of abstract verb- and adjectivederived abstract nominals (e.g., construction, adaptability) is a hallmark of mature narrative and more especially of expository discourse, and that these occur more in written than in oral texts (Ravid in press; Ravid & CahanaAmitay 2004). Children’s avoidance of derived nominals in monologic text construction is consistent with findings from structured elicitations, as detailed in Ravid (this volume). Late preschool 5-year-olds generally understand such forms in context, 9-year-olds produce them consistently in a completion task, but only older children are able to coin novel nominalizations from unfamiliar

Between emergence and mastery

verbs. This avoidance of derived nominals until well into adolescence is not due merely to morphological complexity. Although alternations between source verbs and their related action nominals are not fully regular or predictable in form/meaning mapping, these are similar to other derivational processes in Hebrew which, as noted for the other domains considered in this chapter, are largely mastered by children as young as age 4 (Berman 1995b). I suggest, rather, that difficulty with nominalization has several sources. One is syntactic complexity, defined here by ‘degree of deformation’ of simple clause structure entailed by a particular syntactic process. Compare the three different levels of complexity illustrated for English nominalizations and their Hebrew counterparts in (8a) to (8d). In English, the nominative-initial, SVO order of the simple clause is violated in (8c) by omission of overt subject and of tense marking, and in (8d) by a possessive pronoun followed by a gerundive or nominalized form, with genitive of and no overt tense marking. Besides, in the Germanic wordstock, gerunds and derived nominals take the same -ing form as in building, whereas the Latinate construct splits between constructing and construction. In Hebrew, normal SVO word order is further violated, since the possessive suffix -o ‘his’ follows the nominalized form; it is required in the gerundive form in (8c) but optional in action nominals as in (8d); and in both cases, accusative case-marking is retained, but verb-tense is neutralized in the feminine-form derived nominals bniya ‘building’, hakama ‘construction’. These various types of structural complexities are one explanation for why children defer use of these forms till a later, more literate stage in their development. Besides, in pragmatic terms of language use, nominalizations tend to be restricted to formal registers of expository and academic discourse in many languages (Comrie & Thompson 1985). Here, too, structural complexity – in the sense of both syntactic density and semantic opacity – interacts with language typology to determine what options children will select for combining two or more tensed clauses in a single syntactic package. In our elicited narratives, preschool children typically string simple, tensed clauses one after another across their texts. School-age children use the typologically most accessible options for syntactic packaging. In Hebrew, although rather less than in Spanish, they rely on subordination with še-, que ‘that’ for this purpose, while Turkish children use the obligatory nonfinite nominalized forms from preschool age (Berman 1998; Slobin 1988). English-speaking children, who use both to infinitives and -ing forms of verbs correctly in simple clauses by age 2, fail to exploit these same constructions for the complex purposes of nominal embedding illustrated in (8b) through (8d). These represent more formal, literate rhetorical options for English speaker-writers.





Ruth A. Berman

This again points to the multiplicity of factors affecting ‘what comes later’ in language acquisition and development. In developing form/function relations for clause linkage and discourse connectivity, the factors of syntactic complexity and typological preferences interact with the expressive options available to speaker-writers of different languages. Textual cohesion can be achieved by strings of simple clauses, as in (8a), favored by young children, and common in casual speech; by infinitivals, which are readily available to children by late preschool age, although not when extraposed, as in (8b); or by finite subordinate clauses as used increasingly by schoolage children, more so in Spanish and Hebrew than in English and German (while structurally largely unavailable to Turkish-speakers). These interrelated factors combine to explain the finding from experimental elicitations in Hebrew and from extended texts in different languages, that nominalizations are a late acquisition, not readily accessible to preschool children.

. Conclusion The ideas presented in this chapter were motivated by an attempt to demonstrate how linguistic, cognitive, and social factors combine in shaping language development. From this perspective, it appears that the giants of our domain, Chomsky, Piaget, and Vygotsky, are all right about what is involved in learning a language, as a complex enterprise that involves the focal points of each of their concerns: grammar, cognition, and social interaction respectively. The challenge for developmental psycholinguistics is to achieve an integration between these disparate perspectives. Such an endeavor would serve to specify more precisely the ‘multiple bootstrappings’ that apply at the initial phases and to constrain the many factors involved at more advanced phases. Ideally, this would make it possible to define the relative weight of the different factors involved in early acquisition and later mastery across different developmental stages – in various linguistic domains and contexts of use. A tentative first stab at such a weighting – deriving from the phase-based model of Section 2 and the research findings of Section 3 – is proposed in (9).

Between emergence and mastery

(9) Ordering in the developmental route from emergence to mastery: 1. Language typology: Universal > Typologically Pervasive > Language-Particular 2. Social-cultural factors: Phase I =

context-bound, communicatively motivated Phase III = context sensitive, discourseembedded

3. Structural complexity: Phase II = structure-dependent, grammaticization

First, in terms of inter-language variation and target language impact, the prediction is that children will start out with linguistic universals, in the sense of properties and categories that are shared across languages, they will subsequently acquire knowledge of processes and features central to and pervasive in a particular type of language, and only later will proceed to more highly specific knowledge restricted to particular syntactic contexts, morphological classes, or lexical items in a given target language. That is, when they initially encounter a particular target language, children come equipped with a strong basis of shared knowledge about the structural categories and semantic distinctions that are in principle available to all and any languages.9 For example in the domain of predicate types, (9.1) implies that children do not have to learn the semantic distinction between states and activities or changes-of-state, since these distinctions are universal and apply across all languages.10 What they do need to learn is which morpho-syntactic encodings mark these distinctions in their language. Similarly, in the domain of syntactic transitivity: Children from the start use predicates that require either one, two, or more arguments, irrespective of their particular target language. But they must learn how verb-argument structure is encoded in their language, by word order, case-markings, and/or the morphological shape of verbs. A third example, in the area of temporality: children recognize semantic distinctions of aspect (e.g., duration versus punctuality) very early on; and subsequently they mark distinctions of tense by inflections, before they acquire other verb inflection such as markers for gender, number, and person agreement. From the point of view of ‘early’ versus ‘late’ acquisitions, this leads us to predict that, across languages, children will acquire person marking later than tense marking on verbs; they will acquire semantically universal distinctions between classes of predicates before they gain command of the typologically relevant inflectional distinctions for marking grammatical aspect in their language; and in a language which marks transitivity alternations morphologically, command





Ruth A. Berman

of verb-form alternations to express these distinctions (cf. the nonproductive alternations of English rise/raise, lie/lay, sit/seat) will emerge later than syntactic encoding of transitivity distinctions through word order alternations, verb-argument configurations, and/or adpositional case-markings. On the basis of such shared linguistic universals, children soon come to attend to key typological properties of the input language. Even very young children recognize ‘where the action is at’, so to speak, with respect both to which categories are formally distinguished, and how these distinctions are typically marked. Crosslinguistic research reveals the influence of languagespecific effects on speech perception and babbling in the first year of life (Jusczyk 1997: 178–179); on children’s early encoding of spatial distinctions in languages like English and Dutch compared with Korean or Tzeltal (Bowerman 1996); on their strategies for new-word formation in English compared with Hebrew and other languages (Clark & Berman 1984, 1987); and on their narrative development in different languages (Berman & Slobin 1994; Hickmann 2003). Such findings converge to show that children are early on attuned to the language-particular way of encoding form-meaning relationships in their language. When this type of sensitivity finds expression will depend on developmental factors, of the kind that are the burden of this chapter. But in each case reported, the way young children encode form-meaning relations accords with how this is done by adult speakers of the same target language rather than by children of the same age in other languages. That is, young preschoolers reveal strong sensitivity to the ‘typological imperatives’ of their language. Second, as suggested in (9.2), social factors of intra-language variability (communicative contexts, discourse registers, genres, and levels of usage) will reveal a superficially U-shaped development. During the period of initial emergence of knowledge, defined as Phase I in Section 2, young children will focus on pragmatic factors of communicative efficacy, and their use of language will be highly context-bound and restricted to the extra-linguistic speech situation. Eventually, with mastery, speakers will again become critically attentive to factors of linguistic appropriateness, manifested in increased sensitivity to discourse-based needs and constraints of different registers, modalities, and genres. The link between these two periods in development of language knowledge is provided, as suggested in (9.3), by structural command of form/meaning relations in the domains of grammar and semantics. Here, attention focuses on grammaticized knowledge of syntactic structures like passives and nominalizations, of lexical classes like adjectives, of morphological processes of derivation, and of how to structure discourse so that it constitutes a well-formed conver-

Between emergence and mastery

sation, narrative, or expository text. Again, in line with the non-monolithic view of acquisition advocated here, these cycles of developmental phases recur across different domains rather than one time across-the-board. And in each case they proceed from both the most general and universal and from the most highly specific, item-based, and context-bound knowledge to being both more narrowly constrained by target-language typology and more broadly discourse-sensitive and text-based. So development in linguistic form proceeds from general to specific; while concurrently, in language use, it proceeds from narrowly local to broadly global. And this leaves us with yet another paradox.

Notes * Support for studies cited here was provided by grants from: the Israel-U.S. Binational Science Foundation (BSF) to R. Berman and Dan I. Slobin for the crosslinguistic study of developing temporality and to R. Berman and Eve Clark for the study of lexical productivity; the German-Israel Foundation (GIF) to R. Berman and Juergen Weissenborn for the crosslinguistic study of early grammar; the Israeli Science Foundation (Israel Academy of Science) to R. Berman and Dorit Ravid for the study of the oral/literate continuum; and by a Spencer Foundation major grant to R. Berman for the crosslinguistic study of developing literacy. I am indebted to Brian MacWhinney, Bracha Nir-Sagiv, and Liliana Tolchinsky for their careful reading and helpful comments on an earlier draft. They are in no way responsible for the inadequacies that remain. . This view corresponds closely to Tomasello’s observation that there “are three major types of problems that face a language learner: Problems of speech perception and production . . . ; problems of communication and meaning; . . . and problems of grammar and creativity” (2001a: 1–2). . Most of these studies include monolingual, educated adult subjects as criteria for ‘mastery’ of each of the domains that were analyzed. . By Semitic convention, root consonantal elements are represented by C and the interdigited affixal elements by relevant consonants and vowels (Shimron 2003). Thus the abstract form CaCuC stands for words like sagur ‘closed, shut’, katuv ‘written’, tafus ‘caught’, meCuCaC for mesugar, mekutav, metufas. . Stop / spirant alternations between p / f, b / v, and k / x are irrelevant in this analysis. . The term ‘productivity’ is used here as accepted in linguistic analysis, to refer to structural productivity or lack of constraints on the application of a rule or process. For example, the Hebrew -i suffix is more productive than English denominating suffixes -al, -ic, -ive, -ary, etc., because it does not have any other counterparts. On the other hand, Hebrew passive formation (Section 3.2) is syntactically less productive than its English counterpart, since it does not apply to verbs which govern prepositions other than the accusative et, and only tensed verbs have a passive counterpart, not infinitives or gerunds. Two other, rather different senses of ‘productivity’ are: (1) in language acquisition, children’s knowledge is said





Ruth A. Berman

to be ‘productive’ when it is rule-bound rather than based on rote-learning of unanalyzed strings; and (2) in language use, processes and structures that are relatively productive are those that are preferred by speakers for expressing particular form/function relations under given circumstances or at a particular point in time (Berman 1987c, 1993b; Clark & Berman 1987). . Beard (1993) notes an interesting semantic difference between two classes of denominal adjectives, which he terms ‘possessional’ and ‘similitudinal’. In the former, a bearded man is a man who has a beard, whereas in the latter, a friendly man is one who is like a friend. And, more relevant to the contrasts illustrated in (4), compare buttered toast ‘toast which has butter on it’ versus a buttery cake ‘a cake which is like butter’, iced coffee / icy hands, flowered material / flowery style. . The role of Hebrew verb-pattern morphology in mastering systems of transitivity and voice, detailed in Berman (1993a, 1993b) is disregarded in the present analysis. . Jisa et al. use the term in a rather different, but complementary sense to that of the ‘competition’ model as articulated, for example, in Bates & MacWhinney (1987), MacWhinney (1985: 1089–1149). . This is not a claim that this knowledge is ‘innate’ in the strong sense of being genetically encoded in the organism without any interaction with the environment. Rather, this kind of knowledge accords with aspects of brain structure, cognition, and behavior that have been termed ‘primal’ by some developmental theorists (Elman et al. 1996; Johnson & Morton 1991), in the sense that it is shared by or common to all members of the species, but it develops in interaction with the environment. . The term ‘universals’ as used here does not refer to a given model of grammar or to formal syntactic principles and constraints (so is not identifiable with a Chomskian model of ‘UG’). Examples of what were originally termed ‘substantive’ universals by Chomsky (1965), akin to Keenan’s (1975) ‘naive universals’ include: the distinction between consonants and vowels, between nouns and other word classes, and the preference for encoding certain distinctions in predicates (like those relating to aspect, tense, and mood) and for marking other categories (such as case, gender, and definiteness) on nominal elements (Bybee 1985).

Lexical acquisition in the early school years* Julie E. Dockrell and David Messer

.

Introduction

Children’s first words mark the beginning of a lifelong lexical journey. During this journey they move from apprentice word learners to competent vocabulary users at a remarkable rate. The apparent ease of this process has led to the suggestion that “learning vocabulary is a relatively simple affair” (Plunkett & Wood in press). However, we adopt a different stance, to argue that lexical acquisition is a complex and extended process involving the integration of phonological, semantic, and morpho-syntactic knowledge with cognitive and social processes. Vocabulary knowledge is a strong predictor of academic success, and it plays a central role in cognitive development, especially in relation to literacy and learning (Cunningham & Stanovich 1997; Stanovich & Cunningham 1993). The lexicon provides a unique domain for studying the interaction between context and cognition, and the ways in which this interaction changes with development. We address these issues by considering the factors that play a role in early lexical development (Section 1), examining the ways in which different assessment procedures provide contrasting views of children’s abilities (Section 2), considering the support for vocabulary learning in school (Section 3), and the challenges and difficulties encountered by later vocabulary learning (Sections 4 and 5). . What needs to be acquired? When children acquire a new word, they must identify the sound in the speech stream to encode a phonological representation and then establish a mapping between the word and world; ultimately a detailed semantic representation is developed for the new term with knowledge of its morpho-syntactic features. Inaccurate phonological representations reduce the accuracy of children’s lex-



Julie E. Dockrell and David Messer

ical productions and may also hamper the initial establishment of semantic representations (Section 4.1). Learning a new word also involves the formation of or links to a conceptual domain. A child must learn that ‘sheep’ denotes a specific set of animals, and that ‘happy’ denotes a particular class of emotions. The key to lexical acquisition is the mapping between world, meaning, and form. The question of how children establish a word-world mapping has been the subject of considerable research (Bloom 2000). Many of the studies imply that this is a simple process, where all that is needed is the selection of the referent from an array of stimuli. However, a match between an object or set of objects and a form is insufficient on its own to develop a semantic representation; children must also integrate the new term with their existing lexicon (Anglin 1993; Clark 2003; Dockrell & Campbell 1986). As yet we have no clear understanding of the ways in which word learning in young children differs from word learning in older children and adults. Answers to these questions require more sophisticated evaluations than those commonly applied with younger children. The word learning skills of older children depend on both the exposure they receive and their cognitive and linguistic competence. . ‘Fast mapping’ and constraints in early lexical acquisition In experimental settings, young children acquire information about the meaning of a novel term after a single exposure to its use (Carey 1978; Heibeck & Markman 1987). This process, often described as ‘fast mapping’ or ‘quick incidental learning’ (Rice 1990) has led researchers to focus on the cognitive factors that underlie such learning (Markman 1989; Markman & Hutchinson 1984). These investigations generally involve word learning in situations where contrasts are drawn between a named and an unnamed object, and in which the novel term is contrasted with a known term. One prominent explanation of children’s early success in these mappings is that inbuilt constraints restrict the hypotheses that children entertain (Golinkoff et al. 1994). Some of these constraints may eventually be relinquished, but it is never specified when and why such constraints cease to operate. Clark (2003) presents a careful analysis that questions the viability of each of these constraints in turn (see also Nelson 1988) Various proposals have been made about constraints on early meaning assignment, but there is little agreement on (1) where they come from, (2) when

Lexical acquisition in the early school years

they start to apply and how long they last, and (3) why they are abandoned. (Clark 2003: 138). Later vocabulary acquisition poses further challenges for such accounts. Much of later vocabulary learning involves terms for entities and relations that are not definable using explanatory mechanisms derived from studies with concrete vocabulary terms. Older children encounter words that are abstract, low in frequency, domain-specific and domain-general, and that involve nonliteral meanings (Nagy et al. 1993; Nippold et al. 1988). Furthermore, no single set of semantic relations or organizational structure is adequate for the entire lexicon (Miller & Fellbaum 1991). As children mature, these different aspects of semantic representation become increasingly important, but are not easily explained by the presence of constraints (Section 4.2). . Beyond fast mapping and constraints Conceptual distinctions influence the meanings of some words (Soja et al. 1991). However there is increasing evidence that in some lexical domains the construction of categories occurs under linguistic guidance (Bowerman & Choi 2003). Children as young as 17 months of age talk about spatial events in language-specific ways (Choi & Bowerman 1991) and language-specific patterns are also evident in comprehension (Choi et al. 1999). These differences are consistent with differences in the way target languages partition the semantic domains. Thus, the word to world mappings that children begin to establish are sensitive to the semantic and statistical properties of the target language (Bowerman & Choi 2003: 402). By corollary, specific linguistic input may direct children’s hypotheses about the intended referent of a novel term. Preschoolers who were exposed to objects labeled with a novel word where the objects were described in vignettes as having properties typically associated with artifacts, generalized the term on the basis of similarities in shape. In contrast, children extended the same labels on the basis of similarities in both shape and texture when the same objects were described as having properties typically associated with animate kinds (Booth & Waxman 2002). Only the conceptual information provided in the vignettes could be responsible for the differences in extension patterns that were observed since the same objects were presented in both conditions. Thus, the conceptual distinctions that guide children’s interpretations of new terms are affected by both the language they are learning and the ways in which the new term is introduced. The wider social context is an important source of information to guide word learning and provides a more complex range of information than is im-





Julie E. Dockrell and David Messer

plied by constraints. Building a vocabulary depends on hearing words and interpreting the meaning of a term in a specific setting (but see Hoff & Naigles 2002, for a re-analysis of research that casts doubt on the sufficiency of social approaches to word learning). Infants as young as 18 months actively gather social information to guide their inferences about word meanings (Bloom 2000) and this process continues to serve as an important source of information about the meanings of new terms as children develop (Clark & Wong 2002). For example, children learn a new verb best if it is introduced when the event is impending rather than when it is already ongoing (Tomasello & Kruger 1992). This points to the importance of match in timings of exposure, which varies for different word classes. It also relates to young children’s reliance on the intentions of their interlocutors (Baldwin 1991) and their ability to understand the knowledge that others possess to guide their word learning (Sabbagh & Baldwin 2001). Another problem with studies of fast mapping is that they generally fail to consider how semantic representations are established, particularly in the case of abstract terms. There is more to learning the meaning of a word than pairing it with an observable referent, since words clearly refer to a range of entities, only some of which are observable. This is demonstrated by a simulation study of vocabulary learning (Gillette et al. 1999), where adults were presented with different types of linguistic and non-linguistic information about a motherchild interaction that introduced either a noun or a verb as a target word. A key factor in the accuracy of identifying the target word was its ‘concreteness’ or ‘imageability’. The authors interpret this as revealing an advantage for nouns over verbs in the early vocabulary of children (Gillette et al. 1999: 154) and conclude that only a limited stock of nouns can be identified solely in terms of reference to their standard extra-linguistic contexts of use (that is, by ‘word-toworld pairing’); verb identification, in contrast, also requires inspection of the standard linguistic contexts of use (‘sentence-to-world pairing’). These findings are consistent with the fact that syntactic complexity in the input appears to support lexical learning (Hoff & Naigles 2002). Fast-mapping studies illuminate important aspects of early strategies for establishing reference, but they fail to provide in-depth insight into the nature of semantic representations and how these change over time. Children’s initial hypotheses about word meanings arise from a range of factors including the initial cognitive strategies that are used to limit possible referents, pragmatic factors, extra-linguistic context, and features of the input and target language typology. Consideration of this range of factors is critical as, with increasing age, vocabulary becomes both more extended and more detailed, and children

Lexical acquisition in the early school years

encounter the same word in varying circumstances. To provide adequate descriptions of acquisition processes it is also important to consider the ways in which lexical competence is evaluated.

. The assessment of vocabulary There is a general consensus that significant vocabulary gains are made in childhood, yet there is considerable debate about both vocabulary size and the rate of learning ‘new lexical items’ (Anglin 1993). Estimates of vocabulary size depend on a number of factors, specifically on the criteria for establishing that a child ‘knows’ the meaning of a word. A basic distinction can be made between measures of comprehension and measures of production. Younger children take much longer to establish accurate representations for production (Clark 1993), and this asymmetry between production and comprehension appears to be particularly marked for verbs (Gillette et al. 1999). However, initial representations for comprehension may not be detailed and in some cases differential understanding lags behind productive use (Nelson 1996: 306–307). Below we review some of the problems involved in assessing children’s growing vocabulary in relation to the asymmetry between comprehension and production (Section 2.1), and we propose alternative methods for coping with these difficulties (Section 2.2). . Comprehension and production in assessment Comprehension measures aim to tap children’s semantic representations. They may be used to establish the word boundaries for later acquired vocabulary or the relationships within semantic domains. Comprehension is needed for the recognition of words and to provide templates for production. Clark (1993) argues that representations from comprehension consist of an acoustic template to which children then add progressively more information; so one aspect of understanding vocabulary development is to consider the way in which information is associated with the acoustic template (Section 4.1). Measures of comprehension have the potential to contribute to our understanding of semantic structures. Appropriate production draws on a wider range of skills than comprehension including selection of the appropriate semantic representation for the item, instantiation of a phonological representation, and use of the word in its appropriate linguistic form and context. It has been argued that production





Julie E. Dockrell and David Messer

based measures of vocabulary size may underestimate what the child actually knows (Dapretto & Bjork 2000), because such measures rely on phonological accuracy in addition to semantic and grammatical knowledge. In addition, when considering production it is important to realize that children practice new words prior to the establishment of detailed, accurate semantic representations (Clark 2003; Nelson 1996) and there are situations where words may be produced with little or no understanding of their meaning. Indeed it is argued that, for abstract terms, children initially engage in “use without meaning” in specific contexts (Levy & Nelson 1994; Nelson 1996). Production is also vulnerable to other task demands. There may be a range of alternative and acceptable labels for an item or action. A picture of a poodle can be labeled as a ‘poodle’, but ‘dog’ or ‘animal’ are equally correct. Actions can be labeled with general all-purpose verbs, such as ‘do’, and while these are less precise than specific verbs, they are often acceptable productions. Thus, the pragmatic context in which productions are assessed will direct the types of lexical items that children produce and this is particularly evident in conversation. Nonetheless, one might view the ability to produce a word in the appropriate context as the ‘gold standard’ of lexical knowledge and errors of production as particularly revealing about a child’s semantic representations (Braisby & Dockrell 1999; McGregor & Waxman 1998). As we outline below, vocabulary knowledge can be assessed in a range of ways so that comprehension and production data offer complementary insights about lexical representations. . Alternative methods of investigating word knowledge A typical way of measuring word knowledge is by means of a multiple-choice format, in which the child selects a picture for a target word from among several pictures (for younger children) or written words (for older children). These procedures are problematic, since they are open to guessing or the use of non-linguistic strategies to identify items, particularly when the distractors are not selected by clear criteria (Anglin 1993). It is also the case that children’s choices in such tasks may not provide useful information about the nature of their semantic representations. As early as 1942, Cronbach pointed out the need to determine how complete a student’s understanding of a word was. Forced-choice comprehension tasks give a ‘flat’ view of vocabulary, as if all the words are either unknown or known to the same level, to the extent that such measures can be viewed as “useless at best and dangerous at worst” (Kameenui et al. 1987: 138). Ideally, word knowledge should be assessed by a range of different measures (Beck & McKeown 1991), and these

Lexical acquisition in the early school years

should take into account both the quality and the quantity of children’s lexical knowledge. Definitions are a way of assessing children’s explicit knowledge about a word’s semantic attributes. Prior to age seven, children’s definitions are simple, and tend to focus on perceptual or functional information (Benelli et al. 1988; Storck & Looft 1973), lacking in superordinate terms (Watson 1995). In contrast, older children produce definitions that are more precise, include conventional social information (Benelli et al. 1988), and increasingly refer to superordinates (Curtis 1987; Snow 1990; Watson 1995). Taken together, this research points to an increase in semantic content and syntactic form across word classes. Interestingly, children generally provide higher-quality definitions for nouns than for verbs and adjectives, and for roots and compounds than for morphologically inflected or derived words (Anglin 1993; Johnson & Anglin 1995). Definitions thus provide a potentially powerful investigative tool, since they assess semantic representations, demonstrate developmental progress, and show word-class effects. Yet definitions place a high cognitive load on the respondent. Initial understanding of a lexical entry is implicit and there are significant difficulties in expressing this knowledge. Thus, definitions require meta-linguistic awareness and are dependent on other linguistic skills beyond knowledge of the lexicon. With age, children make greater use of morphological analysis in processing unfamiliar vocabulary, a process described as morphological problem-solving (Anglin 1993). Some words may appear to have been learnt but in effect the child may construct a meaning from morphological information alone (Ravid, this volume). Similar more general processes can be seen in children’s developing ability to use words explicitly to express their ideas about other domains such as science (Karmiloff-Smith 1992; Pine & Messer 1998, 2003). Other attempts to apply techniques that tap various aspects of vocabulary knowledge have generally been limited to small-scale tests of their effectiveness in experimental studies (e.g., Jenkins et al. 1984; McKeown et al. 1985; Nagy et al. 1985). Such approaches have often involved investigations of the breadth of children’s semantic representations including knowledge of relevant antonyms, synonyms, hyponyms (Heibeck & Markman 1987); the semantic attributes of words (Funnell et al. submitted); and children’s inferences about categorical dimensions of new lexical items (Keil 1983). Taken together, different methods of measuring vocabulary indicate that word knowledge needs to be defined along a continuum that involves several dimensions (Section 3.1); evaluating a child’s vocabulary means specifying where along the continuum a particular lexical item lies in relation to each dimension.





Julie E. Dockrell and David Messer

Context also provides children with potential cues for inferring word meaning. For example, Rudel et al. (1980) asked children aged between 5 and 11 years to generate names in four different conditions: picture naming, naming to definition, sentence completion, and naming to touch. The oldest children’s performance did not differ across the contexts, but for younger children sentence-completion and tactile naming were the most supportive while naming to definition the hardest. These data add further evidence to the view that once knowledge is fully established, children’s responses are less task-dependent and less vulnerable to context (Section 4.2). Online tasks offer new ways to assess lexical representations, particularly the organization and links that are present between items in the lexicon. Online tasks involve largely automatic, non-conscious processes and so may be less vulnerable to metalinguistic knowledge. Simple picture-naming tasks demonstrate that lexical access becomes quicker with increasing age and is related to word frequency and age of acquisition. The presence of a semantically related spoken word also increases naming latency, while the presence of a phonologically related word shortly before production will decrease naming latency (Jerger et al. 2002). Thus, words that are similar in meaning interfere with search while words similar in phonological form facilitate naming. Such findings could provide information about the way in which items are linked together in the lexicon. In sum, the nature of the assessment protocol is central to the conclusions we draw about children’s skills and the representations that they develop. As children develop, a range of techniques are available to study vocabulary acquisition. We will argue that when different dimensions of lexical knowledge are considered, a multifaceted picture of vocabulary development emerges. Such lexical profiles can also provide important insight into how context and cognition support ensuing representations.

. School-based lexical development By the end of the preschool period, children have a range of cognitive, linguistic, and social processes to support vocabulary learning. In addition, an average 6-year-old knows about 10,000 words (Anglin 1993) which will provide a basis from which to extend their vocabulary and to “capitalize effectively on the information-rich social context within which word learning occurs” (Baldwin & Moses 2001: 318). Yet there is considerable lexical learning to be done; the six-year-old possesses 1/6th of the words that will be known by the

Lexical acquisition in the early school years

end of formal schooling (Bloom 2000: 12); and not all children are equally well equipped for the task (Section 5). The rate of vocabulary growth differs significantly between children, with larger vocabularies supporting the faster acquisition of new words than ones with smaller vocabularies (Elley 1989; Leung & Pikulski 1990). . Oral language input Both the amount and the nature of the language input children receive impact on the subsequent development of their lexicon (Hart & Risley 1992, 1995; Hoff & Naigles 2002). The relationship between input and acquisition holds across a range of situations including bilingual acquisition (Pearson et al. 1997). The type of lexical items that parents use plays an important role, and for some children the environmental opportunity to develop vocabulary is less rich than for others (Weizman & Snow 2001). This is an important factor in school achievement: vocabulary assessed in first grade predicts over 30% of the reading comprehension variance in 11th grade (Cunningham & Stanovich 1997). Thus, for children from less educated or less advantaged backgrounds, the exposure to language received in educational contexts is of critical importance. As children develop through the school years, they increasingly encounter new words in educational and leisure contexts. Support for lexical acquisition can be provided explicitly, in the form of direct instruction or dictionary use, or incidentally, in varied types of exposures and contexts. Little time appears to be devoted to vocabulary instruction in schools (Graves 1986, 1987; Nagy & Herman 1987). Moreover, dictionaries are inappropriate for younger children and the definitions provided are not always easily understood even by older children (Scott & Nagy 1997); much of schoolchildren’s vocabulary exposure tends to occur incidentally. Children are not necessarily given enough opportunities to build a rich vocabulary in the early school years. Findings from longitudinal studies indicate that gains in receptive and expressive language skills occur as a result of participation in quality nursery provision (McCartney et al. 1985). Yet on the whole, preschool settings are not sensitive vocabulary learning environments (McCathren et al. 1995). As with younger children (Section 1), the linguistic context where children encounter new words influences the resultant semantic representations. In a study of 130 five- and six-year-old children over a period of six weeks, Ralli (1999) collected baseline data about vocabulary knowledge and then had children exposed to new vocabulary items in one of five conditions: (i) repetition





Julie E. Dockrell and David Messer

of the unfamiliar word, (ii) an ostensive definition, (iii) an introduction of a word as a lexical contrast, (iv) a definition which included a description of the item, and (v) a no-exposure condition. The following week the children heard a story that included the ‘new word’, and their knowledge of the item was assessed on a variety of tasks. Both the lexical contrast group and the definition group developed detailed semantic representations of the new terms. The definition group had the most detailed knowledge and this knowledge was maintained overtime. Ralli also found that children’s performance varied with the measure used to assess word knowledge: all groups performed well on the comprehension measure, but only those in the lexical contrast and definition groups provided categorical information about the new term and used the term productively in a story telling task. The repetition and control group could not be differentiated, and both performed poorly on the semantically more demanding tasks. This highlights the importance of the types of exposures children typically receive in schools and indicates that informal oral definitions can provide children with enduring knowledge about a new lexical item. Word learning has been examined in other contexts, such as when teachers are reading to children (Elley 1989; Penno et al. 2002). As was shown by Ralli’s (1999) study, explicit highlighting of word meanings (Penno et al. 2002) and direct instruction (Stahl & Fairbanks 1986) facilitate vocabulary acquisition. Vocabulary acquisition is also facilitated if children are actively engaged in a story rather than simply hearing a ‘good’ reading of it (Elley 1989). Interactional reading facilitates vocabulary acquisition of both six- and eight-year-olds (Brabham & Lynch-Brown 2002). Explicit vocabulary instruction is not always provided. Carlisle et al.’s (2000) observations of 4th and 8th grade science lessons indicate that teachers seldom offer formal vocabulary instruction about unfamiliar words. These results are corroborated by studies with younger children. Best (2003) recorded science lessons for five- and six-year-olds. Teachers also rated the vocabulary used in these lessons along a scale, from definitely known to unknown for both comprehension and production. There was no explicit vocabulary teaching, but Best found subtle differences in how teachers introduced new terms. It was the combination of linguistic and non-verbal cues provided by the teachers that discriminated between items believed to be new and those believed to be familiar. Explicit teaching of word meanings can therefore explain only a limited amount of classroom-based word learning.

Lexical acquisition in the early school years

. Written language input Once children start formal schooling, written language becomes increasingly important for learning about language generally and vocabulary specifically (Nagy et al. 1985; Ravid this volume; Sternberg & Powell 1983; Tolchinsky, this volume). There is evidence that words are learned in roughly the same order in school, although the rate of development is determined by a child’s initial root word vocabulary (Biemiller & Slonim 2001). Some estimates indicate that ten-year-olds may be exposed to as many as a million words of text in a year, between 15% to 55% of which may be unfamiliar. The ability to learn words incidentally from written texts increases with age. Children who know more words are better at reading comprehension than children with poorer vocabularies (Nagy & Herman 1987; Nagy et al. 1985; Sternberg 1987) and good readers develop larger vocabularies than poorer readers (Carnine et al. 1984; Nagy et al. 1985). Studies that have investigated word learning from written texts have manipulated the types of exposure, the rationale provided to the participants, how word learning is evaluated, and the characteristics of the participants. Swanborn & de Glopper (1999) carried out a meta-analysis of 20 experiments of incidental word learning during normal reading; students’ attention was not drawn to the lexical item and there was a single exposure of the target item. They found substantial variation in the results; assessment methods sensitive to partial word learning showed higher word learning gains and students learnt more words when the ratio of text to target words was higher (Swanborn & de Glopper 1999: 277), yet the mean rate of incidental word learning was low. Can children and young people be sensitized to key unknown terms and trained how to use context effectively? Probably yes (Fukkink & de Glopper 1998). Exposure to new words, oral and written, in school offers an important source of later vocabulary acquisition.

. The challenges of later vocabulary learning Initial word exposures provide the basis for developing a semantic representation of a term but input is often insufficient to establish meaning and learning is not inevitable. In this section we consider how phonological form and semantic and morphological complexity impact on later word learning.





Julie E. Dockrell and David Messer

. Phonological processes in lexical acquisition As noted earlier, children’s ability to establish an initial phonological representation is central to the development of subsequent semantic representations. During early and middle childhood there is a close link between children’s ability to retain new phonological information for short periods of time and their vocabulary knowledge (Gathercole & Adams 1994; Gathercole & Baddeley 1989). The nature of the relationship between vocabulary and phonological memory tasks is a matter of debate. Phonological sensitivity can enhance the acquisition of phonologically unfamiliar words (Bowey 1996, 2001; de Jong et al. 2000), yet phonological awareness as such may not be independent of oral language skill (Cooper et al. 2002); that is, children with larger vocabularies construct more detailed phonological representations and may therefore be more successful on phonological tasks. The role of phonological regularities in lexical acquisition indicates that common sound sequences are learnt more rapidly than rare sound sequences, and that the word learning of older children (10- and 11-year-olds) continues to be influenced by phonotactic features (Storkel 2001; Storkel & Rogers 2000). Older children may use their receptive vocabularies as a scaffold to assist in the encoding and retrieval of non-words with word-like characteristics. When four-year-olds were exposed to similar-sounding words after they had heard a novel word, their production of the item was enhanced (Demke et al. 2002). Exposure after learning helped maintain the phonological traces of the new word in working memory, thereby leading to more durable long-term representations (Section 3.1). New words with many phonological neighbors (phonologically similar competitor words), then, should be learnt more quickly than those with fewer phonological neighbors (Metsala 1999). Phonological information alone is not sufficient to establish a lexical mapping. For example, phonological memory predicts the acquisition of explicitly taught lexical items but not of items introduced in an incidental fashion (Michas & Henry 1994). This is consistent with studies showing that schoolchildren’s ability to acquire the meaning of complex scientific terms was limited by difficulties in establishing the semantic representations of the new terms. No relationship was found between phonological memory and patterns of acquisition for comprehension, production, or assessments of wider domain knowledge (Braisby et al. 2001).

Lexical acquisition in the early school years

. Semantic factors in lexical acquisition Some of the issues concerning the acquisition and assessment of semantic representations are noted earlier (Section 3). In this section we consider the way that acquisition of the semantic representations continues to present challenges beyond the early years. A number of studies have pointed to the difficulties that children experience in developing the meanings of relational terms and mental state terms (Nelson 1996). In cases where there is a non-obvious relationship between a term and its possible meanings, the development of semantic representations is difficult and extended. The semantic representations of objects also change with development. The complexity of these processes is illustrated by a study of object naming and object knowledge in 288 children between the ages of 3;7 and 11;6 in relation to four different categories of object – implements, fruits, vegetables, and vehicles (Funnell et al. submitted). Different patterns of performance were evident for the children above 6;6 from those who were younger. For younger children, their ability to name exceeded their object knowledge, while for older children the reverse pattern was true, suggesting that the older children’s knowledge was more conceptually based. This shift in performance varied across category. Funnel et al. argue that older children often develop their knowledge of new objects in contexts where the object is not present. Thus, an older child may have a rich semantic representation for the word ‘yacht’ but be unable to accurately name a picture of a yacht. A change in representational status of lexical items is also supported by a study conducted by Keil & Batterman (1984). They presented children with verbal descriptions and then asked them to judge whether a particular term, such as ‘robber’ or ‘island’, could be applied to the description. One description contained characteristic information and the other description contained defining features. There was a focus on characteristic features during the preschool period moving towards an emphasis on defining features by the age of 9. This shift from individual exemplars to defining features occurs at different ages for different words, suggesting that experience and familiarity play a central role in the evolution of children’s understanding of terms. The nature of these changes depends on the semantic domains concerned. Words that stand for many meanings also pose challenges. Preschool children understand the primary meaning of polysemous words, but acquire secondary meanings gradually over subsequent years (Durkin et al.1985). This is consistent with more recent work where low-level meanings are mastered rapidly while higher-level meanings show a more protracted profile of devel-





Julie E. Dockrell and David Messer

opment (Booth & Hall 1995). Similarly, the overlap between psychological and physical properties (such as sweet, hard) is a particularly late attainment with mastery being achieved only in the mid-teens (Schecter & Broughton 1991). These studies demonstrate that when the cognitive system is taxed by the complexity of the semantic representations, lexical acquisition is a more protracted affair. . Morphological factors in lexical acquisition The structural complexity of word forms also influences children’s vocabulary acquisition. Compounding occurs when two or more root words are combined to create a new word. Relatively young children know a lot about the rules of compounding, yet this knowledge appears to be restricted to morphologically simpler combinations (Clark & Berman 1987). Morphologically more complex terms also take longer to acquire. Anglin (1993) studied children’s definitions for four different morphological relations – inflected words, derived words, literal compounds, and idioms. Skilled word learners appeared to be particularly good with literal compounds and could use derivational knowledge to infer the meanings of new words (Freyd & Baron 1982). The ability to use derivational morphology appears to be “a quite gradual development extending throughout the school years” (Anglin 1993: 33). This ‘morphological problem solving’ ability is influenced by two factors, the knowledge of root words and the productivity of the suffix (Bertram et al. 2000).

. Difficulties in vocabulary learning We have seen that phonological and semantic factors impact on later vocabulary learning. Studies of children with difficulties in vocabulary acquisition (Dollaghan 1987; Rice et al. 1990; Rice & Woodsmall 1988) offer further evidence of how these factors can impact on lexical development. The problems identified can be attributable to difficulties with either the process of acquisition itself or to a failure in consolidating and retaining information in the lexicon. In the following three sections we consider the evidence supporting these different views.

Lexical acquisition in the early school years

. Children with language difficulties Children with Specific Language Impairment (SLI) have language skills below the level that would be expected on the basis of their non-verbal abilities, and their level of language is not attributable to hearing or neurological disabilities (Bishop 1997). The problems that these children face in comprehending and using language are varied, and as yet there is no generally agreed typology of the condition. Given the varied nature of these children’s language problems it is significant that many appear to have difficulties in acquiring new words (Dollaghan 1987). A good example of the problems with lexical acquisition faced by children who have SLI has been reported by Rice (1990). She and her colleagues used a story presented via television that included novel words and assessed lexical acquisition by means of a comprehension test. The performance of children with SLI was compared with peers matched for chronological age (CA) and mean length of utterance (MLU). The 5-year-old children with SLI performed less well than both CA and MLU matched controls on comprehension tasks. Rice et al. (1994) report that children with SLI, unlike their CA and MLU peers, failed to acquire words after 3 exposures, but like peers were successful after 10 such exposures. Thus, children with SLI have a greater difficulty than their peers in acquiring new words and verbs pose particular problems (Kelly 1997; Windfuhr et al. 2002). It would appear that some children with SLI are less able to make use of word-world mappings. This is likely to make the further development of the language system during the pre-school and school years more difficult and protracted. The reasons why these children have difficulties with lexical acquisition are not yet understood. In the next two sections we consider findings which suggest that problems with lexical entries may occur as a result of inadequate phonological or semantic representations. . Lexical acquisition and phonological representations The ability to name pictures can be used as a basic assessment of whether a lexical item has been acquired (Section 2). Children with literacy difficulties make more errors on discrete picture naming tasks than skilled readers (Scarborough 1989; Snowling et al. 1988). Why should there be this link between lexical acquisition and literacy abilities? It has been argued that imprecise or inadequate phonological representations make it more difficult to learn the mapping between graphemes and phonemes (Snowling et al. 1988; Swan &





Julie E. Dockrell and David Messer

Goswami 1997). Evidence of imprecise or ‘fuzzy’ phonological representations is provided in these studies by a high rate of phonological errors when naming, more naming errors with longer than shorter words, and occurrence of more naming errors in comparison to children with similar semantic abilities. Thus, failure to acquire the full phonological specifications of words in the lexicon appears to have consequences that go beyond difficulties in oral communication. Low scores on phonological awareness tasks and on assessments of phonological short-term memory implicate the perception and storage of phonological information as the cause of the children’s difficulties. In other words, information-processing limitations interact with exposure to external information to influence lexical acquisition, and this has significant consequences for development. . Lexical acquisition and semantic representations Children with word-finding difficulties (WFDs) (German 1984) are able, on hearing a word, to identify the appropriate picture on a comprehension test, but are sometimes unable to produce the same word spontaneously during discourse or when presented with a picture of the item. Even when they do produce the appropriate word, it often has longer latency than in children of the same chronological or language age (Dockrell et al. 2001). A range of findings supports the view that WFDs are a consequence of inadequate or incomplete semantic representations (Kail & Leonard 1986). Seven-year-olds with WFDs have been found to give less accurate definitions of object names than control groups and provide fewer semantic features in their definitions (Dockrell et al. 2003). When producing verbs they make greater use of general all-purpose verbs and inappropriate choices from other semantic domains (Dockrell et al. 2001). Children with WFDs also perform poorly on tasks where they are asked to name as many items as possible that correspond to an identified target (Messer et al. in press). The problems of these children in accessing a name of an item can be attributed to a failure to fully acquire semantic representations. Research on a related group of children reinforces this interpretation. Studies by Nation, Snowling, and their colleagues have identified a group of children that they term ‘poor comprehenders’. Children with these characteristics who are aged 10 years have difficulties making use of semantic information when reading (Nation & Snowling 1998). They also tend to be slower and less accurate at picture naming than their peers (Nation et al. 2001). The children perform relatively well on phonological tasks and at decoding when reading. This suggests

Lexical acquisition in the early school years

that inadequate phonological representations of words is not the cause of their difficulties. Instead, their problems are attributed by Nation & Snowling (1999) to inadequate semantic representations. Thus, in this group of children, as in children with WFDs, it appears that impaired lexical acquisition results in difficulties when dealing with information where knowledge of semantic relations could help the children be more efficient and more accurate. These studies suggest that children’s development is put at risk by impairments involving lexical acquisition. The studies also suggest that these effects can occur in sub-components of the lexical acquisition process, those involving phonological information and those involving semantic information. . Summary This section has concerned challenges to lexical acquisition, and it contains a number of messages about the development of children with disabilities, about typical children, and about models of lexical acquisition. Children with SLI are less likely than language-age matched peers to be able to pick out the referent of words that they previously have heard in an appropriate context. Not only are such problems likely to directly delay vocabulary development, but the presence of an impoverished lexicon may effect further lexical acquisition (Section 3). New words may be more difficult to integrate into a less sophisticated lexical network. Other studies of the naming process indicate that there are children who have impoverished phonological or semantic information in their lexicon. The findings draw attention to the general consequences of a failure to acquire stable representations of a word, and raise questions about the mechanisms responsible for these difficulties. The findings also point to the vulnerability of specific components of lexical acquisition.

. Conclusions This chapter has considered the ways in which context and cognitive factors support later lexical learning. Word learning has sometimes been viewed as a simple word-to-world pairing that is established predominantly through access to the extralinguistic context. But such a view fails to consider the range of factors that have an effect on and play a part in acquisition. These include the more general social context of the situation, the linguistic information supplied with the new word, the sophistication of the child’s existing lexicon, and the child’s ability to detect and retain relevant information from the exposure to the new





Julie E. Dockrell and David Messer

word. A simple view about acquisition also fails to consider the nature and types of lexical information that children acquire and by corollary the need for a range of assessments to fully understand the nature of the acquisition process. As children progress through the school years it is clear that the increasing sophistication of the lexicon enables children to acquire new words more easily and with less need for contextual support. As a result, those children who have more developed lexical structures are better able to add to these already developed structures and make further gains. Another feature of development is that even though acquisition results in more information and increasingly complex information being stored in the lexicon, this is associated with easier access to the information. For example, children become quicker at naming, and better at providing more complex definitions. It also appears that the experience of schooling can facilitate these processes, though teachers do not always provide optimal circumstances for the learning of new words. Anglin stated, “Vocabulary acquisition is a remarkable process and one that we need to understand better” (1993: 185). To achieve this objective we need to collect information about the influences on lexical acquisition and to develop more comprehensive models of the process (Gentner 2003). Considerable advances have been made in understanding the acquisition of phonological information about words and developing models about this process (Section 4.1). In contrast, there are still many uncertainties about the nature of semantic and of morphosyntactic representations in the lexicon. Lexical acquisition is a complex process, and the lexicon increases in complexity as children become older. Further research is needed to produce a model of lexical acquisition that addresses the complexity of the process.

Note * We would like to thank Eve Clark and Ruth Berman for constructive criticisms of an early draft of this chapter. Discussions with Dr. Assimina Ralli and Dr. Rachel Best during the completion of their doctorates have further contributed to our understanding of the issues raised here.

Derivational morphology revisited Later lexical development in Hebrew Dorit Ravid

Introduction Language development is a lengthy process that interacts intensively with the acquisition of literacy. Language acquisition during the school years, here, ‘later language development’, takes place at all linguistic levels – lexical, grammatical, and pragmatic – and is accompanied by increasing metalinguistic awareness and more abstract and explicit representation of linguistic knowledge. This chapter concerns morpho-lexical knowledge, with the goal of investigating the connection between derivational morphology and development of the mental lexicon and delineating psycholinguistically motivated procedures for evaluating lexical development in different school-age populations. The chapter focuses on derivational morphology as a semi-productive domain that involves both lexical and pragmatic knowledge (Nagy et al. 1989). A richer and less obligatory domain than inflection, and one that depends directly on an extensive vocabulary, derivational morphology affords rich potential for linguistic evaluation of schoolchildren and adolescents who have already mastered obligatory inflections (Ravid 1995a). Additionally, as discussed in this chapter, derivational morphology requires integrated knowledge of the interrelations between lexical convention, semantic content, and formal structure – literacy-related domains that continue to develop well into school age. Yet to date, derivational morphology has been largely untapped as a source of information on later language development. The chapter starts with consideration of the mental lexicon (Section 1.1), followed by an overview of factors that contribute to later vocabulary growth (1.2) and discussion of ‘the literate lexicon’ (1.3). Part 2 focuses on Hebrew morphology (2.0), describing experimental studies with schoolchildren and



Dorit Ravid

adolescents compared with adults, on the one hand, and with their languageand learning-impaired peers, on the other, in three domains: passive constructions (2.1), derived nominals (2.2), and denominal adjectives (2.3). I conclude by considering implications of these Hebrew-based studies for general models of lexical and morphological development (3.0).

.

Psycholinguistic and developmental perspectives on the lexicon

This section reviews research relevant to later lexical development in relation to the mental lexicon (Section 1.1), the interface between cognitive development and literacy (1.2) and the nature of the ‘literate lexicon’ (1.3), as perspectives that go beyond insights from conventional printed dictionaries and accepted educational practices. . The mental lexicon The mental lexicon is the repository of words that speaker-writers make use of in language production and comprehension, and which they rely on in trying to make sense of novel linguistic input. The mental lexicon of a literate adult is a rich and powerful system, which provides the underpinning for conceptual and linguistic knowledge in any language. It contains tens of thousands of words standing for objects, people, and places, properties, activities, processes, and states, although what can be called ‘a word’ may differ across languages (Anderson 1985; Berman 2001). The ‘mental lexicon’ differs from conventional printed dictionaries, which fail to take account of psycholinguistic factors such as the ability of native speakers to recognize a word of their language and to reject non-word sound sequences, slips of the tongue and of the ear, tip-ofthe tongue phenomena, as well as priming and prototype effects and lexicaldecision tasks that show faster response times for high-frequency and familiar words (Aitchison 2003; Anderson 1995). These all testify to a key feature of the mental lexicon: the fact that words are not stored in separate compartments in the mind. Rather, they coexist in an elaborate network of both distinct and interrelated associations with conceptual, semantic, syntactic, morphological, and phonological representations that give rise to systematic connections of varying strengths between the items (Waxman 1994; Williams 1994). Current lexical models attempt to account for word learning, storage, and access by reference to the multiple relationships in the mental lexicon. A stepping stone model (Hastings & Lytinen 1994) relates two different, but not

Derivational morphology revisited

entirely separable aspects of words: the lemma – combining abstract meaning and word class – and the sound sequence. It enables new word learning from context, assuming that a large enough initial lexicon and parsing system are available. The waterfall model allows people still to be thinking about the meaning as they select the sound, taking into account that any specification is initially incomplete and usually undergoes frequent changes because of upcoming new information (Parnas & Clements 1986). All the information activated at the first stage is still available at the next stage, suggesting that word selection is not just a matter of following one word through from beginning to end, but of controlling and narrowing down a cascade of possible words that are related in various ways. A spreading activation model in which information can flow both ways captures both the way lexical knowledge is represented and also how it is processed (Anderson 1995). Concepts are represented in memory as nodes, and relations between the concepts as associative pathways between the nodes. When part of the memory network is activated, activation spreads along the associative pathways to related areas in memory. This spread of activation serves to make these related areas of the memory network more available for further cognitive processing. Speed and probability of accessing a word is determined by its level of activation, which in turn is determined by how frequently and how recently it has been used. The mental lexicon is thus a highly complex apparatus that interacts with phonology, morphology, and syntax, and is strongly dependent on semantic and procedural memory – encyclopedic, lexical, and conceptual (Baayen 1991; Talmy 2003). Lexical knowledge is a crucial component of any higherorder cognitive activity, and so is lexical development for academic achievement across the school years (Carlisle 2000; Mahony, Singson, & Mann 2000; Ravid in press). Research shows that during the period of later language development, vocabulary knowledge increases exponentially along with growth in morphological skills (Clark 1993; Nagy & Anderson 1984; Nagy & Herman 1987; Nippold 1998). By grade school, children have a large and varied vocabulary with complex hierarchical lexical and morphological connections (Anglin 1993; Carlisle 1995). Growing familiarity with written language both as a notational system and as a special discourse style contributes to increased ‘linguistic literacy’ during the school years, together with acquisition of less canonical structures typical of written language (Light & Littleton 1999; Ravid & Tolchinsky 2002). Anglin (1993), a pioneer in demonstrating the role of morphological analysis in children’s lexical expansion, showed that at least half the words they know are acquired through morphological form-to-meaning mapping. Today, the importance of lexical knowledge in later language acquisition and linguistic





Dorit Ravid

literacy in both typically developing and language-impaired children is widely acknowledged (Dockrell & Messer, this volume; Nippold 1998; Scott & Windsor 2000). This knowledge is critical for speakers of a highly synthetic language like Hebrew (Berman 1987c). . Cognition and literacy in later lexical development The core lexicon – the basic vocabulary acquired by children of preschool age – derives mainly from the spoken language. It then increases exponentially during the school years to become the advanced or literate lexicon, that is, the mental lexicon of educated adults. Children begin to learn words before their first birthday, and by their second they hoover them up at a rate of one every two hours. By the time they enter school, children command 13,000 words, and then the pace picks up, because new words rain down on them from both speech and print. A typical highschool graduate knows about 60,000 words; a literate adult, perhaps twice that number. (Pinker 1999: 3)

Development of the advanced lexicon reflects major cognitive and linguistic changes en route to maturely flexible language knowledge. True, children have access to the bulk of the morpho-syntax of their language before they reach school age. Yet a five-year-old hardly matches an adult or even a twelve-yearold in linguistic proficiency in general and lexical command in particular. One critical feature of this difference is the protracted acquisition of lexical items and morpho-syntactic forms that are rare in the spoken, everyday linguistic input to children, such as derived nominals (Berman 1997c; Ravid & Avidor 1998; Ravid & Cahana-Amitay in press). Another is the consolidation and diversification of less salient aspects of already existing linguistic systems as in the case of the English past perfect or middle-voice and passive verbs in the Hebrew system of binyan verb-patterns (Berman 1993b; Berman & Slobin 1994; Comrie 1986). A third related change is the increasing ability to make flexible use of varied linguistic resources for diverse communicative purposes (Ravid & Zilberbuch 2003a). Taken together, such maturation in knowledge and use of morpho-lexical structure underlies the adult mental lexicon, which is large in size and densely organized, yet flexible and accessible in use (Baayen 1991, 1994; Baayen & Renouf 1996). Features of linguistic structure and use are paced by at least two other factors – cognitive development and linguistic literacy – that account for differences between the lexical inventories and abilities of children and adults.

Derivational morphology revisited

Models of later cognitive development note that adolescents are increasingly able to handle abstraction, and have greater attentional, memory, and informationprocessing resources. In Piagetian terms, adolescents begin to represent and manipulate operations in order to formulate and test hypotheses and to deal with abstract concepts like probability, proportion, and analogies (Ginsburg & Opper 1988). Information processing models of cognitive development (Goel 1991; Velmans 1991) track changes in problem-solving behavior, decisionmaking, information gathering, and storage (Rumelhart & McClelland 1986) that yield a growing ability to take in and focus selectively on information, increased attention spans and capacity for memory storage. Recent research on executive control processes (Kluwe & Logan 2000; Rubinstein et al. 2001) focuses on processes underlying goal-directed action that control the preparation and execution of human activities. Executive control is shown to emerge later than basic cognitive abilities (Williams et al. 1999), underlying the ability to monitor and coordinate actions and thoughts. Taken together, these studies show that formation of the mature mental lexicon relies on later-emerging cognitive processes that enable storage, selection, and retrieval of lexical items in their appropriate contexts. A second factor in advanced lexical development is linguistic literacy, a constituent of language knowledge characterized by the availability of multiple linguistic resources and the ability to consciously access one’s linguistic knowledge and to view language from various perspectives (Ravid & Tolchinsky 2002). Developing linguistic literacy means increased control over a larger, more flexible linguistic repertoire and awareness of one’s own spoken and written language systems (Olson 1994). Being ‘linguistically literate’ involves a linguistic repertoire encompassing numerous registers and genres. Once literacy is established, it interacts with other components of linguistic knowledge to shape the key property of rhetorical flexibility: the ability to access and produce varied linguistic forms attuned to different communicative contexts. This adaptability develops together with other facets of advanced language knowledge and a growing ability to think about language so as to create “flexible and manipulable linguistic representations” for metalinguistic reflection (Karmiloff-Smith 1992: 32). Linguistic literacy evolves in the school years through growing command of written language, including exposure to expert texts, with use of textbooks, dictionaries, encyclopedias, the web and other reference sources providing experience with genre-specific constructions in authentic contexts (Light & Littleton 1999; Paltridge 1997). The language of school texts is typically formal in register with complex syntactic constructions (Abedi et al. 2001; Shalom





Dorit Ravid

2002). Higher-register lexical items and marked morpho-syntactic structures are in general typical of texts encountered in the course of formal schooling (Nippold 1998; Ravid & Shlesinger 1995). Since written language is betterplanned and more coherent than spoken language, and encourages editing and revision (Ochs 1979), schoolchildren have the opportunity to reflect on and retrieve appropriate lexical items and morpho-syntactic structures in the course of writing. . The literate lexicon Generative linguists have tended to make a sharp distinction between syntax as the primary language component and the lexicon as a secondary, rote-learned component (Chomsky 1970). But the study of real-world language in developmental perspective suggests that this distinction is less clear-cut. First, the lexicon is essential for development of syntax, since syntactic constructions depend crucially on (open-class) words for content. Second, the higher the occurrence of content words, especially nouns and adjectives, the higher the complexity of the syntactic architecture that frames them (Ravid et al. 2002; Ravid in press). Third, idiomatic constructions held in memory include syntactic phrases and whole clauses like and then some, the eleventh hour, wet behind the ears, it’s raining cats and dogs (Jackendoff 2002). In fact, the syntactic texture of mature language derives largely from the frequent occurrence of such ‘prefabricated’ multilexemic units (Wiktorsson 2000, 2003). Development of the advanced lexicon thus interacts with syntactic complexity, on the one hand, and with contextual factors of genre (narrative, descriptive, expository) and modality (speech or writing), on the other. The literate lexicon is encyclopedic in nature, and contains high-register words from various knowledge domains of a kind rare in everyday conversation, requiring specialized, school-type world knowledge (Baayen & Renouf 1996; Biber 1995; Ravid & Berman 2003a). Also, extant words expand their meanings metaphorically (Aitchinson 2003), particularly in restricted multilexemic contexts, e.g., a hot temper, the road to peace, jump the queue. The literate lexicon thus differs from the core lexicon in both size (Anglin 1993; Clark 1995) and quality. Conventional ‘words’, as items separated by spaces in print (Berman & Verhoeven 2002b), tend to be longer in letters, syllables, and morphemes (Smith 1998; Strömqvist et al. 2002). In English, this involves a move to multimorphemic derived words, often of Romance stock (Bar-Ilan & Berman, in revision; Carlisle 2000; Mahony et al. 2000); this is also true to some extent of Hebrew, where most content words are at least bi-morphemic,

Derivational morphology revisited

since even everyday nouns are internally complex, made up of a consonantal root and morphological pattern (see Section 2 below). More literate words are even longer and more complex, with multiple affixation expressing deverbal and deadjectival attributes and states, e.g., English ungentlemanliness, Hebrew yald-ut-iy-ut ‘child-hood-ish-ness = childishness’. Change in the quality of the advanced lexicon entails change in definition of a ‘word’ (Di Sciullo & Williams 1987), inter alia to accommodate multilexemic items like couch potato, drip-dry, change of heart, trial and error, while in Hebrew such lexicalized expressions are often derived from the scriptures and other classical texts, including words not used in current Hebrew (e.g., naxalat ha-klal ‘legacy-of the-whole = public property’, š’at néfeš ‘despisal-of soul’ = ‘repugnance’, ose nefašot ‘make souls = convert’, kše óref ‘hard-of-neck = stubborn’, kal da’at ‘light minded = frivolous’.) Even in everyday colloquial usage, the mature mental lexicon is rich in semi-lexicalized, rote-learned, habitually collocated multilexemic expressions. And it contains many ‘international terms’, particularly in academic and media discourse, in Hebrew in the form of loan words like manipulátsya ‘manipulation’, konkréti ‘concrete’, often with non-native multisyllabic stress, yielding lengthy consonantal roots that depart from more classical lexical norms, e.g., metalgref ‘send a telegram’, mesubsad ‘subsidized’ (Berman 2003a; Cohen-Gross 2000). The advanced lexicon is further enriched by items lying between the major categories of noun, verb, adjective and closed class function words – adverbials, connectives, and discourse markers like in short, for example, first and foremost, as a result. These specify and interrelate linguistic units and segments of discourse, and so enhance textual coherence and cohesiveness (Hickmann 2003) and provide personalized communicative perspectives on the referential content of linguistic materials (Jucker & Ziv 1998; Schiffrin 1995). Many of these intermediate items are syntactic constructions that have undergone functional and categorical changes (Brinton 1996; Ravid & Shlesinger 1999). Using them appropriately involves integrating morpho-syntactic, semantic, pragmatic, and social types of knowledge. Semantically, words in the advanced lexicon express less concrete, less familiar, and less ‘imageable’ concepts. Thus, conceptual development combines with a dramatic increase in words referring to abstract and categorical entities and to internal, cognitive and affective states. In Hebrew, as in other languages, these are typically derivationally complex (e.g., tna’im ‘conditions’, medake ‘depressing’, mezuyaf ‘hypocritical’, marshim ‘impressive’) and reflect complex semantic and structural properties in different lexical categories. Nouns show an increase in generic, categorical, collective, and abstract derived nominals





Dorit Ravid

(Ravid 2003); adjectives – more structural classes including denominals, resultatives, and diminutives, and those denoting cognitive and affective states (Berman, this volume; Ravid & Zilberbuch 2003a); and Hebrew verbs show increasing verb pattern diversity which has both morphological and semantic implications (Berman & Nir-Sagiv 2004; Rabukhin 2003).

. Developmental trends in the advanced lexicon of Hebrew This section analyzes findings for how speakers acquire an ‘advanced lexicon’ in different domains of Hebrew morpho-syntax: passive constructions, derived nominals, and denominal adjectives (Sections 2.1 to 2.3). As background, I start by outlining some Hebrew-specific features of the lexicon. Lexical knowledge depends on analytical insight into how derivational morphemes network in content words together with recognition of when to abandon analysis to accommodate idiosyncratic multilexemic elements (Aronoff 1976; Baayen 1991, 1994; Carlisle 2000). In Hebrew, the analytic task is compounded by a rich derivational morphology that encodes diverse semantic notions, combined with a complex array of root, stem, and affix allomorphy (Berman 1987c; Bolozky 1999; Ravid 1990). For example, in relation to the topic of word-length noted earlier, two Hebrew-specific phenomena have an effect on the literate vocabulary. One is use of bound suffixal morphemes as higher-register, more formal alternants of periphrastic alternatives (Schwarzwald 1998). For example, the word ma’asiyuto ‘his practicality’ consists of the stem ma’ase ‘deed, practice’ (from the root ‘-s-y ‘do, make’ plus an affixal pattern) plus the adjectival suffix -i to yield ma’as-i ‘practical’ and the abstract noun suffix -ut, to which is added the high-register bound possessive suffix -o. This alternates with the referentially synonymous but more colloquial ha-ma’asiyut shelo ‘the-practicality of-him = his practicality’. Another word-lengthening factor is that Hebrew words are often preceded by orthographically affixed morphemes standing for items that are separate words in European languages. For example, the string pronounced as uxeshebitmunata – meaning ‘and when in her picture’ and rendered in conventional Hebrew orthography as WKŠBTMWNTH – consists of four bound prefixal morphemes, the stem tmuna ‘picture’ and a suffix as follows: u+ke+še+be+tmunat+a ‘and+as+that+in+picture+her’ (Ravid 2001). In Hebrew as in other languages, increased length of so-called ‘words’ reflects the changing nature of the advanced lexicon, with more complex internal struc-

Derivational morphology revisited

Table 1a. Verbs related by roots k-l-t ‘take in, absorb’, and r-š-m ‘register, impress’ Verbs

Gloss

Pattern

Function

kalat rašam niklat niršam hiklit hiršim huklat huršam hitrašem

absorb write down, register be absorbed be registered record impress be recorded be impressed get impressed

P1 CaCaC

Basic, nonderived

P2 niCCaC

Passive, inchoative

P5 hiCCiC

Causative

P6 huCCaC

Passive

P7 hitCaCeC

Inchoative

Note: Verbs are entered here and across the paper in the morphologically simplex form of past tense, 3rd person, masculine, singular, 3rd person, unless otherwise specified.

Table 1b. Nouns and adjectives based on the roots k-l-t ‘take in, absorb’ and r-š-m ‘register, impress’ Nouns and Adjectives

Gloss

Pattern

Category

rašam koltan kalit rašum maklet miklat miršam taklit taršim klita rešima kélet róšem

registrar receptor absorbable registered receiver shelter registration (office) record chart absorption list input impression

CaCaC CoCCan CaCiC CaCuC maCCeC miCCaC

Agent noun Agent noun Potential adjective Passive adjective Instrument noun Place noun

taCCiC

Derived nominal

CCiCa

Derived nominal

CéCeC CóCeC

Derived nominal Derived nominal

ture, more morphemes, and more content per word, laying the basis for complex syntactic architecture. Hebrew makes use of two major devices for word-formation: non-linear root-and-pattern and linear stem-and-suffix structures. All verbs and most nouns and adjectives are constructed by Semitic root and pattern morphemes, e.g., the root s-g-r ‘close’ yields the verb sagar ‘to close’, the noun séger ‘closure’, and the passive resultative adjective sagur ‘closed, shut’ in three different patterns (Berman 1994; Ravid 2002). Consonantal root plus affixal patterns like





Dorit Ravid

Table 2. Linear stem plus suffix word-formation in Hebrew Hebrew

Derived Word

Gloss

mada-an iton-ay ir-iya pa’ot-on samxut-i kos-it lamdan-ut enoš-i-ut

science – AGENT newspaper – AGENT city-PLACE / COLLECTIVE toddler – PLACE / COLLECTIVE authority – ATTRIBUTE glass – DIMINUTIVE scholar – ABSTRACT human – ATTRIBUTE-ABSTRACT

scientist journalist city hall nursery school authoritative small wine glass scholarship humaneness

these subdivide verbs, nouns, and adjectives into syntactic-semantic categories clustering around shared roots. The seven binyan (literally ‘construction’) verb patterns indicate transitivity and valency-changing features of verbs such as causativity, reflexiveness, reciprocality, and voice (Berman 1993b), while the several dozen noun patterns indicate ontological categories (Clark 1993) such as agent, instrument, place, or abstract nominals. Table 1a, b illustrates this for two consonantal roots. Hebrew words can also be formed by linear stem-and-suffix combinations, e.g., mal’ax-i ‘angel-ic’. Linear derivation is based on a bound stem or free word (e.g., mada ‘science’, kos ‘glass’) plus a syllabic suffix (e.g., agentive -an, diminutive -it), yielding a concatenated form (mada-an ‘scientist’, kos-it ‘wine glass’). Stems often undergo morpho-phonological changes (e.g., késef/kaspi ‘money/financial’, cf. English five/fifth), but their boundaries mark discernibly distinct entities. This linear type of derivation is restricted to nouns and adjectives, as illustrated in Table 2. Below, we analyze two nonlinear late-emerging Hebrew constructions critical to both lexical and syntactic development – passives, associated with the binyan verb-pattern system, as in tigen/tugan ‘fry/be fried’ (Section 2.1) – and derived nominals, abstract nouns related to verbs that share a common root, e.g., the root h-l-x ‘go ‘walk’ yields nouns like halixa ‘walking’, hilux ‘(a) move’, tahalix ‘process’ (2.2); and one linear stem-plus-suffix construction, denominal adjectives (2.3). All three share the property of being ‘late-emerging’ aspects of Hebrew lexical development, and each has been the subject of specially devised experiments with native Hebrew-speaking school-age children.1 Participants’ ages are referred to by school grade, thus: 1st graders are 6–7 years old, 4th graders – 9–10, 10th graders – 15–16 years old, etc.

Derivational morphology revisited

. Passive constructions Passive voice is a means for demoting agency, by assigning the agent background status while foregrounding the event and/or the undergoer (Keenan 1985; Shibatani 1985). Passives also serve to express a distanced, less personalized ‘discourse stance’ in texts (Berman et al. 2002; Jisa et al. 2002b; Tolchinsky & Rosado in press). The non-canonical mapping of semantic content onto syntactic structure makes it a rather late acquisition in many languages (Budwig 1990; de Villiers & de Villiers 1985), explained in generative terms as due to the need to move NPs in the sentence (Borer & Wexler 1987; Pierce 1992; Pinker et al. 1987). Others suggest that passive development is delayed in certain languages (including Hebrew) due to its restricted occurrence in child-directed input (Berman, this volume; Gordon & Chafetz 1990; Tomasello et al. 1988), citing languages where the passive has higher frequency and is acquired much earlier (Allen & Crago 1996; Demuth 1989; Pye & Quixtan Paz 1988). Hebrew passives are formed in the verbal binyan system by the classical Semitic pattern affixation, e.g., tiken ‘fix’/tukan ‘be fixed’.2 They thus take the form of morphological rather than periphrastic passive formation with an auxiliary verb plus participle in European languages (Keenan 1985), and in this sense, they are formed similarly to other types of verbs in Hebrew. Like English, Hebrew has lexical or adjectival passives with participial forms (e.g., shavur ‘broken’, mesudar ‘tidy, well-arranged’) as well as verbal or syntactic passives (e.g., nishbar ‘got-broken’, yesudar ‘will-be-arranged)’. Syntactically, passive sentences are similar to English constructions, e.g., ha-texnay tiken et ha-radyo ‘The technician fixed the radio’/ha-radyo tukan al yedey ha-texnay ‘The-radio was-fixed by the-technician’. However, Hebrew has quite a few other devices for agent demotion, including numerous subjectless constructions that compete with the passive form (Berman 1980). Besides, only verbs that take direct objects passivize in Hebrew, so the construction is structurally more restricted than in a language like English (Berman 1978). Adjectival passives are acquired between ages 4–6 (Berman 1994; Ravid & Yagev 2003), but verbal passives, the focus of this section, take much longer (Berman 1997c; this volume). Like other word-formation processes in Hebrew, passive involves both structural and lexical knowledge, but its relative regularity, transparency, and predictability and the fact that it is a syntactic operation make it a unique meeting-point for morphology, syntax, and the lexicon. This section traces the acquisition of the morphological form of verbal passives and presents experimental data from a number of studies, mainly with gradeschool children.





Dorit Ravid 100 90 80 70 60 50 40 30

Repetition

20

Comprehension

10 0

Production K

Grade 1

Grade 3

Grade 4

Figure 1. Mean percentages of correct passives on repetition, comprehension, and production tasks (Ravid et al. 2003) in gradeschoolers

Figure 1 shows the correct results on repetition, comprehension, and production tasks with 60 kindergartners, 1st, 3rd, and 4th graders (Ravid et al. 2003). Children had to repeat sentences such as ha-tsiyur hudbak ba-maxbéret ‘The-drawing was-stuck into-the-notebook’; to produce the active forms of passive sentences, e.g., ha-tapuzim niktefu me-ha-ets ‘The-oranges were-picked off-the-tree’ >> mišehı kataf tapuzim me-ha-ets ‘Someone picked oranges offthe-tree’; and to produce passive sentences in response to their active counterparts, e.g., aba gihets et ha-xultsa ‘Dad ironed the-shirt’ / ha-xultsa guhatsa al yedey aba ‘The-shirt was-ironed by Dad’. Kindergarteners showed large between-task differences: Repetition started at about 80% in kindergarten and reached ceiling by 3rd grade; Comprehension was only 40% in kindergarten, improved drastically in 1st grade, and closed the gap with the repetition task by 3rd grade; and passive Production took off dramatically only in 3rd grade, with children still at only 70% success by 4th grade. What happens next? Results of a shorter and easier passive production task administered to 118 children and adolescents (Saban 1999), as shown in Figure 2, yielded somewhat higher scores, but here, too, learning to produce passive voice occurs across grade school, with a cut-off point around 5th grade, and reaches ceiling only in high school.

Derivational morphology revisited

100 90 80 70 60 50 40 30 20 10 0

K

Grade 2

Grade 5

Grade 7

Grade 11

Adults

Figure 2. Mean percentages of correct scores on a passive production task (Saban 1999) across the school years

Figures 3 and 4 present data from four different studies using the same passive task as in Saban with different populations. Lapidot (2002) tested 24 gradeschoolers with SLI, showing an increase from over 40% in 2nd to 60% in 4th grade. In Appel-Mashraki’s work (2001), 51 4th graders from average to low SES backgrounds reached the same 70% as the children in the Ravid et al. study (Figure 3). Geiger’s (2000) three-month intervention study used linguistic humor as a tool to improve morphological skills in 4th graders, with a focus on derivational morphology. Figure 4 shows that the high-SES 4th graders in Geiger’s intervention study reached the same 70% in the pre-test. The post-test showed a significant increase to over 90%, even though passive was not one of the structures directly taught on the intervention. Finally, Stiassnie (2004) tested 16 5th graders with Auditory Perception Deficit compared with their typically developing peers. The children with APD reached about the same scores as the younger children in the Appel-Mashraki and Geiger studies, lower by about 20% than the same age group in Saban’s study. However, the normally developing control group reached exactly the 90% mark as Saban’s 5th graders. Each of these studies reveals the multifunctional nif’al verb pattern as relatively early acquired with both the highest scores and the most errors (see, too, Schwarzwald 1981). As such, out of the three passive binyan patterns, nif’al forms constitute a bridge between fully syntactic passives and intransi-





Dorit Ravid

100

71.8

90

62.85

80 70

46.18

60 50 40 30 20 10 0

Grade 2 SLI

Grade 4 SLI

Grade 4 Control

Figure 3. Mean percentages of correct passive scores in 2nd and 4th grade children with SLI (Lapidot 2002) compared with typically developing 4th graders from a middle/low SES background (Appel-Mashraki 2001)

92.08

100 90 80

93.14 74.216

68.47

70 60 50 40 30 20 10 0

Grade 4 Pre

Grade 4 Post

Grade 5 APD

Grade 5 Control

Figure 4. Mean percentages of correct passive scores in 4th graders, pre- and postintervention (Geiger 2000); and in 5th graders with APD compared with a control group (Stiassnie 2004)

Derivational morphology revisited

tive, inchoative or change-of-state verbs, along lines consistent with studies on Hebrew verb usage in conversational and narrative contexts (Berman 1993b, this volume; Berman & Slobin 1994). Passive verbs, as noted, are closely linked to syntactic context, and yet constitute a focal area of later lexical development in Hebrew. They emerge with the onset of literacy skills and consolidate over the grade school years. Their acquisition throws light on the lexically and syntactically denser language knowledge of pre-adolescents and their increased ‘rhetorical flexibility’, in the sense of access to and the ability to deploy a range of alternative linguistic constructions. . Derived nominals Acquisition of ‘derived nominals’ (DNs) can be defined as the heart of later lexical development. This heading goes beyond nominalized abstract nouns derived from verbs (Chomsky 1970), to include the entire range of nouns related to verbs and adjectives with which they share a root or stem (Comrie & Thompson 1985), like English gift as well as giving for the verb give, or delivery, deliverance as well as delivering from deliver. In Hebrew, a subset of DNs, so-called shmot peula ‘action nominals’, are related to verbs in a formal and systematic way, through binyan-specific patterns, e.g., gdila ‘growing’ in CCiCa (in the so-called P1 conjugation), gidul ‘growth’ in CiCuC (P3), and hagdala ‘magnification’ in haCCaCa (P5); but Hebrew also contains numerous other nominal patterns that are more or less systematically related to verbs (Berman 1976; Ravid 1999; Schwarzwald 2002). Together these create morphological chains of verb-related DNs whose meanings range from the abstract, regular, semantically transparent, compositional gerundive ‘the act or the state of Ving’, across activities, states, and events to idiosyncratic, concrete, discrete, and countable meanings (Zubizarreta & van Haaften 1988; Zucchi 1993). For example, derived nouns related to the Hebrew verb halax ‘go, walk’ include the regular action nominal halixa ’walking’, and also hilux ‘move’, halix ‘procedure’, tahalix ‘process’, tihalux ‘processing’, mahalax ‘move’, tahaluxa ‘parade’, haloxrúax ‘way-of spirit = popular opinion’, halaxa ‘religious law’. Chains of nouns referring to verbal entities are very common in the Hebrew lexicon (Ravid 1999; Seroussi in press). Derived nominals play a central role in later language development (Ravid & Avidor 1998). They demand a firm grasp of Hebrew verbal and nominal morphology; they involve command of lexical items referring to a wide array of concepts and entities (Seroussi 2003); and they function importantly





Dorit Ravid

in construction of syntactically dense NPs, of complex subordinating constructions, and multilexemic prefabricated units (Ravid et al. 2002; Ravid & Cahana-Amitay in press). Relatedly, nominalizations inside of compounds allow speaker-writers to express a relationship between processes that can be backgrounded in constructing an argument to advance, package, and integrate the flow of information in discourse (Halliday 1988). It is not surprising, then, that DNs are less common in everyday spoken usage, but that they figure widely in maturely literate written texts, especially in the expository genre (Ravid 2003, in press). Focus here is on occurrence of DNs in experimental contexts during the school years, based largely on ‘the Machine Test’ as an effective means for eliciting derived nominals in Hebrew. This requires subjects to complete a noun compound with the bound feminine noun stem mexonat ‘machine-Gen’ as the initial, head element and to give a derived nominal as the second, modifying element. For example, ‘a machine that loses – source-verb mafsida-Fem – all the time is mexonat- . . . for the expected response mexonat hefsed ‘machine-of loss = losing machine’.3 The task consisted of 20 items, both morphologically regular and irregular, administered to 100 participants across the school years (Ravid & Avidor 1998). Responses were scored for accuracy, that is, for example, hefsed ‘loss’ rather than regular, well-formed but non-occurrent hafsada. The ‘total success’ curve in Figure 5 indicates a protracted period of learning that reaches only around two-thirds among adults, with a marked difference between two groups of items. Transparently predictable and regular items that match a small set of four active binyan patterns start below 50% in kindergarten and reach around 90% in the adults; in contrast, the irregular, unpredictable items formed in a range of idiosyncratic form-meaning matches show a gradual learning curve, up to only 55% in the adult group. This contrast illustrates the two interdependent kinds of knowledge involved in learning derived nominals: systematic morphological patterning, well in place by kindergarten and maturing over the school years, versus sporadic lexical knowledge that forms part of the ‘advanced vocabulary’ which continues to develop across the life span. Seroussi’s (2003a) study modified the Machine Test to include only 15 items, administered to adolescents from grade 6 to adulthood. As shown in Figure 6, Seroussi’s rather less challenging version yielded similar scores for 6th graders, but higher scores for high school and adults. Items were scored twice – leniently, as an average on a scale, including erroneous forms (taken to reflect overall morphological ability), and stringently, labeling as correct only the targeted nominal (to reflect lexical knowledge). For

Derivational morphology revisited 100 90 80 70 60 50 40 Total

30

Regular

20

Irregular 10 0

K

Grade 3

Grade 6

Grade 10

Adults

Figure 5. Mean percentages of correct scores on the machine test across the school years (Ravid & Avidor 1998)

100 90 80 70 60 50 40 Stringent

30

Lenient

20 10 0

Grade 6

Grade 8

Grade 11

Adults

Figure 6. Mean percentages of correct scores on the machine test across the school years (Seroussi 2003)





Dorit Ravid

100 90 80

80.37 66.11

82.22

66.8

70 60 50

Comprehension

40

Production

30 20 10 0

Pre intervention

Post intervention

Figure 7. Mean percentages of correct comprehension and production of derived nominals in 4th grade (Geiger 2000)

example, non-existent but possible hafsada ‘losing’ instead of required hefsed ‘loss’ would be credited on the lenient scale but would score 0 on the stringent scale. The lenient curve in Figure 6 shows that by 6th grade, children have good morphological ability to construct DNs, but are still only at the start of lexical learning of these items, reaching cut-off only in high school. Geiger’s (2000) study tested 4th graders on comprehension and production of derived nominals pre- and post-intervention. The comprehension task was a shortened version of the Agent Task from Ravid and Avidor (1998), for example, “what does a person do who hunts (tsad) – engages in. . . hunting (tsáyid)?” The production task was a shortened version of the Machine Test, with lenient scoring. Although the intervention process did not include work on nominalization, Figure 7 shows that 4th graders significantly improved by about 15% on both comprehension and production, suggesting that analytical morphological skills in one domain may improve through explicit work on other domains. Figure 8 shows how lexical and morphological knowledge of nominalization develops from beginning to end of gradeschool (Avivi-Ben Zvi 2002) on a task where children were asked to produce DNs in modified versions of the Sensations and Agent Tasks (What would you call the action / sensation of a person who Vs?). Findings again show that even 1st graders have adequate analytical morphological skills to approach nominalization, and that this ability greatly improves in grade school. But the mental lexicon of 1st graders does not

Derivational morphology revisited

100 90

74.17

80 70 60 50

42.22 Lenient

35.14

Stringent

40 30 20

4.44

10 0

Grade 1

Grade 6

Figure 8. Mean percentages of correct production of derived nominals (lenient and stringent scoring) in 1st and 6th grade (Avivi-Ben Zvi 2002)

contain derived nominals, while the scores of 6th graders still show the same rather low level as those in the Ravid and Avidor study. A final analysis of DNs is shown in Figures 9 and 10, for application of derived nominal tasks to special populations. On the left are shown correct scores (lenient) on a task requiring the production of a derived nominal from a given verb administered to 3rd to 4th graders with SLI compared to age-matched, normally-developing peers and to younger, language-matched controls. While normally-developing 3rd to 4th graders scored about 60% on this slightly more difficult task, younger 2nd to 3rd graders scored far lower, with the task proving almost impossible for 3rd to 4th graders with SLI. The 7th graders with learning disabilities (Putter-Katz et al. in press) scored much lower on Seroussi’s short version of the Machine Test than all the other same-grade groups. Finally, the 5th graders with APD in Stiassnie’s (2004) study achieved the same level of lenient scores as the controls did on stringent scores (Figure 10). Here, too, the difference between analytical morphological skills and lexical knowledge was apparent in both groups, with the scores of the normally developing children very similar to those of Seroussi’s (2003) study. Several observations emerge from these studies of derived nominals. In terms of learning strategies, prior to consolidation of both formal and lexical knowledge, children’s errors indicate reliance on phonological cues and perceptual similarities. For example, instead of relating the noun te’ura ‘lighting’





Dorit Ravid

100 90 80 70

60.5

60.5

60 50 35

40 30 20

14.5

10 0 Grade 3/4 age matched

Grade 7 LD

Figure 9. Mean percentages of correct production of derived nominals in children with SLI compared with age-matched and with language-matched peers (Ravid et al. 1999); and in learning-disabled 7th grade girls (Putter et al. in press)

86.107

100 90 80

72.563

69.789

70 48.954

60 50 40 30 20 10 0

APD Lenient

Control lenient

APD Stringent

Control Stringent

Figure 10. Mean percentages of correct production of derived nominals in 4th and 5th grade children with APD and control peers – stringent and lenient analysis (Stiassnie 2004)

Derivational morphology revisited

to the verb he’ir ‘give light’, many 7th and 8th graders recalled the unrelated verb te’er ‘describe’; and instead of relating taxanunim ‘pleading’ to hitxanen ‘plead’, they connected it to tixnen ‘plan’. In terms of the leitmotif of this chapter, DNs are a late-emerging linguistic phenomenon, learning of which takes place mostly between middle childhood (4th grade, age 9 to 10) and adulthood. The combination of abstract concepts, verbal content expressed in nominal form, and morphological variety and inconsistency together contribute to the delay in acquisition. This in turn mitigates against the complex syntax typical of proficiently mature, literate texts in the context of extended discourse, where nominalizations figure prominently (Ravid & Zilberbuch 2003a, b; Seroussi in press). In methodological terms, structured elicitation tasks of derived nominals such as the Machine Test are good diagnostic tools of cognitive and language difficulties in later childhood and adolescence, when easier tasks no longer differentiate across populations. In principle, derived nominals call on both types of knowledge involved in the ‘literate lexicon’ – mastery of the relevant morphological devices and fine-tuned knowledge of the conventional lexicon. Both are needed for retrieving the exact nominal response to a verb or adjective in the face of morphological irregularity, and for latching onto the interrelated networks of ‘word families’ that form part of a mature lexicon. . Denominal adjectives Denominal i-suffixed adjectives (DAs) constitute the most productive type of adjective formation in Modern Hebrew, e.g., xašmal-i ‘electr-ic’, iš-i ‘person-al’, comparable to English adjectives of Romance origin like effective, domestic, industrial. Structurally, DAs involve the linear formation of a nominal stem (e.g., xašmal ‘electricity’) and an adjectival suffix, often with a morpho-phonological change in the stem, e.g., báyit ‘house’ → beyt-i ‘domestic’ (Ravid 2004). The challenge posed by DAs is not, however, mainly morpho-phonological, since the kinds of stem changes that nouns may undergo in the shift to adjectival status are common to a range of morphological operations on noun stems in Hebrew, including: number and gender inflection, compound and genitive formation, and numerous other inflectional and derivational processes of suffixation (Ravid 1995b; Levin et al. 2001). What is relevant is the rather complex semantics of DAs: in forming a denominal adjective, the base noun must be ‘dissolved’ into its component features to select the specific property that will carry over to the derived adjective. This is not always predictable, as in other cases of denominal derivation (Clark & Clark 1979). The problem thus resides in the semi-productive, opaque derivational character of these adjectives, es-





Dorit Ravid

pecially when they enter an NP-modifying construction (N-Aden ). Compare, for example, the specific meaning of beyti ‘domestic’ from báyit ‘house’ in óxel beyti ‘food domestic = home-cooked food’ and in tipus beyti ‘type domestic = a stay-at-home type’. Besides, N-Aden s typically designate complex sub-categorization to abstract entities (e.g., tafkid tsiburi ‘public role’, tšuva xelkit ‘partial answer’) – in the Hebrew order of Noun Adjective. As such, they constitute a structural alternative to Noun Noun compounds – in the same order of Head Modifier (e.g., compare the N-Aden s string bad mešyi ‘silken fabric’ with the N-N compound bad méši ‘silk fabric’). Interestingly, the more classical Noun-Noun compound construction is far more prevalent in extended discourse (Ravid & Zilberbuch 2003a, b) and the comparison between the two constructions contrasts interestingly with their English counterparts (e.g., water bug ∼ acquatic insect, as discussed in Berman 1988c). DAs, and especially those in the NP-modifier function, are typical of higher-register, written Hebrew, such as literary prose and journalistic and expository texts (Ravid & Zilberbuch 2003b; Shlesinger & Ravid 2003). Apart from lexicalized forms such as xagigi ‘festive’, they are completely absent from child-directed speech. Accordingly, denominal i-suffixed adjectives are the last type of adjectives to emerge in Hebrew child language around age 6, often in the form of lexical innovations, e.g., childish xatani ‘groom-like’ (Berman, this volume; Ravid & Nir 2000). Longitudinal observation reveals a few examples where young children produce unconventional N-Aden compounds, either morpho-phonologically or lexically ill-formed. For example, Assaf (5;2) described a sports car (adult N-N compound mexonit∧ sport) by the N-Aden compound óto spórti ‘sportive car’, and at age 4;9 he termed a mountainous area ezor hari for adult ezor harari (from har ‘mountain’); and Sahar (6;8) defined crying about a funny situation as béxi cxoki ‘laughy crying’. Recent work on the distribution of adjective categories in text production shows that DAs have high frequency in expert texts, especially in complex NP constructions, but in texts produced by non-experts, they emerge only around 11th grade, confined mainly to expository-type essays (Rabukhin 2003; Ravid & Zilberbuch 2003a, b; Shlesinger & Ravid 2003; Zilberman 2003). Experimental studies of DA development used tasks that required changing a noun into an adjective, e.g., xiyux shel mal’ax hu xiyux mal’axi ‘the smile of an angel is an angelic smile’. Figure 11 shows DA production in the Levin et al. (2001) longitudinal study from kindergarten to first grade. The improvement is interesting, since children at this age range are not exposed to the types of texts typically rich in DAs, nor do they receive instruction in analytical morphological skills. Other

Derivational morphology revisited

100 82.78

90

71.11

80 70 60 50 40

40.5 34.17

33.75

30 12.22

20 10 0

K

Grd 1

Grd 1 lenient

Grd 1 strict

Grd 6 lenient

Grd 6 strict

Figure 11. Mean percentages of correct production of denominal adjectives in kindergarten and grade 1 (Levin et al. 2001); and from 1st to 6th grade (Levie 2002)

data show an increase in DA production between 1st and 6th grade (Levie 2002); when scored leniently as an average on a scale, children display good morphological skills from early on, reaching 90% by the end of grade school, but when scored stringently for accurate production only, first graders show almost no lexical knowledge of DAs, and this is not mastered by the end of grade school. These studies were extended to children with special needs, as shown in Figures 12 and 13. Figure 12 shows that 2nd graders with SLI (Lapidot 2003) have much lower scores than normally developing 1st graders, nor do they improve with age and schooling by 4th grade – unlike their improvement on passive formation. The normally-developing 4th graders from a middle to low SES background (Appel-Mashraki 2001) and the control 5th graders in Stiassnie’s (2004) study scored between 50–60%, as can be expected from the scale set by Levie’s (2002) study. But the 5th graders with APD did much worse, close to the 1st graders in Levin et al. According to Figure 13, normally developing children improved from about 70% in 2nd to 3rd grade to over 80% by 3rd and 4th grade, but SLI children scored significantly less than their age-matched peers and than younger normally developing children (Ravid et al. 2003). All analyses show that DAs which do not require stem change are easier to produce (com-





Dorit Ravid

100 90 80 62.7

70

51.563

60 50 32.813

40 30

20.83

16.67

20 10 0

Grd 2 SLI

Grd 4 SLI

Grd 4 ND

Grd 5 APD

Grd 5 control

Figure 12. Mean percentages of correct production of denominal adjectives in 2nd and 4th grade children with SLI (Lapidot 2002); Normally developing 4th graders (AppelMashraki 2001); and 5th graders with APD compared with controls (Stiassnie 2004)

pare dimyon / dimyon-i ‘imagination / imaginary’ and késef / kasp-i ‘money / monetary’). The experimental evidence shows that denominal adjectives emerge earlier than derived nominals, yet they are harder than passive constructions, which have less complex semantics. Rule-bound morphology is fully in place by the end of grade school, but lexical convention and semantic irregularity take longer to consolidate, while data from extended discourse demonstrate that it takes until late adolescence before DAs are correctly produced in a context-appropriate fashion. Although structurally transparent, DAs are not easily mastered, since – like all adjectives – they lexicalize in relation to nouns (Markman 1989). Besides, their favored ‘landing site’ is constituted precisely by high-register, lexically specific derived nominals in association with which they label complex sub-categorizing entities – such as hitnahagut koxanit ‘powerdriven behavior’ from koax ‘strength’, miflaga yemanit ‘rightwing party’ from yamin ‘right’ (Shlesinger & Ravid 2003). Given their complex semantics and their proclivity for lexicalization, it is no wonder that DAs are highly diagnostic for gradeschool children with cognitive and language disabilities. Their low scores indicate that their ability to acquire new words, a critical factor in later-

Derivational morphology revisited

100 90

82.57

80

69.71 64.57

70 60 50 40 30 20 10 0

Grade 3/4 SLI

Grade 1/2 language matched

Grade 3/4 matched

Figure 13. Mean percentages of correct production of denominal adjectives in children with SLI (grades 3–4) compared with age-matched normally and language-matched peers (Ravid et al. 2003)

language development, is impaired, and that they demonstrate difficulty with productive application of morphological knowledge (Nagy et al. 1989).

. Conclusions This chapter has focused on three late-emerging forms: passive constructions, derived nominals, and denominal adjectives. Below, I try to show in what sense all three share features that delay their acquisition to gradeschool and beyond (Ravid 1997). They are higher-order and low-frequency structures as defined by Nippold (1998), rare in the spoken input to children (Snow 1995). They are pre-empted by more transparent and accessible, early-emerging, alternative structures. The forms considered demonstrate low transparency and iconicity, and they require prior integration of lower-order linguistic forms and constructions. Further, all three conspire to express the type of distant, depersonalized, and abstract ‘discourse stance’ characteristic of written school-type academic or expository style (Berman et al. 2002), and absent from the input to children.





Dorit Ravid

In Hebrew as in most languages, passives do not occur in child-directed speech. They are pre-empted by active constructions available to children early on (Berman 1993b) that express dynamic scenarios iconically, e.g., ha-yalda daxafa et ha-agala ‘the-girl pushed ACC the-wagon’. Active constructions are lower-order and iconic in the sense of being dynamic, agent-oriented, and highly transitive (Hopper & Thompson 1980). They clearly identify agent with syntactic subject, mark agent/subject by verb inflection, and direct objects by the accusative marker et. In contrast, passive constructions have a low transitive value, are agent-demoting, identify syntactic subject with semantic patient, and inflect verbs accordingly. Not only is such oblique expression inaccessible to young children, it is also absent from child-directed speech, and is not typical of the narrative genre that is the main source of written discourse style in the input to children (Ravid & Tolchinsky 2002). The easy availability of generic and impersonal subjectless constructions in the spoken Hebrew vernacular (Berman 1980, 1990; Ravid 1995a) further delays acquisition of the passive construction. Passive voice is expressed in Hebrew through the binyan verb pattern system, and so poses no special morpho-phonological problem to children, who can manipulate this system by age four. But development of the verbal passive is mediated by the earlier acquisition of adjectival passive forms (e.g., banuy ‘built’, muklat ‘recorded’) that require fewer syntactic operations and have different pragmatic functions. Adjectival passives, in the form of presenttense passive verbs, consolidate between the ages four to six (Berman 1994) and serve as a bridge to acquisition of genuine non-participial passives by the end of grade school. But spontaneous use of passives in naturalistic production is delayed even further to late adolescence (Seroussi in press): It awaits onset of the mature ability to take varying perspectives on the same scene in the communicative context of abstract academic discourse. Derived nominals, like passives, are largely lacking in child-directed speech, except for lexicalized forms such as hafska ‘school-break’ or tiyul ‘hike’. As abstract and morphologically complex nominals, they are pre-empted by verbs (and adjectives), which are the more natural choice for expressing verbal or predicative content (Ravid & Avidor 1998). Verbs contribute to a personalized, concrete, and dynamic discourse stance both semantically and by marking syntactic subjects clearly, while derived nominals contribute to a detached, abstract, and static stance (Berman et al. 2002). Derived nominals require integration of two types of prior knowledge: (1) the rich and complex aspectual distinctions expressed in the binyan verb conjugation system which underlies most derived nominals; and (2) the syntactic system of complex nominals and

Derivational morphology revisited

compounding structures as favored sites for derived nominals (Berman 1988c; Ravid & Shlesinger 1995). In fact, Ravid & Cahana-Amitay’s (in press) analysis of personal-experience narratives shows that only in late high school do derived nominals emerge alongside verbs and adjectives to express predicative content, such as gneva ‘theft’, hit’alelut ‘abuse’, and these are nearly always embedded in complex syntactic constructions. Seroussi (in press) and Ravid (2003) likewise demonstrate the long developmental path to mastery of derived nominals (Berman, this volume) in learning to construct texts, in mastering complex NPs, and in alternating use of these forms to suit contextual factors like text genre and speech versus writing. Denominal adjectives (e.g., klali ‘general’) constitute part of what Ravid (2004) terms ‘secondary’ or ‘word-level’ morphology in the sense that they require already extant words to operate on. In this sense, Hebrew speakers must get a good grasp on the more basic root-and-pattern morphology for constructing nouns before they can tackle further derivational noun-based operations. Denominal adjectives are pre-empted by the nominal system in yet another way, by the Hebrew compounding system which expresses the necessary notions in less dense and more transparent ways, and which is acquired between ages four to six (Clark & Berman 1987). For example, instead of sxiya leylit ‘a nocturnal swim’ with a denominal adjective, first graders said sxiyat layla ‘a night swim’ using two nouns in a compound construction (Ravid 1997). Studies of developing text production abilities and analyses of expert texts show clearly that denominal adjectives do not emerge in actual usage before teenage, and that they always characterize written expository texts, as the hallmark of a detached, static, and abstract stance (Rabukhin 2003; Ravid in press; Ravid & Ziberbuch 2003b; Shalmon 2003). Most interesting, recent work shows that denominal adjectives very often tend to gravitate towards modifier position in complex NP constructions headed by abstract, derived nominals (Shlesinger & Ravid 2003). This means that not only do later-acquired morphological constructions cluster together to form complex syntactic structures, but also that these complex clusters further delay their acquisition. This chapter has gone beyond traditional investigation of child language to present experimental findings for a range of morphological constructions as evidence for the emergence and consolidation of the advanced lexicon of Hebrew. Over childhood and adolescence, children have numerous opportunities to learn the intricacies of the inflectional and, usually rather later, the derivational morphological processes in their language. Yet extreme and/or idiosyncratic stem changes and a high degree of lexical specificity in formmeaning mappings are difficult even at high school age. We have seen that





Dorit Ravid

the internal morphological structure of Hebrew words and the syntacticosemantic classification of affixes in nouns, adjectives, and verbs play a major role in lexical knowledge and development. The vocabulary of adults, high school students, and even grade school children is thus very different from the core preschool lexicon. As discussed in the introduction, factors that distinguish early from late language development in this respect include cognitive developments, acquisition of literacy, and extended real-world experience and school-based learning, which combine to create a richly packed and highly complex yet accessible body of knowledge. Although the chapter focuses on Hebrew-specific structures, its findings have more general implications for later language development. The constructions under investigation may be matched by corresponding, though not identical, low-frequency constructs that characterize the advanced or literate lexicon in other languages and are targeted in high-register, academic usage, such as Romance vocabulary in English (Bar Ilan & Berman, in revision; Smith 1998), or written forms of the classical language in Arabic (Holes 1995). Some general conclusions emerge from the findings surveyed in this chapter. At the conceptual level, they provide psycholinguistic evidence to support Berman’s model of ‘a confluence of cues’ in language acquisition and development (Berman, this volume; Berman & Verhoeven 2002a): Later lexical development is realized by a range of different devices and subsystems which are often considered in isolation in linguistic analysis – verbs, nouns, and adjectives. From a linguistic perspective, the chapter argues for a rich interaction between linguistic components, most particularly the lexicon/syntax interface, by showing how lexical growth and diversification, the consolidation of new, more complex morphological devices, and syntactic development go hand-inhand. At the methodological level, I have tried to demonstrate where and how we need to go about testing language constructions in different age groups, levels of schooling, and populations. Derivational morphology proves to be an important component of language development and linguistic knowledge that has not received adequate focus to date. And it adds an important facet to the picture that is emerging of the changes in children’s and adolescents’ language during the school years.

Derivational morphology revisited

Notes . Details on research method and statistical analyses are given in the studies cited. . When cited out of context, Hebrew verbs are given in the morphologically simplex form of past tense, 3rd person masculine singular. . English and Hebrew contrast markedly in this respect. English has numerous compounds with agent nouns as the head (e.g., fireman, policeman, washerwoman), but these are rare in Hebrew, which has numerous instrument noun compounds (e.g., mexonat ktiva ‘writing machine = typewriter’, mexonat yibush ‘drying machine = clothes dryer’, mexonat yeria ‘shooting machine = machine gun’). See Clark & Berman 1984.



Discursive constraints on the lexical realization of arguments in Spanish* Ekaterina Khorounjaia and Liliana Tolchinsky

.

Introduction

In natural language use, messages contain both new and given information. If we say only what our interlocutor knows, the communicative act is pointless and boring, while if we say only unknown things, communication becomes impossible. As a result, we usually combine the two kinds of information in a single message. On the other hand, people tend to minimize the amount of new information they transmit in an utterance. This is because processing new information is highly demanding (Givón 1975): it needs to be activated, since it is provided for the first time by the interlocutor or is currently in the listener’s long-term memory. As a result, new information recruits more cognitive resources than given information, which is ‘active’ and readily accessible, since it is in the listener’s focus of consciousness (Chafe 1987).1 Moreover, speakers tend to distribute new and given information in different syntactic positions of an utterance, apparently in order to facilitate processing. For example, new information is usually introduced in object rather than in subject position. The study described in this chapter examines the relationship between information flow and the syntactic patterns of argument structure. Following the work of Du Bois (1987), we argue that the choice of argument position for introducing new information is not random, but is governed by discursive constraints. Do children need explicit instruction on these patterns of information distribution, or do these patterns evolve as part of their discursive development? How specific are these patterns to particular communicative contexts? To address these questions, we explored how patterns of new and given information are distributed in oral texts produced by Spanish-speaking children,



Ekaterina Khorounjaia and Liliana Tolchinsky

adolescents, and adults, when telling a personal-experience narrative and when discussing the topic of interpersonal conflict. Research on text production in young children reveals early sensitivity to genre differences (Hicks 1991; Hudson & Shapiro 1991; Sandbank 2002). Studies in different languages indicate that 9-year-old fourth-graders distinguish narratives from expository texts at all levels of linguistic expression – lexical, phrasal, and clausal, as well as discursive (Berman & Nir-Sagiv 2004). The questions addressed here in this respect are: Does genre affect the relationship between information status and argument structure, or does information manifest itself similarly irrespective of genre? The goal of this study is twofold: to explore whether choices of argument structure made by schoolchildren and adolescents observe the same constraints as govern adult usage in different languages, and to examine the effect of discourse genre on these choices. . Argument structure As background, we start by discussing the notion of argument structure, the key theme of the present study. A predicate expresses a relation (or relations) among participants, termed the arguments of the predicate. Consider the examples in (1): (1) a. b. c. d. e. f.

The teacher put the book on the table The clown laughed She suggested this idea *The teacher laughed the book on the table *The clown put *She put this idea

In these examples, the predicates are the verbs put in (1a), laughed in (1b) and suggested in (1c).2 In (1a), the participants the teacher, the book, and on the table each constitute one of the arguments of put. Similarly, in (1b) the clown is the argument required by laughed, and in (1c) she and this idea are the arguments required by suggested. If the number of participants does not match the number of arguments specified by a given predicate, ungrammatical sentences result, as in (1d) to (1f). The correspondence between grammatical functions (Subject, Direct Object, Indirect Object, etc.) and the arguments of a predicate is relatively predictable, and is partially determined by the semantics of the predicate. These principles are so general and yet so constrained that even novel or nonce verbs will conform to them. Thus, the argument structure

Discursive constraints on the lexical realization of arguments in Spanish

of a predicate is the minimal information needed for determining its syntactic frame (Alsina 1996). Our study focuses on two universal types of argument structure: transitive and intransitive.3 Transitive sentences require at least two arguments – a subject and an object – whereas an intransitive requires only one. In some languages, word order is fixed: English, for example, is a strictly SVO language, so that the subject in English is always preverbal and the object postverbal (Thompson 1978). Other languages, including Spanish, have freer word order, and argument positions relative to the verb are more dependent on discursive factors. Because the number and type of arguments are syntactically defined, it is of interest to investigate the possible effect of discursive factors such as information flow on speakers’ choice of argument structure. Such an effect would mean that pragmatics plays a role not only in the more flexible, optional aspects of grammar, but also in ones that are more fixed and obligatory. Arguments may also vary in surface expression. Subjects can be expressed by a lexical noun or noun phrase, by a pronoun and – in so-called ‘null subject’ or pro-drop’ languages like Spanish, Italian, or Hebrew – they may have zero expression. An overt subject need not appear in two main types of constructions: (1) In canonic ‘pro-drop’ contexts, the verb is inflected for person, obviating the need for an overt pronoun subject because the subject is identified by verb morphology; and (2) in impersonal subjectless constructions, where the verb has generic rather than personal reference. In the first type, an overt subject is optional, while in the second type it is prohibited. In ‘non pro-drop’ languages like English or French, non-overt subjects occur in very restricted contexts, typically beyond the simple clause, for example, in same-subject coordination. Use of different positions (preverbal or postverbal) as well as different realizations (e.g., lexical nouns or pronouns) has both discursive causes and consequences. For example, postverbal nouns are often used to introduce new referents, whereas preverbal NP’s are more often used to maintain reference to existing entities (Kail & Sánchez-López 1997; Karmiloff-Smith 1979). To recapitulate, the number and type of arguments is general across languages for transitive and intransitive sentences, while the position of arguments with respect to the verb and the realization of subjects in the form of noun phrases, lexical nouns, pronouns, or zero may vary. We further suggest that this is affected by the typological distinction between languages that allow zero subject marking to a greater or lesser degree. Another dimension relevant to the relationship between grammar and discourse is the distinction between the three ‘core arguments’: the two obligatory arguments of a transitive sentence (subject and object) and the one obligatory





Ekaterina Khorounjaia and Liliana Tolchinsky

argument of an intransitive sentence (subject). In languages like English and Spanish, speakers treat the subjects of transitive and intransitive sentences similarly. The unmarked position for both is sentence-initial; both subjects control agreement features of the verb; and both subjects may be replaced by the same kind of pronouns (Keenan 1976). These features distinguish the two types of subject from the object, as shown in the examples in (2). The subject in (2a), She, is expressed by the same pronoun as the subject in (2b), although the first sentence is transitive and the second intransitive, differing from the object pronoun her in (2a); and in (2a) it is the singular subject She that determines the morphological marking on the verb. (2) a. She has never met her. b. She has slept for hours.

Nominative-accusative languages like English, Spanish, Hebrew, and other European or Semitic languages treat transitive and intransitive subjects similarly and have an opposition between the grammatical functions of ‘subject’ and ‘object’. In contrast, ergative-absolutive languages involve an opposition between the subjects of transitive and intransitive clauses, and group the subjects of intransitives together with the objects of transitive sentences. This yields a three-way distinction between Transitive Subjects/Intransitive Subjects and Transitive Objects, which is typically morphologically marked on nouns and/or pronouns. Dixon’s (1987) proposal to specify these distinctions by the letters A, S and O – where A is the ‘transitive subject’, S the ‘intransitive subject’, and O the ‘transitive object’ – assigns distinct labels to transitive and intransitive subjects, and so neutralizes the traditional categories based on nominativeaccusative languages.4 The question of concern to the present study is to what extent the syntactic opposition of the three core arguments – S and A versus O – is also a discursive opposition: that is, whether the discursive functions of arguments fall into the same tripartite pattern of grouping. . Flow of information and argument structure The idea of analyzing the patterns of relationship between the distribution of new information and argument structure was first proposed by Du Bois (1987). His analysis of narrative texts in Sakapultec Maya, an ergative-absolutive language, led him to conclude that, in parallel to syntactic and morphological ergativity, the distribution of new and given information also tended to follow an ergative pattern. Finding that new information was introduced in the positions of intransitive subject (S) and object (O), with given information encoded

Discursive constraints on the lexical realization of arguments in Spanish

in the transitive subject (A), Du Bois suggested that the main syntactic distinction between languages (nominative versus ergative) is motivated discursively: On the one hand, there is a strong motivation to group together S and O as the argument position for new information while, on the other hand, S and A can be combined as a position for topical (or non-new) information. A language is then defined as nominative or ergative depending on which tendency is the stronger. Moreover, because speakers tend naturally to minimize the amount of new information in a given interchange, sentences with two obligatory arguments – transitive sentences – are likely to contain only one new argument, not two. But the choice of which argument should carry the new information is open: either the subject or the object can do so. In sentences with only one obligatory argument – intransitive sentences – the possibilities are obviously more limited. Nonetheless, Du Bois found that subjects of intransitive sentences are more likely to carry new than known information. To describe this tendency in distribution of new and given information, Du Bois proposed a ‘Preferred Argument Structure’, defined by four constraints that regulate the informational status and surface form of core arguments. Two of these constraints are general: (1) Avoid more than one lexical argument per clause and (2) avoid more than one new argument per clause. The other two are specific to the A-argument: (3) Avoid lexical A’s and (4) avoid new A’s. These constraints on information flow are thought to help listeners in processing incoming information. Not only is the distribution of information among the different types of arguments discursively motivated, so is the specific surface realization of the arguments. As noted, these can be realized by several types of referring expressions: lexical descriptions, personal pronouns, demonstratives, or zero, as illustrated for Spanish in (3). (3) a.

el niño duerme ‘The boy sleeps’ b. él duerme ‘He sleeps’ c. ese duerme ‘That (one) sleeps’ d. duerme ‘(He/she/it) sleeps’

The first formulation (3a) is the most explicit, since it gives a fuller picture of the situation than the second (3b), which could refer to a man, and much more





Ekaterina Khorounjaia and Liliana Tolchinsky

so than (3d), which could refer to a boy or a girl, a man or a woman, or a dog or cat. In terms of Ariel’s (1990) notion of ‘accessibility’, full lexical realizations are low accessibility markers, since they serve as indicators of new, less active information. Pronouns are higher accessibility markers, and zero marking still higher. The decision to use a certain type of realization depends on a range of cognitive and discursive factors, implying that speakers (or writers) recognize the addressees’ need to identify, activate or “recover” (Halliday 1967) the discourse referents. Speakers’ attempt to reduce the use of lexical arguments as part of the tendency to minimize new information was articulated by Chafe’s (1994) proposal of the ‘Light Subject Constraint’. Chafe argues that subjects in conversational discourse are either not new or of trivial importance, they refer to given or accessible information, and are realized mainly as pronouns. He makes no distinction, however, between transitive and intransitive subjects, ascribing the absence of ‘heavy’ realizations to a general avoidance of new information in any subject position. Du Bois, in contrast, argues that differentiation of the subject position is crucial, because of its double role in the distribution of informational flow. This distinction is supported by Assayag’s (1999) analysis of narrative and expository texts produced by Hebrew-speaking university students asked to tell a story about an incident of interpersonal conflict and to discuss the problem of violence in schools (Berman & Ravid 1999). Assayag found that S-subjects in both genres were realized by low accessibility markers (lexical NP’s) more frequently than A-subjects, whereas A-subjects were realized by medium and high accessibility markers (pronouns and zero marking) more frequently than S-subjects. In other words, it is not the subject position as such that determines the use of lower or higher accessibility markers, but the position within a transitive or intransitive clause. Because these positions carry information of different status, their surface realizations differ. Thus, a further issue addressed in our study is the surface realization of subjects and objects. Full noun phrases, as low accessibility markers, were expected to appear more frequently in S and O argument positions carrying new information, and to be avoided in A. . Genre, age, and argument structure Assuming that the selection of argument structure is constrained by the flow of information and that speakers do indeed manifest a Preferred Argument Structure for marking the difference between new and given information, the question arises how generally this applies across discourse genres. Do speakers

Discursive constraints on the lexical realization of arguments in Spanish

choose similar argument positions for new information whether telling a story, describing an object, or analyzing a topic? There is no general consensus on use of the term ‘genre’, and it has been used interchangeably with the term ‘register’ in different research domains (e.g., Biber 1995; Guenthner & Knoblauch 1995; Miller 1984). Here, we adopt Ferguson’s (1994: 21) sociolinguistic characterization of the notion as A message type that recurs regularly in a community (in terms of semantic content, participants, occasions of use, and so on) will tend over time to develop an identifying internal structure, differentiated from other message types in the repertoire of the community.

to address the question of whether this ‘identifying internal structure’ will involve a Preferred Argument Structure. Participants in the study were asked to produce texts in two contrasting genres: an account of an incident related to a conflict at school in which they had been personally involved, that is, a personal-experience narrative, and a consideration of the topic of interpersonal conflict from an analytical point of view, that is, an expository discussion.5 Various analyses deriving from the crosslinguistic project in which this study is embedded indicate that the youngest subjects (9- to 10-year-olds) were able to distinguish narratives from expository texts in the linguistic devices they used, including in lexical usage (Gayraud 2000), types of subjects (Ravid et al. 2002), tense-aspect markings (Ragnarsdóttir et al. 2002), lexical modality (Reilly et al. 2002), types of verbs and clause constructions (Berman & Nir-Sagiv 2004). The question is whether and to what extent genre also has an effect on Preferred Argument Structure. In line with Givón (1983), Du Bois’ explanation for the appearance of lexical and new mentions in S and O but not in A relates to topic continuity in narrative discourse. Human protagonists as key participants are maintained in successive clauses and referred to by means of pronouns rather than full NPs. The O position, instead, is occupied by changing participants, usually inanimate, that will not be maintained across many clauses. The changing status of these referents explains why O position is used for new information and why they are referred to by more explicit means like full NPs. As for S, Du Bois relates the preference of this role for new information to the function of intransitive verbs in the flow of information, mainly the way they function for introducing new referents. Indeed, he found that most intransitive predicates were used for introducing referents that were maintained by topical (or given) reference in the A position.





Ekaterina Khorounjaia and Liliana Tolchinsky

The proposals of Du Bois and Givón provide support for a consistent finding for the surface realization of subjects and objects across narrative and expository discourse. Expository discussions are typically concerned with concepts, ideas, and processes that are either unknown to addressees or must be activated from their long-term memory or general world knowledge. Narratives, in contrast, are about particular people in specific temporally and spatially defined circumstances, who can be referred to pronominally, for reasons noted earlier. Thus, encoding of grammatical subjects, lexical subjects – particularly heavy ones with phrasal and clausal modifications – are commoner in expository than in narrative texts. Expository texts and narratives also differ along the dimension of personal involvement (Berman et al. 2002; Chafe 1994). First person narratives – like those elicited in the present study – represent the highest point of involvement, whereas expository texts tend to be more detached and impersonal. One indicator of different levels of personal involvement is the type of pronouns used in each genre. Speakers of different languages at different ages preferred use of personal pronouns in narratives as against impersonal and generic pronouns in expository texts (Jisa, this volume; Ravid et al. 2002; Reilly et al. in press; Ragnarsdóttir & Strömqvist in press; van Hell et al. in press). Referring expressions are not the only discursive devices that characterize narratives compared with expository texts. As Longacre notes, narratives are ‘agent-oriented’ where expository texts are ‘topic-oriented’: “While narrative discourse is agent oriented and, furthermore, deals with the actions of particular agents, expository discourse lacks this agent orientation and deals more with generalities” (Longacre 1996: 245). This leads to the prediction that genrespecific demands will yield different grammatical constructions. For example (as demonstrated by Berman & Nir-Sagiv 2004, for Hebrew), stative copular clauses rather than lexical verbs are more likely in referring to static situations, and impersonal clauses can be expected in non-agentive contexts. Studies of the influence of genre on the relation between argument structure and informational flow have reached contradictory conclusions. While Assayag (1999) concludes that the PAS constraints act similarly across genres, O’Dowd’s (1990) analysis of expository texts shows that the PAS pattern is characteristic only of narratives, where a human protagonist is likely to appear in the A position in the role of actor. This conclusion, however, is limited since O’Dowd analyzed a very specific text type: a training session in cardiopulmonary resuscitation (in terms of genre – a lecture or instructional dialogue). Kumpf examined the relationship between genre and argument structure in conversation versus narrative (Kumpf 1992) and in high school science class-

Discursive constraints on the lexical realization of arguments in Spanish

room discourse (Kumpf 2003). She found that PAS is maintained across genres, although teacher lecturing yields a high proportion of lexical S’s and O’s. In light of these findings, we decided to focus on possible effects of both genre differences and animacy. The two discourse types used for our comparison – a personal-experience account and an expository discussion – are both about the shared theme of ‘problems between people’, and so involve human protagonists. In the selected sub-genre of narrative, however, the narrator is also the protagonist, a feature that in a sense magnifies the properties of A as a role to be filled by given mention of a topical human protagonist, favoring a pronoun rather than a full NP. What concerns us here is a pattern of information distribution that has been found in typologically different languages like Hebrew and Sakapultec and that seems to reveal a basic discursive mechanism reflected in the grammar of different languages (Du Bois 1987; Durie 1988; Dutra 1987). And there is growing evidence that the speech of young children acquiring different languages exhibits PAS similar to that of adults, in studies on Venezuelan Spanish (Bentivoglio 1994), Korean (Clancy 2003), and Inuktitut (Allen & Schroeder 2003). Given the crosslinguistic generality of this pattern and the fact that the youngest participants in the present study were nine-year olds who have already acquired basic grammatical and discursive processes (see Tolchinsky, this volume), we expected this pattern of information distribution to appear from the youngest age group, with no significant age-related differences. That is, the constraints which comprise Preferred Argument Structure, appear to us good candidates for a stable aspect of linguistic competence that does not change as a function of (later) language development. In sum, the study reported on below concerns the effects of genre and development in expression of Preferred Argument Structure. We expect to confirm the correlation between type of argument and the status of the information that it conveys: the preferred argument for introducing new information will be O, followed by S, while the A position will be largely avoided for expressing new information. We also expect to find a general avoidance of lexical realizations of subjects, especially in transitive clauses, although there might be genre differences in this respect: For example, a higher occurrence of lexical nominals in expository texts might yield more lexical nouns in A position. And we expect to find a preference for certain types of constructions, such as copular and impersonal clauses in expository compared with narrative texts. In developmental perspective, we predict that the general tendency to avoid lexical subjects and new arguments will be common across age groups.





Ekaterina Khorounjaia and Liliana Tolchinsky

. Description of study . Corpus Our data-base is a sample of Spanish-language texts collected in Cordova, Spain. Each participant produced four texts relating to the same content. Subjects were shown a short wordless video depicting situations of interpersonal conflict in a school setting, and were then asked to talk and to write about “something similar that happened to you” and “your reflections on the topic of problems at school”. (For details, see Berman & Verhoeven 2002a). Here, we analyze only oral texts, one narrative and one expository, produced by 10 participants at each of four levels: grade school, junior high, high school, and university, giving a total of 80 texts divided equally between the two genres (narrative and expository). . Coding All the texts were coded for: argument types, linguistic realization, animacy and position with respect to the verb (for subjects), and informational status. Since the four constraints defining Preferred Argument Structure (Section 1.2) apply only to core arguments – subjects and (direct) objects – our analysis is confined to transitive and intransitive clauses with finite verbs. Several other clause types were also considered but, in contrast to other studies (Ashby & Bentivoglio 1997, 2003; England & Martin 2003) without separate coding for argument structure: constructions with copular, impersonal, or nonfinite predicates, and gapped or verbless clauses. This enabled us to examine whether there is a genre-based preference for certain clause types, on the assumption that expository texts would contain more copular and impersonal constructions than narratives. In finite-verb clauses, we also looked for a relationship between information flow and argument types and specified the surface realization, position, and animacy of Subject arguments. .. Argument type For each clause, the surface grammatical role of its ‘core’ arguments was classified as A, S or O, disregarding oblique NP’s and adjuncts as outside ‘direct’ argument structure (Du Bois 1985). S is the sole argument of intransitive verbs as shown in the example in (4b), while A and O are the arguments of transitive verbs – in (4a) and (4c) respectively. Purely syntactic criteria were applied in defining these categories (transitive and intransitive) because, except for the

Discursive constraints on the lexical realization of arguments in Spanish

clitic se that often changes a transitive construction to intransitive, Spanish verbs are not morphologically marked for transitivity (Hidalgo 1994). Besides, in semantic terms, transitivity should be viewed along a continuum rather than as binary (Thompson & Hopper 2001). (4) a.

Porque mis padres (A) no me han enseñado a hacer otra cosa because my parents not me (clitic O) have taught to do other thing ‘Because my parents haven’t taught me to do anything else’ [pA08fes]6 b. Y estas cosas (S) pués surgen ‘And these things, well, appear’ [pA05mes] c. Y necesitas mucho el apoyo de los demas (O) and need-2ndsg much the support of the others ‘And you very much need/one needs support from other people’ [pH02fes]

Although Spanish nouns are not marked for case, the subject is readily identifiable, since it controls number and person agreement on the verb. With objects, case-marking is encoded by means of clitic pronouns (see below) which help decide whether an object is direct or indirect. .. Morphological forms Three main types of surface forms were distinguished: zero, pronouns, and full noun phrases, as illustrated in (5) through (8). Spanish verbs agree with the subject in person and number, for which they are inflectionally marked, making the zero form the unmarked case of (pronominal) subject reference. ... Zero. Applies only to subjects, as shown in (5). (5) (S) Hemos empezado por una serie de preguntas have-1stpl started with a series of questions ‘(We) began with a series of questions’ [pA06mes]

... Personal pronouns. Personal pronouns agree for person and number in first- and second-person singular and for person, number, and gender in third-person singular and plural and in first- and second-person plural.7 They are also marked for case: nominative (e.g., yo, ‘I’; tu, ‘you’, el, ‘he’); accusative and dative (e.g., me 1stSg, nos 1stPl), or oblique (e.g., conmigo ‘with me’). In addition, there is a phonological distinction between stressed and unstressed personal pronouns – with each having different syntactic and semantic properties. Pronouns in nominative and oblique case are always stressed; pronouns





Ekaterina Khorounjaia and Liliana Tolchinsky

in accusative and dative case are unstressed. Further, stressed pronouns may have only human referents, as in (6b), whereas unstressed pronouns may have plus or minus human as well as plus or minus animate referents, as in (7b). El alumno escribe ∼ Él escribe ‘The pupil writes’ ∼ ‘He writes’ b. El lápiz escribe ∼ *Él escribe ‘The pencil writes’ ∼ *He/It writes

(6) a.

La niña escribió las cartas ∼ La niña las escribió ‘The girl wrote the letters’ ∼ the girl (clitic O) wrote = ‘The girl wrote them’ b. La niña beso a su madre ∼ La niña la besó ‘The girl kissed her mother’ ∼ the girl (clitic O) kissed = ‘The girl kissed her’

(7) a.

As in other pro-drop languages, stressed pronoun subjects are not required; when used, as in (8), they have a contrastive or emphatic connotation, particularly where verb morphology serves to identify the subjects. (8) que ellos (S) tienen that they have+3rdpl [pA09fes]

In contrast, unstressed or clitic pronouns depend phonologically on their host verbs and are required in cases of object ellipsis. They cannot be omitted, since case is not marked on the verb; and they are typically preverbal when the verb is finite, as in (9), and postverbal in non-finite forms like infinitives and gerunds (Fernández-Lagunilla & Anula-Rebollo 1995). (9) Antiguamente pués te (clitic O) educaban más o menos los padres Formerly well you-Sg (clitic O) educated more or less the parents ‘Well, formerly your parents sort of educated you’ [pA01mes]

... Other pronouns. Demonstrative and relative pronouns like those in (10) were treated separately from the personal pronouns in (6) to (9) for both syntactic and discursive reasons. Syntactically, they can occupy both subject and object positions, so that their syntactic behavior is different from that of clitics and stressed personal pronouns. Discursively, they are less accessible, and more marked.

Discursive constraints on the lexical realization of arguments in Spanish

(10) a.

Y pueda eso generar violencia and can this to-generate violence ‘And this could generate violence’ [pA13mes] b. Los que el día de mañana van a seguir haciendo lo mismo those that the day of tomorrow are-going to continue doing the same ‘Those who tomorrow will continue doing the same’ [pA06fes]

... Full noun phrases (lexical arguments). Yet another kind of encoding of arguments is provided by noun phrases with lexical heads, as in (11) and (12). (11) Los profesores pueden ayudar en eso (NP+Det) the teachers can-3rdpl to help in that ‘Teachers can help with that’ [pH03mes] (12) Y al poco llegó un profesor jesuita (NP-Det) ‘And soon came a Jesuit teacher’ [pA05mns]

Core arguments may be realized by full noun phrases of different complexity: bare nouns, nouns with only a lexical head, and noun phrases with various kinds of modifiers – adjectival, or prepositional. However, since the PAS constraints disregard internal noun phrase complexity, we did not distinguish different levels of internal complexity in coding lexical arguments, and all possible lexical realizations were considered equivalent. The only distinction in coding lexical arguments was whether they were definite (as in 11) or indefinite (as in 12). Here, too, surface grammatical criteria applied: all noun phrases with a definite article or possessive pronoun were coded as definite, all others as indefinite.8 New information can be expected to be introduced by indefinite NP’s, and old or given information by definite. Our analysis showed, however, that this correlation does not always obtain, as noted below. ... Clausal complements (for objects). Object nominals can also take the form of complement clauses as in (13), but these were not counted as lexical arguments because the PAS constraints function within clause boundaries, while complement constructions are themselves clauses, as in (13).9 (13) Podría decir que es una consecuencia lógica (Complement Clause) ‘I could say that (it) is a logical consequence’ [pA01mes]





Ekaterina Khorounjaia and Liliana Tolchinsky

.. Animacy Only subjects were coded for animacy. We further distinguished animate as plus or minus human, since – as expected from personal-experience narratives and discussion of ‘problems between people’ – referents in both types of texts were all human. (In this, our study contrasts with use of animal-based pictures for eliciting narrative, e.g., Berman & Slobin 1994; Hickmann 2003). Our coding for animacy aimed to examine the relation, if any, between animacy and morphological realization (zero, pronoun, noun phrase), on the one hand, and subject animacy and genre, on the other. (14) que los niños (+Anim), pues, roban ‘That the kids, well, steal’ [pA08fes] (15) y que la cabina (–Anim) no responda ‘And that the (phone)booth (does) not work’ [pA05mes]

.. Subject position In Spanish, subjects can occur both sentence-initially and in postverbal position as in (17) and (16) respectively. The position of the object is more fixed than of the subject, with postverbal the unmarked position for the object, and where the object precedes the verb, it is duplicated by a preverbal clitic pronoun. We therefore analyzed clause position only for subjects, to examine (1) whether and how far Spanish is basically SVO and (2) the relationship between the position of the subject and its informational status, that is, whether subjects that introduce new information tend to appear postverbally position. (16) venian sus amigos (Post verbal S) came his friends [pH02mns] ‘His friends came’ (17) este video refleja nuestras vivencias (Preverbal A) ‘This tape reflects our own experiences’ [pA05mes]

.. Informational status All arguments were coded as given or new according to the type of information they conveyed. This decision was always text-based, according to the previous mention or lack of mention of information in the text. For an argument to be considered given, the information to which it referred must have been mentioned in the text at a distance of less than five clauses, that is, from one to four clauses, before the clause being coded. Arguments that seemed recoverable only

Discursive constraints on the lexical realization of arguments in Spanish

from context (for example, from scenes in the video shown prior to text production) but were not previously mentioned in the texts were coded as ‘new’. Recall that subjects were asked to tell a story illustrating a personal experience and to give a talk discussing ‘problems between people.’ If a speaker began by saying trata de los problemas ‘deals with the problems,’ as did a 7th grade girl giving an expository discussion, the clause was coded as containing new information although the interviewer understood that she was talking about the topic given in the elicitation instructions and illustrated in the video. Contextual information is always needed for interpreting utterances, but our concern was whether subjects could ‘textualize’ the required information, referring to it in their discourse without resorting to the extra-textual context as part of their referring strategies. In other words, we were interested in their ability to create an ‘autonomous’ piece of discourse (Tolchinsky et al. 2002). Arguments realized by a first-person pronoun (referring to the speaker) were coded as given, because any discourse always implies a speaker who does not need to be introduced – especially in the context in which the texts analyzed here were produced.

. Analyses and findings All the oral texts were transcribed and divided into clauses following the conventions specified by Berman and Slobin (1994: 660–663), as adapted for the cross linguistic study on which the present study is based (Berman & Verhoeven 2002: 11). Our focus was thus on monologic texts produced by speakers from age nine years across adolescence and into adulthood. This was the territory within which we defined given and new information, so that we excluded from analysis preparatory dialogues between speaker and interviewer or remarks made by the speaker to the investigator during the interview. The texts varied in length as measured by number of clauses. Our calculations thus took the form of the internal distribution of a given category out of total occurrences of a given target structure or of proportions, that is: total number of occurrences of a target structure divided by total number of clauses. Our first analysis looked for the distribution of all the types of clauses in our sample. This analysis shows that what we defined as transitive and intransitive clauses had the highest frequency in both genres. They account for 76% of the total coded clauses in narratives and for 67% in expository texts. As predicted, Copular, Impersonal, and Nonfinite clauses were more frequent in





Ekaterina Khorounjaia and Liliana Tolchinsky

Table 1. Breakdown (in number and percentages) of argument type used in narrative and expository texts, by age group [n = 40]

Narrative Expository

Argument Type O

A

S

48 (428) 46.3 (408)

51.5 (458) 48.9 (433)

48.5 (433) 51.1 (454)

Total clauses 100 (891) 100 (887)

expository than in narrative texts. Of these, Copular clauses were most frequent and account for 19% of the clauses in expository texts and under 14% of the clauses in the narratives. Nonfinites and Impersonals account for less than 5% in narratives and almost double for expository. There were very few occurrences of verbless clauses (less that 1%). Except for verb gapping, use of construction-type is genre-dependent, confirming findings for written narrative versus expository texts produced by the Hebrew-speaking participants of the larger project (Berman & Nir-Sagiv 2004). Our next analysis focuses on the transitive and intransitive clauses for the distribution of the different types of arguments (A, S and O) by age group and genre. The figures in Table 1 show the percentage of each type of argument over the total of transitive and intransitive clauses in each type of text (raw number of occurrences appear in parentheses). Narrative texts show a similar proportion of argument types (between 48% and 51% for each type) unlike expository texts, where the percentage of S and O arguments is also similar, but A arguments are less frequent. Note that the difference between number of A’s and of O’s in expository texts is related to the frequent use of existentials (e.g., hay muchos niños ‘have [=there are] a lot of children’. Although the NP following the existential verb is the sole argument, syntactically it is an O.10 As genre demands would lead us to predict, agentive subjects occur less often in expository texts than in narratives, but the other two argument types present a very similar distribution. A series of separate ANOVA’s with repeated measures on the mean proportion of each type of argument over total number of clauses showed that the effect of genre is significant only for A arguments (F(36, 1) = 7,052; p < .01) (X for narratives 35. 37 and X for expository texts 27. 35). The distribution of argument types was similar in all age groups, indicating that choice of argument type is not a developmental phenomenon. As to whether there is a preferred argument type for introducing new information and whether, following Du Bois (1985), there is a general avoidance of new A’s, we classified all arguments as providing new or given information. No effect was found for the factor of Age on selection of arguments. Accord-

Discursive constraints on the lexical realization of arguments in Spanish

Table 2. Breakdown of argument type (A, S and O) by new and given information [n = 40] Given

New

Total

91.3 (391) 72.0 (330) 43.4 (188)

8.7 (37) 28.0 (128) 56.6 (245)

100 (428) 100 (458) 100 (433)

85.6 (345) 73.9 (320) 35 (159)

15.4 (63) 26.1 (113) 65 (295)

100 (408) 100 (433) 100 (454)

Narratives A: Subject of transitive verbs S: Subject of intransitive verbs O: Object of transitive verbs Expositories A: Subject of transitive verbs S: Subject of intransitive verbs O: Object of transitive verbs

ingly, Table 2 presents the breakdown of argument types (A, S and O) by new and given information grouped by genre alone. In narrative texts, despite the similar proportion of arguments of each type, speakers prefer to introduce new information through the O argument, the object of transitive clauses. Second in order of preference for this purpose is S, the sole argument of intransitive clauses, and last is A, the subject of transitive clauses. We found this same pattern in every age group, and for both genres: that is, the order of preference for introducing new information is similar in narrative and expository texts and there is a general avoidance of new A’s independent of discourse genre. This general pattern of information distribution is part of children’s discursive performance from as early as age nine and does not develop with age, in line with findings for discursive performance of even younger children (Clancy 2003). As for reference to previously given information, the order of argument preference shows some variation with age and school level but not with genre. Among the youngest speakers, the preferred argument for location of given information is S, the subject of intransitive verbs, but for the two oldest groups it is A, the subject of transitive verbs. In this sense, subjects of transitive clauses differ both from objects and from subjects of intransitive clauses. Our next analysis concerned the position of the subject (preverbal or postverbal) and the semantic features of subject-referents (animate or inanimate). Recall that Spanish has flexible word order, but the preverbal position is considered unmarked for subjects. Postverbal positioning is associated with more formal registers (Green 1988) and with the introduction of new information. Table 3 presents the breakdown of pre- and post-verbal S and A subjects, over the total of overt subjects, by school-level and genre.



 Ekaterina Khorounjaia and Liliana Tolchinsky

Table 3. Breakdown of transitive and intransitive clauses according to subject position by genre and school level [n = 40] A

S

Preverbal

Post-verbal

Preverbal

Post-verbal

Narratives

Primary Junior High Adults

95.2 (20) 78.2 (18) 97.1 (34) 89.6 (26)

0.8 (1) 1.8 (5) 2.9 (1) 10.4 (3)

83 (25) 56 (23) 61 (42) 68 (32)

17 (5) 44 (18) 39 (27) 32 (15)

Expository

Primary Junior High Adults

89 (9) 80 (16) 90 (54) 77 (30)

11 (2) 20 (4) 10 (6) 23 (9)

79 (19) 42 (13) 95 (40) 68.5 (50)

21 (5) 58 (18) 5 (2) 31.5 (23)

In general, preverbal position is much more frequent than postverbal, particularly in the youngest age group. This finding supports the idea that the preverbal position of subjects is the unmarked choice. A series of analyses of our data showed that age but not genre is significant (F (3, 35) = 5,384, p < .004) in this respect. Except for the high school expository texts, the postverbal position is more frequent in S subjects than in A subjects, confirming the correlation between postverbal subjects and the introduction of new information (see also Dutra 1987, for Portuguese). Consider next the semantic features of subject-referents and the effect of animacy on distribution of new and given information. If texts containing animate protagonists, such as personal-experience narratives, show the same pattern of distribution as texts containing mainly inanimate subjects, such as expository texts, we could claim a general pattern of relationship between argument structure and information status, independently of text type. Our analysis showed an overwhelming preference for animate subjects in both genres; but inanimate subjects in S position were significantly more frequent (F (1, 34) = 26,411, p < .000) in expository (X = 11,1653) than in narrative texts (X = 4,4722). As for A arguments, those referring to inanimate subjects were more frequent in expository texts (X = 3.16) than in narratives (X = 0.46). A-type arguments were not only less frequently used in expository texts than in narratives, when they did occur, they referred to inanimate subjects (e.g., la discriminación ‘(the) discrimination’ in la discriminación provoca problemas ‘(the) discrimination provokes problems’, el teléfono no devuelve dinero ‘the telephone (does) not return money’). In other words, each argument type carries a different kind of information. How is this difference between what is grammatically possible and what is discursively useful reflected in speakers’ selection of lexical

Discursive constraints on the lexical realization of arguments in Spanish 

Table 4. Breakdown of clauses according to number of lexical arguments by genre and school level [n = 40] No lexical

One lexical

Two lexical

Total

Narratives Grade Junior High Adults

67.2 (92) 77.4 (138) 71.3 (256) 66.2 (143)

31.4(43) 21 (38) 23.9 (86) 31.9 (69)

1.4 (2) 1.6 (3) 4.7 (17) 1.8 (4)

100 (137) 100 (179) 100 (359) 100 (216)

Expository Grade Junior High Adults

64.7 (68) 65.5 (99) 68 (213) 63.5 (202)

33.3 (35) 30.4 (46) 27.1 (85) 32 (102)

1.9 (2) 3.9 (6) 4.7 (15) 4.4 (14)

100 (105) 100 (151) 100 (313) 100 (318)

arguments? To address this question, we examined the distribution of clauses by number of arguments that were lexically realized, so as to explore the proposed PAS constraint that speakers tend to avoid using more than one lexical argument per clause. Table 4 presents the proportion of clauses containing zero, one, or two lexical arguments out of the total of transitive and intransitive clauses in each age group and genre. Table 4 shows that there are very few clauses with two lexical arguments and relatively more with only one, and both are far less common than clauses with no lexical arguments at all. True, only transitive clauses can have two arguments but, as shown in Table 1, the different argument types are very similar in frequency. The low proportion of clauses with two lexical arguments is therefore not due to the infrequency of transitive clauses, but derives from pragmatic reasons: Speakers avoid lexicalizing both arguments of a transitive clause. What in traditional Spanish grammar is considered a canonical transitive sentence – AVO, with lexical realization of A and O – is almost nonexistent in normal speech. Clearly, it is not the case that clauses without lexical arguments are uninformative; rather, they convey information via the predicate rather than the arguments. The excerpt in (18) is from the text of a nine-year-old girl who managed to tell an entire story using only one clause with a lexical argument (6 clauses out of a total of 23). (18) Pues que yo estaba jugando con una amiga. well (that) I was playing with a girlfriend Entonces nos peleamos. then (we) quarreled

 Ekaterina Khorounjaia and Liliana Tolchinsky

Entonces ella no quería más jugar conmigo. then she didn’t want to play with me Entonces se fue con otras. then (she) went with others Entonces nos empezaron a insultar y todo eso. then (they) started to insult us and stuff like that Entonces nosotras también les insultábamos. then we also insulted them [pG04fns]

Although these six clauses contain only one lexical realization of a nominal element (con una amiga in the first clause) and none in argument position, the text is sufficiently informative for the reader to get the full story. The fact that the girl is telling a personal-experience narrative in which she herself is protagonist might be facilitating her disregard for lexical arguments. The first lexical realization of a noun slot (Ravid, et al. 2002) occurred four clauses later, around halfway through her narrative, when she introduced her sister as another character in her story. This demonstrates how speakers can construct narrative accounts whose major content is transmitted through the predicates. If, then, speakers are reluctant to use lexical arguments which, if any, arguments are preferred for lexical realization? For example, if full lexical expressions are preferred as subjects of intransitive rather than transitive clauses, the proportion of lexical NP’s should be higher for S than for A. Besides, if low accessibility markers like full NP’s are used in an argument position dedicated to new information, the informative value of the position is increased; conversely, use of high accessibility markers like pronouns or zero in a position dedicated to given or known information reduces its informative value. Tables 5a and 5b show the proportion of each type of realization (zero, personal pronoun, definite and indefinite noun phrases) out of total noun slots in each argument position, S, A and O, by age level and genre. The figures in Tables 5a and 5b show that zero is the preferred means of reference for A and S, particularly in the case of transitive sentences, and this is not affected by age or genre. For Spanish speakers, null reference is the preferred means of referring to participants in a situation. As for O (objects of transitive clauses), preferences were more divided, and changed with age. Use of clitics is preferred by the two younger groups, but this decreases somewhat with age, while use of full lexical noun phrases increases. In other words, the O position becomes increasingly informative with age. In narrative texts, the percentage of lexical realizations in the S and O arguments was twice that of lexical realizations in the A argument. A separate

Discursive constraints on the lexical realization of arguments in Spanish 

Table 5a. Breakdown of referring expressions used in three positions in narrative texts, by age group [n = 40] A

S

O

Zero Pro Other Lex- Tot Pro ical

Zero Pro Other Lex- Tot Pro ical

Cli Pro Lex- Compl. Tot ical

Grade 69.5 19 1.5 (48) (13) (1)

10 100 55 18 3 (7) (69) (37) (12) (2)

24 100 43 0 30 27 (16) (67) (30) (21) (19)

100 (70)

Junior 67 22 0 (48) (15)

11 100 62.7 22 0 (8) (71) (69) (24)

15.3 100 39 10 24 27 (17) (110) (27) (7) (16) (19)

100 (69)

High 80.5 10 2.3 (145) (18) (4)

7.2 100 61.2 9 12.3 17.5 100 41 9.4 32.6 17 (13) (180) (109) (16) (22) (31) (178) (74) (17) (59) (31)

100 (181)

Adults 73 15.7 7.5 (79) (17) (8)

3.8 100 54.4 6 10.6 29 100 29.2 8 34.5 28.3 (4) (108) (56) (6) (11) (30) (103) (33) (9) (39) (32)

100 (113)

Table 5b. Breakdown of referring expressions used in three positions in expository texts, by age group [n = 40] A

S

O

Zero Pro Other Lex- Tot Pro ical

Zero Pro Other Lex- Tot Pro ical

Cli Pro Lex- Compl. Tot ical

Grade 71.8 20.5 0 (28) (8)

7.7 100 54 9.5 11.5 25 100 45 9.4 40 5.6 (3) (39) (28) (5) (6) (13) (52) (24) (5) (21) (3)

100 (53)

Junior 63 15 11 (34) (8) (6)

11 100 69 1 7 (6) (54) (69) (1) (7)

100 (52)

23 100 34.7 3.8 44.3 17 (23) (99) (18) (2) (23) (9)

High 64 15 9 12 100 59 7 18.8 15 100 17.7 8.2 40.8 33.31 (107) (25) (15) (20) (167) (75) (9) (24) (19) (127) (33) (15) (76) (62)

100 (186)

Adults 73.6 9.5 5.5 (109) (14) (8)

100 (163)

11.4 100 53 5 19 23 100 20.8 12.3 46.6 20.3 (17) (148) (82) (8) (29) (36) (155) (34) (20) (76) (33)

ANOVA performed on the mean proportion of lexical arguments (including definite and indefinite noun phrases and proper names) revealed no significant differences either by age or genre. That is, speakers produce a similar total amount of lexical realization, but they are selective regarding which types of argument are lexically realized. Positions used for new information receive more lexical realization than positions used for given information. Thus, the morphological realization of the argument reinforces its informative status.

 Ekaterina Khorounjaia and Liliana Tolchinsky

Spanish speakers’ preference for null forms applies even where a lexical argument might have been required for communicative reasons.11 And in fact, use of pronouns or zero rather than lexical nouns may be an obstacle to understanding, especially when the listener does not share the speaker’s background knowledge. Consider the example in (19) of how a nine-year-old girl starts her text: (19) Que está mal hecho that it is badly done Y que los tenían que llevar al director and that them had+3rdpl to take to the principal ‘And they had to take them to the principal’ [pg04fes]

The second clause contains two arguments, neither realized lexically. The subject is marked by the suffix -an on the verb and the object is a preverbal clitic pronoun (los ‘them’). Neither is identified for the listener, although the child is obviously talking about something she saw on the elicitation video. There were many such cases in which a listener who was not present at the interview would find it impossible to understand the text. These problems of reference might be at least partly due to the elicitation conditions. Although participants were asked to produce a text describing their own experience of events and to discuss conflicts like those depicted in the video, the interviewer was present both when the video was shown and when texts were produced. Respondents thus may have been taking this shared knowledge into account when constructing their discourse. Research shows that young children improve their reference devices when there is a genuine situation of unshared knowledge between speaker and interlocutor (Hickmann 2003). Moreover, situations where the absence of shared knowledge is more obvious, such as written texts, might also entail fewer such problems of reference.

. Conclusion Spanish-speaking children and adolescents have a preferred argument structure that is identical to that of Spanish-speaking adults or, for that matter, of Sakapultek-speaking adults (Du Bois 1987). It may not be an exaggeration to say that what we have here is a discursive universal. The four constraints proposed by Du Bois are supported by our data. Concerning the constraint on the number of lexical arguments per clause, we found a clear avoidance of lexical arguments (Table 4). The number of

Discursive constraints on the lexical realization of arguments in Spanish 

clauses without lexical realization of arguments is at least twice that of clauses with one lexical argument, while clauses with two lexical arguments are extremely rare. The youngest subjects produced only two clauses containing two lexical arguments across their texts. Speakers find other ways to get their message across. One way we noted was use of a series of predicates for conveying the relevant information in a story – and indeed entire texts in both genres were developed without resorting to lexical arguments. Non-argument positions, which were not analyzed in the present study, may also play an important role in establishing the content of a text. Further research should explore different means that speaker-writers use to transmit information without lexicalized core arguments (A, S, or O). Sentence-peripheral components (adjuncts and oblique NP’s in general) might well be more informative than sentence-core components. Another line of research needs to explore how far the pattern we found applies beyond personal-experience narratives, where the speaker is protagonist. Our data provide indirect support for the constraint on the number of new arguments, which could only have been violated in transitive sentences in which both arguments carried new information. The small number of cases where new information was introduced in A position can thus be taken as evidence for the “New Argument Constraint”. As for the two constraints specific to the A argument, our results clearly demonstrate that speakers avoid lexical arguments in A position. We found that only about 10% of arguments in A position are lexical, as against about 30% in narrative or 40% in expository texts in O position (Table 5a and 5b). Lexical arguments in the position of transitive subject are clearly more marked. The same preference was found for the location of new information: Although speakers produce a similar amount of transitive and intransitive clauses, they avoid introducing new information in the subjects of transitive sentences (A position), and prefer to do so in O and S. This finding needs to be interpreted in the context of the more general relationship between grammar and discourse genre, as discussed below. Concerning the effect of genre, we need to distinguish between general differences that arise from the demands specific to a personal narrative versus discussion of a topic and those that concern ‘Preferred Argument Structure’. Our analysis showed that certain constructions – Impersonal, Copular, and Nonfinite – occurred more frequently in expository than in narrative texts. This confirms previous findings for the use of a range of depersonalization devices in expository texts in different languages (Berman in press; Jisa, this volume; Jisa et al. 2002; Tolchinsky & Rosado in press). Our finding that copular and

 Ekaterina Khorounjaia and Liliana Tolchinsky

impersonal constructions were used more frequently in expository texts closely parallel the findings of Berman & Nir-Sagiv (2004) for Hebrew, and correspond to the Reilly et al. (2002) analysis for differences in use of modal expressions in expository versus narrative texts. It could be, however, that the differential use of copular versus lexical verbs is more marked in speech than in writing. This, too, deserves further investigation. By attending to genre distinctions in the spoken modality alone, we ignore the factor of modality, which has been shown to be crucial in other aspects of text construction (see Berman & Ravid, submitted). Genre and modality are the two dimensions through which later language development should be analyzed. Another genre-related difference concerns the use of A arguments. As discussed, subjects of transitive verbs were commoner in narratives than in expository texts, while the major distinguishing characteristic of these two genres – agent versus topic orientation (Longacre 1996) – was reflected in the differing distribution of A arguments. This general distinction bears directly on the location of new and given information. Because the protagonists (agents) of a narrative usually remain the same throughout the story, once they are first mentioned they are given. Thus, reference to them is normally reference to given information. They occupy positions reserved for given information, are referred to by pronouns or by zero, and remain active – known – throughout the text. In contrast, the objects that the protagonists encounter and the events in which they participate change as the text progresses, and are therefore located in positions used for new information – O or S. By the same reasoning, they should be more lexicalized. This is precisely what we have found, in support of Du Bois. What about expository texts? Here we found consistently fewer A arguments – the position for given information. This is because new topics for discussion may be introduced at any point in an expository piece of discourse, not only at the beginning. As a result, and in light of the constraints on information distribution, we can expect more arguments in positions for new information, and higher lexical realization of these arguments. Again, this is precisely what we found. Besides the intrinsic interest of identifying a general pattern of interrelations between argument structure and information status, the question arises as to why speakers should tend to avoid full lexical arguments and why they prefer to locate new information in object and in postverbal position. A possible answer to the first question is that speakers understandably avoid verbalizing information that can be deduced from context, in this case textual context. As

Discursive constraints on the lexical realization of arguments in Spanish 

long as the speaker’s references are clear to the listener, there is no need to keep realizing them by means of full lexical expressions. On the other hand, in some of the children’s texts, referents were not lexically identified even when this was necessary for comprehension. That is, although the general trend to avoid lexical realization is justified by the principle of economy, children need to learn when they must be more explicit – not only how little but also how much they need to specify – in order to enable the interlocutor to understand the text or to satisfy the demands of the task. Analysis of NP complexity in subject position in several of the languages in the larger cross-linguistic project showed that in order to meet the demands of genre and modality, speakers in higher age groups increasingly use ‘heavy subjects’ in positions reserved for given information (Ravid et al. 2002). Heavy subjects take the form of complex NP’s that may include prepositional phrases, and relative clauses. Because expository and written texts must be more precise and explicit than spoken personal narratives, they require more lexicalized subjects. This trend to increasing explicitness and lexical specificity develops with age and increased literacy, as shown by several of the chapters in this volume. In this respect, use of indefinite and definite articles for marking new versus given information needs to be more carefully examined. Here, we applied strictly surface, grammatical criteria: all noun phrases with a definite article or possessive pronoun as modifier were coded as definite, while the rest were coded as indefinite. But, as noted, in Spanish like French or Hebrew but unlike English, the definite article can be used with generic nouns without having the sense of old information at all. Future research should consider not only the definiteness of articles, but also whether they are used for specific or generic reference. The second question is more complicated. Why should speakers locate new information in object position? The first explanation that occurs involves positioning. The final part of an utterance is more salient and remains more active in memory since it is the last to be spoken. Indeed, ‘paying attention to the end’ is one of the operating principles suggested early on by Slobin (1973), and many early acquisitions can be explained in terms of positioning. Moreover, in discourse there is a strong attraction between local newness marking and postverbal position. . .This attraction is predicted by the general tendency for new information to occur towards the end of the sentence, obligatory in Chinese, optional in other languages and it emerges with the advent of local marking at around age seven. (Hickmann 2003: 234–236)

 Ekaterina Khorounjaia and Liliana Tolchinsky

We consider this a preliminary attempt at explanation. Further research on Oinitial languages like German might help clarify whether the syntactic function of positioning is the main factor in speakers’ decision to locate new information in O. This is a far from trivial issue in the explanation of the phenomenon we are investigating. A major feature of later language development is an increasing sensitivity to the demands of different communicative circumstances and interlocutor needs. With age and literacy, in our culture a consistently important factor in the development of linguistic abilities, children learn to attune their messages to diverse kinds of circumstances and to control discursive variety (Tolchinsky, this volume). The participants in our study undoubtedly related differently to narratives and expository texts. This capacity for differentiation and for adaptation of linguistic devices to genre-specific demands was evident at every level of text organization, from overall structure down to clause linkage, internal construction of clauses, and lexical choices. The fact that children so clearly differentiate between genres and are able to adapt their output to differing communicative demands strengthens the claim that we are concerned here with a universal pattern of inter-relations between grammar and discourse.

Notes * The study was supported by DGES Grant from the Spanish Ministry of Education and Science B-502000-076 and by a major research grant from the Spencer Foundation, Chicago, for the study of developing literacy, Ruth Berman (IP). We are grateful to Ruth Berman for her careful editing and enlightening comments. . This dichotomous presentation oversimplifies the differing degrees of informational status that are involved in actual use of language (Ariel 1990), but is adequate for present purposes. . We use only verbs as examples of predicates since we deal here with the argument structure of verbal predicates. Lexical items from other syntactic categories like adjectives can also be considered to be predicates and so take arguments (Di Sciullo & Williams 1987). . By ‘intransitive’, we refer to two-place predicates only, disregarding bitransitives like give, show for present purposes. . Comrie (1978: 332) suggests using A, S, and P. . Data were collected in the framework of a crosslinguistic project in which 80 subjects at four levels of age and schooling (gradeschoolers aged 9 to 10, junior high students aged 12 to 13, high schoolers aged 16 to 17, and graduate student adults), native speakers of seven different languages, were asked to tell and write a personal-experience narrative and

Discursive constraints on the lexical realization of arguments in Spanish 

to give a talk and write an essay on the shared topic of interpersonal conflict. For details of motivations and methodology, see Berman & Verhoeven (2002a). . Speaker identity in examples is specified as follows: the first, lowercase letter stands for the language (p = Spanish); the second, uppercase letter stands for grade or age level (G = grade school, J = junior high, H = high school, A = adult); the two digits give the subject number (05 is the fifth subject in that age-group); the next lowercase letter stands for sex (m = masculine, f = feminine); and the last two letters identify text type (es = expository text, ns = narrative text). Thus pG05fes stands for a Spanish-speaking grade school child, the fifth subject in that age group, a girl, and her expository spoken text. . Personal pronouns are a nominal category in Spanish [+N–V] that behave syntactically like proper rather than like common nouns. . The definite article can be used with generic reference in Spanish – like in French or Hebrew but not English – and this may have an effect on the distinction between new and given information. . Other studies may count complement clauses like lexical NPs inside their matrix clause, but we feel this needs to be supplemented by analysis of differing NPs complexity, a topic beyond the limits of the present study. . Spanish hay comes from Latin habet ibi, with the NP that follows considered as the object, since in Latin it was marked for accusative case and in modern Spanish exhibits features of a direct object in pronominalization and in interrogation. Other studies of Spanish treat it as an S, since it is the sole argument taken by the verb (e.g., Ashby & Bentivoglio 1993, 2003) but, as noted, we consider it to be an O. . This is exactly in accord with the finding from picturebook based oral narratives produced by children aged 3 to 9 years and adults. Thus Berman & Slobin (1994: 540n) note that in contrast to Hebrew and even more markedly to English, “subject pronouns are almost non-existent in the Spanish texts, where null subject is the norm for successive same-subject clauses. There are only 14 subject pronouns in the entire Spanish corpus, five of them used by one 4-year-old.”

Syntactic ability in children and adolescents with language and learning disabilities Cheryl M. Scott

.

Introduction

By the time children enter formal schooling around the age of six, they produce grammatically well-formed, multi-clausal sentences in the course of conversing with family and friends. They can tell stories, argue a point, and provide explanations for things they observe. To the casual observer, it might appear that children have learned the grammar of their native language and all that remains is learning new words and uses of language. Inquiry into syntactic development in older children and adolescents in the last 30+ years, however, demonstrates that a great deal of syntax remains to be learned (Berman & Verhoeven 2002b; Chomsky 1969; Scott 1988). Early studies of later syntactic development focused on measures of language production that would be sensitive to syntactic growth. Loban’s work (1976) is a good example. Using a longitudinal design, Loban gathered spoken (interview) and written (school composition) naturalistic language samples from a large group of children from kindergarten through the 12th grade. Two measures of overall syntactic complexity, average utterance/sentence length in words and average number of clauses per sentence, increased slowly but steadily throughout the school years. Loban and other investigators also documented increasing frequencies of later-developing structures at phrasal, clausal, and sentence levels (see Scott 1988; Nippold 1998 for reviews). It is now possible to describe later language development in terms of the increasingly frequent use of particular forms (e.g., modal auxiliaries, passive voice, relative clauses) and variants of forms (e.g., center-embedded relative clauses). As inquiry into later syntax development was proceeding, it became apparent that simply listing structures that increase in frequency contributes little to



Cheryl M. Scott

an understanding of factors that drive later language development. If a child uses a later-developing structure in a sentence, for example, the past perfect form of a verb, the motivation for using that form derives from the broader discourse context. In the sentence in (1), the 9-year-old narrator uses the past perfect to signal temporal backgrounding relative to the story as it unfolds (from Berman & Slobin 1994: 142): (1) The boy, who had climbed the tree, looked in it.

This child has learned to (1) point backwards and forwards in a text and signal the current relevance of information, in this case, climbing the tree, and (2) judge what information is required to properly orient the listener. Recently, Berman (2003c) documented a steady increase in the use of past perfect markers in written narratives of students from late grade school through high school. Other studies citied in this chapter are likewise designed to examine the ability of older children and adolescents to deploy these and other more marked grammatical structures in a discourse-embedded context. The key word is deploy. Syntax is seen as a tool for participating in the full array of discourse types that characterize adult levels of ‘linguistic literacy’ (Ravid & Tolchinsky 2002). Another topic that has been explored in older children’s language development is grammatical well-formedness. Studies of naturalistic language reveal that young children have mastered basic grammatical morphemes prior to formal schooling (Brown 1973). However, children’s performance on more decontextualized comprehension and production tasks reveals that it takes many more years to acquire adult-like grammatical representations (Chomsky 1969). The fact that children are more error-prone in writing than in speaking (Scott & Windsor 2000) underscores the influence of contextual and processing variables on grammar. Metasyntactic abilities, the conscious awareness of syntactic structure, develop throughout the school years, as shown on grammaticality judgment/correction tasks, and by children’s ability to properly punctuate and edit their writing. Against this background, the topic of this chapter is syntactic development in children with developmentally-based language impairment, a group for whom language is the primary domain affected.1 Epidemiological research confirms that a large number of young children, as many as seven to eight percent of all children of kindergarten age, experience difficulties understanding and producing spoken language, even when they appear to be progressing normally in other developmental domains (Tomblin et al. 1997). A core feature of their impairment is difficulty learning syntax, a problem that is quite transpar-

Syntactic ability in children and adolescents

ent in younger children, whose utterances are shorter, less complex, and more likely to contain grammatical errors such as omitted inflectional morphemes and function words. To adults, the preschool child with this type of impairment ‘sounds like’ a younger child. An extensive literature documents the fact that developmental language impairment is a long-lasting condition. Preschoolers with language impairment are at high risk for reading and writing disorders (Catts et al. 1999) and substantial numbers experience academic difficulties requiring special education resources as they progress through school (Stothard et al. 1998). Once the academic consequences of language impairment were known, research on older children intensified. This chapter reviews recent findings in two distinct areas of syntactic development in school-age children with language impairment who are monolingual speakers of American English. The first topic concerns difficulties with clause-level obligatory structures which, when omitted, result in ungrammatical sentences (Section 2). Recent research has shown that morphosyntactic difficulties, a core feature of language impairment in young children, continue into the grade school years. Other problems with sentence construction that result in ungrammatical sentences are briefly reviewed, and the significance of error rate is considered. The second topic is clausal connectivity (Section 3). The ability to combine clauses inside sentences is a major contributor to syntactic complexity. Studies that analyze connectivity in the broader contexts of the influence of linguistic genre and modality are of particular interest. In analyzing both grammatical errors and connectivity, concern is with language production rather than comprehension. For reasons of space and focus, studies that require conscious (metalinguistic) awareness and manipulation of language structures are not addressed, despite their importance and relevance to the present volume. For a recent review of such studies, see Scott (2004).

. Grammatical deficits . Command of grammatical morphemes A central finding in the language impairment literature is that young children, when compared with age peers and with language peers, show a lower accuracy of grammatical morphemes in obligatory contexts (summarized in Leonard 1998). The kind of grammatical morphemes that have been investigated for English (after Brown 1973) include markers of tense and aspect on verbs, number





Cheryl M. Scott

and possessive inflections on nouns, copular and auxiliary forms of be, articles, and the prepositions in and on. Whereas typically developing children by the age of four produce these grammatical morphemes with near-perfect accuracy, same-age children with language impairment (LI) are less accurate, particularly for grammatical markers attached to verbs. For example, verbs were produced with less than 50% accuracy compared to nouns (70% accuracy) in a study of preschoolers with specific language impairment (SLI) by Bedore & Leonard (1998), while Rice & Wexler (1996) reported an even lower accuracy rate of 33% for verbs in 5-year-olds with SLI. This disproportionate difficulty with verb tense has been interpreted as supporting a competence-based explanation – the ‘Extended Optional Infinitive’ (EOI) account – in which children with SLI extend a period of optional marking of verb finiteness (tense or agreement) that is characteristic of typically developing children but for a much shorter period. EOI results in a longer period of incomplete specification in the underlying grammatical representation of children with SLI. The question of interest here is how much longer. Rice and her associates have now followed a group of children with SLI from the age of 2;6 through 8;9 years, extending previous accounts by approximately three years (Rice et al. 1998), with the methodological advantage of a longitudinal design. They measured children’s use of a group of structures tied together as obligatory markers of verb finiteness, including third person singular -s, past -ed, and copular and auxiliary forms of be and do, in both spontaneous and elicited contexts. Growth modeling was used to track change over time for each child, and for the group as a whole. By the age of 8, children with SLI had clearly improved across all morphemes to reach a composite accuracy rate of 89%. Still, there were no ‘spurts’ of growth, and in spite of continuous enrollment in language intervention, the SLI group had not yet caught up entirely to either age- or language-peers whose difficulties on these forms were negligible. The authors interpreted their finding of similar patterns of growth across this age range for different morphemes that share a common tense-marking function as supporting the EOI account. A recent investigation by Windsor et al. (2000) extended the age envelope for the investigation of morphosyntax three years beyond the data on 8-yearolds in Rice et al. (1998). Windsor et al. studied 20 children with language learning disabilities (LLD) with a mean age of 11;5 (range 10;0 – 12;11) who had significant oral language and reading difficulties, but scored within normal limits on a test of nonverbal cognition. Subjects were shown video films depicting narrative and expository content, and then asked to provide both spoken and written summaries of each.2 In order to preserve the distinction between

Syntactic ability in children and adolescents

grammatical morphemes that attach to nouns versus verbs, separate analyses were conducted for noun composite (regular plural -s and articles a, an, the) and verb composite (regular past -ed, third person singular -s, and be used as a copula or auxiliary) morphemes. As expected for typically developing children (both age peers and language peers who were several years younger than the LLD group), the mean percentage of error across all noun and verb forms was close to zero in the spoken samples. Although the LLD group registered 2.6% noun and 3.6% verb error rates when speaking, for all practical purposes, they had mastered these forms. In the written samples, however, the comparable LLD error rates were 4.3% for the noun composite, and reached a high rate of 13.9% on the verb composite. Specifically on regular past tense -ed, writing exacerbated the error rate by a factor of six (3.9 % in spoken contexts versus 26.2% in written contexts). Writing also raised the error rate for regular plural -s from 1.2% to 12.2%. In marked contrast, writing had no such effects for ageand language-matched children, whose highest error rate of all the items tested was 5.0% on regular plurals (by the younger, language-matched group). That is, unlike language impaired children, these typically developing children had essentially mastered English verb and noun morphology in written language by age nine. Examples of typical morphosyntactic errors in texts produced by children in the research population are shown in Table 1 (Nos. 1–6). Studies designed to shed light on underlying causes of morphosyntactic errors suggest that if grammatical errors are distributed differently in relation to factors like discourse context, lexical features (e.g., frequency or word-level phonology), subject characteristics, and linguistic task (e.g., elicited tasks such as generalization or learning paradigms), then perhaps performance factors explain the problem. Evans (1996) designed a study to examine the effect of subject characteristics (SLI subtype) in a comparison of 7-to 10-year-olds with expressive-only impairments versus those with expressive-receptive impairments. She reported significant effects of group on error rate and type in naturalistic spoken language. The study also showed that a discourse variable – the position of an utterance in a discourse sequence – affected error rate. Previous research had shown that the first utterance of a conversational turn was more demanding than later utterances, perhaps because placement immediately following another speaker’s utterance imposes additional timing, topic, or paralinguistic demands. For both SLI subtypes studied by Evans (1996), most grammatical errors were omissions of obligatory items, but the children with expressive-only SLI were more likely to omit bound morphemes (72% of all omissions) over closed-class words (pronouns, prepositions, articles, conjunctions, and auxiliary verbs). In contrast, children with mixed problems (ex-





Cheryl M. Scott

Table 1. Examples of grammatical errors in spoken and written language of school-age children (age 11 and 12) with LLD. Source: Scott et al. (1998). Example

Description

1.

Yanis lived in a small village/ his father was a farmer/he help out around the house a lot (narrative, written, age 11)

Regular past tense: omission

2.

And then no one really accept him because they already had help from their sons (narrative, spoken, age 11)

Regular past tense: omission

3.

And he was walk up on the mountains (narrative, written, age 12)

Progressive aspect: omission

4.

There’s one animal that’s likes eating cactus (expository, spoken, age 11)

3rd sing: double marking

5.

They would say that it is 60 year old (expository, written, age 12)

Regular plural: omission

6.

They talked about how wide * snake’s mouth can go (expository, spoken, age 11)

Article: omission (*a)

7.

And he lost some sheep/ so he went off to find it (narrative, written, age 12)

Pronominal reference: number agreement

8.

Plants and animals had learned to adapted to the scarceness of water (expository, written, age 12)

Improperly marked nonfinite verb

9.

It talks about how cactuses are structure (expository, written, age 12)

Passive voice verb structure: omission of past participle

10.

But it was his first time * sleep anywhere else

Infinitival (*to): omission

11.

So *just stuck with him (narrative, spoken, age 11)

Argument structure: omitted pronoun (*it) as obligatory subject

12.

Or else it will just go into * if it think there’s food in it if it’s hungry (expository, spoken, age 12)

Pronoun head (*it) of prepositional phrase: omission; 3rd sing: omission

13.

When he went to sleep that * all he thought of was going to sea (narrative, written. age 11)

NP head (*night) of adverbial (optional) constituent: omission

14.

One day when he brought the * to the mountains a goat ran away (narrative, written, age 12)

NP head (*goats): omitted

15.

The animals have a way of protecting themselves by color * places to hide (expository, spoken, age 12)

Phrasal coordinator (*and): omission

Syntactic ability in children and adolescents

Table 1. (Continued.) 16.

And one day when he made his own fishing pole and went fishing (narrative, spoken, age 11)

Incorrect use of subordinate clause and coordinate clause together

17.

When the cool air from the ocean goes over the mountains it either drops summer rain *

Omitted compound coordinated clause (*or + clause)

pressive and receptive) were more likely to omit closed-class words rather than bound morphemes (34% of all omissions). Similarly, the discourse constraint of utterance placement interacted with language subtype so that across-turn utterance locations exacerbated errors for the expressive-only group alone. Marchman et al. (1999) asked 8-year-olds with SLI and their age-matched peers to participate in a cloze task designed to elicit verbs that take both regular and irregular past tense and also differ in frequency, stem-final phonology, and phonological context features. The SLI group not only made significantly more errors overall, but differed from age peers in terms of error pattern, producing more nonvalid errors (e.g., present progressive for past) and zero-marking (omission) errors. Suffixation (e.g., the bell ringed) did not distinguish the groups. Although several word features exerted similar effects on both groups (e.g., word frequency and stem attributes predicted zero-marking), suffixation was predicted by stem attributes only in the SLI group, suggesting hypersensitivity to phonological features in this group. Other studies have tested morphosyntax with nonsense words, on the assumption that the ability to generalize bound morphemes to novel items is a powerful demonstration of the rule-governed nature of morphosyntactic competence. As early as 1977, Vogel showed that second-grade boys with dyslexia had difficulty inflecting nonsense words, particularly those that required a syllabic morpheme (e.g., cubashes, foozes, howted). Bellaire et al. (1994) also found group differences on a morpheme generalization task between schoolage children with LI (age 10) and age-matched peers. In an extension of the generalization task, Bellaire et al. first taught their subjects novel words within a story context, and later cued the children to affix novel inflections to the same words. Again, the LI group showed significantly more difficulty. These findings were interpreted as suggesting that school-age children with LI retain a core morphosyntactic deficit – difficulty in learning the rule-governed aspects of inflectional morphology. The gradual progress made by these children in spoken language contexts may in fact reflect only repeated practice with a small closed set of bound morphemes on high frequency words (Bellaire et al. 1994: 275).





Cheryl M. Scott

In sum, there is ample evidence that school-age children with LI continue to experience significant difficulty with morphosyntax, at least through the mid-grade school years, and throughout grade school in written language contexts. As with younger children with language impairment, a strong case can be made that a signature feature of the problem in older children is difficulty with verb finiteness marking. Evidence includes (1) differential accuracy rates of verb morphosyntax as opposed to nouns (reflecting a general difference in the processing demands of verbs compared with nouns, as discussed by Dockell & Messer, this volume); (2) differential error rates of the omission type specifically for verb tense (as predicted by the EOI account);3 and (3) difficulties on morphosyntactic generalization and learning tasks, all of which have been documented in school-age children with LI. Further, when studies have included language-matched controls in addition to an age-matched group (as in Rice et al. 1998; Windsor et al. 2000), the children with LI consistency made more errors on verb morphosyntax as compared to both matched groups. Yet, despite the evidence that has been marshaled to argue for an underlying competence-based account of such difficulties, performance-based processing accounts cannot be dismissed out of hand. Results of research revealing effects of such factors as verb frequency, phonological complexity, and task are difficult to explain without taking into consideration possible contributions of processing capacity (see Norbury et al. (2001) for pertinent data and discussion). The exacerbation of difficulties in the written language of children with language impairments is of particular interest here. To date, studies tracking written as well as spoken contexts have not been designed to address the debate about the origin of error, but it could be argued that there is indirect support for aspects of both competence and performance-based accounts. Windsor et al. (2000) reported that accuracy rates for regular past -ed differentiated children with LLD from both age and language controls, a finding all the more striking because typically developing second and third grade children are accurate spellers of verbs marked for past -ed (Carlisle 1996). The fact that morphosyntactic errors in writing, particularly finite tense marking, seem to assume the status of a true clinical marker has traditionally been interpreted in the SLI literature as support for a focused problem of grammatical representation. As Bellaire et al. (1994) reasoned, if the core representational deficit remains, it is not surprising that the problem surfaces when children’s knowledge must be revealed in different ways, as required by spelling. Performance-based, limited capacity accounts can also be supported because writing is frequently cited as the premiere example of a behavior that requires simultaneous atten-

Syntactic ability in children and adolescents

tion to multiple domains. Writing calls for the coordination of handwriting (or keyboarding), spelling, sentence formulation, and text construction, combined with concurrently imagining the level of audience knowledge about the topic – all capacities which are beyond the abilities of language impaired children even at late school age. Given these demands, as Windsor et al. concluded (2000: 1332), morphosyntactic errors in writing may reflect different processes than those of speaking, and thus entail explanations beyond either account. . Other types of grammatical error Although morphosyntax has been the focus of research on continuing grammatical difficulties in school-age children with language impairments, there are a few studies of other grammatical problems. One area of investigation is argument structure, an inherent semantic property of lexical verbs that determines the number of obligatory semantic/syntactic constituents entailed – e.g., subject, direct object, indirect object (see Khorounjaia & Tolchinsky this volume). For example, omission of a grammatical subject in a subject-requiring language like English has been investigated. Further, King & Fletcher (1993: 351) reported that SLI subjects between the ages of 7 and 10 produced argument errors in conversational speech – like those in examples (1) and (2) below – at a slightly higher (but statistically nonsignificant) rate than typically developing children age 3 to 7, thus indicating an overall delay in this domain: (2) You can take over there. (3) There it’s something white but I can’t find.

In both sentences, a direct object complement required by the verb (take, find) is omitted. Argument errors also applied across a wider range of verbs for the SLI group compared with typically developing children, whose errors were largely limited to the verb put. Although the overall rate of argument errors in SLI was low (3%), King & Fletcher argued that their findings refute the claim that the grammatical difficulties of children with SLI are limited to the domain of morphosyntax. Schuele & Nicholls (2000) tracked the emergence and use of subject relative clauses in three children with a diagnosis of SLI, members of an extended family. Five siblings without the diagnosis from the same family were also observed, while two of the three children with SLI were followed longitudinally between ages 6;6 and 8;0. The SLI children had high rates of omission of obligatory subject relative markers (that or WH pronouns), but this had disappeared



 Cheryl M. Scott

by the last sampling period. The third child with SLI, followed from 3;0 to 7;11, did not use the relative marker until age 5;3 and had not mastered the form at the last sampling point. Although one of the five typically developing relatives had some inconsistent omissions, the other four always used obligatory relative markers. Schuele & Nicholls interpreted their findings as consistent with a ‘functional categories deficit’ explanation (Eyer & Leonard 1998), a variation of a competence-based account. Support for this interpretation included the inherent properties of the relative marker (it serves as the head of a functional category in the grammar), its omission as unique to SLI, and the operation of genetic heritability. Rather than concentrating on one particular problematic structure or set of structures, Scott et al. (1998) constructed an inclusive typology of all errors in their database of spoken and written, narrative and expository discourse produced by children with LLD (average age of 11;5 years, range 10;0 – 12;6), with analyses extended to age-matched and language-matched peers for purposes of comparison. (See earlier reference to this same database in the study by Windsor et al. 2000, in Section 2.1 above). Scott et al. (1998) found that morphosyntactic errors accounted for two-thirds of all errors for both LLD and age-matched groups, and half of all errors for the language-matched group. Even within the morphosyntactic category, however, errors extended beyond the group of structures defined as finiteness markers highlighted by Rice et al. (1998). Several items in Table 1 provide examples. In #8, a child over-marks a nonfinite verb for past; the passive voice past participle marker is omitted in #9; and in #10 the infinitive marker to is omitted in a nonfinite complement clause. Other grammatical deficits found by Scott et al. (2000) included: difficulties with pronominal reference, verb-argument structure omissions, and problems with coordination (phrasal and clausal) and subordination. Similar to verb-argument errors described by King & Fletcher (1993), the LLD group omitted obligatory subjects or objects (cf. Table 1, #11, 14); and there were cases where children dropped the head in noun phrases functioning as adverbial constituents (cf. Table 1, #12, 13). On the surface, omissions in examples (12), (13), and (14) appear to be performance rather than competence-related since at least some part of the relevant constituent is produced in each case (the preposition into, the demonstrative that, and the article the respectively). Another category of errors revealed difficulties with coordination and subordination as illustrated in Table 1, #15 to #17. Summing across all error types, the LLD group made significantly more errors than either control group, indicating clearly that grammatical error is a persistent feature of language impairment in older school-age children that clearly sets them apart from those with

Syntactic ability in children and adolescents

typical language development. Once again, and for all groups, errors were exacerbated in writing. Genre and modality interacted to bring about a higher error rate in the expository compared with the narrative texts, but only in written language. . Error rate While grammatical errors are thus clearly characteristic of later language development in LI populations, the question remains as to how prominent or noticeable such errors actually are in naturalistic language use, for instance in conversations and in school language contexts (groups discussions, oral reports, written homework). The prominence of ungrammatical utterances/sentences is of interest beyond statistical analyses that establish group differences, the latter being the typical research concern. Two variations of error rate are noted in the literature. Researchers concerned only with morphosyntactic errors mainly report percent correct occurrence in obligatory contexts, where the numerator is the number of correct tokens of a grammatical morpheme and the denominator is the number of contexts/opportunities for the morpheme to occur (by extrapolation, % error can also be determined). In monologic speech samples of children with language impairment, error rates are reported to decline between 8 and 11 years of age from 11% to 3% (Rice et al. 1998; Windsor et al. 2000). However, such errors are a larger presence in written language for the 11- to 12- year-olds, where in one study error rates on all verbs reached 13.9% and as high as a quarter (26.%) of all regular past tense -ed contexts (Windsor et al. 2000). Analyses limited to morphosyntactic errors in obligatory contexts fail to accommodate a full range of errors. Thus, another group of studies of older children with language impairments have reported error rates that include all discernable instances of grammatical error. In most of these investigations, an error count has been converted to an error rate (errors per utterances/ sentences) to control for differences in sample size. Scott & Windsor (2000) calculated overall error rates on the same database described earlier for Windsor et al. (2000). Table 2 shows comparisons of error rates across groups, genre (narrative and expository), and modality (spoken and written). In each of the four comparisons (spoken narratives, spoken expository, written narratives, and written expository), children with LLD produced three to four times the amount of error recorded for the two control groups (chronological and language-age peers). In spoken texts, significantly higher error rates distinguished the LLD and normally developing older children, whereas in written



 Cheryl M. Scott

Table 2. Error rates (means and standard deviations) reported by Scott & Windsor (2000) for LLD (language learning disabled), CA (chronological age match) and LA (language age match) groups. Error rates were calculated by dividing total errors by the number of utterances/sentences in the units. LLD

Narrative CA

LA

LLD

Expository CA

LA

.128 (.090)

.031 (.028)

.041 (.035)

.145 (.130)

.029 (.033)

.061 (.068)

.357 (.224)

.119 (.142)

.153 (.138)

.438 (.347)

.152 (.176)

.178 (.134)

Spoken Error rate Written Error rate

texts, the LLD group had a higher rate than both the control groups. Genre differences failed to reach statistical significance, although mean error rates for expository texts were higher than for narratives. The main effect for modality was significant, confirming the damping effect of writing on grammatical accuracy. Gillam & Johnston (1992) considered the effects of both modality and complexity on grammatical error in narrative texts produced by children with language impairments aged 8 to 12 years, defining a complex utterance or sentence as one containing two or more clauses. Both variables had a significant impact on grammatical accuracy. Whereas relatively few (12.3%) single-clause spoken utterances produced by LLD subjects contained errors, over threequarters (78.3%) of all complex written sentences had one or more errors. Roth & Spekman (1989), who found grammatical errors in 14% of complex sentences produced in spoken narratives, reported that this was one of the few measures that distinguished 8- to 12-year-olds with language impairment from their age peers. Accordingly, there is agreement across studies on overall error rates for spoken monologic discourse of school-age children with language impairment between the ages of 8 and 12 years, with three studies reporting rates that exceed 10% of all utterances. Further, in the two studies that compared spoken and written texts, both report significantly increased error rates when children write. In answer to the question posed at the outset of this section, educators and clinicians no less than researchers are sure to notice such errors when listening to these children. In written language, with error rates as high as 44% of all sentences in expository texts (Scott & Windsor 2000) and 78% of complex sentences in narrative texts (Gillam & Johnston 1992), there is no question that grammatical error would be considered a major problem.

Syntactic ability in children and adolescents 

. Connectivity in literate contexts Of the many ways that sentences and extended discourse can gain in complexity at word, phrase, and clause levels, inter-clausal connectivity – combining clauses by coordination or subordination – is a major contributor. In monologic discourse, the sole responsibility for connectivity lies with the options selected by speaker-writers for this purpose. Speakers can juxtapose two clauses without any connective, leaving it to the listener to infer the logical relation between them, or they can choose to explicitly mark the relation via a subordinate clause headed by a conjunction like because, provided that, unless. The challenge to children is learning what forms the language makes available for this purpose and which are best suited to particular types of meanings, texts, and spoken versus written modalities (Berman 1998). The analysis of connectivity in child language generally and language impairment specifically differs fundamentally from the issue of morphosyntactic errors (Section 2) that render a clause ungrammatical. In the absence of grammatically obligatory marking, there is no ‘gold standard’ for what constitute desirable frequencies and varieties of linguistic forms, and many of the literate constructions of greatest interest are low frequency structures in the ambient language (Ravid this volume; Scott 1988). Two lines of research, however, provide a foundation for the analysis of connectivity in children. First, discourse linguists interested in form–function interactions have described structural features that are broadly characteristic of different discourse genres and modalities. For example, in his studies contrasting spoken and written language, Halliday (1985, 1987) argued that written sentences are denser both lexically (a higher proportion of open-class, content words) and phrasally (a larger number of complex nominals). He illustrates this by comparing the sentences in (4) and (5) (1987: 62): (4) Every previous visit had left me with a sense of the risk to others in further attempts at action on my part. (More written) (5) Whenever I’d visited there before I’d end up feeling that other people might get hurt if I tried to do anything more. (More spoken)

Finite verb predications in the spoken version in (5) are nominalized in the written version in (4) yielding a denser, more compact form of expression (e.g., other people might get hurt versus the risk to others; I’d visited there before versus previous visit). In addition to differences attributable to modality (speaking or writing), others have investigated modality and genre in combination. One question in

 Cheryl M. Scott

this respect is whether a particular type of discourse is more or less complex than another. Biber and his colleagues have addressed this issue (e.g., Biber et al. 2002) by means of multi-dimensional analyses of several genres of written and spoken language. They found variation in the frequencies of structures in both spoken and written texts, with the potential for structural variety, including connectivity structures, particularly striking in writing. In fact, Biber (2001) concluded that the fundamental difference between speaking and writing is not a matter of complexity per se, but rather has more to do with the potential for variation in complexity. Writing affords considerable freedom to draw on an elaborate array of complexity features called for by the type of writing (e.g., newspaper language versus academic prose). Speaking, by virtue of its on-line temporal constraints, is more restrictive, regardless of genre. The applicability of work like that of Halliday, Biber, and their colleagues to the study of syntax in school-age children with language impairment is apparent. What criteria should be used to say that a child or adolescent has mastered a particular structure or category? Mastery in this context means more than the occasional use, but rather the effective use as needed across the full range literate genres (Berman 2001 this volume). Miller (2002) illustrated the challenge involved in accounting for full mastery by detailing the varieties of relative clauses that would have to be tracked during advanced periods of language growth – many that are found almost exclusively in written language. Miller’s view that written language is not “just a small add-on” (2002: 472) for a nativist account of first language acquisition may find increasing support as more is learned about literate forms of language and the lengthy acquisition period needed to master them (Ravid & Tolchinsky 2002). The second type of research effort that applies even more directly to the study of the syntactic aspects of literacy acquisition in LI populations are studies of typically developing older children and adolescents that span a wide range of school years and sample language in a variety of literate contexts. Ideally, these contexts would be further designed to control for the independent effects of genre and modality. Results are beginning to be published from a largescale cross-linguistic study that meets all three criteria – a wide age span, literate language variety, and a design that permits study of the independent effects of genre and modality (Berman & Verhoeven 2002a; Jisa this volume; Khorounjaia & Tolchinksy, this volume). Several papers from this project on typically developing children, adolescents, and adults in seven languages (including English) deal specifically with clausal connectivity and provide a backdrop for the following section, which explores connectivity in school-age children with LI,

Syntactic ability in children and adolescents

based on recent findings of this author’s research on discourse-related syntactic complexity (Scott 2002; Scott 2003; Scott & Windsor 2000).4 . Clause connectivity Clause connectivity measured by the average number of clauses per sentence has often been used as a measure of overall syntactic complexity in schoolage children with and without language impairments.5 Scott (1988) reviewed findings for typically developing school-age children and adolescents, which generally show slow but steady increases for both spoken and written language across the entire grade-school and secondary school years. Findings for children with LI are less clear. Several studies of spoken narrative discourse have reported significantly lower clausal connectivity (Gutierrez-Clellen 1998; Simms & Crump 1983), while others report no difference (Gillam & Johnston 1992; Roth & Spekman 1989). Morris & Crump (1982) reported lower clausal connectivity in the written language of children and adolescents with learning disabilities. Three studies reporting quantitative results on connectivity have compared spoken and written language in the same subjects. Loban’s research (1976) on typically developing students identified a ‘crossover’ period in which the number of clauses per sentence is at first greater when speaking but then is overtaken in writing as students gain fluency (although there is some debate about the timing of this crossover; see Scott 1988). This expected crossover had not occurred for 4th through 6th grade children with LI studied by Singer (1997). In fact, 9- to 12-year-old subjects with language impairment studied by Gillam & Johnston (1992) showed reduced connectivity in writing compared to speaking, whereas the opposite – more clausal connectivity in writing – was true for age and language-matched groups. Scott (2003) applied a fine-grained analysis of clause connectivity to the Scott & Windsor (2000) data-base of narrative/expository and spoken/written discourse (described in Section 2). The analysis centered on determining frequency rates for five types of connectivity that included: adverbial subordination (temporal and logical varieties) [ADV] nonfinite complementation [NFIN] finite complementation [FIN] coordination [COORD] relativization [REL] and other types of noun post-modifying clauses [PMOD]



 Cheryl M. Scott

In addition to the frequency and distribution of major types of connectivity, the analysis was constructed to reveal the extent to which children create connectivity depth, defined as those instances when a clause is subordinate to a clause which is itself a subordinate clause. To illustrate, in the sentences in (6) to (9) below – from the Scott & Windsor (2000) data-base – successive clauses are labeled for type and depth by a code inserted after the head of each clausal verb. The matrix (independent) clausal verb is marked as [MVERB]. For example, the code [ADV1] signifies an adverbial clause that is immediately subordinate to the matrix clause (a depth of 1); a code of [ADV2] signifies an adverbial clause that is subordinate to a clause that is itself subordinate to the matrix clause, etc. (6) Sometimes because the hot big earth can’t absorb [ADV1] that much water because they’re not used to [ADV2] it just runs off [MVERB]. (spoken, expository) (7) And then there’s [MVERB] a badger that usually gets [REL1] underground animals that you have to like dig out [REL2] from underground because they like go [ADV3] underground so they don’t get eaten [ADV4]. (spoken, expository) (8) Finally a woman pitied [MVERB] Yanis and told [COORD] him of her husband who was [REL1] a sailor and told [COORD] him to go back [NFIN1] to his (um) mother ‘cause she was getting [ADV2] worried. (spoken, narrative). (9) One day when Yanis was tending [ADV1] the goats in the mountain, because the city that they live [REL3] in was [ADV2] on a mountain, one of the goats ran away [MVERB]. (written, expository)

The analysis revealed a significant main effect of group on the rate of (1) relatives and other post-position noun-modifying clauses, (2) finite complement clauses, (3) coordinated clauses, and (4) overall connectivity rate (summed across all types). Post hoc analysis confirmed that in each case, subjects with language impairments produced significantly lower rates of connectivity than their age peers. Adverbial clauses were also produced at a lower rate by LI subjects, although the difference did not reach statistical significance. The LI group did not differ from their language-matched peers, who were 2.5 years younger. Nonfinite complements (e.g., he started to talk to his dad) are early developing forms of connectivity (Loban 1976), so it is not surprising that this type did not distinguish the groups. Genre affected frequency such that adverbials and relative/post-modifying clauses occurred at significantly higher rates in expos-

Syntactic ability in children and adolescents 

itory texts, while all other types were more common in narratives. The fact that adverbial clauses as a group were more common in expository texts, in spite of the frequency of temporal and causal adverbials in narratives (Verhoeven et al. 2002), can be attributed to the rich variety of logical relations found in expository discourse including reason, causation, condition, and purpose. Modality effects were significant only for coordination, which was more characteristic of writing.6 A number of interactions of genre and modality with group were evident. For the typically developing group only, there was (1) a higher overall rate of connectivity in writing, (2) a higher rate of coordination in writing, and (3) a higher rate of adverbial clauses in expository texts. The fact that there was only one interaction of group and genre indicates that genre effects are generally pervasive across groups, a finding that agrees with Verhoeven et al. (2002) across a much wider age range. The intensity and reliability of genre effects on a wide variety of syntactic structures from the age of nine on has been documented in comparisons of narrative and expository discourse in different languages (Berman & Nir-Sagiv 2004). As expected, subordinate clauses with a depth of 2 or more (that is, subordinate to other subordinate clauses), occurred significantly less in texts produced by the language impaired group who resembled their language-matched, but younger peers. Subordination depth was more typical of narratives, but unaffected by modality. An analysis of the number of clauses that carry both coordination and subordination markers at a depth of 1 (or greater) revealed higher rates (1) for children with typical language, (2) for expository discourse, and (3) for spoken texts. To illustrate this type of connectivity, in the following sentence the bolded clauses are coordinated via the conjunction and while concurrently functioning together as post-modifying relative clauses for the noun animals: Cactuses have spikes that protect them from animals that would bite into them and get their moisture. Scott (2003) concluded that children with LI produce texts that fall short of their age peers in several ways including lower rates of later developing adverbial and relative clauses and less depth of subordination/embedding. Moreover, the typically developing 11-year-olds were beginning to show modality effects for connectivity favoring writing, but these were not yet apparent in LI children. Taken together, results suggest that the clause ‘packages’ constructed by LI children are not as complex as those of their age peers if complexity is defined as the number, depth, and variation of coordinate and subordinate clauses. The results are particularly revealing in relation to the findings of Verhoeven et al. (2002) who employed a similar design but compared two groups free of

 Cheryl M. Scott

language impairments and separated by ten years of development (9- to 10year-olds and graduate students). Language impairment has a negative impact on the same connectivity systems that are developmentally significant over the course of many more years. The finding that language impaired and language matched groups were indistinguishable on measures of connectivity contrasts with findings for clauseinternal features of grammar (morphosyntax) discussed in Section 2. Grammatical features that reliably distinguish children with LI from children who are their broad language peers (although younger) have attained the status of a clinical ‘marker’ in representational accounts of specific language impairment (Rice 2000). Within the bounds of simple clause structures and at a young age when language is in a period of rapid growth, competence models may provide a viable explanation for grammatical growth and difficulties. On the other hand, when older children’s use of language is examined in rich discourseembedded contexts, particularly monologic, we see that it takes many more years to acquire a flexible and effective repertoire of structures that allow for full linguistic literacy (Berman & Slobin 1994; Berman & Verhoeven 2002b; Ravid & Tolchinsky 2002), For language-impaired children, this process is even more protracted, but not ostensibly different, at least based on the small amount of evidence now available. A preliminary conclusion is that a competencebased model is not a particularly good explanation for clausal connectivity, which appears to engage different kinds of linguistic, perhaps processing, and undoubtedly cognitive, faculties. . Modality modifications of clause connectivity Another analysis recently completed by Scott (2002) concentrated on the effects of modality on clause connectivity. An analysis of spoken and written versions of the same propositional content is particularly suited to reveal whether there are independent effects of modality on sentence form (see examples (4) and (5) from Halliday). To reiterate, the methodology for the studies by Scott et al. discussed in this chapter required children with language impairments to speak and write about the same topics, based on narrative and expositorytype video content. Scott (2002) developed an analysis protocol where written and spoken T-units from each participant were matched if they contained similar propositional content and then analyzed for structural categories that have been shown to be cross-modality distinctive (Biber 1986; Halliday 1985, 1987; Perera 1984, 1986).7 Some of the syntactic differences that emerged from the

Syntactic ability in children and adolescents 

Table 3. Examples of syntactic contrasts between written and spoken versions of the same or similar content (from Scott 2002). T-units are separated by slash (/). Spoken version

Written version

1

(and um um) And then one day he was walking his sheep through the mountains/ and (uh he) one of the goats got away/

One day, when Yanis was walking his sheep through the mountain, one of the goats got loose/

2

And he doesn’t really do (an) anything with his friends or anything/ and he doesn’t listen to his dad as much/ and his dad’s (rea) realizing this/

His father notices that he does not play with his friends or listen to his father anymore/

3

And (um) sometimes if you dig far enough down the temperature is 20 degrees below what it is on the surface/

Under the ground it can be twenty degrees below surface temperature/

4

And like birds that lay their eggs some animals eat their eggs/ so they lay them in the plants that have thorns where they can get to but the animals that eat then can’t/

Some animals hatch their young in the middle of a cactus or something prickly so no animal can eat it/

5

And once cactuses die animals move into the cactus to live/

Animals make homes out of dead plants/

6

And they decide to let him go (to the) to the sea and try to be a fisherman/

Yanis’ parents, seeing he was sad, let him go off to the sea to fish thinking then he would see how hard it would be fishing and not want to anymore/

7

And the hawk is on top of the food chain/

The top of the food chain is the hawk/

8

There are rain storms every summer and winter/

Every summer and winter there are big rainstorms/

analysis are shown in Table 3 (both the spoken and written versions of each example were produced by the same subject). Examples (1) and (2) in Table 3 illustrate the replacement of coordination with subordination linkages. In #1, two clauses are joined by the coordinator and in the spoken version, but the subordinate conjunction when connects the clauses in the written match. Similarly, in #2, a sequence of independent coordinated clauses (spoken version) is reconfigured into an ‘umbrella’ finite complement clause (notices that. . .) consisting of two clauses coordinated by or with co-referential subject deletion (. . .or [null subject] listen to his father anymore). Comparisons in examples (3) and (4) illustrate that in writing, the same information can be recast in non-clausal form, yielding a more compact,

 Cheryl M. Scott

more densely packaged version with fewer clauses. This event-packaging results in fewer clauses and is accomplished in a variety of ways including (1) moving constituents into a pre-modifying slot before a noun (e.g., surface temperature in #3), and (2) deleting information that can be inferred from world knowledge (e.g., # 4 explicitly states that birds lay their eggs in a place that they can get to, but this somewhat obvious information needs to be inferred in the written version). Example (5) also illustrates clause savings, partly through nominalization (cactuses die becomes dead plants) and also by means of more specific lexicalization (animals make homes, a wording that implies creating a comfortable place to live, replaces animals move into the cactus to live). All these changes ‘save’ a clause or a phrase in the written version. It was also common for children to add adverbial clauses to the written version, thereby including information that was not in the spoken match or in surrounding T-units (see # 6). Other modality-influenced differences included constituent order changes to achieve different theme and focus relationships. In examples (7) and (8) the grammatical subject (the new information) moves to the end of sentence where clause-final position confers a degree of focus (spoken language can employ intonation to accomplish the same focus). After summing across these types of examples and adjusting for sample size, statistical testing revealed group differences that favored the typically developing group with respect to the frequency of modality modifications. The language impaired group and the younger language-matched group produced these spoken-written variations at only half the rate set by the typical group. Although there was no effect of genre on the total quantity of the variations, genre did affect the types of variations produced. These results yield the same picture as those outlined above for clausal connectivity (Scott 2003) and underscore the point that school-age children with language impairments are learning some of the syntactic connectivity features that are more characteristic of written language, but not on a par with their age peers.

. Causes and consequences Researchers interested in underlying mechanisms that account for the types of language impairments considered in this chapter have looked to fields such as linguistics, sociolinguistics, genetics, psychology, and cognitive neuroscience for explanation. As seen in Section 2, linguistic principles in a formal Chomskian paradigm are helpful in explaining the distribution and patterns of findings for grammatical deficits in children with specific language impairments

Syntactic ability in children and adolescents

(e.g., Rice et al. 1998). Genetic contributions to language impairments are well documented (e.g., Tallal et al. 2001; Tomblin & Buckwalter 1994). Recent family aggregation accounts document significant co-morbidity of language impairments and reading impairments in families of school-age children with these difficulties (Flax et al. 2003), a fact that highlights the continuity of syntactic difficulties long into the school years when fluent reading and writing skills become so critical for academic success. Experiential explanations for the findings in the area of syntactic complexity (clause connectivity and modality variations) are intuitively appealing, particularly in terms of factors like exposure to complex language patterns through silent reading and other literacy-related activities. It is hard to imagine that sheer volume of words read is not an important contributor to a child’s ability to turn that input around into literate output. Accordingly, since many children with LI are not grade-level independent readers, this source of input is limited in comparison with their typically developing age peers. As for other types of input, there is little available information, particularly on spoken input beyond the early years of language learning. A recent set of studies by Huttenlocher et al. (2002) on the effects of parent and teacher syntax is welcome and revealing in this connection. Although children in this research were younger (4–5 years) than the school-age group of interest here, they were nevertheless beyond the stage of rapid acquisition of simple clause structure and grammatical morphology. Huttenlocher et al. (2002) showed that the proportion of multi-clause utterances in parent input was a good predictor of children’s multi-clause utterances both at home and at school when measured a year later and, moreover, that children’s comprehension was highly correlated with their multi-clause production and was also related to parent input. Because parent input was associated with production in another setting (school) and with comprehension as well as production, their results suggest that parent complexity plays an important and basic role in language development, above and beyond a family ‘style’ of speaking, and beyond the earliest years of language learning. As further verification of the importance of input generally, Huttenlocher et al. (2002) reported that teacher input (multi-clause utterances and the number of noun phrases per utterance) predicted the children’s syntactic growth over the course of a year of preschool. Although there is an increasing volume of work on functional neuroanatomical substrata underlying syntactic processing, few studies relate directly to the types of grammatical differences discussed in this chapter. To date, research has emphasized reading comprehension, adult populations, acquired language impairments (e.g., aphasia), and off-line paradigms (rather than on-



 Cheryl M. Scott

line, or real-time paradigms). The complexities of studying brain-language relationships during naturalistic language production are understandable, given that current models of language production propose as many as six successive processing stages (Indefrey et al. 2001). There is good evidence, however, that syntactic complexity matters across a range of processing paradigms, and that it affects, for example, both neural activation and strength (for an adult production task, see Indefrey et al. 2001). Syntactic complexity also affects processing strategies – in comprehension tasks with school-age children (Booth et al. 2000). As methods for studying real-time language production improve, two types of research may be fruitful. These include syntactic priming studies (e.g., Branigan et al. 2000), and fine-grained analysis of sentence disruptions (e.g., Rispoli & Hadley 2001). The consequences of these lingering syntactic difficulties encountered in children with LI are difficult to pinpoint. There is a tradition in education and clinical fields for taking measures in several different language domains (e.g., vocabulary as well as syntax) and relating these to broader domains (e.g., reading achievement scores, school grades). In these efforts, however, syntax measures are typically derived from norm referenced tests (which stress word and sentence-level processes) or based on very general measures of syntax (e.g., mean length of utterance). Another approach is to investigate the relationship between isolated language measures and overall quality judgments by trained raters. Neither strategy has yet incorporated syntactic error and complexity measures of the type highlighted in this chapter. However, with large numbers of school-age children and adolescents failing to achieve adequate scores on state and national reading and writing tests (U.S. Department of Education 2003), which require discourse-level comprehension (reading) and composition (writing), there is reason to believe that syntactic abilities make an important contribution to academic success, as was shown early on by Loban (1976). Finally, information in this chapter has focused on naturalistic language production – only one domain of language ability. This restricted focus makes it difficult to integrate findings, in a direct way, with these same children’s grammatical comprehension skills or their metalinguistic capabilities. However, there is reason to think that production syntax may be indicative of performance on more metalinguistic tasks (e.g., Rice et al. 1999), and in comprehension (Huttenlocher et al. 2002). Confirmation of these possibilities awaits further research.

Syntactic ability in children and adolescents

Notes . Various terms are used as labels for children and adolescents who have primary language impairments, that is, impairments that are not caused by hearing loss, mental retardation, emotional/behavioral disturbances, or other major health conditions. Common research and educational placement criteria for children with primary language impairment include (1) performance within normal limits on a measure of nonverbal cognition, and (2) normal neurological, behavioral, and medical history, and (3) normal sensory and motor structure/function. The term specific language impairment (SLI) is most common in the research literature for preschool and young school-age children. For older children whose language difficulties impact reading and writing, the term language learning disability (LLD) is often used. For research cited in this chapter, the authors’ terminology/labels will be preserved; otherwise, the more generic term language impaired (LI) is used. . The expository video depicted a description of plant and animal desert life in the southwest United States. The children produced four language samples in all, a spoken and written narrative summary over the same content (a narrative video), and a spoken and written expository summary, also on the same content (an informational video). Because the genre difference triggered very different distributions of obligatory tense contexts (simple past tense in the narrative and generic present in the expository context), the genre distinction was collapsed for this analysis. . A recent study by Leonard et al. (2003) showed that children with SLI have even greater difficulty with past tense -ed than with passive participle -ed, consistent with an EOI account of selective tense-marking difficulties. Participants with SLI were between the ages of 4 and 6 years, younger than the ages of interest for this chapter, however. . The design of the studies by Scott et al. also controls for the independent effects of genre and modality because children spoke and wrote about the same topic. . The T-unit (terminable unit, after Hunt 1965) is the commonly used unit in language research with school-age children and adolescents. The T-unit is defined as the main clause and any attached dependent clauses. Use of the T-unit provides researchers with structurally-defined criteria for segmenting language into sentence-like units. . Coordination in this study was tallied only within T-units, and did not include the use of and in the initial position of the T-unit. Thus, in the sequence – His father wanted Yanis to carry on the family tradition and be a farmer like everyone else. And so he talked to his son – only the first instance of and was counted. . Whether speaking or writing the summary, subjects managed to convey the same basic content at a textual level. However, spoken summaries were considerably longer (sometimes twice as many words). At the level of individual sentences, then, there were many spoken sentences that had no written counterpart (in terms of basic content). The opposite, written sentences with no spoken version, also occurred, but, given the shorter overall length of written summaries, much less frequently.



Growing into academic French* Harriet Jisa

The term ‘academic French’ in the title refers to the type of language that is typical of, although not necessarily confined to, scholarly contexts including writing of dissertations, research reports, articles, and grant proposals; reading of such materials and also of expert writings in various disciplines; making oral presentations of scientific topics; and listening to lectures on such topics. Mastery of academic French occupies a center-stage position for students and teachers at all levels of education. Legislation concerning educational programs is instituted on a national level and results in official published texts which subsequently serve as fuel for hot debates among teachers, administrators, and in the mass media. In the present context, ‘academic French’ refers first and foremost to the type of language use necessary for school-based academic success from grade school across high school and on to the university level. The chapter aims to demonstrate how, in French, this involves a complex interaction between the variables of modality (writing versus speech), register (level of usage from formal and distant to everyday colloquial), and genre (monologic versus conversational discourse and expository compared with narrative text production). ‘Register’ is conceived of here as a key facet of linguistic expression – and an important feature of communicative competence, since it involves the ability to vary linguistic forms of use to suit the circumstances of their use. In a developmental perspective, mastery of the register appropriate to a given communicative context is one of the many facets involved in acquisition of communicative competence in a given speech community and the sign of a well-educated, literate individual in any society is that he or she has access to and command of a wide range of registers.

 Harriet Jisa

.

Introduction

Producing a text requires that speaker/writers make a number of computations in order to ensure that the discourse they produce is appropriate to the particular communicative situation. Any given communicative setting influences the forms that the propositions making up the text will take at all levels – phonological, morphological, lexical, and syntactic. Competent speakers of French know, for example, which situations require bagnole (slang for ‘car’) as against the more neutral, normative voiture ‘car’ or the more formal véhicule ‘vehicle’; j’ai pas faim ‘I’m not hungry’ (omitting the initial negative marker ne) as opposed to the normative version of the same utterance je n’ai pas faim; or vous voulez un café? ‘want some coffee?’ as opposed to voulez-vous un café? ‘Would you like some coffee?’ It is probably impossible to predict exactly when one form or construction will be chosen over other potential competing forms. Speaker/writer decisions relevant to such computations involve consideration of a multitude of factors, including: the interlocutors, their social status, and the relations between them, the institutional setting, the physical context, the goal and type of the discourse, and most important for the present study, the mode of production, written or spoken. . Learning to speak as you write: The key to academic success Cutting across differences due to modality of production are differences due to relative level of formality, ranging from informal to formal, which can be characterized as representing four register prototypes: informal spoken, informal written, formal spoken, and formal written (Bialystok 1991). Fully-established adult competence involves being able to establish correspondences between a given linguistic form and its appropriateness to a given communicative situation and hence to a particular choice of register. The knowledge necessary for constructing such correspondences, of course, does not emerge full-blown from one day to the next. Rather, it constitutes a long developmental process that continues across many years, possibly across the entire life-span. An important step in this developmental process is the achievement of literacy, learning to use language to encode information in writing and to extract information in reading (Ravid & Tolchinsky 2002). While it is interesting to study how spoken language affects children’s early writing, this study aims to show that it is no less interesting to consider how writing affects children’s spoken discourse. No less importantly, the ability to use lexical and morphosyntactic forms characteristic of highly planned, formal written discourse while

Growing into academic French

operating under the time constraints of spoken production is an important facet of academic success – for example, in such school-based activities as classroom discussion, oral class presentations, giving talks or lectures, and conducting debates or interviews. The present chapter investigates various aspects of French children’s language use, both written and spoken, with the goal of illustrating how learning to read and write has a profound impact on developing knowledge of language. To this end, different types of monologic texts elicited from French-speaking schoolchildren and adolescents are examined – picturebook-based oral narratives (Section 2.1) and two samples of expository texts, both oral and written (Sections 2.2 and 2.3). . ‘Standard’ French Though the literature dealing with register proposes many different definitions, there is a general consensus that register correlates with situationally defined language usage (Andersen 1992; Ferguson 1994; Biber 1995). Register variation is both a question of circumstances (the discourse context and communicative goal underlying a given piece of discourse) and of expression (the linguistic means used to encode a given discourse content) (Berman 2003b). Register differences are the subject of considerable controversy in sociolinguistic research in general, and in relation to variation in French usage in particular. Recent research in French linguistics has devoted considerable attention to the problems inherent in applying the terms ‘standard’ and ‘non-standard’ language usage. Lambrecht (1981: 13–14) uses ‘Non-Standard French’ to refer to the spoken language, so as to avoid reference to the notion of ‘Standard French’, which has generally been associated with the written language. Blanche-Benveniste (1990: 207) suggests using the idea of grammaire première ‘first grammar’ to refer to the grammar that all speakers acquire before going to school and grammaire seconde ‘second grammar’ to refer to the language to which speakers are exposed in school. While all French speakers will acquire a ‘first grammar’, mastery of the ‘second grammar’ varies from one individual to the next, and across different groups of speakers. The problem of where to draw the line between standard and non-standard French is compounded by the confounding variables of differences in dialect (due to social class and geographical origin), in register (literary or formal versus informal or colloquial), and in modality (speech versus writing). While in theory it is relatively easy to identify structures which represent the extremes of a continuum ranging from spoken informal at one end to written formal at the other, in practice the categorization of a given form or usage



 Harriet Jisa

as more or less characteristic of a particular type or situation of use requires careful analysis of the form, its variants and its functions in many contexts. Only after examining a large corpus, ranging over several different situations, is it safe to consider a given form as an index of a given level of formality. Such careful, empirical study of French in both the spoken and written modalities is just beginning. The present chapter aims to contribute to this endeavor by examining empirical evidence from the development of written and spoken French by monolingual, non-expert writers. . Variable usage: Past participle agreement marking and two forms of future tense Blanche-Benveniste (1995) points to a number of structures that show variation in French usage, whose distribution has, in her view, been oversimplified as a simple choice between formal or informal register. Two examples of such variable usages are agreement marking on the past participle (1.3.1) and distribution of the so-called ‘simple future’ versus the periphrastic future (1.3.2). In the first case, there is evidence that the rule governing past participle agreement differs in formal compared with informal French, whereas in the second, the two future markings which might appear to constitute no more than a register distinction in usage, in fact also reflect a difference in meaning. .. Past participle agreement Sweeping generalizations such as ‘French speakers no longer indicate the agreement between past participles and their objects in spoken French’ or ‘Only highly educated speakers continue to mark agreement between the object and the past participle’ are often not entirely accurate. In normative varieties of French, the past participle requires obligatory agreement with the direct object when the direct object, or a direct object clitic pronoun precedes the participle. For example, a sentence such as Le garçon avait apparemment capturé une petite grenouille qu’il avait mis dans un bocal ‘The boy had apparently caught a little frog that he had put in a jar’ uttered by a French-speaking adult would be considered ungrammatical in normative French, and the relative clause should be replaced by une petite grenouille qu’il avait mise dans un bocal, with the past participle mise ‘put, placed’ marked for feminine singular to agree with the feminine object noun grenouille ‘frog’. For some verbs, this distinction is audible, as in the case of masculine mis, pronounced [mi] (unless followed by a vowel initial word) versus feminine mise, pronounced [miz]. For most verbs, however, agreement marking on the past participle – for feminine gender or plural num-

Growing into academic French 

ber – fails to yield any audible distinction (see Pacton & Fayol, this volume). For example, the past participle ‘kissed’ can be written in four different ways depending on its direct object: masculine singular embrassé, masculine plural embrassés, feminine singular embrassée, and feminine plural embrassées, as in la/les fille(s) qu’il a embrassée(s) ’the girl(s) that he kissed’ compared with le(s) garçon(s) qu’elle a embrassé(s) ’the boy(s) that she kissed’. Despite these different spellings, the four different written forms (embrassé, embrassée, embrassés, embrasséés) are pronounced identically. Gibier’s analysis of spoken French (1988, as cited in Blanche-Benveniste 1990) examined over 500 cases where agreement on the past participle is obligatory in written French, showing that factors pertaining to the educational status of the speaker and the speech situation (conversational interviews versus radio, television, and school situations) are relevant, but insufficient to explain the data. The examples in (1), from Blanche-Benveniste (1990: 204–206) show that when the position following the participle is filled, there is an overall tendency not to mark the agreement in spoken French, as in (1a); on the other hand, when the position following the participle is not filled, there is a quite generally tendency to mark the agreement, as in (1b). The past participles in question are bolded. (1) a.

Je l’ai pris parce que je croyais que personne la voulait [university student, talking about une gosse d’aïl ‘a piece of garlic, feminine gender] ‘I it took because I thought that no one wanted it’ b. pour les raisons que j’ai dites [university student, talking about raisons ‘reasons’, feminine gender, plural] ‘for the reasons that I said’

When the direct object refers to a feminine third person, even if the position following the past participle is filled, agreement is marked, as in (2a). If, however, the feminine direct object refers to the speaker or the listener, the agreement is usually not marked (2b). (2) a.

Ils l’ont mise à l’hôpital la fille [talking about a female university student] They him/her/it put in the hospital the girl They put her in the hospital, the girl’ b. des choses qui ne m’ont pas satisfait [speaker is a female university student]

 Harriet Jisa

things that me did not satisfy ‘things that didn’t satisfy me’

In written language, if the direct object (or a pronoun referring to a direct object) precedes the participle, agreement is obligatory, irrespective of what follows the participle, or of who is speaking to whom. Thus, it turns out that in fact there are two standards, one standard oral and one standard written, and they have different rules. .. Simple versus periphrastic future The two futures in French, the so-called simple, inflected future (e.g., il descendra ‘he will come down’) and the periphrastic future with the auxiliary verb aller ‘go’ plus infinitival verb (il va descendre ’he’s going to come down’) provide another interesting instance of variable standards and characterizations of register distinctions. Some scholars treat the choice between the two forms as a question of register (e.g., Lambrecht 1981), and in many cases this is true. However Jeanjean (1980) has shown that the two are not always interchangeable. For example, for the expression of general truths only the simple future can be used, as in Une femme sera toujours une femme ‘a woman will always be a woman’ versus the less acceptable Une femme va toujours être une femme ‘a woman is going to always be a woman’ – not entirely unlike the case in English will versus going to future forms, in fact (Bybee 1985). The periphrastic future is used when the projected action is either under way at the moment of speaking or will occur shortly after the moment of speaking. Compare, for example, the sentences in (3a) and (3b). (3) a.

je vais avoir un enfant I’m going to have a baby’ b. j’aurai un enfant ‘I will have a baby’

In uttering (3a), the speaker is probably already pregnant or is attempting to become pregnant, whereas in (3b) the speaker is projecting an action onto some indefinite or distant future. The choice between the simple and periphrastic future is thus not purely a question of register variation. . Lexico-syntactic indicators of spoken versus written French In contrast to these two examples, certain linguistic features can be relatively clearly categorized as either written or spoken forms. Several such examples

Growing into academic French

have been noted in the literature. While the negative particle ne is rarely used in spoken French, its absence in written French would shock most readers; and colloquial ça ’that’ is typically represented by cela in written French (BlancheBenveniste 1990). Another clear distinction is represented by the analytic or composite past tense passé composé, with an auxiliary verb plus past participle which is typically oral but also used in writing, as against the synthetic or inflected simple past tense form passé simple, which is found almost exclusively in written discourse. Some non-standard forms of relative clauses that are common in spoken French would also be considered totally unacceptable to most adult French readers (Gadet 1990). Another feature which seems to differentiate spoken from written French is use of nouns and pronominalization, particularly in subject position. .. Lexical nouns and pronominals French is described as having a basically SVO order (Hawkins 1983), yet in spoken French, the subject position is only rarely occupied by a lexical noun. The preferred, unmarked clause structure in spoken conversational French typically involves a subject clitic pronoun and often other pre-verbal clitics as well. François (1974) found a total of 1,550 nouns in a long corpus of conversation between members of a working class family, out of which only 46 (some 3%) were lexical subjects as compared with the vast proportion of 1440 clitic subjects (Lambrecht 1987). Jeanjean (1980) confirmed this finding on the basis of a corpus of casual conversation where she found an average of 11% out of total lexical nouns in subject position. Blanche-Benveniste’s (1990) comparison of a variety of discourse types gathered from a wide range of speakers revealed that an increase in lexical nouns in subject position is associated with a more elaborate discourse code. Gayraud (1998) shows that children as young as seven years of age are sensitive to this variable, and that they use significantly more lexical noun subjects in written than in spoken texts. Not only are nouns in subject position associated with written French, but nouns in other syntactic positions, too, contribute to a higher noun per verb ratio in written as opposed to spoken texts in general (similarly to what has been shown for other languages as well, cf. Ravid et al. 2002). Thus, BlancheBenveniste (1995) compared two different types of accident stories, one consisting of stories recorded in oral interviews, the other of accidents reported in the press. Both types of stories obviously contained nouns, but more lexical nouns and various types of nominalizations were observed in the press reports (Ravid & Cahana-Amitay, in press). Where verbs with single arguments were common in the spoken stories – for example, il est tombé et il est mort ‘He fell



 Harriet Jisa

and he died’ – written stories collapsed equivalent information into a single clause – e.g., un homme a fait une chute mortelle ’a man took a fatal fall’. Crucial to understanding the forms used in a text is the time allotted to text planning. Producing language in writing alleviates some of the time pressure involved in online spoken language production, and so allows more time for the work of converting information into words (Strömqvist et al. 1999). Becoming a proficient writer involves gaining mastery over more compact means of establishing the flow of information, resulting in texts that show more densely integrated packages of information (Chafe 1994). For example, syntactic subjects in written expository discourse do not necessarily obey Chafe’s (1994) ‘light subject constraint’ which is characteristic of spoken discourse. It is for this reason, for example, that written French shows more lexical nouns phrases than pronouns (Blanche-Benveniste 1990, 1995; Lambrecht 1987). Heavy subjects, often the result of syntactic packaging through nominalization or subordination, are characteristic of mature written expository discourse (Ravid et al. 2002). In addition, written texts generally show more lexical diversity than spoken texts, since writing allows more time for planification and hence more time to search one’s mental lexicon for different and less frequent lexical items (Ravid & Tolchinsky 2002; Strömqvist et al. 2002). .. Prosodic information as a source of difference While it thus does seem possible to make relatively clear distinctions between spoken and written French for some constructions, several French linguists consider broad, sweeping generalizations contrasting written French and spoken French to be premature at the present stage of research in this domain (Berrendonner 1990; Berrendonner & Reichler-Béguelin 1997; BlancheBenveniste 1990, 1994; Gadet 1997). These well-motivated objections to drawing hasty distinctions between spoken and written French have motivated researchers to take considerable pains in their description of relevant constructions as well as in how they characterize their distribution. Prosady is an important candidate for diagnosing modality distinctions in French, as an area that is inherent in the very nature of spoken compared with written language, since speech alone marks syntactic, semantic, and pragmatic distinctions through prosodic means which are unavailable in writing. Consider the following examples, from Gadet (1990: 15).1 (4) a. moi / j’ai faim / je mange b. moi / j’ai faim. je mange ‘Me I’m hungry I eat’

Growing into academic French 

The semantic interpretation of the relationship between the two clauses in the utterances in (4a) and (4b) differs depending upon the intonation given to j’ai faim ‘I’m hungry’. Rendering the meaning of (4a) in written French would require something like dès que j’ai faim, je mange ’As soon as I’m hungry, I eat’; in contrast, a written version of (4b) would be something like je commence à avoir faim. je vais manger ‘I’m beginning to feel hungry, (so) I’m going to eat’. Even the assignment of subject versus object status to a given nominal in a clause can vary by prosodic contour (Lambrecht 1994). For example, the string of words Marie Nicole elle ne l’aime pas, literally ‘Mary Nicole she her not like’ can mean either ‘Marie doesn’t like Nicole’ – pronounced as Marie / Nicole/ elle ne l’aime pas – or ‘Nicole doesn’t like Marie’ – when pronounced as Marien/ Nicole elle ne l’aime pas (Gadet 1990). Such prosodic distinctions lack a clear punctuation convention in the written mode, and thus ambiguous constructions like these tend to be avoided in writing. In speech, however, they are heavily exploited and have a marked effect on the size and the linear ordering of constituents in utterances produced in the spoken mode. Given these marked differences in the two modalities, it is of interest to consider how children of different ages and levels of schooling and hence literacy develop the ability to modify the texts they produce to take full advantage of these distinctions, on the one hand, and to manipulate them appropriately in a given communicative context, on the other.

. Developmental perspectives on text production This section examines evidence for the development from grade school age across adolescence of sensitivity to different facets of monologic text production in both narrative and expository texts and in speaking compared with writing. Production of monologic texts requires that speaker-writers engage in planning at different levels (Levelt 1989). Individual messages must be elaborated and encoded into a linear form for articulation in a propositional format, and the resulting propositions must then be packaged together by the various syntactic means available for clause-combining in a particular language. These inter-clausal packages of information also need to be structured into more global text components, such as the opening and closing segments of a piece of discourse – in the setting and coda of narrative texts and the introduction and conclusion of expository texts (Berman & Katzenberger 2004; Tolchinsky et al. 2002). The ability to plan and organize a monologic text demands a complex

 Harriet Jisa

interweaving of linguistic and cognitive abilities, and is a capacity that develops gradually over the years of childhood, into and even beyond adolescence. Clearly, both speaking and writing call upon a number of shared cognitive abilities. However, writing typically allows people to allot more time and hence greater cognitive and linguistic resources to planning activities than can be accessed in the course of rapid online speech output. As a result, the study of what children know about language can be fruitfully approached by observing their text production in both written and spoken modalities. Once children are over the major hurdles of letter formation and spelling and other facets of what Ravid & Tolchinsky (2002) refer to as ‘writing as a notational system’, writing may actually facilitate the use of less frequent and more complex constructions, and thus give a somewhat different picture of what children know about language and how to use it. Since children’s knowledge of language changes as a function of their experience with how it is used in different circumstances, close examination of how children make use of language in both the spoken and written modalities seems necessary for understanding later language development in general, and the process of ‘moving into academic language’ in particular. To this end, the following sections summarize a range of studies which reveal how French children’s spoken language is impacted by their learning to write. . Syntactic packaging and maintaining reference in narrative discourse In spoken French, new referents are generally introduced in either clefted presentational structures, such as in (5a), or in other post-verbal positions such as in (5b) (Hickmann et al. 1996; Lambrecht 1994). (5) a.

y a un homme ‘There is a man’ b. hier j’ai vu un homme ‘Yesterday I saw a man’

The newly introduced referent can be promoted to subject of the following clause by means of four structural options: repetition of the noun as in (6a), repetition of the noun with a clitic pronoun (6b), use of a pronoun alone (6c), or use of a subject relative pronoun (6d). (6) a.

(et) l’homme m’a donné un bonbon ‘(And) the man gave me a candy’

Growing into academic French

b. (et) l’homme i m’a donné un bonbon ‘(And) the man he gave me a candy’ c. (et) il m’a donné un bonbon ‘(And) he gave me a candy’ d. qui m’a donné un bonbon. ‘who gave me a candy’

Two additional structural options are available for maintaining a subject argument in subsequent clauses after it has been introduced and promoted: subject ellipsis as in (7a) and non-finite subordination (7b). (7) a.

(et) est parti au travail ‘(and) left for work’ b. avant de partir au travail ‘before leaving for work’

While all of the structures illustrated in (6a) to (6d) and (7a) and (7b) are grammatical in French, they are not all equally appropriate in all situations. Blanche-Benveniste’s (1990) comparison of a wide range of discourse from a variety of speakers revealed, as noted earlier, that anaphoric subject clitic pronouns abound in spoken conversational French. In more formal registers of French, however, they are avoided through the use of such devices as lexical noun substitutions, pronoun ellipsis (7a), or non-finite syntactic packaging (7b). Maintaining referents in subject position, then, involves numerous different potential structures, including full noun phrases with or without a detached pronoun, anaphoric pronouns, subject ellipsis, subject relative pronouns, and non-finite subordination. Relative pronouns and subject ellipsis, as opposed to full nouns and anaphoric subject pronouns, exhibit denser packaging of events by establishing a tighter dependency relationship between the two clauses (Berman & Slobin 1994: 515–554). Non-finite connectivity represents perhaps the most tightly packaged type of clause combining, since in such constructions, the subject and the tense of the subordinate clause are totally dependent on the principal clause (Foley & Van Valin 1984), and it has been shown to be a late development in different languages (Berman 1998). Forms for maintaining referents as subjects show variation both in compactness of information and in register appropriateness. The more compact forms, for example, use of optional subject pronoun ellipsis or non-finite ellipsis, indicate a more formal register (Blanche-Benveniste 1995), and these, too, are known to develop late, beyond preschool age (Berman 1990; Jisa 2000).



 Harriet Jisa

The structures used for maintaining reference across propositions were examined in narratives produced in speech and writing by French-speaking children aged 5, 7, and 10 years compared with adults, in a study that replicated the Berman & Slobin (1994) oral picture book task with speakers of five other languages (Jisa 2000). Participants were shown a booklet of 24 pictures without text that relate the adventures of a boy and his dog in search of their runaway frog, in the course of which they encounter several other characters (an owl, a mole, bees, a deer). Here, findings are noted for a group of children aged 9 to 10 years (the equivalent of 4th graders in the U.S.), twenty of whom were asked to tell the story orally and twenty of whom in writing. Results revealed that the children who wrote their stories used significantly more indications of formal written register than their age- and background-matched counterparts who told the story orally. For example, they made significantly more use of the passé simple and of the preverbal negative particle ne. More interestingly, they also used more of the two types of highly compact syntactic packaging – pronoun ellipsis and non-finite subordination – to maintain referents in subject position than did children who produced their texts in the spoken modality. These differences are illustrated in the excerpts in (8) and (9), written and spoken respectively, both recounting the final episode of the story, where the boy finds his missing frog or a substitute for it. (8) Ils virent Marguerite avec son fiancé. Ils avaient fait des petits. François et Flippeur prirent un petit et___ partirent___ en disant au-revoir. [girl, aged 10;6, written] ‘They saw Marguerite with her fiancé. They [=the frog and the fiancé] had had babies. François and Flippeur took a little one and ___ left, while ___ saying good-bye’ (9) Il voit toute la famille. Et il retrouve sa grenouille. Et il dit au revoir à sa famille. [girl, aged 10;3, spoken] ‘He [= the boy] sees the whole family. And he finds his frog. And he says good-bye to his family’

Even these young gradeschool-age writers show variation in the grammatical encoding of events in spoken versus written narrations. In (8) reference to the two protagonists, the boy and his dog, François et Flipper, is maintained in subject position by a range of devices – first by subject ellipsis (et partirent) and subsequently through non-finite syntactic packaging (en disant au revoir). In contrast, the narrator in the spoken text (9) re-uses the same subject clitic il ‘he/it’ to maintain referential cohesion to the boy as protagonist across the three

Growing into academic French 

clauses in a way that is typical of children’s oral “frog story” texts in French as in other languages. . Non-finite syntactic packaging in expository texts Against this background, we undertook a second study which specifically examined the use of non-finite syntactic packaging among schoolchildren at different levels of schooling compared with adult speaker-writers of French as a native language (Gayraud et al. 1999, 2001). Four groups of subjects aged 9 to 10 years, 12 to 13 years, and 15 to 16 years of age, and adult university students – with twelve subjects in each group – were asked to discuss their ideas about violence in schools in the form of an oral class presentation and a written essay, with tasks balanced for order. In this study, in contrast to the picturebook “frog story” narratives discussed in the preceding section, the same subject produced both a written and spoken version of an expository text. Findings showed that non-finite forms of clause-combining or syntactic packaging were more frequent in the adult than in the children’s texts. Moreover, non-finite subordination was more frequent in the texts written by the adults than in their spoken versions, regardless of the order of production. An even more interesting result was yielded by the youngest group of children, 9to 10-year-old grade schoolers: for them, the order of production proved to be an important factor. If children performed the task first in writing and then in speech, both their texts – spoken as well as written – showed a significantly higher percentage of non-finite syntactic packaging (t = 2.32, 10df, p < 00;02), as illustrated in Figure 1; but this was not the case for their peers who first produced a spoken text and then wrote an essay on the same topic. In other words, producing a written text before discussing the same topic in speech had an effect on the syntactic structures deployed by children in both their spoken and written versions. It might be claimed that the young subjects who produced the texts in the order written-then-spoken were simply repeating from memory what they had written when they switched to the spoken modality. There are two reasons to doubt this interpretation. First, the children were administered a questionnaire in the time interval between the productions of their two texts – written and spoken; this questionnaire was intended to gather information about their reading practices, but it also afforded an activity that would reduce the possibility of rote memorization of the written text being carried over to the spoken text. A second reason suggesting that children did not simply repeat their written texts in the spoken modality stems from comparison of the written and

 Harriet Jisa 25

20

15

10

5

0 Order: Spoken-Written Spoken Expository

Order: Written-Spoken Written Expository

Figure 1. Mean percentage of non-finite subordination in 9-10-year-old expository texts

spoken texts produced by individual children. This is illustrated for the French data base by the excerpts in (10) and (11) from the texts of a French girl (Charlotte, aged 10;2), where (10a) and (11a) are from her written version, compared with the corresponding excerpts from the spoken text produced by the same child in (10b) and (11b). The relevant pieces of her texts are bolded. (10) a.

Mais aussi elle peut commencer par un jaloux d’une simple note de poésie faisant un croche-pied. [Written version] ‘But also it can begin by (someone) jealous simply over a grade in poetry, tripping up (someone)’ b. ya aussi euh quand on est jaloux euh pour nos notes ou euh (2”) des choses comme ça (1”) quand on fait un croche-pied [Spoken version] ‘There is also when one is jealous over grades or things like that when one trips (someone) up’

(11) a.

Alors punissons ces gens-là pour arrêter la violence. [Written version] ‘So let’s punish those people in order to stop violence’

Growing into academic French 

b. ben euh pour qu’ tous ces gens là recommencent plus faut qu’i soient punis [Spoken version] ‘Okay / well er in order that those people don’t start again, they must be punished’

In the written version, (10a) and (11a), Charlotte uses non-finite packaging, whereas in her spoken version, (10b) and (11b), she combines the same information in a more linear fashion, by stringing together finite subordinate clauses. This clearly shows that comparing written with spoken texts produced by the same subject fails to provide evidence for a transfer by rote repetition of form from the written to the spoken versions. This conclusion is strongly confirmed by findings of other studies that elicited similar sets of data comparing spoken and written expository texts discussing the problem of violence in schools in Hebrew (Berman in press-a; Berman & Ravid 1999). And it is also strongly demonstrated in research comparing the written and spoken versions of narative texts produced by children, adolescents, and adults, native speakers of American English and of Israeli Hebrew, who had been asked to write and tell a story about an experience of interpersonal conflict (Berman & Ravid submitted; Ravid & Berman 2003b). (See, too, Section 2.3 below). It might be tempting to argue that there is a communicative basis to the differences observed in the spoken and written version. For example, some semantic relations might be preferentially encoded by non-finite packaging while others might favor alternative means. In such a case, distinct semantic relations should also be encoded in the written and spoken texts. This idea, however, is not supported by comparison of the written and spoken versions of texts produced on the same topic by the same person. Thus, for instance, in both (10a) and (10b), Charlotte gives the same hypothetical examples of a potential trigger to a dispute, while in (11a) and (11b) she proposes the same possible goal or solution for stopping violence. This, again, suggests that differences in referential semantics or thematic content do not seem to be what underlies choice of different forms for packaging information in texts. Rather, the difference in syntactic packaging of information in the spoken and written versions appears attributable to the order in which the child produced the texts. When children start with the written text, both the written and the spoken versions show tighter syntactic packaging than when they start by speaking. Strömqvist et al. (2003) suggest that in very early stages of learning to write, speaking solidly underpins writing. This first stage is followed by a strongly marked differentiation between writing and speaking as children be-

 Harriet Jisa

gin to acquire what Ravid & Tolchinsky (2002) call “written language as a special discourse style”. It can be argued that the children in our study are in this second period of development. This would contribute to explaining why those who produced their written texts in the written-first order condition showed more non-finite subordination – a clearly prototypical feature of written French. Similar results were observed in Swedish 15-year-olds by Strömqvist et al. (2004) using the frog story as the basis for written and spoken narratives produced in either a written-first or spoken-first condition. Characteristics of written Swedish narratives carried over into subsequently produced spoken narratives, whereas written narrations produced subsequently to spoken narrations remained unaffected by the previous spoken narration. A syntactic priming effect may be responsible for the children’s use of nonfinite subordination constructions in their spoken texts. Recent research in syntactic priming (Bock 1986; Branigan et al. 2000; Pickering & Branigan 1998; Scheepers 2003) has revealed that adult speakers (Smith & Wheeldon 2001) and writers (Corley & Scheepers 2002) produce utterances that are structurally consistent with an initial prime. The same effect is observed in experiments where participants are constrained to use a particular prime structure in one trial, but are free to choose between two or more alternative target structures in the following trial. In addition, over consecutive trials, temporal latencies for selecting the structurally consistent utterance decrease. It is argued that when a structurally consistent form is produced over consecutive trials, the cognitive effort associated with its generation is reduced. Results from syntactic priming experiments are regarded as providing important insights into the mechanisms of grammatical encoding (Bock & Levelt 1994). Clearly, further research on syntactic priming in children is needed, but it may be argued that the results we obtained from children in the writtenfirst condition show a priming effect with the non-finite structures used in the written versions serving as primes for the non-finite structures in the spoken versions. Our results also illustrate one way in which “thinking for writing” can influence “thinking for speaking” (Slobin 1993, 1996). A construction mobilized in the written version “primes” that construction in the spoken version. They also show how literacy impacts grammatical performance in the spoken mode: Children show that they are in fact able to speak using forms which are more characteristic of written planned discourse when required to produce a more formally structured oral monologue as in the study in question, and this impact is most strongly evident when they can use a version they have already

Growing into academic French

produced in writing, if not as a model, then as a clear point of reference for how they speak. . Discourse stance A final example of French children’s exploitation of forms more characteristic of formal, planned discourse in their later language development revolves around the notion of competition between constructions. Languages provide mature speakers with a variety of grammatical options, the choice among which depends on speakers’ individual conceptualizations or point of view and their communicative intention in a given discourse context. The notion of ‘competition’ adopted here differs somewhat from how the term is used in the acquisitional literature (see, for example, Bates & MacWhinney 1987). Rather, in the present context, competition refers to the idea that there is no single way to verbalize the contents of any given situation in the world (of reality or fantasy), and that speaker/writers have a range of options for describing the selfsame scene (Berman & Slobin 1994: 516–517; Slobin 1996, 2001). The third study discussed here, like the one referred to in the preceding section, concerns monologic expository texts in which French children and adults were asked to discuss a socially relevant topic, in a talk and an essay about interpersonal conflict or problems between people in general. Expository discussion is a discourse genre that requires the speaker/writer to package information in a generic, generalized fashion. In contrast, for example, to personal-experience narratives, where speaker/writers report on highly individualized and specific experiences, expository texts require a generic discourse stance as a means of talking or writing about quite general and abstract ideas with a certain degree of objectivity or personal detachment between the speaker/writer and the content of the propositions he or she encodes (Berman et al. 2002; Berman in press b). As noted, any event – whether veridical or not – can be expressed in various ways. This is illustrated in (12), where different renderings of the situation of ‘resolving a conflict’ can be ranked on a continuum of speaker involvement in, or responsibility for, the contents of the utterance, from highest in (12a) to lowest in (12e). (12) a.

J’ai résolu le problème. ‘I resolved the problem’ b. On a résolu le problème. ‘(Some)one∼We resolved the problem’





Harriet Jisa

c.

Le problème a été résolu (par moi). ‘The problem was resolved (by me)’ d. Le problème est/était résolu (?par moi / par la législation). ‘The problem is/was resolved (by me / by the law)’ e. Le problème s’est résolu ‘The problem resolved (itself)’

In (12a) the speaker takes full responsibility for the information in the predicate by use of the first person pronoun as subject; in the on construction (12b), the agent of the activity encoded in the event is necessarily human, but on can either attribute responsibility to the speaker or not – that is, it may, but need not have a reading that is close to (12a); the passive construction in (12c) enables the speaker to shift responsibility for the action encoded in the predicate to an agent, or omit all mention of an agent; the resultative passive construction (12d) is very close to the process passive construction (12c) (Croft 1991: 248). Process passives describe the same process as the corresponding active, but they do so from the perspective of the affected entity rather than the agent or initiator. In contrast, resultative passives describe only the resulting state of the affected entity and, in fact, do not accept an agent. A paraphrase of par in par la legislation (12d) is grâce à ‘thanks to’ or ‘due to’, indicating a means, but not an animate agent. (This contrast corresponds largely to adjectival passives compared with verbal or syntactic passives, as discussed for Hebrew in Berman, this volume). Finally, in (12e), the middle voice construction with the reflexive clitic se avoids any explicit mention of or implicit attribution of an agent potentially responsible for the resolution of the problem. Note that each of these constructions shares certain features with the others, but also differs in critical ways. Thus, the on construction illustrated in (12b) shares some of the functional load carried by agentless passives and middle voice constructions in French and in other languages (Ashby 1992; Berman 1980; Jisa et al. 2002; Koenig 1999; Lyons 1995; Tolchinsky & Rosado in press; Vinay & Darbelnet 1995). Agentless passive constructions as in (12c) and middle voice constructions as in (12e) have in common the fact that the patient participant is foregrounded and the agent participant backgrounded. A human agent is implied in both cases, but explicit reference to this participant is typically absent in passive constructions and is disallowed in middle voice constructions. Use of on as in (12b) does not eliminate the agent, but serves to clearly downgrade its individuation. Thus, the on construction resembles the passive in foregrounding the patient and downgrading the agent; and it contrasts with the middle

Growing into academic French

voice since on is confined to (an implied) human agentivity, whereas middle voice is not. Each of the constructions illustrated in (12) contributes differentially to expression of what Berman et al. (2002) have termed the ‘discourse stance’ adopted by a speaker-writer, and all serve as crucial markers of two important facets of this notion: orientation (to sender, recipient, or product of a piece of discourse) and generality of reference (ranging from specific to generic and impersonal). A competent speaker-writer of French can deploy all these alternatives to encode a highly involved, personal stance as in (12a) or a removed, more generalized stance as in (12c), (12d), or (12e) in the interest of expressing different perspectives on events. The question is when and how these distinctions are manifested developmentally in accordance with different discourse and communicative settings. To address this issue, focus in this section is on the distribution of passive constructions as in (12c) and (12d) compared with use of the on subject (12b) in expository texts produced in both speech and writing by French schoolchildren, adolescents, and university-educated adults. As background to this analysis, and in order to elucidate the notion of ‘competition’ between superficially similar, and often referentially synonymous, constructions, consider first, the passive, as illustrated in (12c). The passive construction is said to be used less frequently in French than in English (Jones 1996), for two reasons: syntactically, French passives are more constrained than in English, and discursively, French has a wider variety of alternative constructions that serve functions carried by the passive. The main syntactic constraint is that only direct objects of transitive verbs can be promoted to subject in French, ruling out a sentence such as *Pierre a été donné un livre par Marie ‘Pierre was given a book by Mary’; relatedly, objects of prepositions cannot occupy subject position in the passive, so that a sentence like The doctor was sent for would require either a rather different verb that takes a direct object, e.g., Le médecin a été appelé ‘The doctor was called’ or the generic pronoun on, as in On envoya chercher le docteur ‘One / we / they sent to-seek the doctor’ (Vinay & Darbelnet 1995: 140). In other words, if a French speaker wants to promote an argument other than a direct object to subject position, some other grammatical means must be employed. These could take the form of a topicalizing construction, e.g., C’est à Pierre que Marie a donné le livre ‘It’s to Pierre that Mary gave a book’; by dislocation as in Pierre, Marie lui a donné un livre ‘Pierre, Marie gave him a book’ (Lambrecht 1994) or in a closely related ‘as for’ construction as in Quant à Pierre, Marie lui a donné un livre ‘As for Pierre, Marie gave him a book’ (Kuno 1972; Reinhart 1982); or else by using an infinitival pronominal verb construction with the clitic se, which is limited to



 Harriet Jisa

a small set of verbs (e.g., faire ‘make’, laisser ‘let’, voir ‘see’.) that can have either a passive sense, as in (13) or a benefactive interpretation, as in (14) and (15) (Creissels 1995). (13) Jean s’est fait attraper (par la police). John got (himself) caught (by the police)’ (14) Jean s’est fait construire une maison (par l’architecte). ‘John got (himself) a house built (by the architect)’ (15) Jean s’est vu donner un livre (par Marie) ‘John saw (himself) given a book (by Marie)’

Clearly, as noted by Jones (1996), French thus has several grammatical alternatives for achieving some of the patient topicalizing effects of passives. Passive constructions have not only a patient topicalizing function, they also contribute to backgrounding the agent of an event, either by demoting it to an oblique argument position or by eliminating it altogether through use of an agentless passive as in (16). In this connection, we find the indefinite or generic on construction illustrated in (17) as ‘coming into competition’ with the agentless passive, as shown in some detail in cross-linguistic terms by Jisa et al. (2002) and from a developmental perspective in Jisa & Viguié (in press). (16) Les documents ont été volés. ‘The documents were stolen’ (17) On a volé les documents. ‘Someone stole the documents’

The chameleon character of on has been studied from many different angles, including its social and demographic distribution in everyday discourse and in interviews, its use and perhaps abuse in the mass media, both for Canadian French (Laberge 1978; Laberge & Sankoff 1980) and for European French (Ashby 1992; Atlani 1984; Koenig 1999; Simonin 1984). An important conclusion emerging from such analyses is that on is extremely multifunctional, and that its reference varies depending on the particular discourse context and communicative setting. As a colloquial alternative to nous ‘we’, on has first person plural reference, as in sentences like on a passé les vacances dans le Midi ‘We spent our vacation in the Midi’ (Jones 1996). As a generic form, on refers to people in general, e.g., en France on mange les escargots ‘In France one eats / people eat / they eat snails’, in which case it corresponds approximately to English ‘one’ or impersonal ‘they’ and to French impersonal ils ‘they’ or to other generic expressions such as tout le monde ‘everyone’. In yet a third use, on cor-

Growing into academic French

responds to indefinite quelqu’un ‘someone’, e.g., on a volé mon stylo ‘someone stole my pen’ or to the understood subject of a passive construction, e.g., mon stylo a été volé ‘my pen was stolen’. In the latter case, on indicates a change of verb valence by eliminating an agent without promoting any other participant, a use which serves to foreground the predicate (Ashby 1992). In all cases – except as a variant of first person plural nous – reference is non-specific, but restricted to human referents. It is not always easy to classify different uses of on, but several studies note that features of the verb with which it is associated are critical for how it is interpreted. Verb tense, for example, is important for determining the type of on. The generic interpretation is available only when the verb is non-punctual, e.g., in the present or imperfect tense, denoting a state or habitual event (Jones 1996: 287). When used with a verb in the specific past tense (French passé composé corresponding roughly to English simple past), as in on a volé son sac ‘someone∼we stole her/his purse’, on can have either a first person plural or an indefinite interpretation – as shown by the gloss. Verb semantics is also important in determining the indefinite interpretation of on. For example, Koenig (1999: 238) argues that the referent of indefinite on must be an active, volitional participant in the situation encoded by the sentence in which it has the subject role, as shown in (18b) compared with (18a). (18) a.

On a reçu des lettres d’insultes ‘*Somebody∼We received insulting letters.’ b. On lui a envoyé des lettres d’insultes ‘Somebody∼We sent him insulting letters’

The subject of recevoir ‘to receive’ in (18a) does not entail agentivity, since no causal role of semantic agency is needed in order to ‘receive’ something. In contrast, indefinite on can occur as the subject of a verb such as envoyer ‘to send’ in (18b), which does involve volitional agentivity. This means that in order to be interpreted as an indefinite subject, the clitic on “must be the subject of a verb whose agentive or actor semantic role it satisfies” (Koenig 1999: 237). In sum, on can be characterized as having three basic functions: (1) to refer to first person plural nous ‘we’, (2) as a generic referent, particularly when used with a verb in a non-punctual tense, and (3) as an indefinite variant of quelqu’un ‘someone’ or of an agentless passive. In order to trace the developmental underpinnings of these constructions, we examined the distribution of generic and indefinite on and passives in spoken and written texts produced by French monolinguals in the same agegroups as those in our earlier study (Section 2.2): gradeschoolers (aged 9 to 10



 Harriet Jisa

years), junior high (12 to 13), high school (15 to 16), compared with universityeducated adults. The texts examined here form part of a larger cross-linguistic, developmental study of spoken and written text production in seven languages including French (the chapters by Khorounjaia and Tolchinsky and by Wengelin and Strömqvist in this volume deal with Spanish and Swedish materials from this same project).2 In each country, twenty subjects in these four age groups were shown the same short video clip depicting different types of interpersonal conflict (moral, social, physical) and they were then asked to tell and write a story describing an incident where they themselves had been involved in a situation of conflict with someone and also to give a talk in class and write an essay discussing the topic of ‘problems between people’. Thus, closely comparable methods of elicitation were applied so that each participant across age groups and countries in the project produced four monologic texts: a personalexperience narrative – both written and spoken – and an expository text – both written and spoken – on a socially relevant topic, with performance on the four tasks balanced for order across the population. (Details on the goals, population, and procedures of the project are given in Berman & Verhoeven 2002a). The French school-going participants were recorded in two private schools in Lyon, and the adults were graduate students attending two universities in Lyon. The following discussion deals with the expository texts produced by the French-speaking subjects in both speech and writing. Expository discourse appeared particularly appropriate to the aims of this analysis, since it requires that speaker/writers generalize across individuated experiences and events, to present information as objective generalities (Berman & Katzenberger 2004; Tolchinsky et al. 2002). The two constructions targeted here – on and passives – can both be used by speaker/writers to step back from attributing unequivocal, clearly specified responsibility for the information contained in their utterances, and as such they are important indicators of speaker/writer stance. Consider, first, the findings for use of on, as shown in Figure 2. Overall, Figure 2 shows that on is more frequent in the spoken modality than in written (F(1,152) = 5.61, p < 0.01) and the higher frequency of on in spoken texts is attested at all age groups. Age is a significant variable in the distribution of on (F (3,152) = 2.59, p < 0.05) with generic, indefinite on being more frequent in the two younger groups as opposed to the two older groups. Thus, the use of on decreases with age and this is particularly true of the written modality. As will be shown, one of the reasons that the use of on decreases is that the passive which is in competition for the same depersonalizing function becomes more productive.

Growing into academic French 40 35 30 25 20 15 10 5 0

9–10-year-olds

12–13-year-olds Spoken: on

15–16-year-olds

Adults

Written: on

Figure 2. Distribution of on constructions in spoken and written expository texts (in %)

Because the overwhelming majority (91%) of the passive constructions were agentless, passives with and without agent arguments are not distinguished. In addition, no distinction was made between process and resultative passive constructions. The distribution of passive voice constructions given in Figure 3 shows that this construction is more frequent in the written than in the spoken modality (F(1, 152) = 19.24, p < 0.0001) and this applies across all age groups. The use of passive constructions increases with age (F(3, 152 = 5.06, p < 0.002), and this is particularly clear in the written modality. In comparing these two means of expressing a less personalized, more generic and hence distanced discourse stance, it turns out that on is more frequent in spoken French and decreases with age, whereas the passive is more frequent in written French and increases with age. This illustrates a wellestablished pattern in the domain of language acquisition (Slobin 1973) as in other areas of cognitive development (Werner & Kaplan 1963: 60): new forms take on old functions and old functions receive new forms. New forms taking



 Harriet Jisa 18 16 14 12 10 8 6 4 2 0

9–10-year-olds

12–13-year-olds Spoken: passif

15–16-year-olds

Adults

Written: passif

Figure 3. Distribution of passive constructions in spoken and written expository texts (in %)

on old functions is illustrated by the development of passives taking over some of the functional load of on constructions. Old forms taking on new functions can be observed by considering that, with age, earlier use of on as the spoken French equivalent of non-personal nous ‘we’ comes to be replaced by more generic and indefinite uses of the same form. In both cases, we find expanding repertoires along the dimension of form–function mappings as a function of increased age and schooling. Of course, not all indefinite uses of on are replaced by a passive construction. Rather, what emerges is a gradual development of control over the multiple options provided by the language. One of the advantages of the methodology adopted in this study (and across the cross-linguistic project in which it is anchored) is that the written and spoken texts are produced by the same subject. This allows for comparison of the same or similar text content with contrasting forms of information packaging. The excerpts in (19a) and (19b) contrast use of an on construction in a spoken expository text and an agentless passive in the written text produced by the same woman. (19) a.

Il y a d’autres problèmes qu’on a tendance à négliger [A 11 Exp, Sp]3 There are other problems which one tend(s) ∼ we tend to neglect’

Growing into academic French

b. Les autres difficultés de rapports entre les personnes au niveau collège sont par contre un peu oubliées [A 11, Exp, Wr] ‘Other difficulties in personal relations at junior high are on the other hand somewhat forgotten’

In the on construction in (19a), responsibility for the negligence can be ambiguous between a specific or a generic agentive meaning. The passive construction in (19b) leaves responsibility for forgetting unassigned. In this respect, the agentless passive eliminates the sender role, whereas the on construction leaves it somewhat more ambiguous between a generic ‘one’ and an inclusive first person plural ‘we’. This modality contrast was observed consistently across all age groups. That is, when there is a change in construction choice, it is always the case that the passive is opted for in the written modality. There was not a single case of on in the written modality corresponding to a passive in the spoken modality. One of the goals motivating the present study was to ascertain how learning to write modifies children’s use of grammatical constructions. Nine-year-olds use some passive constructions in written discourse, but almost never in spoken discourse (Figure 3). Starting from 12 years of age, the subjects used the passive in spoken, as well as written expository texts, while across age groups, passive constructions are more frequent in the written modality. It is often claimed that children write as they speak. These results suggest that children also learn to speak the way they write. That is, it may be the case that experience with use of passive constructions in writing increases their accessibility in speech.

. Conclusion The results reviewed here from several different studies of monologic text production highlight the importance of studying (later) language development on the basis of children’s performance in both the written and spoken modalities. Generalizations based on just a single modality may fail to do justice to the developing linguistic knowledge of school age children, particularly at more advanced stages of general cognitive development, on the one hand, and of schooling, on the other, when subjects have had extensive experience with literacy-based activities and with reading and writing different types of academic discourse. Some forms, such as non-finite syntactic packaging or passive voice constructions, are more characteristic of the written modality. Relatedly,



 Harriet Jisa

pronoun ellipsis for topic maintenance and use of non-finite subordination for the purpose of inter-clausal information packaging develop later than the use of anaphoric pronouns as a competing means of expressing the function of reference maintenance; and the use of passive constructions to downgrade the agent of an event develops later than one of its competitors, the on construction. That is, later language development and development of written language as a special discourse style (Ravid & Tolchinsky 2002) coincide in the ability to use more sophisticated, more academic or higher register forms of expression. The findings discussed here are typical of French, a language which is noted for a rather extreme divide between everyday colloquial usage compared with the higher, more formal linguistic register associated with the written modality. However, findings for related as well as other data bases suggest similar interactions between later language development and mastery of written language as a hallmark of literacy in different languages across a range of domains, including vocabulary (Ravid, this volume; Strömqvist et al. 2002); clause length (Ravid in press); clause-combining (Berman 1998); use of passive constructions (Jisa et al. 2002; Reilly et al. in press; van Hell et al. in press); as well as of other means for downgrading of agency (Tolchinsky & Rosado in press). These late developing forms are all more characteristic of formal, planned discourse typical of writing in general and of academic language in particular (hence, too, of expository discussion rather than personal-experience narratives). Exposure to these constructions is crucial for their emergence, but experience with applying these constructions in a range of different contexts is crucial for their mastery (see Berman, this volume). Written language constitutes a highly appropriate communicative medium for exercising the ability to use such less frequent, higher-register constructions. It is thus not surprising that children’s knowledge of and ability to use language is profoundly affected by learning to produce written discourse and by their increased command of written language as a special style of discourse.

Notes * Special thanks are expressed to Ruth Berman for her countless insightful comments and suggestions on different versions of this article. Thanks also to Denis Creissels for his continuous clarifications of the mysteries of French syntax. . I use a slash ‘/ ’ to indicate rising intonation and a period ‘ . ’ to indicate falling intonation.

Growing into academic French . The project was funded by a Spencer Foundation (Chicago, Illinois) major grant for “Developing literacy in different contexts and different languages”, to Ruth Berman as principal investigator. . The subject identification codes given at the end of these examples specifies the following variables: The first number or letter refers to age group: 9 refers to 9-10-year-olds, 12 to 1213-year-olds, 15 to 15-16-year-olds and A refers to adults. The next number makes reference to the individual subject in the age group. Finally, Exp refers to expository, followed by an indication of modality, either Sp (spoken) or Wr (written).



Learning to spell in a deep orthography The case of French* Sébastien Pacton and Michel Fayol

.

Introduction

Even in these days of computerized spelling checkers, learning to spell is an important part of writing. Good writing clearly demands higher level skills than spelling but, just as “the ability to read words, quickly, accurately, and effortlessly, is critical to skillful reading comprehension” (Adams 1990: 3), “the ability to spell words easily and accurately is an important part of being a good writer” (Treiman 1993: 3). Indeed, a person who must stop and puzzle over the spelling of each word, even if aided by a computerized spelling checker, has little attention left to devote to other aspects of writing. Although still far less common than research on reading, studies on writing and most specifically on spelling have burgeoned in recent decades (e.g., Brown & Ellis 1994; Moats 1995; Nunes & Bryant 2003; Perfetti et al. 1997; Treiman 1993). Most researchers have examined development of spelling in English, although some have considered the similarities in the problems and strategies encountered by children who learn to spell in written systems as different as English, French, Portuguese, or Greek (Bryant et al. 1999) as well as Dutch compared with Hebrew (e.g., Gillis & Ravid 2001). Doubts have also been expressed about the extent to which findings from English generalize to other writing systems, for example, in Wimmer & Goswami’s (1994) comparison of reading acquisition in English and German. The study of reading and/or spelling acquisition in various written systems thus appears to be a fruitful approach, since it enables one to determine which aspects of these acquisitions are general and which are language-specific (Harris & Hatano 1999). The present chapter focuses on French, but where relevant, acquisition of aspects of orthog-

 Sébastien Pacton and Michel Fayol

raphy that are not specific to French and have been studied for English are also considered. French, like many other languages with alphabetic systems of orthography, has an incomplete one-to-one mapping between phonemes and graphemes. Using a computer simulation, Véronis (1988) showed that the application of sound-to-spelling transcription rules allows for the correct spelling of no more than one half of all French words. On the one hand, many phonemes can be spelled in more than one way. For example, /o/ can be spelled o, au, eau, ot...; /ã/ can be spelled an, en, and, ant, ent... On the other hand, many letters do not have any phonological counterparts. For example, the final d of the words bavard ‘talkative’ and foulard ‘scarf ’ are unpronounced, and the plural markers -s and -nt are also silent (e.g., Les filles dansent. ‘The girls are dancing’). Consequently, in order to spell correctly, French children must acquire and deploy several linguistic abilities, making use of lexical, morphological, and morpho-syntactic information that goes far beyond the sound-to-spelling transcription rules. Lexically specific knowledge is often necessary, for example, in order to correctly spell words such as pomme ‘apple’ with a double mm compared with a single m as in the word moment ‘moment’, since om and omm are pronounced the same. Moreover, sensitivity to (sub)lexical regularities can also be helpful; for example, awareness that consonants are never doubled at the beginning of French words could prevent children from making spelling errors such as *ppome. Section 2 below reports on a series of experiments designed to explore children’s sensitivity to different untaught regularities of written French, some of which are probabilistic (that is, they cannot be captured by a general, acrossthe-board rule), others of which can be described with abstract, general rules. With regard to morphology, children’s sensitivity to the morphological relationship between words such as bavard /bavar/ ‘talkative’, bavardage /bavardaj/ ‘chattering’, and bavarder /bavardé/ ‘to chatter’ might be of help, since the final -d that is silent in the adjective bavard is pronounced in the morphologically related noun bavardage and the verb bavarder. If children benefit from being able to use such information, ‘morphological’ words (that is, words that are members of ‘word families’ in the sense that they are related in both form and meaning to other words in the established lexicon, see Ravid, this volume) such as bavard should be easier to spell than ‘deep’ words such as foulard ‘scarf ’, which end with the same string and which also have a final silent letter, but lack morphologically related words in which this letter is pronounced. While morphological relationships between words such as bavard and bavarder are lexically specific and so idiosyncratic, other morphological features apply

Learning to spell in a deep orthography 

across large groups of words. For example, although the sound /7t/ can be spelled ette, aite, ête, or ète in French, /7t/ is always spelled ette when it corresponds to a diminutive suffix, as in the morphologically complex words fillette ‘little girl’ or vachette ‘little cow’. Note, however, that although using morphology can be helpful in such cases, it is not essential, since the spelling of words like fillette can be retrieved from memory exactly in the same way as the spelling of morphologically underived words like crevette ‘prawn’, where ette does not stand for a diminutive suffix. In Section 3, we report experiments designed to explore whether and how children use implicitly or explicitly acquired morphological and morpho-syntactic information when this can be helpful but is not indispensable. Cases where correct spellings cannot be produced without relying on morphology and morpho-syntax are reviewed in Section 4, where we focus on the use of two silent, unpronounced inflections -s (as a plural marker on nouns and adjectives) and -nt (as a plural marker on verbs), which are explicitly taught as such in schools in France.

. Acquisition of untaught (sub)lexical regularities When and how do children become sensitive to (sub)lexical regularities? To address this question, we assigned judgment and spelling tasks using pseudowords, since use of words from the established lexicon means there is no way of knowing whether children are sensitive to particular orthographic regularities or simply know how to spell specific words. In a series of experiments, Pacton et al. (2001) explored children’s sensitivity to regularities concerning double letters, taking advantage of the fact that in French some consonants can be doubled whereas others cannot in contexts where the consonant’s ‘format’ (single versus double) has no phonological effect (for example, the strings om and omm are pronounced the same in both the words moment and pomme). We further took account of the fact that, although there is a link between the frequency of a given consonant in a single compared with a double format (m is more frequent than x and mm is more frequent than xx), some consonants are frequent in both single and double formats (e.g., m and l) whereas others are frequent only in single format (e.g., c and d). This allowed us to go beyond previous studies in which it was unclear whether English-speaking children preferred pseudo-words such as yill over yihh because they were sensitive to the frequency of single letters or of doubling (Cassar & Treiman 1997). In our study, children were presented with pairs of pseudo-words such that one pseudo-word in each pair included a consonant frequent in both single and

 Sébastien Pacton and Michel Fayol

double formats whereas the other pseudo-word included a consonant frequent only in single format. The target consonants appeared in single format in half of the pairs (e.g., orile – oride) versus in double format (e.g., abolle – abodde) in the other half. Children were asked to select the ‘invented’ word in each pair that looked most like a French word. From first grade onward, pseudowords with consonants frequent only in single format were more often selected when they occurred in single format (e.g., oride instead of orile) than in double format (e.g., abodde instead of abolle) and this effect increased at the next two grade-levels. Children’s sensitivity to the relative frequency with which a given consonant is doubled in their language was confirmed by completion tasks where they were required to fill in the blanks in pseudo-words such as ‘tuba__i’ with single (e.g., l or k) versus double (e.g., ll or kk) consonants. There is no single high-level rule that provides a precise definition of which letters may or may not be doubled in French, yet the position at which doublets may occur can be described with a general, abstract rule, along the following lines: ‘Consonants can only be doubled in medial position in French’. As early as first grade, children judged nullor (a pseudo-word with a doublet in a legitimate, word-medial position) as more word-like than nnulor (with a doublet in a prohibited word-initial position). This effect further transferred to consonants that do not allow doubling in French. For example, although k and x are never doubled in the language, children found that pseudo-words such as koxxir were more word-like than pseudo-words such as kkoxir. However, the effect for these letters remained substantially lower than for frequently doubled letters from Grades 1 to 6, and this transfer-decrement showed no trend of decreasing across grade levels. This transfer-decrement, also observed with a completion task, indicated that even after as long as five years of exposure to print, children’s orthographic knowledge did not correspond to a general rule such as ‘consonants can only be doubled in medial position in French’. Indeed, an essential prediction of any system using abstract, general rules to represent knowledge about some domain is that performance should be equally good on unfamiliar and familiar items (Anderson 1993; Whittlesea & Dorken 1997). In another series of experiments (Pacton et al. 2002), children from Grades 2 to 4 were instructed to spell ‘invented’ words such as /obidar/, /ribore/ or /bylevo/, all possible words in terms of French phonology, but which do not exist in the conventional lexicon. Pseudo-words were elaborated so that it was possible to assess whether children varied their transcription of the sound /o/ (as o, au, eau, ot, aut, ho, etc.) according to its position in the word – for example, the sequence eau is frequent in final position, rare in medial position, and nonexistent initially – and its consonantal context – for example, eau is more

Learning to spell in a deep orthography 

frequent after the letter v than f . Even the youngest children used a variety of spellings to transcribe /o/. Out of 20 second graders, only one always spelled /o/ with the single letter o; 6 used two different spellings; and 13 used at least three different spellings. Moreover, the variety of spellings used to transcribe /o/ increased in 3rd and 4th grade; and from 2nd grade onward, children’s transcription of /o/ varied according to its position and its consonantal context. For instance, not only were they more likely to spell /o/ as eau in word-final position than in medial and initial positions but, within the final position, eau was more often used after -v(where it is common in French) than after -f (where it is rare). Similar effects were obtained with pseudo-word judgment tasks (Pacton & Fayol 1998). The fact that even the youngest, 2nd grade French children varied their transcriptions of /o/ echoes Treiman’s (1993) observation from her naturalistic study that English-speaking 1st graders in the U.S. used different graphemic forms (e.g., c, k, and ck) in order to transcribe the sound /k/. Similarly, the influence of positional regularities is consistent with Treiman’s finding that children were more likely to use ck and double letters in medial and final positions (where they can occur in English) than in initial position (where they never occur in English). The results of our studies, which used polysyllabic pseudo-words that differed from occurring words more extensively than in most previous studies (Campbell 1985; Nation & Hulme 1996), further confirmed the finding that the size of the sound-to-spelling correspondences on which children base their spellings is larger than the phoneme-grapheme unit.

. Optional reliance on morphology and morpho-syntax This section considers instances in which making use of morphological and morpho-syntactic information can be helpful but is not indispensable in order to spell correctly. Pacton (2001) compared how 8- to 9-year-old poor, average, and good spellers write words with final silent consonant endings that differed in whether they do or do not have morphological derivatives that provide cues to the nature of the silent consonant. These studies compare what were termed ‘morphological’ words like bavard ‘talkative’ which can be related to the verb bavarder ‘to chatter’ versus ‘deep’ words like foulard ‘scarf ’ which lack such morphologically associated forms. The morphological group of words included ones where the silent final letter could be revealed by a related feminine form (e.g., bavard – bavarde, ‘talkative-male – female’) or not (e.g., combat ‘battle’). The results showed that 3rd grade children were more likely to cor-

 Sébastien Pacton and Michel Fayol

rectly spell the final silent-consonant in ‘morphological’ words than in ‘deep’ words, especially in the case of words whose final silent-consonants is revealed by a feminine form (see Sénéchal 2000 for similar results). Interestingly, although good spellers correctly marked silent letters more often than average spellers and even more so than poor spellers, the differences between the three kinds of words – ‘deep’, ‘morphological’ without a feminine form, ‘morphological’ with a feminine – did not differ significantly according to spelling ability. The difference that children showed in how they spelled ‘morphological’ compared with ‘deep’ words, even though they could have retrieved the spelling form of both types of words in the same way from memory, supports the view that spelling acquisition depends on various sources of linguistic knowledge rather than on a simple process of rote memorization (Treiman & Cassar 1997). In other studies, we focused on morphological and morpho-syntactic features that apply to large groups of items. For example, we showed that 3rdand 5th-grade French children benefited from the morpho-syntactic information provided by the sentence when they spell gerunds and adverbs (Pacton & Fayol 2003). In this study, which took the form of an isolated word dictation task, participants were first dictated adverbs ending with /mã/, spelled -ment (e.g., rapidement ‘quickly’) and present participles ending with /ã/, spelled -ant (e.g., comprenant ‘understanding’). Subsequently, the same words were again dictated in isolation to some children (the control group) versus in the context of a sentence to others (the experimental group). For example, the controls were given the participle comprenant ‘understanding’, while the experimental group were given the sentence Elle pleura en comprenant la situation ‘She cried when understanding the situation’. While spelling performance did not differ significantly between the first and second isolated word dictations in the control group, children in the experimental group did better on spelling /ã/ in the context of the sentence dictation than in the isolated word dictation. The fact that this effect was observed as early as 3rd grade, even though only 5th graders had been taught how to transcribe the suffixes /ã/, constitutes further evidence that children learn various aspects of written language implicitly (Perruchet & Pacton 2004). On the other hand, in both grades, spelling performances varied according to the frequency of the words even when a sentence context was provided. Thus, although children benefited from the morphosyntactic information provided by the sentence, they did not rely (or at least not systematically) on a morphological rule such as ‘words ending with /ã/ always take the -ant spelling after the preposition en (e.g., en comprenant ‘when understanding’), even when this rule had been explicitly taught in school.

Learning to spell in a deep orthography 

In order to examine the role of derivational morphology, we exploited the fact that in French, transcription of the selfsame sound can be governed by graphotactic constraints that are probabilistic in nature (for example, the sound /7t/ is more often transcribed ette after -v than after -t) and also by morphological constraints that are describable with a general rule (for example, /7t/ is always transcribed ette when used as a diminutive suffix). This strategy allowed us, first, to explore the combined influence on children’s spelling of the two types of constraints and, second, to assess the degree of abstractedness of knowledge that children have acquired implicitly. Eight- to 11-year-old children’s spellings of pseudo-words were affected by derivational morphology, with pseudo-words such as /soriv7t/ more often spelled with ette when they were dictated inside sentences like ‘a little /soriv/ is a /soriv7t/’ than when dictated in isolation or embedded in sentences in which /7t/ did not function as a diminutive suffix (Pacton et al. in press). Children’s spellings were also influenced by graphotactic regularities when no morphological rule could be applied, with pseudo-words such as /soriv7t/ more often spelled with ette than pseudo-words such as /sorit7t/. Importantly, this effect remained unchanged in conditions triggering the application of the morphological rule, and it persisted over grade levels, without any falling off as a function of age. Pacton et al. (in press) argue that the fact that, even after massive exposure to print, children did not rely on abstract, rule-based, morphological knowledge is consistent with non-abstractionist accounts of implicit learning according to which statistical learning processes can apply to features that are abstracted away from their sensory context (Perruchet & Vinter 2002a, b; Redington & Chater 2002). That is, children will learn associations between abstract features like ‘diminutiveness’ and ‘end in ette’, leading them to use ette more often when they encounter pseudo-words like /soriv7t/ in the context of a sentence like ‘a little /soriv/ is a /soriv7t/’. But associations between other features may also be established; specifically, since words with final ette preceded by -v are far more frequent than final ette proceeded by -t, the association between -v and ette will be stronger than between -t and ette. That is, the effect of such graphotactic constraints will persist in the ‘diminutive’ condition as well, since even though children encounter more and more words ending with ette where the final /7t/ stands for a diminutive suffix, the proportion of words with ette after -v to which they are exposed will always remain higher than the proportion of words with ette after -t. Pacton et al. (in press) also noted that pseudo-words could be spelled through analogies to real words, so that the effect of the graphotactic constraints would persist in the ‘diminutive’ condition because the seman-

 Sébastien Pacton and Michel Fayol

tic notion of ‘diminutiveness’ is not necessarily a better cue for analogy than phonological factors. Along very similar lines, Kemp & Bryant (2003) found that 5- to 9-year old children and adults did not usually base their spellings of plural real-word and pseudo-word endings on the morphological rule that all regular plurals are spelled with -s in English. Instead, they appeared to rely on their sensitivity to untaught spelling patterns, based on the frequency with which certain letters co-occur in written English (e.g., /z/ is always spelled -s after /b/ or /d/ but it is not always spelled -s after long vowels). Researchers have stressed the need to develop theories of spelling that take account of the relationships between the different sources of information (phonological, morphological, and lexical) that influence spelling (e.g., Lennox & Siegel 1994; Snowling 1994; Treiman & Cassar 1997). For this reason, it is crucial to concurrently study the impact of different sources of information and their potential interactions, as done by Pacton et al. (in press) and by Kemp & Bryant (2003). The interest of these two studies lies in the fact that they show that children’s sensitivity to graphotactic regularities, already well documented when explored as a separate issue (e.g., Cassar & Treiman 1997; Pacton et al. 2001), continues to have an influence on how children spell even when they can rely on a simple, morphology-based rule.

. Reliance on obligatory morpho-syntax In French, the number markers for nouns and adjectives (zero in the singular form versus -s in the plural) and verbs (zero in the singular versus -nt in the plural) are not audible. As a result, French conceals many homophones that are not homographic. For example, the word timbre ‘(postage) stamp’ is written in the plural with -s if it is a noun (les timbres ‘the stamps’), with -nt if a verb (ils timbrent ‘they (put on a) stamp’; and these two plural forms are pronounced identically, as also are the singular noun le timbre ‘the stamp’ and the singular verb il timbre ‘he stamps’. The silent character of much of the inflectional morphology of their language means that French children need to discover features that are specific to the written system which are not reflected in speech. Totereau et al. (1997) and Totereau et al. (1998) studied how grade-schoolers mark the plurals of nouns and verbs. They found that children first spell the endings of plural nouns and verbs according to sound, omitting the silent -s and -nt in sentences like Les filles dorment ‘The girls are sleeping’, along lines similar to English-speaking children who spell the past tense of verbs like kiss

Learning to spell in a deep orthography

as kist, instead of kissed (e.g., Nunes et al. 1997). Later, French children begin to use the -s ending for noun plurals, but they then also over-extend it to verb plurals (e.g., both the correct plural noun nuages ‘clouds’ and the incorrect verb form dormes ‘sleep’). Children taught the rule ‘If it is the plural of a noun, add -s’ seem to apply the rule ‘If it is a plural, add -s’. This over-generalization echoes the spelling of non-verbs produced by English-speaking children when they begin to use -ed, for example, spelling soft as sofed (Nunes et al. 1997). Later on, children learn to add -nt to the end of verbs in the plural, but then instead of using it only for verbs, they may also add it to nouns, especially in cases where there is a homophonous noun. Learning to use this second rule requires that children discriminate between verbs and words belonging to other syntactic categories that are not inflected with the -nt ending. Although French grade-schoolers are typically subjected to systematic and quite extensive grammar lessons and given explicit exercises where they need to apply grammatical rules, this fails to prevent occurrence of substitution errors, such as adding -s to a verb (Fayol et al. 1999). These errors are particularly common when there are pairs of noun/verb homophones, with the nouns of higher frequency than the related verbs resulting, for example, in the spelling of *ils timbres for required ils timbrent ‘they stamp = they put stamps on’, a type of error that, as further noted below, occurs even among adults. The question of what causes errors of omission (use of zero instead of -s or -nt) was also investigated by Fayol and his associates. They found that the younger children would make such errors because they do not have a declarative knowledge of the function of the -s and -nt markers. When presented with pictures depicting one or several objects, persons, or actions, and asked to select from two pictures – one that corresponded to the linguistic item provided along with the drawing (e.g., poisson ‘fish-Singular’ versus poissons ‘fish-Plural’, or le poisson ‘the fish’ versus les poissons ‘the fishes’), children’s selections at the beginning of 1st Grade were at chance level (Totereau et al. 1997). Slightly older 1st grade children, after about eight months of schooling, succeed in processing the number markers (zero versus -s and -nt respectively) in comprehension, as tested by linguistic item / picture matching tasks, but not in production tasks. At this stage, children have enough knowledge to understand the meaning of the presence or absence of these markers, but they are not yet able to mobilize the number markers in production. At a later stage of development, successful use of the plural markers -s and -nt in production depends on the demands of the task at hand. In second grade, even the graphic transcription, that is, being required to write an entire word is enough to disrupt the agreement process (Bourdin & Fayol 1994, 2000). And in fact, omission errors are far less com-



 Sébastien Pacton and Michel Fayol

mon when 2nd graders need to complete sentences by writing down missing words or parts of words (e.g., les filles dans___ dans la rue ‘the girls dance in the street’) than when they have to themselves write sentences in full (Totereau et al. 1997). Similarly, at this stage children who make many omission errors when writing sentences are able to detect errors in agreement that are shown to them – both in the form of omission of the plural markers -s and -nt when required and their insertion when not (Largy 2001). These studies show that when task difficulty is reduced to more local levels, there is a lower rate of omission errors as compared with numerous spelling omission in writing entire sentences. Conversely, increasing the difficulty of the task (for example, with a dual-task paradigm) increases the rate of omission errors among 5th graders (Fayol et al. 1999) and even among adults (Fayol et al. 1994) who do not make omission errors when asked to write sentences in full. Other studies have considered what determines errors of substitution (e.g., -s instead of -nt or vice versa). Earlier, we noted that substitution errors made by spellers when they begin to use -s to mark the plural of all words irrespective of grammatical category result from incorrect specification of the condition for application of the rule. These errors soon decrease (Totereau et al. 1998), but other substitution errors are not so systematic and persist even in adults. These are more frequent and possibly restricted to cases of items with homophonous counterparts. Under standard writing conditions, most educated adults inflect nouns and verbs correctly, whether they have homophones or not. Yet substitution errors on items with homophones may arise with adults when they are more focused on the meaning of their message than on its orthographic correctness (for example, university students when writing in exam situations) or in experimental conditions as in dual-task situations (Hupet et al. 1998; Largy et al. 1996). And similar findings have been observed for Dutch (Sandra et al.1999). Similar effects have been reported between adjective/verb homophones with 11-year-old children, indicating that substitution errors on homophones that cover different grammatical categories constitute a quite general phenomenon (Pacton et al. in press; Thévenin et al. 1999). Substitution errors may also depend on the syntactic position of an item. Pacton (in revision) compared 11 year-old children’s writing of adjectives with and without verbal homophones (e.g., bavarde, ‘talkative – female’ / bavarder ‘to chatter’ versus pauvre ‘poor’) in post-nominal position where both adjectives and verbs can occur in French (e.g., Les femmes bavardes ‘The talkative women’ / Les femmes bavardent ‘The women chatter’), and in post-verbal position where adjectives can but verbs cannot occur (e.g., Les femmes du quartier sont bavardes ‘The women

Learning to spell in a deep orthography 

of the neighborhood are talkative’). Adjectives that cannot be verbs were very rarely inflected with the verbal plural marker -nt, whether they occurred in post-nominal or post-verbal position. By contrast, substitution errors occurred on adjectives with homophone verb counterparts, and they were more frequent for post-nominal than for post-verbal adjectives. That substitution errors are mainly limited to items with homophonous counterparts indicates that writers do not systematically rely on the morphological rules they have been explicitly taught. Although adults know the rule and they know how to apply the rule ‘If plural and a verb, then -nt’, they tend not to systematically perform a syntactic analysis in order to identify the grammatical status (verb, noun, or adjective) of the item to be marked. Instead, they may retrieve from memory associations between a given stem and inflection combination (e.g., the association between the word timbre ‘stamp’ and the plural marker -s) or whole instances (the plural word timbres). For a stem such as trouve ‘find’, which can only be a verb, or nuage ‘cloud’, which can only be a noun, whatever the syntactic structure in which they occur, the retrieval procedure and application of the explicitly taught grammatical rules work towards the same response. However, with words that can be either nouns or verbs, these two procedures may elicit different responses, since the writer can retrieve from memory the nominal instead of the verbal form (e.g., timbres instead of timbrent) and vice-versa. Homophone spelling errors raise the crucial issue of the interactions between implicit and explicit learning. True, children explicitly learn a morphological rule that stipulates the correct agreement of every item in a given grammatical category, but they also implicitly learn (through reading and spelling) that the associations between certain stems and inflections are particularly frequent. A related question concerns the effect of having encountered a word almost always in the same form (for example, always in the singular versus always plural) before or after being explicitly taught a grammatical rule such as the nominal plural. Cousin et al. (2002) had children from 2nd to 5th grades complete sentences by writing down missing words, either previously encountered or not, either in the singular or in the plural form in the textbook used to teach reading. In line with previous findings, the younger children made many omission errors when they had to spell words in the plural form (zero instead of -s), but they made few mistakes of erroneously adding -s to a singular form. Nonetheless, although such errors were rare, they were more frequent for words previously encountered mostly in the plural form (e.g., writing le parents ‘the+Singular parents’ in place of required les parents or le parent) than for new words or words previously encountered mostly in the singular form (e.g., ma-

 Sébastien Pacton and Michel Fayol

man ‘mommy’). This effect persisted with the same intensity after several years of schooling, suggesting that children had memorized previously encountered items as such and that memorized instances would sometimes interfere with rule application. Further study is necessary to ascertain whether the persistence of this effect should be interpreted as an ‘age of acquisition’ phenomenon, since words like parents are more frequent in the plural form not only in children’s first readers but also in the texts they are exposed to after the relevant rule has been taught.

. Conclusions The lack of one-to-one mappings between phonemes and graphemes makes gaining mastery of writing as a notational system (Ravid & Tolchinsky 2002) particularly difficult for French-speaking children, and many features of written French must be learnt somewhat independently from the spoken language. French children are from an early age sensitive to sub-lexical regularities relating to the identity of the letters that can be doubled and where they can occur as well as to the statistical distributions of some transcriptions according to their position in a word and the letters that precede or follow them (Pacton et al. 2001, 2002). The specific nature of the regularities to which children are sensitive is nevertheless open to question. They could be attending to graphotactic regularities relative to the order of letters, for instance that eau is frequently attached to -v but not to -f , independently of any regularity on a phonological level. Support for this account is provided by empirical evidence of statistical learning of the order of visual shapes (e.g., Fiser & Aslin 2002). However, children could also be sensitive to the association between eau and certain sounds (e.g., /o/ is more often associated with the voiced fricative /v/ rather than /f/). If that is the case, it would be more appropriate to say that children are sensitive to ‘phono-graphotactic’ regularities. The choice between these two accounts calls for further specifically devised studies. Also relevant to our analyses is the integration of these graphotactic (or phono-graphotactic) regularities with higher level, morphological regularities. For example, we found that French children transcribed /7t/ more often as ette when /7t/ stands for a diminutive suffix than when it does not. However, despite the possibility of relying on a morphological rule specifying that /7t/ is spelled ette when it corresponds to a diminutive suffix, children’s performances varied according to whether /7t/ followed a letter after which ette is frequent or rare in French, and this effect persisted in a very stable way over grade levels (Pacton

Learning to spell in a deep orthography

et al. in press). This result, coupled with findings reported by Kemp & Bryant (2003) for inflectional morphology in English, indicates “that early sensitivity to orthographic frequency patterns could be an important part of learning about written language” (Kemp & Bryant 2003: 273). Graphotactic regularities were found to have an impact on transcription of regular plurals in English even though the rule is discussed at school (Kemp & Bryant op cit). To the best of our knowledge, the impact of an explicit rule being taught relative to the transcription of the suffixes /o/ and /7t/ in French has not been explored. However, we have reported many other cases in which neither children nor adults consistently make (full) use of explicitly taught morphological rules. For instance, French children use the morpho-syntactic information provided by the sentence when they spell adverbs and present participles ending with /ã/ before and after the morpho-syntactic rules have been taught explicitly. However, even when the rules have been explicitly taught, children do not rely on them (systematically). Similarly, even though French children are explicitly taught how to mark the plural of verbs, nouns, and, adjectives, they are more likely to correctly write the plural of adjectives without a verbal homophone than the plural of adjectives that can also be verbs. This raises the issue of the relations between implicit and explicit learning. In cases like the one just cited, the effect of implicit learning may need to be counteracted. This may be particularly difficult since there is no way to prevent a child from encountering the plural verb bavardent more often than the plural adjective bavardes. On the other hand, certain learning situations that lead to children’s (implicitly) acquiring erroneous information should be discouraged or even disallowed. Thus, giving opportunity for errors during a training program is certainly beneficial for formation of explicit knowledge, but could be detrimental for implicit learning. Accordingly, explicit processes should be applied to identify and eradicate errors during a learning episode. As against this, implicit learning might encode statistical cues without differential processing of correct compared with incorrect information. There is considerable empirical evidence that people implicitly learn correct as well as incorrect associations, as shown, for example, by the literature on errorless versus errorful learning (as in Baddeley & Wilson 1994). In the case of spelling, exposure to incorrectly spelled words has been shown to harm subjects’ orthographic knowledge (e.g. Brown 1988; Dixon & Kaminska 1997). This questions the relevance of multiple-choice tests in which the correct spelling of a word is presented together with one or more incorrect (pseudo-homophone) spellings (e.g., street / streat in English) where students are required to select the correct spelling. Situations that do not expose students to incorrect spellings, but



 Sébastien Pacton and Michel Fayol

instead lead them to focus on the different letters that go to make up words might be particularly effective. For example, oral spelling tasks (spelling the word aloud) would seem to be more efficient than copying tasks (writing the word down into a notebook) and even more so than reading tasks (Bosman & Van Orden 1997). Such recommended types of tasks would be relatively independent of learners’ intention to learn how to spell words, for example in order to prepare for an upcoming test. The crucial factor would be to assign tasks where learners pay attention to processing each of the constituent letters of the target word. Finally, it is worth stressing that cases where implicit and explicit learning have contradictory effects are rather rare. For the most part, these two modes of learning have complementary effects, and the question is, rather, whether and how to make joint use of the two together. The issue of the interaction between implicit and explicit learning remains largely unexplored, although in the paradigm of artificial grammar learning, it has been suggested that learning is improved when the two modes (implicit and explicit) alternate with one another (Matthews et al. 1989). From an educational perspective, a preliminary form of implicit learning could facilitate the subsequent explicit learning of rules in a given domain. Prior familiarity with the domain should facilitate subsequent grasp of the rules, since examples illustrating these rules are available to learners as the subject-matter for acquisition of new knowledge.

Acknowledgment We would like to thank Ruth Berman, Peter Bryant, and Arnaud Rey for their helpful comments on a previous draft of this chapter.

Note * This research was supported by the Universities René Descartes, Paris V, and Blaise Pascal, Clermont-Ferrand, and the Centre National de la Recherche Scientifique (CNRS).

Text–writing development viewed through on-line pausing in Swedish Åsa Wengelin and Sven Strömqvist

.

Introduction

Analyzing the ‘the flow of thought and the flow of language’ in relation to spoken narrative discourse in English, Chafe (1979) finds the sentence or clause to be a unit which is rarely broken up internally by pauses. Instead, pauses tend to come in the boundary between sentences or clauses. Chafe argues that the sentence/clause provides an adequate unit for encoding information about an event and that the flow of speaking – reflecting the flow of thought – therefore tends to proceed sentence by sentence when a narrator is relating an event. The present study is concerned with writing, not with speech. It is concerned with the flow of thought and the flow of language in relation to written narrative discourse in Swedish. Writing, too, is a dynamic activity. In order to fully appreciate that fact, writing needs to be analyzed as it unfolds in real time, with its fluent and hesitant phases as well as its editing patterns (Strömqvist et al. 2004). Further, writing is a complex activity, where the writer has to manage the major processes of planning, execution, and monitoring in order to successfully arrive at a final edited text. From a developmental point of view, the management of these processes can be expected to differ at different ages and levels of schooling and literacy. Thus, in earlier phases of development, more cognitive resources are typically allocated to low-level processes, such as spelling and punctuation. Later, as these processes become more habitual or even automatized, cognitive resources are freed for concentrating on higherlevel processes concerned with, for example, content, structure, and style. Learning to write texts is a long-term process involving several aspects and components of different character. From a linguistic point of view, learning to write is not only a question of learning notational specifics, ranging from or-

 Åsa Wengelin and Sven Strömqvist

thography to style sheets, but also a question of learning new genres and new aspects of the lexicon and grammar (Berman & Verhoeven 2002a; Strömqvist 2003). That is, it is not enough to gain command of “writing as a notational system”, literacy also involves command of “written language as a special discourse style” (Ravid & Tolchinsky 2002). As a result, the acquisition of text-writing belongs solidly in the domain of later language development – the focus of the present volume. Further, from a communicative point of view, learning to write means adapting to and exploring new communicative conditions (Strömqvist et al. 1999). For example, the absence, typical of written communication, of on-line feedback on how the message is interpreted by the addressee calls for an advanced level of theorizing of mind (Wimmer & Perner 1983) on the part of the sender. Besides, learning to write means adapting to and exploring new processing constraints. In particular, the tight on-line constraints of speaking – with its extremely high speed and short duration of the signal – are lifted in writing, which allows, among other things, for more planning, for conscious reflection and revision during text production (Strömqvist et al. 1999). The rapidly expanding paradigm of keystroke-logging (see, for example, Berglund & Sullivan, in preparation), that is, the registration and analysis of writing performed on a computer, underlies an increasing number of psycholinguistic studies of writing, including ones that consider developmental patterns (Alves 2002; Alves et al. in press; Erskine 1999; Johansson 2000; Nordqvist 2001; Nordqvist et al. in press; Pelli & Sinimäki 2004; Solheim & Uppstad 2003; Strömqvist et al. 2004; Wengelin 2002; Wiktorsson 2000, 2003). The present study describes a method for analyzing the distribution of pauses in on-line text-writing, on the assumption that the observed distribution of pauses reflects the distribution of cognitive effort during text-writing (Strömqvist et al. in press; Wengelin in press). The method was applied to a corpus of on-line text-writing of native speakers of Swedish in four different age-groups (4th, 7th, and 10th graders and graduate level university students). The results are interpreted as a window on developmental change in the allocation of cognitive effort to different levels of text-writing.

. Methodology The empirical data for this study consist of computer-logged text-writing activity in the Swedish sample of materials elicited in the framework of a largescale crosslinguistic project on ‘Developing literacy in different languages and different contexts’ funded by a Spencer Foundation major grant to the editor of this

Text–writing development viewed through on-line pausing in Swedish 

volume, as described in detail in Berman & Verhoeven (2002a). (Analyses of the French and Spanish samples of this project are provided in the chapters by Jisa and by Khorounjaia & Tolchinsky, respectively, in this volume). The Swedish sample analyzed here included 78 subjects in four age groups: 20 in each of the three younger groups of 4th-graders (aged 10), 7th-graders (aged 13), 10thgraders (aged 17), and 18 university student adults, with an equal number of males and females in each group. None of the subjects had any known history of reading and writing difficulties or other disorders. Each subject produced both a spoken and a written version of two texts, a personal-experience narrative and an expository discussion. The present analysis is confined to the narrative texts produced in writing. In advance of the writing task, the subjects were shown a short video clip, depicting scenes from a school day, including various interpersonal conflicts and predicaments. The video-clip served to constrain the subjects to a shared theme across their narrative and expository texts, both spoken and written. The Swedish subjects all wrote their texts on a computer, and the research tool ScriptLog (Strömqvist & Karlsson 2002; Strömqvist & Malmsten 1998) was used to record the writing activity. The data consists of 142,969 key strokes in the 78 computerlogged text-writing activities and 22,308 words in the corresponding 78 final edited texts. As noted, the text-writing activity was recorded by means of the computer tool termed ‘ScriptLog’, developed specifically for supporting research concerned with on-line writing. ScriptLog has three main modules: a design module, where the conditions of the writing task are defined; a recording module, which produces a bin-file containing a full representation of the events occurring during the writing activity together with their temporal distribution; and an analysis module, allowing the researcher to derive selected patterns from the recording. To illustrate, consider examples (1) and (2) below. (1) shows the beginning of the final edited text, a personal narrative, written by a Swedish 10-year-old while (2) presents the corresponding ‘linear’ text with selected information about its pauses and editings.

 Åsa Wengelin and Sven Strömqvist

(1) [Swedish 10-year-old, final edited narrative] Detta hände för ungefär 3 månader sen This happened about 3 months ago. jag och min, bästis Ane I and my pal Ane (hon kommer från norge) (she’s from Norway) var på en klätterstälning we were climbing det var efter skolan it was after school så det var bara vi där. only we were there.’ (2) The writing activity corresponding to (1), with pauses and editings: det var en dag Detta hände för unge fär 3 måmnader sen. jag och min bästis anAne ( hon komer från norge) varpå en klätter stälning det var efter skolan så det var bara vi där.

The text in (2) reveals, among other things, that the writer makes a long pause (105.48 secs) after she activates the editor window (‘START’), whereupon she writes det var en dag ‘it was one day’. During the production of this first sentence, she makes two short pauses. (By ‘sentence’, we mean a written utterance which contains at least one finite verb and whose end can be marked by a major delimiter). Then she deletes the 14 characters – including spaces – she has just written (‘DELETE14’) and instead writes Detta hände för unge fär 3 månader sen. ‘This happened about 3 months ago’, marking the end of her first sentence with a period (full stop). When writing this sentence, she makes a longer pause (9.36 secs) after Detta hände ‘This happened’ and she corrects a misspelling (måm → månader). When, eventually, she reaches the completion point of her next sentence – ending with the word klätterställning – she makes a pause of 12 seconds and then moves the cursor 22 characters back in the text produced so far (‘LEFT22’) – as if aiming at a revision – and then moves it back to its original position (‘RIGHT22’). ScriptLog produces a variety of alternative output structures, all highlighting selective information patterns derived from the original recording. Some of these structures – such as the linear file illustrated in (2) – are pre-defined and offered as standard output from ScriptLog. Others can be interactively defined. For example, the researcher can define a certain target string, such as a particular word form or expression, and then ask ScriptLog to search for it in a recording and analyze all instances found in terms of their production rate

Text–writing development viewed through on-line pausing in Swedish

Table 1. A subset of the types of micro-contexts which can be defined by means of ScriptLog micro-context

notation/symbols

inactivity after a space preceded by a letter and before a letter (In less technical terms: before starting a new word within a phrase)

a_^a

inactivity between two consecutive letters (In less technical terms: within a word)

a^a

inactivity after the last letter of a word and before a space followed by a letter (In less technical terms: after finishing a word within a phrase)

a^_a

inactivity after a major delimiter (that is, any of the delimiters belonging to the set {.?!}) followed by a space and before a letter (In less technical terms: before starting the first word of a new sentence)

._^a

inactivity after a letter and before a major delimiter (In less technical terms: after the last word of a sentence and before typing a full stop, a question mark, or an exclamation mark)

a^.

inactivity after a major delimiter and before a space followed by a letter (In less technical terms: after finishing a sentence by typing a full stop, a question mark or an exclamation mark)

.^_a

characteristics. Another possibility is to define a search string in terms of a type of micro-context with variables signifying objects such as letters and delimiters. ScriptLog will then search for instantiations of the type of micro-context(s) defined and provide an analysis output in terms of ‘transition time values’, that is, production rate characteristics for the targeted micro-context. Table 1 provides an overview of a subset of the types of micro-context which can be defined by means of ScriptLog. The symbol ‘^’ marks the transition (the transition time value of which ScriptLog provides a calculation) targeted in the micro-context. The first three micro-contexts are associated with the construction of words. ‘a_^a’ is associated with the initiation of the construction of a word, ‘a^a’ with the continuation of the construction of a word, & ‘a^_a’ with the closing of the construction of a word. The last three micro-contexts are associated with the construction of sentences, thus: ‘._^a’ is associated with the initiation of the construction of a sentence, ‘a^.’ with the closing of the construction of a sentence, & ‘.^_a’ with the decision to write another sentence. In the present study, we use the first three of these micro-contexts as a window on the allocation of cognitive resources to the construction of words within phrases in the course of text writing and we use the latter three micro-contexts



 Åsa Wengelin and Sven Strömqvist

as a window on the allocation of cognitive resources to the construction of sentences during text writing. The detection of the micro-contexts is fully automatic and exhaustive. This does not mean, however, that there is an exhaustive detection of word and sentence boundaries in the texts analyzed. The detection depends entirely on the presence of the symbols referred to in the definition of the search strings, as shown in Table 1. As a result, word boundaries that are not marked by a space or sentences not marked by a major delimiter will not be detected by the analysis program. In effect, the analysis runs the risk of underspecifying the number of word and sentence boundaries that actually exist in the text corpus.

. Analysis of data Table 2 presents a summary overview of the analysis data. The table shows age groups, micro-contexts and transition time values for the contexts in question. The transition time values – in seconds – are calculated as ‘mean across medians’, that is, group mean across the median transition values for individual subjects. The reason for using medians and not means is that the distributions of transition time values are skewed in the direction of very large values, that is, there are many long pauses which affect the mean, although not the median. For further details, see Strömqvist et al. (in press) and Wengelin (in press). Note the low number of instantiations of the micro-contexts ‘.^_a’ ‘ ‘._^a’ in the younger age groups. For 9 of the 20 4th-graders and for 6 of the 20 7th-graders, these micro-context are entirely lacking. Further, for 3 of the 4thgraders and one 7th-grader, the micro-context ‘a^.’ is missing. This does not necessarily mean that these children do not ever use major delimiters in their writing. It may simply mean that they do not write them immediately after a letter or that they fail to press the space bar immediately after the delimiter. In some cases, such writers make a space before the delimiter, in some cases they delete something immediately after the space. This could also be the case for some of the older writers. The micro-contexts we have analyzed do not give us any information in this respect, so that hasty conclusions should not be drawn from automatic analyses of micro-contexts. However, it seems likely that punctuation behavior changes across age groups, specifically, the more skilled the writers, the greater the likelihood that their delimiters will occur immediately after the last letter of the sentence and will be followed by a space and a (capital) letter.

Text–writing development viewed through on-line pausing in Swedish 

Table 2. Age groups, micro-contexts and transition time values age group

microcontext

total N

mean N per text

4th-graders 10 yrs (N = 20)

transition times in secs (mean across medians)

a_^a a^a a^_a ._^a a^. .^_a

2050 7218 2050 59 67 59

102.50 360.90 102.50 2.95 3.35 2.95

1.825 0.562 0.450 3.565 3.380 3.767

7th-graders 13 yrs (N = 20)

a_^a a^a a^_a ._^a a^. .^_a

3139 11458 3139 82 120 82

156.95 572.90 156.95 4.10 6.00 4.10

0.637 0.295 0.247 1.873 2.140 0.682

10th-graders 17 yrs (N = 20)

a_^a a^a a^_a ._^a a^. .^_a

9074 36002 9074 377 519 377

453.70 1800.10 453.70 18.85 25.95 18.85

0.470 0.220 0.178 2.175 1.175 0.302

University students (N = 18)

a_^a a^a a^_a ._^a a^. .^_a

7160 30370 7160 339 440 339

397.78 1687.22 397.78 18.83 24.44 18.83

0.358 0.181 0.160 1.753 0.846 0.290

. Intra-word transition times Consider first the micro-context ‘a^a’, that is, transitions between letters within words. This is by far the most frequently instantiated type of micro-context, one which is assumed to be minimally associated with planning and monitoring activity. These two properties make ‘a^a’ a candidate measure of a subject’s keyboard skills. It should be noted, however, that production rate in the context ‘a^a’ can be sensitive to the internal structure of a word form. For example, the prolonged intra-word transition time values could be concomitant to word-internal morpheme boundaries (Bertram et al. in prep.) and to the double spelling of consonants (Nordqvist et al. in press). Spelling problems may also result in an increase of word-internal pausing (Wengelin 2002). Fig-

 Åsa Wengelin and Sven Strömqvist 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 4th-graders

7th-graders

10th-graders

University stud

Figure 1. Mean median transition times within words in different age groups

ure 1 shows the mean median transition times (group mean across individual subjects’ median transition values) for the four age-groups. Figure 1 shows that the youngest age-group, the 4th-graders, have a far longer mean median transition time (0.562 secs) than the other groups (between 0.181 and 0.295 secs). The remaining three groups are very similar to one another. An ANOVA (with Bonferroni/Dunn posthoc tests) showed a significant main effect for age (F(3,24) = 30.320, p < 0.0001), due mainly to the fact that the 4th-graders are significantly slower than all the other groups (p < 0.0001). There is also a tendency for the 7th-graders to be slower writers than the university students (p = 0.013). No other differences were found. We interpret the finding that the 4th-graders are so much slower than the others as indicating a clear improvement at the level of word construction from 4th to 7th grade. This suggests that the 4th-graders still need to allocate relatively large resources to low-level processes associated with word construction, such as spelling and typing, something which has implications for higher-level planning and monitoring processes (for example, the construction of sentences or sentence-transcending structures such as narrative events). A variant of this

Text–writing development viewed through on-line pausing in Swedish

line of reasoning is suggested by Alves (2002) and Alves et al. (in press), in line with the ideas of Scardamalia & Bereiter (1987). They suggest that there is a cognitive cost of effortful execution in text-writing such that writers often lose track of what they planned to execute. According to Alves and Alves et al., writers are forced to pause in order to re-establish their plans. In fact, one effect of the cognitive cost of effortful execution surfaces as frequent pausing, not least between words within phrases. Conversely, the relatively fast and apparently effortless word-level writing of subjects from 7th grade up frees cognitive resources, and so facilitates a smooth interplay between planning, execution, and monitoring of the writing process. . Inter-word and inter-sentence pauses Consider, next, the micro-contexts associated with the transitions between words and sentences, respectively. In a study of pause distributions, Wengelin (in press) applied two different measures: (1) mean duration of the pauses (> 2 secs) associated with a given micro-context and (2) relative frequency (percentage of total number of occurrences) of a given micro-context associated with a pause of a duration of at least 2 seconds. With respect to measure (1), Wengelin had two reasons for using means instead of medians (which were used for measuring the transition times – see above). First, the distributions of pause durations are less skewed than the distributions of transition times and therefore raise less of a problem. Second, most studies of pause durations use the mean rather than the median (see for example Spelman-Miller 2000). Wengelin found that, although there is a tendency for pauses to be longer at sentence than at word boundaries, this difference is not so consistent as to be statistically significant. In other words, type of micro-context was not found to be reliably predictive of pause length. In contrast, Wengelin found the second measure to have a reliable predictive value; pauses are relatively more frequent at sentence than at word boundaries. That is, type of micro-context is predictive of where writers prefer to make a pause, although not of the length of the pause. Figure 2 (from Wengelin in press) provides an illustration of how the two measures compare. The figure shows the results when the two measures are applied to ScriptLog recordings of textwriting activities from 10 university students (different subjects from those in the present study). The bars in Figure 2 show the mean pause duration in the six microcontexts described in Table 1. The durations – in seconds – are shown on the left Y-axis. The dots on the graph show the percentages of the respective micro-



 Åsa Wengelin and Sven Strömqvist 35,00%

10 9

30.00% 25.00%

7 6

20.00%

5 15.00%

4 3

10.00%

Chance of pause in transition

Mean duration in seconds

8

duration Relative pause frequency

2 5.00% 1 0

a_^a

a^a

a^_a

._^a

a^.

.^_a

0.00%

Figure 2. Relative pause frequencies versus mean pause duration in different microcontexts

contexts where pauses of longer than 2 seconds occur. In other words, the dots show the relative frequencies of pauses in these contexts expressed as percentages. These percentages are shown on the right Y-axis. For example, there is a pause in a little more than 30% of the instantiations of the micro-context ‘-.^a’. Figure 2 confirms very clearly that the graph (relative pause frequencies) differentiates between the micro-contexts whereas the bars (pause duration) do not. We have combined the graph and the bars in a single figure in order to highlight this contrast. These results from applying the two measures led us to select the measure of relative pause frequencies (percentage of micro-contexts with pauses) rather than pause duration for our developmental analysis. As in the previous study, here, too, we adopted a pause criterion of 2 seconds. Discarding transition time values of less than 2 seconds allows us to minimize, if not eliminate altogether, the risk of including transition times that may occur for motor rather than cognitive reasons. As our analysis of transition times in the micro-context ‘a^a’ indicated, transition time values around 1 second may well be due to lack of keyboard proficiency.

Text–writing development viewed through on-line pausing in Swedish  80% 70% 60% 50%

4th-graders 7th-graders 10th-graders University stud

40% 30% 20% 10% 0%

a_^a

a^a

a^_a

._^a

a^.

.^_a

Figure 3. Relative pause frequencies in six micro-contexts across four age groups

Applying the selected measure to our age group data shown in Table 2 produces the results summarized in Figure 3. A first observation is that the curves in Figure 3 reveal a very similar overall profile. The similarity of the profiles is noteworthy, considering that the four groups have very different typing skills. Thus, the 4th-graders type more slowly than the writers in the other groups – as shown by the distance between their curve and the curves of the other groups. And yet the overall shape of their curve is very similar to the others. That is, all age-groups show the same basic preferences for pausing: They all prefer pausing between sentences rather than between words. We interpret this tendency as supporting the idea that we tend to plan our discourse in chunks corresponding roughly to sentences, a notion that has been around in one way or another from Aristotle to modern-day cognitive linguists. Chafe (1994) quotes a particularly apt formulation of this idea by reference to William James’ remark about the rhythm of language ‘. . . where every thought is expressed in a sentence and every sentence closed by a period.’ (James 1890, Vol. 1: 243). A second observation to emerge from Figure 3 is that the 4th-grade children exhibit much more pausing than the other three age-groups. On average, the 4th-graders make a pause in almost 40% of the micro-context ‘a_^a’, that is, immediately before starting the construction of a new word within a phrase.

 Åsa Wengelin and Sven Strömqvist

The corresponding figure for the three older age groups is less than 10%. This frequent pausing at word boundaries is an expected side effect – a cognitive cost – of the effortful execution and struggling with low-level word construction processes observed in our analysis of intra-word transition times in the 4th-graders (see Figure 1). Moreover, both the 4th- and 7th-graders show a higher relative amount of pausing in the context ‘a^.’, that is, immediately before a major delimiter, than in the context ‘._^a’. This pausing might reflect monitoring, due to the time taken by writers to read the sentence they have just produced. This hypothesis can be tested with the help of eye-tracking data, a technique we are currently exploring in a new series of studies (see Strömqvist et al. in press). On the other hand, it might be more likely that this pre-delimiter pausing reflects hesitation on the part of writers due to lack of mastery of punctuation. A third possibility is that younger, less experienced writers need more time than other subjects to decide whether they should end or continue the sentence, for example, with a conjunction and a new clause. That is, this type of pausing could reflect a point of planning. A final observation is the substantial decrease in inter-sentence pausing from the 4th-graders to the older age-groups. The 4th-graders make pauses in about 65% of the micro-contexts ‘._^a’, whereas the other age groups make pauses in only 30–40% of the same contexts. We know from previous studies of text writing that grade school children tend to spend less time on discourse planning than older students, and that their writing tends to proceed in an additive fashion, one simple sentence at a time (Strömqvist et al. 2004). In contrast, more advanced writers tend to spend more time on planning, and subsequently they often write a longer chunk of discourse with little or no hesitation. The pausing pattern observed in the present study accords well with such a pattern. The more frequent pausing at sentence boundary observed in the 4th-graders is likely to be a reflection of the fact that these young writers seldom plan more than one sentence ahead at a time. Further, the fact that their word-level writing is still relatively effortful probably increases their tendency to make a pause at sentence boundaries. Later in development, low-level processes such as spelling and typing become habitualized, if not fully automatized, freeing, as noted, cognitive resources for higher-level processes such as planning of content structure. Besides, later development also entails internalization of models for text construction, such as a well-established narrative schema in the data-base considered here, and this, too, helps support global planning during text-writing. Planning larger chunks of discourse beyond the level of an isolated sentence in advance of execution combined with effortless

Text–writing development viewed through on-line pausing in Swedish 

spelling and typing will result in a more frequent crossing of sentence boundaries at a high production rate with little or no need for pausing. And this, in fact, as shown by Figure 3, is the pattern revealed by the three older age groups in our sample.

. Conclusion Taking advantage of the sophisticated means now available for keystrokelogging of on-line processes of writing, we analyzed pause distributions in micro-contexts as a method for detecting developmental change in the allocation of cognitive effort to different facets of text-writing. Our results indicate that at an early phase of text-writing development – in our study represented by the 10-year-olds – more cognitive resources tend to be allocated to lowlevel processes such as spelling and punctuation. Later – in our study, from 13 years onwards – as these processes become habitualized or even automatized, cognitive resources are freed to allow for concentrating on higher-level activities concerned with such features as the content, structure, and style of the text being written. Further, we found that schoolchildren, adolescents, and adults alike prefer to make pauses at sentence boundary rather than at word boundary. By implication, this preferential tendency overrides differences in keyboard skill and typing speed typically associated with age, and it therefore calls for a deeper, semantic and cognitive explanation. We suggest that this explanation must make reference to the fact that a sentence tends to encode a (relatively) complete piece of propositional information that a writer can easily conceptualize and plan in advance of executing it. Indeed, the kind of idea unit encoded by a sentence seems to be a powerful determinant to ‘the flow of thought and the flow of language’, not only in spoken narrative discourse, as Chafe (1979) concluded, but also in written narratives. Our findings, further suggest that more advanced writers have developed the skill of conceptualizing and planning even larger chunks of discourse, since they cross sentence boundaries without pausing in a majority of cases. In all likelihood, this skill is shaped largely through the experience of text-writing, a condition of language use where the on-line constraints of spoken language can be lifted, so facilitating the allocation of time and cognitive resources to the planning and conscious reflection on larger chunks of discourse structure. Our observations of the development of text-writing skills – from 10-yearolds via 13- and 17-year-olds to university students – are part and parcel of later language development. Learning to use written language, however, is not

 Åsa Wengelin and Sven Strömqvist

merely an additive enterprise, where written language skills are added to the spoken language skills already achieved. Learning to use written language is a dynamic process which interacts with and shapes the further acquisition and development of spoken language skills (Nordqvist 2001; Strömqvist et al. 2004). An interesting thrust for future research would therefore be to examine whether and how the ability to plan larger chunks of discourse is also manifested in spoken language as a property of later language development.

The role of peer interaction in later pragmatic development The case of speech representation Shoshana Blum-Kulka

.

Introduction

This chapter argues for peer talk as a crucial site for pragmatic development on the basis of available literature, and demonstrates this by examining the case of speech representation. In the view proposed by socio-cultural and pragmatic socialization theories of language development, children are believed to acquire their linguistic skills through active participation in meaningful interactions with other children and adults in their culture (e.g., Blum-Kulka 1997; Rogoff 2003). While the manifold ways that children are socialized to use language appropriately by talking to and conversing with adults have been widely documented (e.g., Blum-Kulka & Snow 2002), the area of peer interaction remains relatively less chartered, particularly with regard to later development (Blum-Kulka & Snow 2004). Peer interaction has been of interest for scholars from various perspectives – as promoting (or not) cognitive development (Rogoff 1990); for its role in social development (Handel 1988); and from a sociolinguistic viewpoint that focuses on children’s evolving interpretive reproduction and co-construction of cultural patterns through language (e.g., Corsaro 1997; Goodwin 1990). Language use figures in almost all studies of peer interaction, but reference in the present discussion will be mainly to ones that (1) focus specifically on language and (2) are concerned with pragmatic, rather than grammatical aspects of language development. Ninio & Snow (1996) define the study of pragmatic development as “concerned with how children acquire the knowledge necessary for the appropriate, effective, rule-governed employment of speech in interpersonal situation” (1996: 4). Pragmatic, or communicative, competence is thus taken here as the

 Shoshana Blum-Kulka

ability to produce socially and culturally appropriate spoken and written discourse in a variety of participation configurations, including the interpersonal mode. Below, the later development of pragmatic competence is discussed with respect to two major domains of pragmatic ability: conversational skills – which encompass turn-taking and dialogicity – and discursive literacy skills. Turn-taking concerns the ability to take turns in dyadic and multi-party conversations effectively and rapidly, abiding by the situational and/or cultural norms of turn-taking. This involves, for example, knowing when to ‘chime in’, how to maintain and relinquish a turn, to show involvement as a listener, and to keep interruptions and overlaps within limits of cultural or situational acceptability (Sacks et al. 1974; Tanaka 1999). Dialogicity concerns the ability to interpret and formulate speech acts in a range of direct and indirect forms and to initiate and maintain topics of talk in adherence to the Gricean maxims of relevance, clarity, informativeness, and manner. This ability is manifested by topically related, understandable contributions that provide as much (but no more) information as required and that are phrased in the appropriate register. These maxims can be also understood as the ability to tune the type of response required to the conversational demand of the previous turn; this involves attending to both the function or illocutionary force and the (propositional) content of a previous turn (for example, reject or accept an offer or request made directly or indirectly) and the ability to formulate contributions in cohesive ways that manifest topicrelatedness and also as awareness of context and the other’s perspective (Dorval & Eckerman 1984). Discursive literacy concerns the ability to interpret and construct extended discourse in genre-appropriate forms (Ninio & Snow 1996; Pellegrini & Galda 1998; Wells 1985). The skills involved include the ability to construct sequentially cohesive and globally well-structured, coherent stretches of autonomous discourse that are sufficiently ‘decontextualized’ – that is, their level of explicitness is tuned to the audience’s degree of familiarity with the topic at hand – and are appropriate in style and structure to situation and genre. The two essential features of discursive literacy – decontextualization and sequential organization – may be manifest to varying degrees in spoken and written texts of different length and genre. For instance, formal definitions are brief but require a high level of decontextualization and an Aristotelian format (x is y that ...), while narratives are typified by their sequential organization (e.g., Berman & Slobin 1994; Labov 1972; Labov & Waletzky 1967). How do children acquire the different features of pragmatic competence? Do the domains postulated above have ecological validity, that is, can they be

The role of peer interaction in later pragmatic development 

shown to differ in their developmental pathways? Can we map the developmental path for any specific skill from infancy to late adolescence? Are there specific domains of pragmatic competence that can predict success in school? Do contexts of use (family, peers, school) differ in the types of opportunities they provide for learning? The following discussion addresses some of these questions: First, by considering major findings from available research on pragmatic development (Section 2); second, by discussing the merits of peer talk for pragmatic development of school-age children (Section 3); and third, by focusing on speech representation as a test case in pragmatic development from preschool to pre-adolescence (Section 4).

. Pragmatic development: An overview This section reviews available research on pragmatic development from the point of view of the threefold distinction suggested above. . Turn-taking In many western cultures, children are treated as conversational partners from birth, so that it is not surprising that by the time they are producing their first words, British and American children can sustain long stretches of well-timed turn-alternations with their mothers (Kaye & Charney 1981; Snow 1977) and that by the age of three they largely follow the rules of turn-taking with each other (Ervin-Tripp 1979). Nonetheless, the precise timing of an adult conversationalist is beyond young children. They have difficulty in projecting possible turn-completion points for their own speech and in identifying transitionrelevant points in the speech of others, so that their peer conversations contain fewer overlaps and longer gaps than that of more proficient conversationalists (Garvey & Beringer 1981). For the same reasons, children may experience difficulty in joining in multi-party, inter-generational talk around a family dinner-table (Blum-Kulka 1997). . Children’s dialogue Estimates of young children’s dialogic skills cannot readily be divorced from their situational and cultural contexts of use. In interactions with adults, the adults tend to take on the bulk of conversational responsibilities and to support the flow of conversation by initiating topics, asking questions, providing

 Shoshana Blum-Kulka

interpretations, and expanding unclear utterances, as well as by challenging children to observe relevance and informativeness (e.g., Blum-Kulka & Snow 2002). Western conversational norms calling for active participation by all partners are by no means universal. In Japan, for example, children are socialized by their mothers for intent listening, and are expected to develop their skills as active interpreters capable of following the indirectly formulated speech acts of their interlocutors (Clancy 1986). The evidence from peer talk is less clear-cut. On the one hand, cross-turn relevance in young children’s talk is often maintained by sound play and partial imitation rather than topical cohesion (Garvey 1974), and 4-year-olds were observed to engage in private speech half of the time during interaction (SchoberPeterson & Johnson 1991). On the other hand, young children have also been shown to employ numerous linguistic means in order to maintain cross-turn coherence in long stretches of talk. For example, McTear’s (1985) longitudinal study of natural interaction of a dyad of two girls from age four to six years provides strong evidence for the nascent dialogic skills of preschool children. The development trends he identified include: (1) enrichment in modes of topic initiation and reinitiation, showing children’s growing awareness of the need for other acknowledgement; (2) richer scope of cross-turn cohesive markers, from repetition and prosodic shifts toward use of anaphoric pronouns, ellipsis, and sentence connectives; (3) a move from brief, closed exchange structures to longer and more open, topically and interactionally related sequences; and (4) increased use of remedial devices, testifying to the children’s growing ability to monitor their own and the other’s speech and to locate and repair conversational breakdowns. Studies of later conversational development are scant. One exception is Dorval & Eckerman’s (1984) comprehensive study of conversational development (see also Dumensil & Dorval 1989). They observed children during interaction in small peer groups with an adult leader whose task was to ensure the ‘one-at-a time’ rule of turn-taking and, if needed, to control disruptive behavior. The study’s cross-sectional design, including five age levels from 2nd to 12th grade (plus a group of adults), allowed for a systematic charting of the developmental path of conversation. Dorval and Eckerman found significant age differences in extent of topic relatedness. While the second graders produced a large number of unrelated turns, adolescents tended to strive for cross-turn topical coherence. Even more importantly perhaps, the older children tended to take the other’s perspective by relating to other speakers’ thoughts and feelings as expressed in previous turns. Yet since the study was semi-controlled, the

The role of peer interaction in later pragmatic development 

patterns depicted might have been context-dependent and need not necessarily reflect the full gamut of children’s conversational skills at the age levels tested. . Discursive literacy: The developmental path of conversational narratives The study of ‘discursive literacy’, with its focus on specific genres such as narratives, explanations, and definitions, suggests that the development of extended discourse skills follows a differential path from that of conversation (Ninio & Snow 1996). For reasons of space, the discussion of discursive literacy here is limited to narratives, the most widely studied genre of extended discourse. As sustained symbolic representations of possible worlds, narratives require the packaging of experience in cognitively and affectively coherent ways, and are thus related to the development of spoken and written literate, decontextualized types of discourse in general (Wells 1985). Oral narratives have two essential features that make them crucial to the development of pragmatic competencies: as prime and early examples of connected discourse, linked to literate uses of language, and as social and cultural enactments of the narrative paradigm (Bruner 1986). And in fact, several studies have shown a link between children’s oral competence in the use of extended discourse in the early years and their written literacy skills in school (Dickinson & Tabors 2001; Pellegrini & Galda 1998). The bulk of research on narrative development comes from psycholinguistic and sociolinguistic studies of elicited or semi-elicited child-adult interactions. This line of research has yielded a wealth of developmental and cross-linguistic information on linguistic forms used to achieve narrative functions such as temporality and causality (e.g., Bamberg 1987; Berman 1996; Berman 1997b; Berman & Slobin 1994) as well as on the development of narrative linearity or macro-structures (e.g., Berman 1995a; Peterson & McCabe 1983; McCabe & Peterson 1991). In the socio-cultural, ethnographic tradition of studying narratives as they emerge in natural talk, the focus has been on other dimensions of narrativity, such as tellability and co-construction, and on cultural patterns in socialization through narratives, rather than on developmental patterns (Blum-Kulka 1997; Goodwin 1990; Heath 1983; Miller et al. 1990; Ochs & Taylor 1992; Shuman 1986). Relatively few studies on young children’s story-telling in natural conversation have considered peer talk. Topics investigated in this regard include narrative genres employed in peer stories told in a car pool (Preece 1987), the construction of gender identities through story-telling (Kyratzis 1999;

 Shoshana Blum-Kulka

Niclopoulou 1997), and preschoolers’ modes of meaning-making in pretend and non-pretend narratives (Blum-Kulka in press-a). Marjorie Goodwin’s (1990) ethnographic study of African-American adolescents and Amy Shuman’s (1986) study of urban adolescents are among the few comprehensive studies of the role of conversational stories in the peer cultures of school-aged children. Goodwin analyzes in detail the way the distinctive structural properties of a story introduced during a dyadic dispute in a group of boys can be used to restructure the social organization of an emerging argument, expanding the participation framework of the dispute, allowing for new recipients and new alignments. Shuman’s (1986) study foregrounds issues of tellership and tellability – what makes an event reportable and who is entitled to tell which stories to whom – and the resulting role of narratives in the social organization of peer culture. The psycholinguistic and ethnographic perspectives on oral narratives have had different concerns. The first looks at stories first and foremost as pieces of connected discourse, judged by well-formedness, and so broadly relating to skills identified here under ‘discursive literacy’, while the latter considers stories as inter-actional achievements, partly associated with dialogic skills, whose success is assessable through audience response. The developmental path of oral narratives is relatively well charted from the first point of view; but remains largely under-studied from the second.

. Peer talk: Peer cultures and peer language Studies of peer interaction derive from different intellectual traditions, with little contact between them (Blum-Kulka & Snow 2004). In the domain of peer talk per se, the emphasis has been mainly on the ways in which peer talk functions in children’s co-construction of their social and cultural worlds, rather than on the development of language through peer talk. Peer culture has been considered from various perspectives: the discursive ways by which children collaboratively co-construct social relationships in play (Corsaro 1985) or how they frame the passage in and out of pretend play (Aronsson & Thorell 2002; Sawyer 1997) as well as how they reproduce and co-construct language-bound cultural practices, for example, food-sharing, peace-making, and exchange rituals among Israeli children (Katriel 1991), patterns of dispute among AfricanAmerican children (Goodwin 1990), or Italian and American children’s styles of argumentation (Corsaro & Rizzo 1990). With a few exceptions (Ervin-Tripp & Gordon (1986) for requests, Dorval & Eckerman (1984) and McTear (1985)

The role of peer interaction in later pragmatic development 

for conversational skills), most sociolinguistic studies lack a developmental agenda, and they are concerned with toddlers and preschoolers rather than with school-age children. Sociolinguistic interest in later development shifts to issues of language and identity, showing how allegiance to youth culture is indexed and coconstructed by choice of language varieties (see Hoyle & Adger 1998b for a review). Many of these studies follow the sociolinguistic traditions established by Labov (1972), who was the first to demonstrate the importance of peergroup membership for Afro-American youth in shaping norms of speech. Later studies in this tradition have looked at the impact of social categories such as social class, ethnicity, and gender on linguistic choices. For example, Eckert’s (1989) ethnographic study of Detroit high schools powerfully demonstrates how affiliation with one of the two prevailing social categories (with the “Jocks” of hegemonic culture or the “Burnouts” of counter-culture) is strongly associated with variant linguistic usage, with the counter-culture usages serving as agents for linguistic change in the wider community. Similarly, Cheshire’s (1991) study of working-class British teenagers in Reading found that the boys conveyed a dominant theme of their culture – aggression – through syntactic and phonological choices, such as their use of a phonological variant of aint (‘in’t’) to conform to the vernacular norm. Hemphill (1989) found that ethnicity and social stratification also affect conversational style, with AfricanAmerican working-class teenage girls using a collaborative conversational style rich in cross-turns and back channeling cohesiveness, while the American middle-class teenagers opted for a more autonomous, competitive, yet orderly style (see Romaine (1984) for a summary of quantitative studies). Studies of gender differences in children’s ways of speaking have yielded mixed results. Tannen (1990), for example, argued strongly, on the basis of videotaped same-gender conversations at different age levels (2nd, 6th, 10th, and 12th grade), for striking gender differences between the girls’ collaborative versus the boys’ more competitive style. On the other hand, Marjorie Goodwin’s (1990) findings for African-American pre-adolescents show the girls as indeed collaborative in one context (joint activity) but quite assertive and hierarchical in another (playing family). . The study: Peer talk of Israeli children We conducted a longitudinal study entitled “Gaining autonomy in genres of extended discourse”, designed to provide a context- and culture-sensitive developmental account of the acquisition of different discourse genres.1 Two cohorts

 Shoshana Blum-Kulka

of 20 children of mixed socio-economic backgrounds – young preschoolers and 4th graders – were studied over a period of three years. The data were collected by ethnographic observations of natural discourse and by semi-structured sociolinguistic interviews, yielding child discourse across three types of speech events: natural peer interactions, mealtimes, and semi-structured adult-child interviews. Peer-group interactions of the young children were observed during free-play time in the preschool, while the older children were asked to invite a friend and were recorded either at home or in a fast-food restaurant of their choice, and occasionally in the car to or from the restaurant as well. Although we did not specify the gender of the friend to be invited, the children adhered to same-sex dyads, except in two instances – one a threesome of girls, and the other of two boys and a girl. Mealtimes were recorded by the parents during the first year of the study, and sociolinguistic interviews were repeated each of the three years. . The merits of peer-talk Of the three contexts studied, peer talk proved the most variable and intriguing (Blum-Kulka et al. 2004; Blum-Kulka & Snow 2004). Peer talk affords children rich opportunities for pragmatic development by virtue of two of its fundamental contextual properties: (a) a collaborative, multi-party, symmetrical participation structure and (b) shared worlds of childhood culture. Peer talk unfolds in pairs or groups of children unhindered by the inherent asymmetry of adult-child interaction and so may be uniquely conducive to both cognitive and pragmatic development. Our preschool observations yielded several instances of direct child-child teaching: a boy teaching a girl how to play a computer game, one boy showing another how to make a snake from dough, one girl teaching another a new song, a boy teaching another a new word. Arguably, the merits of peer talk need to be evaluated in terms of their context-specific gains: Just as multi-party, intergenerational settings offer children a range of opportunities unavailable in or different from those of child-adult dyadic interactions (Blum-Kulka & Snow 2002), peer group participant structures are important in allowing for types of collaboration unavailable in or different from adult-child discourse. The shared worlds of childhood culture emerge as a second fundamental feature of peer talk, apparent especially in fantasy play. Young children in pretend play display an amazing ability to collaborate in developing elaborate story-plots, loosely associated with scripts and characters of popular culture, often freely mixing several sources (e.g., Auwarter 1986; Blum-Kulka in press-

The role of peer interaction in later pragmatic development 

a; Nicolopoulou & Richner 2004; Sawyer 1997). In such speech events, children need to be highly attentive to each other in order to be able to make the next relevant contribution to the plot. The ability to achieve coherence in these episodes rests heavily on shared familiarity with the ‘raw materials’ of the fantasy play – school experience, family life, or the themes drawn from child popular-culture fiction and fantasy. Such familiarity may be important not only for contributing to the richness of pretend play (Sawyer 1997) but also for the development of genres of extended discourse (Pellegrini & Galda 1990, 1998; Vedeler 1997).

. Speech representation: From voicing to reporting This section considers natural-talk ‘speech representation’ at two age levels. . Some preliminaries The ways in which peer talk may enhance the development of discursive skills is illustrated here through the case of speech representation. The ability to represent the speech of others calls into play a wide array of linguistic and pragmatic skills and functions, cutting across the domains of conversation and extended discourse. The phenomenon has been widely studied, particularly as concerns verbatim representation in quotes. For some (Coulmas 1986; Leech 1983), direct quotes entail the speaker’s commitment to words said; for others (Clark & Gerrig 1990; Kothoff 1998), in quoting the speaker is committed only to certain aspects of the words said, or sometimes to none at all, using “constructed dialogues” (Tannen 1989) or “fictive quotes” (Weizman 1998). In line with Lucy (1993), speech representation is considered here as a metalinguistic skill that encompasses iconic (voicing and direct quotes), reportative (indirect, with or without verbs of saying), and descriptive ways of ‘talk about talk’. Children’s manifold ways of speech representation lend strong support to what Clark & Gerrig call the ‘demonstration’ theory in showing the extent to which talk about talk can be presented as iconic while adhering to words imagined, rather than said. Wolf and Hicks (1989) discuss children’s “capacity for intertextuality”, namely their discovery that naturally occurring talk is a rich mix of voices and forms, and their nascent ability to distinguish between voices, such as in play enactments between stage managing, character dialogue, or narrative. Their longitudinal study of children from 2 to 7 years in play enactments with toys

 Shoshana Blum-Kulka

revealed that “the understanding of text as multivoiced emerges early enough to be considered a fundamental aspect of what young speakers learn about discourse” (Wolf & Hicks 1989: 333). By as young as age three, all the children in the study consistently marked the distinction between narrative, character dialogue, and stage managing, by such means as prosody, pronouns, and utterance type. Between ages 3 and 5, children added temporal features to distinguish between the voices, and from 5 years up, they revealed the potential multifunctionality of single voices, such as reporting speech as narrator or recounting events via character speech. Aronsson & Thorell (2002) further demonstrate, in line with other voice-type models of pretend play (e.g., Auwarter 1986; Sawyer 1997) that in pretend play, preschool and school-age children not only coordinate perspectives by role enactments co-constructed on purported ways of speaking of real-life adult characters, they also find prosodic and lexical means for parodying the adult roles, so displaying their own role detachment and agency. These observations are confirmed by our study: both preschoolers and school-age children showed a high capacity for intertextuality, distinguishing between voices, including special means for signaling what we call a ‘subversive keying’ used to parody the adult world (Blum-Kulka et al. 2004). In linguistic form, this development entails enrichment with age in verb inflections, in verb semantics, and in sequential connectives. From a discursive point of view, it entails a growing awareness of the social aspect of language and a developing capacity for exploiting the different modes of speech representation to meet multiple functions in various genres, especially narratives. Hickmann (1993) considers the major developmental feature of speech representation to be the ways in which children learn to transform dialogues into narrative texts. In her study, adults and 4-to 10-year-old children were asked to report dialogues from made-up scenarios. The results showed a clear preference of the younger children for the use of speech modes that did not require marking explicit boundaries between the reported speech and the narrator’s message, such as the re-enacting (voicing) mode, and the non-explicit descriptive mode. In contrast, the 7- and 10-year-olds “framed speech events systematically with direct reports and the adults did so as often with direct and indirect reports” (Hickmann 1993: 83). Hickmann’s (experimentally based) claim for an overall developmental transition from dialogue to narrative text is indirectly supported by our findings from peer talk at two age levels. The developmental shift from voicing to reporting is manifested in choice of narrative mode: One third (32%) of preschool children’s narratives (n=60, in free play) are pretend play enactments in which in-character voicing constitutes a major mode of speech representa-

The role of peer interaction in later pragmatic development 

tion, but by age nine (in dyadic conversation in a restaurant or at home), the proportion of pretend stories diminishes to only 5%, being replaced by stories of genuine experience which combine direct and indirect modes of speech representation (Habib et al. 2003). Although partly attributable to the different circumstances of the recording sessions in the two cohorts, this pattern could be indicative of developmental trends. Differing from ‘on-line’ enactment of a storyline in pretend play, which requires mainly the sociolinguistic skills of register variation, the indirect and direct representation of speech in an ulterior, distanced narration requires several additional text-related discursive skills, such as temporal ordering and cohesion. Thus the transition from narrative dialogue to narrative text can also be interpreted as the transition from utterance to text (Olson 1977).2 More broadly, it can be seen as part of the shift from contextualized to more autonomous modes of discourse (Blum-Kulka et al. 2002; Ninio & Snow 1996).3 . Speech representation and conversational skills: Register A major site for conversational development is pretend play: It provides children with opportunities to perfect their skills in different registers and to rely on register-specific resources to build coherent play events (Blum-Kulka in press-b; Sawyer 1997). Young children deploy a rich repertoire of mainly prosodic and to some extent lexical resources to signal in-character speech within the play frame (Aronsson & Thorell 2002; Auwarter 1986; Corsaro 1985). Among school-age children, the earlier reliance on prosody is replaced by choices of lexicon and grammar (Halmari & Smith 1994) which, as shown by Hoyle’s (1998) study of preadolescent boys, are used to mark different roles of speaking associated with different roles, as well as to sustain these roles collaboratively, exhibiting mutual alignment to pretend selves. On the few occasions when the nine-year-olds in our study engaged in pretend play, they displayed the same type of talk in register as noted by Hoyle for this age group. On one occasion, two girls observed at home spent the entire recording session role-playing ’saleswoman’ and ’shopper’ in a fantasy up-scale women’s clothing store – as shown in the text in (1) (see also Blum-Kulka et al. 2004). As soon as they meet, the host girl suggests the play (“Ok, listen, we can play saleslady”) and the girls then engage (in turns 4 to 88) in carefully setting up the scene for the play, designating a toy phone as the cash register and choosing items of clothing for display. As they switch (in turn 88) to in-frame enactment, they sustain this frame by shifts of register, relying heavily on prosody (intonation contour and pitch) and lexical choices to mark the

 Shoshana Blum-Kulka

shift. When ’in character’, both girls use a style of speech very different from their casual conversational style. This is marked by use of high pitch and a mixture of stock ‘shop talk’ expressions with lexical and morphological choices typical of a formal Hebrew style (185 – “I’ll be with you shortly/”; 187 – “It’s very fashionable’”; 189 – “This dress is very festive / It’s gorgeous. . . amazing, beautiful”) Neither the ‘shop talk’ register nor formal-type Hebrew are necessarily authentic representations of the two styles in actual use, but rather are ‘demonstrations’ of imagined styles (Clark & Gerrig 1990). Yet they achieve a functional differentiation between in-frame and out-of-frame talk, as dictated by the goals of the play. The in-character talk is ambiguous on role divisions: the girls may be both playing saleswomen talking to an imagined customer, or playing saleswoman and customer. Consider the exchange in turns 184 and 185 of (1): Noy’s bid for attention in her role as customer (or as co-saleswoman) is atypical for normal shop-talk; instead of asking for help, she demands help, inflecting the verb ‘le-hitpanot’ literally ‘to-become free, make time’ in the imperative (in future tense serving as imperative). Liat accommodates to Noy by repeating her lexical choice, keeping to the standard inflected form of 1st person future singular prefix et- (instead of the colloquial, non-standard form that collapses 1st to 3rd person to yield yitpane for both ‘I’ll make time’ and ‘he’ll make time’) (Berman 1990) and she maintains a formal high register by choice of od daka ‘another minute = in a minute, shortly’ instead of more colloquial or everyday od shniya ‘in a sec’ or od rega ‘in a moment’. Thus both the lexicon and morphology of an imagined formal shop-talk are enlisted for the play, allowing the children practice in linguistic variants normally associated with written, rather than spoken language. . Speech representation and discursive literacy: Conversational narratives The choice between direct and indirect modes of speech representation has been defined as a choice between depicting and describing (Clark & Gerrig 1990). In this view, direct quotations are selective depictions of previous utterances, while indirect reports are their descriptive summaries. Depictions are fundamentally different from descriptions. According to Clark’s (1996) ‘demonstration theory’, “. . .quotations are intended to give the audience an experience of what it would be like in certain respects to experience the original event. Indirect quotations force the audience to recreate such experience on their own” (Wade & Clark 1993: 808). In contrast to ‘wording theories’ (e.g., Coulmas 1986; Leech 1983), demonstration theory does not assume that direct

The role of peer interaction in later pragmatic development 

Table 1. “The dress shop”4 Situation: Two girls, Noy (aged 9;0) and Liat (9;5) are in Liat’s home, playing dress shop for the full 28 minutes of the recording session. 184 NOY: titpani.

Noy: Attend (literally: make time) ((formal register))

185 LIA: ani etpane od daka. Ken, vakasha.(. . .) u-hu. Simla. ken, smalot erev, hakol. ta’amini li, ◦ yesh po et hakol. kol ma she at roca.◦ hine, ze kova, nora ba ofna. Noy, ma im ha simla ha zot?

Liat: I’ll be with you in a minute. Yes, please. (. . .) a-ha. A dress. Yes, evening dresses, everything. Believe me, ◦ we’ve got everything. Everything you could want.◦ There you are, this hat, it’s very fashionable. Noy, what about this dress?

186 NOY: texef ani axnis ota.

Noy: I’ll put it in in a minute.

187 LIA: hi nora ba-ofna. Kashe eh, (3.0) ha-simla hazu hi nora xagigit.mi- at roca limdod. Az kxi et ze, timdedi, bentayim anaxnu navi lax od. aval kodem kol (...) ma, ma at roca? Beseder. az hine, (...). at ohevet et ze? Beseder, bentayim yesh lax shtey sh- eh, parit exad, shehu ole eh axar kax, ani axarkax, ok, ken, ma? Ah.(...)[ha ze, nora xamudim]

Liat: It is very fashionable. It’s hard hm, (3.0) this dress is very festive. Who- do you want to try it on. Then take this one, try it on, in the meanwhile we’ll bring you more. But first of all (. . .) what, what do you want? Fine. Then, there you are, (. . .). Do you like it? Fine, meanwhile you’ve got two tw- ehm, one item that costs eh afterwards, afterwards I’ll, OK, yes, what? Oh (. . . ) [oh this, they’re real cute]

188 NOY: [(...)] sof ona ze ole paxot.

Noy: [(. . .)] end of season it costs less.

189 LIA: nora yafe al ha-guf ve- ze yafe, ze madhim. az et ze. beseder. (2.0 ) ah? ma xamuda? beseder. ze nora gam ze, ze gam raxav, madhim, yafe. ((laughs))

Liat: Really great on you and- it’s beautiful, it’s gorgeous. Then this. Fine. (2.0) ah? What sweetheart? Fine. This is very also this, this is also wide, amazing, beautiful. ((laughs))

quotations represent the wording of the original speaker. Rather, such ‘constructed dialogues’ (Tannen 1989) may involve reproduction of non-existent utterances and reports of unspoken thoughts or generic utterances attributed to ‘many’ or ‘all’ people. .. Textual indicators of direct quotes as ‘demonstration’ The schoolchildren we studied made extensive use of both direct and indirect modes of speech representation, especially in conversational narratives. The extended sequence of reported speech in (2) reveals that, as predictable by demonstration theory, a range of textual indicators show that the children use direct quotes (a) as a rhetorical device to depict selectively previous utterances

 Shoshana Blum-Kulka

rather than to reproduce them verbatim; (b) to distinguish clearly between different speaker-perspectives; as well as (c) to enhance the tellability and the moral stance of the story that is being told. The first quote in the excerpt in (2) – turn (190) – takes the form of ‘direct reported speech’ or DRS (Holt 1996). It contains three types of indicators signaling its function of being not a literal reproduction but rather a generic depiction of Nadav’s repeated request to come and visit his friend Ofer: The use of a present-tense verb meshagea ‘drives me crazy’ indicating generic habituality combines with temporal adverbials indicating iterative and protracted aspect – ‘every day’, ‘all year’ – and constant use of iconic repetition. These three indicators recur in the two other instances of DRS denoting habitual action. Thus in turn (192), Ofer depicts his own previous utterance to Nadav using present tense and several repetitions (“I say to him ... I say him”) and reinvokes Nadav’s request in turn (198), using all three devices yet again. Turn (199) shows nicely how a DRS can indeed relate to a nonexistent utterance; here Oz engages directly with the reported speaker in the previous utterance – and by enacting Ofer’s imagined response in the first person singular he manages to completely appropriate his friend’s voice as his own, speaking for the other. Ofer’s reformulations in turn (200) – “Got to tell him again and again” and “he talks to me all the time” – serve as metalinguistic comments that further clarify the function of some of the direct reported speech utterance in the previous exchange. Turn (201) illustrates yet another projection of a non-existent utterance, where Oz explicitly instructs Ofer what to say in the future when Nadav calls.5 .. Drawing boundaries between speaker perspectives As pointed out by Holt, DRS allows for “making a clear distinction between reported speaker’s point of view as displayed in his or her talk and the current speaker’s attitude toward the utterance being discussed” (Holt 1996: 230). DRS is an important site not only for the enactment but also for the evaluation of what Koven (2002) calls “socially locatable voices”. Drawing on the Bakhtinian (1981) notion of ‘voicing’, and Goffman’s (1981) idea of “footing”, Koven (2002) suggests (for narratives) a threefold distinction between the authorized, non-evaluative voice of the narrator, the critical distancing voice of the interlocutor, who evaluates the narrated events from his current telling position, and the narrated persona voice of the character, when the teller reanimates or re-inhabits the narrated persona. In stories of personal experience, speakers can juggle all three voices to invite their audience to share their attitudes towards the narrated events and/or their protagonists. Shifts in voice or footing are part and parcel of the repertoire of evaluative devices available in

The role of peer interaction in later pragmatic development 

Table 2. “The hedgehog ball” Situation: The conversation between two boys, Ofer (9;4) and Oz (9;7) took place towards the end of a 23-minute recording in a fast-food restaurant in a town in central Israel. Talking about being recorded, Ofer admits to a former unsuccessful attempt with a classmate called Nadav, who is reported to have failed as a proper conversational partner, preferring play with the computer to talk. What follows is a series of ‘Nadav bashing’ anecodotes. 189 Oz: xaval al ha-zman. U shige’a oti Oz: Awful (lit.”it’s a waste of time”). He im ha-pley steyshen shelo eze karciya drove me crazy with his play-station what (. . . .). a pain (. . . .) 190 Ofer: hu meshage’a oti kol yom, #ef- Ofer: He drives me crazy every day, #can shar la-vo elexa efshar la-vo elexa?# kaxa I come to you #can I can come to you?# shana shlema. ((mocking tone; deep voice)) That way all year. 191 Oz: gam ecli!

Oz: Me too!

192 ani omer lo #lo, ha-horim sheli lo nimca’im, omer l-ani omer lo lo, ima sheli lo marsha-# ani lo roce, hu yaharos li et ha-bayit ze ha-meshuga.

I say to him no #no, my parents are not at home, say t-I say to him no my mother does not allow it-# ((very assertive tone)) I don’t want, he’ll break down our house the crazy . . .

193 Oz: ken, gam li (. . . . . . . Psixolog).

Oz: Yes, me too (. . . . Psychologist ((probably meaning “psixi”, in Hebrew “crazy”))

194 Ofer: (. . . . . .). Sixaknu ba-kadur ha- Ofer: (. . . . . .) We played with my hedgehog kipodim sheli, yesh li kadur (. . . .) ha- ball, I’ve got a ball (. . . .) that little one?= katanchik ha-ze?= 195 Oz: = #ex ze (( ha-xalon)) yaxol le- Oz: # but how can it ((the window)) splinhitpocec? ter? ((asked with amazement. It’s a rubber ball)). 196 OFI: (nadav) ha-meshuga ha-ze xoshev she-hu be-migrash. Hu ose vu-vu-vu pa, bum hitnapec ha-xalo- asa sedek anaki, xatax be-ta’ut et ha-(. . . . . .), hepil et ze ala- alav, (. . . . . . .).

Ofer: (Nadav) that nuts- he thinks thathe’s in a playground. He goes vu-vu-vu pa, boom blown up the windo- he made a huge crack, broke mistakenly th- the (. . . . . .), thrown it on- on it, (. . . .).

197 Oz: (3.3) shu:::ga. (. . . .) =

Oz: (3.3) cra:::zy. (. . . .) =

198 Ofer: =kol ha-zman, kol ha-zman Ofer: =all the time , all the time time he hu sho’el oti #lama-lama-lama-lama- asks me #why-why-why-why# ((whining lama# = tone))= 199 Oz: = #lama biglal she-nipacta li et ha- Oz: = # why because you smashed my winxalon, ima sheli ko’esetdow, my mother is mad - ((‘demonstrating’ a reply. Dramatic tone)).

 Shoshana Blum-Kulka

Table 2. (Continued.) 200 Ofer: amarti lo, aval h::u lo kolet et ze. carix le-hagid od pam ve-od pam ve-od pam. (3.9) lo yode’a ma la’asot ito kol ha-zman medaber iti (. . . .) ata zoxer ma ani amarti lexa.

Ofer: I’ve told him, but he:: doesn’t get it. Got to tell him again and again and again. (3.9) Don’t know what to do with him he talks to me all the time (. . . .) remember what I told you.

201 Oz: (. . . . . . .), hu yitkasher elexa amar e tomar lo, #stom ta pe ha-gadol shelxa kvar!# ((talking assertively in a commanding tone)) Natek lo! Ze nim’as!

Oz: (. . . . . . . . . .), [when] he calls you told, hm tell him, #shut your big mouth up!# ((talking assertively in a commanding tone)) Cut him off! It’s enough!

202 Ofer: #ani agid kol pa’am ze ta’ut. # eze Ofer: #I’ll say wrong number every time.# gadol! ((laughing)) (5.08) nira lxa she- ((whispering)) That’s cool#! ((laughing)). yishmeu ma she-dibarnu? (5.08) Do you think they’ll hear what we said?

story-telling, and are displayed also in conversation at large and markedly in representing the speech of others. In the “hedgehog ball” excerpt in Table 2, the ‘proper’ story-telling episode begins in turn, 194 (“We played with my hedgehog ball...”) and ends with the coda offered by Oz (“cra:::zy”, 197), but the story is embedded in a sequence of direct speech reports that share with this anecdote not only its protagonists, but also its basic stances, expressed via several shifts in voice, turning the sequence into a single coherent narrative event. Oz’s statement that Nadav drives him crazy (189) is taken up by Ofer (in 190) to become the unifying theme of the entire segment. All three voices (or footings) depicted by Koven are manifest in this exchange: In the speech-reporting instances, with a shift from the interlocutor’s explicitly critical voice (“he drives me crazy every day”, 190) to either a mocking or sympathetic reenactment of character voices, and by readopting the narrator’s voice in recounting the anecdote. The prosodic modes of delivery change with every voicing (as marked by # in the transcript) as well in the transition from the in-character to the interlocutor voice (as in 192, from “my mother does not allow it” to #I don’t want it”). Ofer’s comment in 190 – “he drives me crazy every day” – pre-echoes the animation of Nadav in the quote to come, anticipating and justifying its mocking tone of delivery. In re-inhabiting his own narrated persona (in 192), Ofer adopts the more distant footing of the author, using a neutral verb of saying (“I say to him”), and a ‘serious’, assertive tone in impersonating his own in-character narrated self. The story of the broken window is told entirely in the narrator voice, mixing external evaluation (“that nut – he thinks he’s on the playground”) with internal evaluative phono-

The role of peer interaction in later pragmatic development 

logical devices (“goes vu-vu-vu-pa, boom . . .”) (Labov 1972). In the last part of the exchange (turns 198-202), the two speakers collaborate in close synchrony in shifting between the voice of the critical interlocutor (“all the time all the time he asks / I’ve told him but he:: doesn’t get it. . .”) and animating the in-character voices of Nadav, the voice attributed by Oz to the future persona of Ofer, and finally, as a distant echo, even evoking the voice of Ofer’s mother (“my mother is angry”). .. Tellability, moral stance, and collaboration The “hedgehog ball” story, like the other conversational stories of preschoolers and 9-year-olds in our data, display some of the overall characteristics of adult conversational stories. Studies of conversational stories in different natural settings (e.g., Blum-Kulka 1997; Ochs & Capps 2001) reveal such stories as complex, non-linear, multi-dimensional interactional achievements that rarely conform to the Labovian structure of temporally and causally ordered events organized around a point. As argued by Ochs and Capps, instead of considering such stories as flawed renditions of well-formed narratives, they should be considered as adhering “to an aesthetic that calls for openness to contingency, improvisation and revision”, an openness that requires skillful collaboration, in “the quest for experiental complexities and alternative possibilities” (2001: 61). Developmentally, conversational stories allow us to study children’s skills from several perspectives: in terms of conversational skills of turn-taking and dialogicity – for how story-coherence is maintained by co-tellers locally across turns in narratively extended stretches of conversation; in terms of narrative structure – for how a narrative event (or episodes within it) is globally organized; and also in terms of how children construct their criteria for tellability and how they adopt, develop, and display their moral stances. The multidimensionality of conversational stories cuts across the boundaries between ‘conversational’ versus ‘discursive literacy’ facets of pragmatic competence: To be good story-tellers, children and adults alike need to orient to dialogic sensemaking, to narrative performance before and with an audience, as well as to overall narrative coherence. Dialogic collaboration in the excerpt in Table 2 takes several forms. In the first part (189–193) the turns are closely tied to each other cohesively – by co-referentiality (he = Nadav, the other), by repetition (in different grammatical forms) of the same lexical token (drove me crazy / drives me crazy / crazy / crazy), and ellipsis (me too / me too). In the second part (turns 194– 197), coherence is upheld by conversational moves involving the audience and supporting the story-telling. In turn 194, Ofer invites audience acknowledg-

 Shoshana Blum-Kulka

ment of shared background by completing the turn with rising intonation on “I’ve got a ball (. . . ) that small one?=”. Oz’s contribution in the next turn (195) preempts the main event to come, probably signaling a close familiarity with the story. It seems this story has been told more than once before, making it into a ‘friendship-fable’, much like the repeated family-fables of dinner table conversations (Blum-Kulka 1997). The move displays Oz’s high involvement in his amazement at the facts anticipated and as indeed reported, by providing the coda for the story (in 197). Collaboration reaches a particularly high level in the third part (198–202), first with Oz impersonating Ofer in his imagined conversation with Nadav (199), and second by Oz spelling out for Ofer the very words he is supposed to say to Nadav in the eventuality of a phone conversation between the two. According to Bruner (1986), conversational stories focus by definition on some breach of expectations; yet the definition of what is considered a breach, hence judgments of tellability, may vary by age and culture (Blum-Kulka 1997). The tellability of conversational stories may vary with context: A story may be told in order to invite the audience to share an emotional response to a remarkable event, or to project a message that invites them to share a moralizing judgment rather than an emotional response (Martin & Plum 1997). Tellability may also range by orientation to form, from the high tellability of orientation to narrative as a rhetorically effective performance to the low (but not less socially meaningful) tellability of orientation to narrative as dialogic sense-making (Ochs & Capps 2001). The “hedgehog ball” excerpt combines such dialogic sense-making narrative (turns 189–194 and 198–202) with an (albeit brief) orientation to narrative as performance (turns 194 to 197) in a joint moralizing effort condemning a mutual friend. The tellability of the events reported derive from the wider topic of the children’s choice of conversational partners for the recording sessions. In the dialogue that precedes excerpt (2), Ofer admits to having been already recorded once with Nadav, and blames Nadav for the failure of the recording. Oz agrees it was not a good choice and, from turn 189 on, the two boys collaborate in building up this argument. Thus from Ofer’s point of view, by covertly admitting fault, the sequence that follows serves to exonerate him from the blame of having initially preferred another friend; for both boys, the sequence builds up as providing warrants for the specific claim of Nadav’s unsuitability as a conversational partner (for the research situation) as well as his inadequacy as a friend. Thus the window-breaking incident, tellable on one level in its own right as representing a breach of proper behavior, also serves in the broader context as a highly tellable moralizing tale, an exemplum (Martin &

The role of peer interaction in later pragmatic development 

Plum 1997) of Nadav’s improper conduct at large. The quotes inserted in the parts preceding and following the account of the window-breaking incident enhance its tellability in several ways. They display the two speakers’ deep engrossment with the tale and its protagonist, dramatize the speech events that occurred and might still occur, and (in the first part) provide evidence for the claims being made. Peer talk thus affords children the opportunity to negotiate sense-making on different planes: in developing criteria for tellability, in sharing a moral stance, and in learning how to collaborate dialogically.

. Conclusion This chapter has argued for the merits of peer talk in later development on the basis of available evidence from sociolinguistic, ethnographically oriented studies of pre-adolescent and adolescent children (e.g., Hoyle & Adger 1998b), and on the basis of findings from our study of the discourse of Israeli children (Blum-Kulka et al. 2004). We postulated three facets of pragmatic development – conversational turn-taking, dialogicity, and discursive literacy – arguing that although all three have been touched on in studies of later pragmatic development, there are still important gaps in our understanding of the developmental path of each one. The case of speech representation examined here reveals several unique contributions of peer talk to pragmatic development. The transition suggested from voicing to reporting unfolds within the childhood habitus of talk and play, serving social functions of childhood culture while simultaneously fostering discursive learning. Thus the voicing enactments of pretend play serve the immediate needs of the play, while also affording children opportunities for practicing a wide variety of register talk and register shifts, with the possible outcome that they adopt the language of the enacted voices as their own and so enrich their own discursive repertoire. The direct speech reporting practiced in conversational story-telling of real experience offers a different type of challenge. It requires the juggling of different narrating stances, such as from narrator to critical interlocutor and vice versa, and so may foster the skills of perspective taking, with the opportunities offered for practicing (and/or learning) the linguistic means for encoding this. The case of speech representation further illustrates how the study of specific discursive phenomena may cut across the dividing lines of ‘conversational’ versus ‘discursive literacy skills’.

 Shoshana Blum-Kulka

It remains for future work in later pragmatic development to probe theoretically and empirically the various facets of pragmatic competence, including its cross-cultural variation, and the relative weight of peer talk in contributing to its development.

Notes . The project was funded by the American-Israeli Binational Science Foundation (BSF) Grant No. 98031, 1999–2002, and by an Israel Science Foundation (ISF) Grant No. 832/011, 2000–2003. Recordings in the preschool (mostly audio, with some video) took two forms: (1) child-focused recordings of children wearing lapel microphones connected to a small pouch and, in preschools only, (2) setting-focused recordings made by a small tape recorder placed at strategic points at the center of some activity, such as a drawing table. All recordings were supplemented by extensive field-notes. See Hamo et al. (2004) for the methodological rationale of the project and Blum-Kulka (in press-a) and Blum-Kulka et al. (2004) for some findings. . For Olson (1977, 1996) the transition from utterance to text signifies children’s progression from a focus on meaning (including indirect meanings) to a growing commitment to the actual words encoding the message. . Note that pretend play contains by definition an element of decontextualization by constructing hypothetical worlds distanced from the here and now (Harris 2000). But this is a different type of decontextualization from the one textually marked in stretches of extended narrative discourse. . Transcription Conventions: word – emphasis; Wo::rd – stretch; WORD – loud volume; ◦ word◦ – low volume; ↑↓ – pitch changes; – slow rhythm; >words< – fast rhythm; #words# – unique tone; (0) – pause; [words] – overlap; word= – overlatch; wor- – cut-off; (word) – transcription doubt; ((comment)) – comments. Turn numbers reflect the original numbering in the full recorded session the excerpt is taken from. In excerpt (1) the girls’ “shop talk” and formal register are in bold, and their use of conversational style are in italics. . Interestingly, Oz’s self-repair regarding the choice of verb tense in the switch between past and future tense (amar>tomar) is also marked for register. The use of the verb of saying lomar is unmarked for register in past tense, but marked as formal when in future tense. The colloquial choice for this slot would be another verb of saying such as tagid ‘say’.

On reading poetry Expert and novice knowledge Joan Peskin and David Olson

I dwell in Possibility – A fairer House than Prose – More numerous of Windows – Superior – for Doors – Emily Dickinson

This chapter addresses later language development from the perspective of how high school and university students make sense of poetry. We argue that poetry is an important example of how being part of a literate culture in general, and schooling in particular, determines our approach to language in later development. The chapter starts by briefly discussing the special features and unique communicative function of the genre of poetry. This is followed by a review of studies conducted by the first author in which students were asked to express their thoughts when reading various poems. We then relate the findings to the development of the ability to make the say-mean distinction in early childhood and later within the poetic genre. The chapter concludes by discussing the implications of studies on poetry reading for later language development.

.

The nature of poetry versus prose

Prose is typically monologic in intention, if not always in effect. For the most part, the aim when writing a non-literary prose passage is to avoid indistinctness and ambiguity, to try to say what one means so that the reader focuses on the content of the message rather than the words themselves. Poetry, on the

 Joan Peskin and David Olson

other hand, seems to invite interpretation, elaboration, and a consciousness of the linguistic form as well as the content of the message. The reader thinks not only about what among various possibilities is being said, but how the author is saying it. In many poetic texts, as Barthes has put it, poetry “stages language rather than uses it”, sets it aside from ordinary discourse and treats it as an object of attention in its own right (1983: 463). The intentions of a writer in writing non-literary prose are usually similar to those of a speaker in oral communication. According to Grice’s (1975) rules of communication, the listener can assume that the speaker is trying to speak in ways that can be understood and the listener should, therefore, not look for multi-faceted explanations unless straightforward ones will not do. Slobin’s “basic ground rules to which a communicative system must adhere” include the linguistic maxims or charges to “be clear,” and “to be quick and easy”: With regard to the former, the charge means that the “surface structures of language must not be too different in form and organization from the semantic structures which underlie them” (1977: 186). With regard to the latter, the attempt is to minimize effort, partly because of memory constraints: The speaker must ensure that the message is conveyed before the listener loses track of what is being said. While these maxims refer to oral communication, the writer of prose similarly does not wish to engender effort on the part of the reader, or to invite distinctions between the surface structure of language (what is written) and the underlying semantic structure (what is meant). The aim of the prose writer, with the exception of some literary prose, is not to demand invention of meaning, but rather to transparently facilitate recovery of meaning. Roughly speaking, the objective is for the text to mean what it says. It must be noted that, while the intention of the speaker to be clear and minimize effort might be shared by the prose writer, numerous differences between writing and speech have been documented in the psycholinguistic and discourse literature. These differences are associated with the different communicative conditions and processing constraints. For example, Chafe (1994), Chafe & Danielewicz (1987), Halliday (1987), Tannen (1982), and more recently Berman & Verhoeven (2002a), Strömqvist et al. (2002), as well as various chapters in this volume, have addressed such differences in terms of both the online and offline processes differentiating the two acts, and in terms of written language as a special discourse style (Ravid & Tolchinsky 2002). Furthermore, because of the very different acts of speaking versus writing, the intention to communicate exactly what the speaker or writer means is more difficult to achieve in writing than it is in speech (Olson 1996). Intonation, voice quality, and facial expression can signal, for example, whether an utterance is to

On reading poetry

be taken literally or ironically. Because written utterances tend not to have the shared contexts of oral communication, the writer of expository prose toils hard to craft discourse for transparency so that there is only one easily recovered meaning. Poetry, on the other hand, seems to be rather an extra-ordinary form of discourse. In contrast to the conversational postulate that the speaker is trying to be clear and that simpler explanations should supercede complex ones, the poetry reader may look for more complex explanations even when the poem appears to be simple: Poetry often tends to compress ideas, thus pushing the reader to look for more meaning than may be apparent in the mere lexical definitions of the words. Bruner (1986) describes the purpose of literary texts, and poetry in particular, as not to impart knowledge but to explore the breadth and depth of human experience. Unlike expository prose that demands the extraction of information, poetry requires that the reader resolve any ambiguity in the text by creatively adding meaning to it. Poetry, more than any other genre, tends to invite distinctions between what the text says and what it means or conveys. Scholars working in different perspectives have discussed how, when reading poetry, routine reading is replaced by a more active, interpretive effort: Jakobson (1987) noted that the language of poetry calls attention to itself and makes the text “strange.” For Iser, the poetic text must “bring about a standpoint from which the reader will be able to view things that would never have come into focus as long as his own habitual dispositions were determining his orientation” (1994: 142). Miall & Kuiken’s notion of “literariness” is similar to Iser’s: “(L)iterariness is constituted when stylistic...variations defamiliarize conventionally understood referents and prompt reinterpretive transformations of a conventional feeling or concept” (1999: 123). In a similar vein, Bruner wrote that literary texts “are about events in a ‘real’ world, but they render that world newly strange, rescue it from obviousness, fill it with gaps that call upon the reader, in Barthes’ sense, to become a writer, a composer of virtual text in response to the actual” (1986: 24). And Frye described the study of poetic texture as plunging “into a complicated labyrinth of ambiguities, multiple meanings, recurring images, and echoes of both sound and sense” (1994: 274).

. Research on students’ reading of poetry In describing the properties of language, Chafe & Danielewicz (1987: 110) propose that every speaker and writer “possesses a repertoire of devices which are



 Joan Peskin and David Olson

combined in varying mixtures depending on the context, the purpose, and the subject matter of language use. In other words, language adapts to its varying environments.” This is the topic of many of the chapters in this volume on later language development (e.g. Blum-Kulka, Jisa, Khorounjaia and Tolchinsky). But what about the listener and the reader? How do they adapt to varying environments in their textual understanding? In particular, can we describe their developing knowledge of the language and special interpretive processes involved in reading poetry? These questions were addressed in two series of studies which tested hypotheses about how high school and university students read poetry, using the methodology of “thinking aloud.” . Study 1: An expert-novice study Participants in the first study consisted of eight ‘experts’, PhD candidates in the English department at the University of Toronto, and eight relative ‘novices’, three of them undergraduates in their first two years of English studies who had taken one course on poetry, and five of them high school students in their last two years of a private school. Each student was instructed to “think aloud” as they read a metaphysical poem, On a drop of dew by Andrew Marvell and an Elizabethan love sonnet, Lyke as a huntsman by Edmund Spenser, but without the poets’ names on the texts. The subjects’ oral responses were tape recorded and transcribed. One of the questions explored was whether it was possible to identify interpretive strategies which appeared to be useful in making sense of the poems. Five of these strategies which differentiated the Experts from the Novices in their attempts to construct meaning in the two poems are briefly discussed below. (Details of the study are given in Peskin 1998). In the interests of simplicity, illustrative examples are restricted to On a drop of dew (see Appendix A). Since the poem may be difficult to grasp initially, we provide the following interpretation, which was common to all the experts in the study, but to only two of the novices: Using a conceit (an elaborately sustained metaphor), Marvell compares a drop of dew in a rose to a soul in the human body. The dew wishes to evaporate and return to the skies, and the soul longs to return to heaven. Analysis of the protocols revealed, firstly, that the experts showed greater evidence of using cues in the structure of the poems (see Table 1). For example, in part of her protocol, an expert reader ponders aloud as follows.

On reading poetry

Table 1. Interpretive strategies: total number of times operation was performed and, in parentheses, the number of subjects performing the operation at least once (n = 8), when reading On a Drop (Poem A) and Lyke as a Huntsman (Poem B) Experts

Novices

Operation

Poem A

Poem B

Poem A

Poem B

Structure as Cue

21(8)

12(8)

1(1)

5(3)

Meaning at the locus Of Binary Oppositions

11(6)

7(6)

4(3)

1(1)

Language as cue/ Word-play

11(6)

4(3)

1(1)

0(0)

Rhyme or Rhythm as cue

7(4)

9(8)

0(0)

3(2)

Scanning for Patterns

4(4)

3(3)

0(0)

0(0)

(1) The “Sun” ends the first part, “Till the warm Sun pitty it’s Pain”, and ends the poem itself, “Into the glories of th’ Almighty Sun”. Not that this tells me anything at the moment yet. I’m just looking for patterns. [Female, PhD student]

A few of the novices also consciously focused on the endings of each of the poems, for example as in (2). (2) I can’t quite grasp what the entire meaning is. I’m looking at the last four lines because I know that in most poetry they’re usually pretty important. [Female, final-year high school]

However, the experts, but none of the novices, also showed evidence of integrating structural elements with the poem’s meaning in more subtle ways, for example, by focusing on such features as: the relationship between the last line of each section (as in example (1) above from an expert); the division into grammatical units; how the arrangement by stanza corresponds to the sense; the division into lines and stanzas; and the resonance between structure and content as in a grouping of very short lines or a particularly long or circular sentence. This is illustrated in (3) from the protocol of the same student as produced the comment in (1). (3) . . . and then at the end of the sentence the “Skies exhale it back again”. So the whole sentence isn’t closed. It’s circular. It starts at the sky and goes back to the sky. So the whole structure of the sentence and those thoughts are circular, like the dewdrop. [Female, PhD student]



 Joan Peskin and David Olson

Another difference between the experts and the novices is that the experts gave evidence of recognizing that poets often use binary oppositions as important thematic devices. The experts displayed an effective strategy of looking for the meaning of the poem at the locus of the contradictions, juxtapositions or dialectic. This is shown in (4): (4) One of the things that comes right away is the definite sense of polarity, of inside, outside, where elements are first established as something distinct and then at some point dissolve into each other. [Female, PhD student]

In contrast, a novice student reading the same lines observes the binary oppositions but rejects them as too confusing, as shown in (5). (5) So you’ve got sort of an equation, or you’ve got a scale there, but it doesn’t give me any sense of clarity. The one word seems to negate the other somehow, and it just jumbles everything for me. I don’t like lines like these. They just jumble things. [Female, first-year undergraduate]

A third feature distinguishing experts and novices is that most of the experts, but only one novice, commented on word-play and the use of language. For example, after reading the last line, “Into the Glories of th’ Almighty Sun”, an expert comments as in (6): (6) The word, “Sun”, is probably a play on son, like Jesus, and it gets Christian overtones. [Male, PhD student]

Compare this to the comment in (7) of a novice, who fails to recognize the pun: (7) “Into the Glories of th’ Almighty Sun”. So it evaporates into the air when the “Sun” comes out. . .I guess at the beginning it was just talking about when it’s morning and then it gets into the day and the “Dew” goes into the “Sun”. [Female, final year high school]

The excerpt in (7) shows that the student has missed the religious overtones and the extended comparison of the dew to the human soul. Fourthly, the experts used rhythm and rhyme as amplifiers of meaning, as in the example of thinking aloud in (8). (8) ..gg..hh..ii.. So we see that, okay, wherever he’s actually closing an argument he gives us a rhyming couplet to sort of slam the door shut so that we get the idea... [Female, PhD student]

Only two of the novices even looked at the rhyme scheme, and that was only on one of the poems.

On reading poetry 

Finally, some of the experts used scanning the poem for words that would help them contextualize it as an effective technique, as in (9) and (10). (9) . . .as I’m skimming down the poem, um, it looks, I’m seeing words like “Manna”, so it’s going to have a religious meaning. [Female, PhD student] (10) [In] the very first line, “Orient” is positioning the poem for me as a Renaissance seventeenth-century poem, I think. That’s a word that often comes up meaning something precious. [Female, PhD student]

The many hours that expert students had spent reading sixteenth and seventeenth century poetry gave them an advanced knowledge of the kinds of language helpful in categorizing the poems. The novices, without the requisite experience in understanding Period poetry, often expressed frustration and the need for extra help, and five of them demonstrated early closure on at least one of the poems. In other words, the responses of the novices gave clear evidence that they were lacking in many of the requisite strategies and the specific knowledge about poetry and poetic language needed to interpret the poems used in this study. A second question that we explored was whether these novices had at least acquired the three general expectations proposed by the literary theorist, Jonathan Culler. Culler (1976) states that reading poetry is not a natural activity; rather, both writers and readers internalize a shared “grammar of literature,” an underlying special set of conventions, which function as expectations. The study of one poem facilitates the reading of the next, and so proficient readers need to have considerable experience in these conventions. For Culler (1976: 115), the main expectations are (1) the “rule of significance: read the poem as expressing a significant attitude to some problem concerning man and/or his relation to the universe”; (2) the “convention of thematic unity” whereby all the parts of a poem are related to create a unified, coherent whole; and (3) the “conventions of metaphorical coherence”, corresponding to an expectation that “the basis of poetic expression is the metaphor” (see also Frye 1978: 91). Analysis of the protocols in our study made it clear that all three conventions proposed by Culler (1976) guided most novice readings of both poems. To begin with the first of Culler’s conventions, his Rule of Significance, a novice high school student thinks aloud as in (11): (11) I’m not sure if the poet is writing this to point out. . . the importance of “Dew” . . . how the smallest little thing that we take for granted should be

 Joan Peskin and David Olson

considered, and we should take some time to, excuse the expression, stop and smell the roses. [Female, final-year high school]

This speaker-reader misses the conceit but clearly expresses an anticipation that the poet is intending to make a point. She shows evidence of having an expectation that poetry is meant to refer to something over and beyond itself, to mean more than what it says. Culler’s second expectation is that of thematic unity. One “should prefer those (explanations) which best succeed in relating items to one another rather than offer separate and unrelated explanations” (1976: 170). This is illustrated by the comment of a novice in (12a) and (12b). (12) a.

I can get many streams of ideas but I don’t know how to tie them together. What is he telling me? [Female, second-year undergraduate]

And later when she begins to make coherent sense of the poem: (12) b. It seems to be funneling, like pinning each thing on to the previous thing, feeling a little more secure about what I have established.

Another novice ponders aloud as in (13): (13) I don’t know, some of the phrases don’t, I’m sure they, like, fit together, but I can’t see how they fit together. [Female, final- year high school student]

The novices thus clearly verbalize their expectation that the poet intended the propositions to relate to each other so as to create a coherent, unified whole. The third expectation is metaphorical significance. The novices were not always able to make coherent sense out of the extended metaphors, but most of them showed evidence that they recognized the importance of metaphor in poetry, as in (14) (14) It seems to me to equate “Dew” with goodness and innocence and purity and part of the “Skies” and part of the heavens. [Female, junior year of high school]

This girl is attempting to unravel the terms of comparison, but fails to perceive the role of the soul as one of the terms. Not surprisingly, this same girl later expresses dissatisfaction with her reading. Another example from a novice relating to poetic metaphor is shown by (15). (15) The poem seems to be not really filled with metaphors and similes but pretty straightforward, when you first look at it. [Female, second-year undergraduate]

On reading poetry 

While this girl initially misses the conceit, her comment demonstrates that she is aware of metaphor as constituting a vehicle of poetic expression. These excerpts demonstrate that although the novices had difficulty making sense of these obscure Period poems, they had internalized a general poetry schema in the sense of their expectations for poetry as a type of discourse. In keeping with Culler’s (1976) predictions, they ruminated over the poet’s intentions, the unifying theme, and the metaphoric content of the poem. . Study 2: Comparative research on reading prose versus poetry The poems used in Study 1 were chosen by English professors at the University of Toronto according to the criterion that they were difficult enough to be a challenge to PhD students. The result was that we were not able to observe what the high school students could do when reading easier texts. Another more recent set of studies by the first author (Peskin 2002) examined genre effects on text processing by having participants think aloud as they read pairs of simple short texts composed for the study. The texts were worded identically, but one was in prose form while the other was recast in the shape of a poem. All subjects read both versions. Here we consider one part of the study which involved eight students of middle- to upper-middle class in the final two years of high school (aged 16 to 18 years). Three sets of texts were presented individually in counterbalanced order, so that half of the students read the prose text before the matching poem, and half received the poem before the matching prose text. For example, the prose form of the Hunter text was presented as shown in (16a). (16) a.

When the hunter hunts, the animals screech a warning, scurry together into the forest, and hide. Still and silent, they blend into the green grass and brown bark. When the hunter leaves empty handed, the animals begin to look hungrily at each other.

In its pseudo-poetic form it was presented as short run-on lines down the center of the page, as in (16b).

 Joan Peskin and David Olson

(16) b.

When the hunter hunts, the animals screech a warning, scurry together into the forest, and hide. Still and silent, they blend into the green grass and brown bark. When the hunter leaves empty handed, the animals begin to look hungrily at each other. *

Both versions of the other two texts, Drug Addict and Soldier, appear in Appendix B. If, as suggested earlier, poetry invites defamiliarization and a more active and imaginative interpretation rather than a simple truth judgment (Bruner 1986; Frye 1994; Jakobson 1987; Iser 1994; Miall & Kuiken 1999), three predictions emerge for the readings of the pseudo-poems. They should initiate (1) more cognitively effortful attempts at interpretation; (2) a search for polyvalence (multiple meanings) which includes the reader’s awareness of more meaning than may be apparent in the lexical definitions of the words they contain; and (3) a concern with the textual representation of the event, not only with what is being said, but how the author is saying it. With regard to the first prediction, the question of how to operationally define the cognitive effort involved in deliberate and reflective analysis is a difficult one. Following Kuhn (2000) and Olson and Astington (1993), we assumed that critical thinking would be reflected by the number of metacognitive terms referring either to a thinking process – expressed by use of verbs such as analyze, attend, conclude, infer, justify – or a thinking product – in nominal expressions like analysis, attention, conclusion, inference, justification. The metacognitive terms, or “language of thinking” (Tishman & Perkins 1997), used in the transcripts were identified, counted, and compared for responses to both the prose and the poetic texts. Next, protocols were examined for evidence of readers’ awareness that poetic form can be a signal to model alternative worlds, that what is meant may be more than what is actually said. That is, we expected to find that subjects would pay attention to multiple meanings such as possible ambiguity, varying perspectives, and symbolic content. Finally, we also exam-

On reading poetry 

ined the protocols for sensitivity to how the poet uses language, that is, student verbalizations which reflect attention to stylistic elements. .. Metacognitive terms The number of metacognitive words used by the participants when reading each text-type served as the first marker of critical thinking. This is illustrated in example (17), which reveals how a high school student thinks aloud when reading the Drug Addict poem. (Metacognitive terms are in italics): (17) I’m thinking that maybe this is an analogy for something else.. and I’m assuming that I am supposed to interpret for myself who the “drug addict” was. [Male, 17 years]

Table 2 shows that on all three sets of texts there were more metacognitive terms (types, not tokens) used when students read the poems. The Hunter text, in particular, yielded nearly twice as many metacognitive terms when the text was presented as a poem. When tokens were counted, again there were nearly double the amount of tokens for the poetry compared with the three prose texts (134 versus 82, 120 versus 71, and 90 versus 55 respectively). As further evidence of the greater cognitive effort invested in reading poetry compared with prose, we found that students spent more than twice as much time reading the texts presented in poetic as against those in prose form (an average of 3.46 minutes for each poetic text versus 1.88 minutes for each prose one).

Table 2. A comparison of higher-order thinking when students read three prose texts and the matching poems. Hunter Prose Poem

Drug Addict Prose Poem

Soldier Prose Poem

Number of Metacognitive terms (types per poem)

36(8)

69(8)

41(8)

63(8)

30(8)

48(8)

Reference to Multiple Meaning

0(0)

4(3)

3(3)

6(4)

1(1)

4(3)

Reference to Stylistic Appreciation

2(2)

19(6)

3(3)

8(5)

2(2)

7(4)

Operation

Note: The number of responses from all participants appears out of parentheses. The number of participants who showed evidence of higher-order thinking at least once appears in parentheses; N = 8.

 Joan Peskin and David Olson

.. Multiple meanings Protocols were also examined for awareness that more than one interpretation is possible, that there may be multiple possible meanings in poetry. A high school student verbalizes on reading the Soldier poem (Appendix B) as shown in (18). (18) (I)t’s something I could take in different directions. There is a lot of meaning, a lot of significance and you could put it in any context you want, whether it’s what people endured decades and decades ago.. or what people may have to go through today, tomorrow. . .. so you could go many places with it and that’s what I like about it. [Male, 17 years]

Another boy expresses his thoughts as in (19) when reading the Hunter poem: (19) I like the symbolism. It really applies to both human nature and subsequently the way politics and nations have been handled in the past, so despite the simplicity of the literal story, the analogy goes a long way. [Male, 17 years]

The examples in (18) and (19) illustrate how participants demonstrate their awareness that poems are ambiguous and open to multiple interpretations. As Emily Dickenson (1890/1960) points out in her poem that is reproduced at the beginning of this chapter, the house of poetry has many potential “Windows” and “Doors.” In reading the poetry versions, students effortfully thought about their own cognition, going beyond literal interpretations and mulling over alternative interpretations. In contrast, many of their prose readings demonstrate what Bereiter & Scardamalia (1987) call “knowledge telling”, that is, summarizing and fact-stating as distinct from the “knowledge-transformation” that is clearly evident in their readings of poetry. Table 2 above also shows that high school students made more references to multiple meanings when reading the poetic forms than when reading the prose versions. When references to multiple interpretations were totaled across the three matched pairs of texts, they yielded 14 comments made by five of the students on the poetic versions, compared with only 4 comments made by three students on the prose. .. Reference to stylistic elements A third aspect of our comparison of how students read prose versus poetry versions of the same text was to examine references to elements of style. As Langer (1995) notes, students become “critics” when they “objectify” their reading experience and the text itself, focusing on the author’s craft, on the text’s struc-

On reading poetry 

ture, literary elements and stylistic devices, and appreciating the way the text says what it says. Thus in (20), a student comments on the stylistic device of line structure in the Hunter poem and how the structure generates emotion: (20) I like the way at the beginning there’s only a few words in each line which has a scary effect. Then in the two long lines the “animals” are all working for each other, and every thing’s OK. Then they start fighting each other to eat. It returns to one or two words in each line, returns almost to the beginning, in a sense of the scariness. [Female, 18 years]

Another subject, in (21), shows creativity in examining the visual structure of the Soldier poem. (21) I don’t know, it looks like a person maybe with its arms extended, legs in a V shape going out. I guess maybe that’s why from the “soaked field” there are two different lines, maybe their structure symbolizes a person or a soldier (although) the head doesn’t really look like a head. [Male, 18 years]

When reading the Drug Addict poem, a third subject observes a binary opposition or contrast, shown in (22). (22) OK, well the first thing that really pops into my head is the mention of the “beach homes” and I think the “beach homes” and the “drug addict” to me are two, maybe even contradictory words that evoke images. . . (T)hat contrast for me (is) effective in making me picture the images. . .the “dark shadows”, the “beach homes”. [Female, 17 years]

And yet another boy, in (23), refers to the stylistic device of alliteration in the Hunter poem. (23) I’m thinking that it is a poem since it’s starred and centred. “When the hunter hunts”, there’s alliteration. . .. “Still and silent”, more alliteration. . . So what’s this saying? [Male, 17 years]

All eight high school students made a total of 34 comments reflecting their appreciation of stylistic elements when reading the texts as poems, but only 4 students made in all seven such comments (one-fifth the amount) on the texts in prose form. As Fish pointed out on the basis of an impromptu exercise in which he told his students that a short list of scholars’ names on the blackboard was a religious poem: “It is not that the presence of poetic qualities compels a certain kind of attention but that the paying of a certain kind of attention results in the emergence of poetic qualities” (1980: 26).

 Joan Peskin and David Olson

In sum, Study 1 showed that, compared to more expert PhD students, novice high school students and undergraduates still have much to learn about useful strategies and the cues provided by the poet’s use of language in order to make sense of Period poetry. Nonetheless, the relative novices did show evidence of having absorbed the three main expectations formulated by Culler for reading a poem. Study 2, which made use of simple pseudo-poems, further demonstrated how much knowledge older high school students have acquired about reading poetry: The textual shape of a poem immediately triggered effortful cognition; awareness of the indeterminacy of the text, and hence reinterpretive transformations; as well as reflections on the writer’s craft. There remains the developmental question as to when students begin to acquire these expectations as important for reading poetry. Preliminary results from an ongoing study by the first author suggest that 13-year-olds do not yet command these expectations. Unlike the older high-school and college students who take nearly double the time when thinking aloud about a poemshaped text than its prose counterpart, young adolescents spend no more time on the poems than on the prose versions. This suggests that these younger subjects reveal little in the way of expectation that the poetic genre demands reinterpretive transformations, that what is meant will be more or different than what is said.

. Language as an object of thought: Development of the Say-Mean distinction In Rosenblatt’s (1989) terms, when readers attend to a text as a poem, instead of reading it “efferently,” as merely having a referent, they can begin to read “aesthetically.” This is shown by the comment a student made in Study 2: “In poetic form each word they chose has to have a lot of meaning.” At this stage, students are no longer considering a one-to-one correspondence between language and meaning; rather, they seem to expect that the language would ‘mean more than what it says’. On this task especially, students also demonstrated a consciousness of form, of the way poetry says what it says. Eco describes poetic texts as having “two model readers. . .a first level, or a naive one, supposed to understand semantically what the text says, and a second level, or critical one, supposed to appreciate the way in which the text says so” (Eco 1990: 55). When high school students read the prose version of the texts given them in the study, only the naive reading was evident. However, when presented with the poetic form, the same students focused not only on meaning but also on how

On reading poetry 

the components of the poem help to construct and amplify the meaning: They attended, for example, to the structure of the poem, to features of contrast and alliteration. The identification of a text as a poem ‘de-automates’ thinking. It makes interpretation conscious or, as Nicholson (1984) suggests, it brings interpretation from the background, where it is more or less automatic, to the foreground, where it is subject to deliberate, reflective analysis. When reading prose, interpretation is cognitive; when reading poetry it is often metacognitive. In terms of Karmiloff-Smith’s developmental framework, interpretation of simple prose is an “automatic skill” where knowledge is “implicit.” But when the text is graphically manipulated to resemble a poem, domain-specific knowledge about poetry results in “more explicit representations that allow for flexibility and creativity” (Karmiloff-Smith 1992: 16). Karmiloff-Smith views the developmental process as both domain-general and domain-specific, proposing a phase model rather than a (Piagetian-like) stage model, such that within each domain the process is the same, but the timing varies across domains. A metacognitive awareness of the special linguistic and textual properties of poetry along the lines described above can be said to represent the final stage of an approach to understanding language that builds upon an earlier, more general metacognitive awareness of language as an object of thought. This awareness may be addressed in terms of children’s ability to distinguish between ‘what is said’ and ‘what is meant’. To this end, below we review studies which examine the early awareness of this distinction, as background to discussion of this development in relation to the poetic genre. In studying the emergence of the ability to represent what is meant as against what is said, the characteristic research paradigm involves situations in which young children experience non-intentional ambiguity. In a typical experiment, children overhear a speaker give an ambiguous message, for instance telling a listener that a toy is in ‘the red drawer’ when in fact there are two red drawers; or requesting merely ‘the blue flower’ when there is both a small and a large blue flower. When asked whether the listener will know which drawer the toy is in, children under age four insist that the listener will know, whereas most children over 4 years of age say that the listener will not know (Ruffman et al.1990). Further, prior to the age of five or six years, children will blame the listener for giving the small blue flower when the speaker wanted the large blue one (Robinson et al. 1983; Torrance & Olson 1985). It is not surprising that children pass this task at the time when they succeed on Theory of Mind tasks, since both require them to rely on their recently established ability to co-ordinate two different representations or perspectives. Thus, by age 5 years,

 Joan Peskin and David Olson

children can think about both their own knowledge state and the listener’s lack of knowledge (Astington et al. 1988; Peskin 1992, 1996) and they can maintain a representation of both what was said and what was meant. They now become conscious of the gap between the utterance and the intended meaning, in the sense of “I said X but I meant Y”. The utterance has become an object of the child’s thought. While by second or third grade, children can clearly differentiate what is said from what is meant, they still have difficulty when this involves more complex tasks such as identifying inferences made in text understanding (Beal 1990). And they are still in the early stages of developing the ability to say what they mean in their own writing and editing. Bransford et al. (2000: 222–223) report a study where fourth graders (aged 9 to 10 years) were asked to draw pictures of, say, a monster, and then to write a paragraph describing the content of their drawing (e.g., ‘Under his body he has four purple legs with three toes on each one’) which they e-mailed to same-aged partners at another elementary school. The resulting pictures, drawn by these partners based on the e-mails, provided feedback on the discrepancies between what the original partners in the study had said and what they had meant. As one of them reflected, “We said, ‘Each square is down a bit.’ What we should have said was, ‘Each square is all the way inside the one before it,’ or something like that.” This type of comment clearly draws attention to the gap between the written form and the intended meaning. To address the relationship between the say-mean distinction and different discourse genres, Lee et al. (2001) examined the development of 3- to 7-yearold children’s ability to distinguish a verbatim statement in a text (what was said) from a paraphrase with the same intended meaning (what was meant) in prose narratives and in nursery rhymes respectively. In narratives, meaning is more important than form, whereas in nursery rhymes, the sound patterns and wording are what matter most. As predicted, all children were more attentive to wording in the nursery rhymes than in the narratives and more attentive to the meaning in the narratives than in the nursery rhymes. Nursery rhymes are children’s introduction to the poetic genre where form is highlighted and the one-to-one correspondence between linguistic form and meaning is undermined: Rhyme, rhythm, and alliteration play an important role in this special genre, while the often nonsensical meaning of some of the words they contain makes conventional meaning less salient than in prose. During the grade school years, nursery rhymes are replaced by collections of more realistic types of poems which in the high school years give way to poems which have increasingly more figurative elements, where what is

On reading poetry 

meant will often be more than or different from what is said. While, as noted, young children become increasingly competent at considering both the surface form of the words and their intended meaning, in the special context of poetic discourse, even adolescents have difficulty distinguishing literal words from possible figurative meanings. Indeed, English teachers complain that young adolescents appear to have difficulty even in accepting that words can have symbolic meanings (personal communication). Thus, for example, when Keats, at the end of his forlorn poem, La Belle Dame Sans Merci states that “no birds sing,” young adolescents, while well able to explain the literal meaning of such expressions are reluctant to accept that this could be a metaphor. Older adolescents, on the other hand, do acknowledge the existence of symbolic content in the poetic genre (as shown in Study 1 above), even though they may acknowledge this in what is at times a rather a derogatory fashion, like saying “I wondered why the damn poets couldn’t just say what they meant” (Wexler 2003). Yet imaginative interpretation of metaphorical and other symbolic features of a poem remains a challenge for many of them. This is shown, for example, in a study by Svensson (1987) who gave short, highly symbolic passages to participants aged 11, 14 and 18. For example, one passage translated from the Swedish original in (24) read as follows: (24) Light-on everything Even the smallest beetle has a Shadow which he cannot run away from.

Svensson coded students’ responses on four levels: at the lowest level, an interpretation such as “It’s about a beetle that can’t run away from his shadow” was coded as literal descriptive; the next level was interpretive, e.g., “Even though the beetle is small it has a shadow like people have”; then thematic, e.g., You should pay regard to animals even though they are small”; with the highest level being symbolic, e.g., “It’s about an evil power that you can’t counteract” or “The light is God, the beetle people, and shadow their sins.” Svensson found that only 8% of 11-year-olds, 18% of 14-year-olds, and 43% of 18-year-olds assigned symbolic or partly symbolic meanings to this piece of poetic language. In sum, the say-mean distinction is fundamental to all conscious or deliberate interpretation, since it allows the separation of the speaker’s or writer’s intention (what was meant) from the literal representation (what was actually said). The metacognitive understanding of this distinction in very simple tasks begins to emerge after about 4 years of age, when children are able to co-ordinate different perspectives. Theory of Mind research shows that at this stage, children begin to take into account the knowledge state of their inter-

 Joan Peskin and David Olson

locutors, and also to think about an utterance as separate from the intended meaning of the speaker. Developmentally, this ability to bring into consciousness both what the literal utterance says and what the speaker means by this is first manifested with simple, less cognitively complex tasks. Thus, for example, it takes many years of schooling and increased literacy for students to reduce the gap between the forms in which they write and their intended meaning when editing their written work. This and related understandings of the complex interrelations and non-correspondences between say-write versus mean-intend continues to develop throughout the school years, as students encounter more complex content and increasingly more challenging tasks. In the specialized context of poetic discourse, the cognitive load imposed by distinguishing what is meant from what is said is particularly challenging. The studies described in Section 2.1 and 2.2 above showed that older adolescents have developed the understanding that in poetry the literal words may be different from, or mean more than, what they say, that is, they may have multiple meanings and metaphoric import. However, when students need to actually make sense of symbolic poetic texts, their interpretations may not be rich in symbolic content even in adulthood. This issue is addressed in the next section, where we consider the implications of studies of poetry reading for the domain of later language development.

. Poetry reading and later language development Consider, first, children’s early awareness of the distinction between what is said and what is meant: Is this simply a developmental achievement or is it a specialized competence attributable to literacy? Age does seem to be a necessary condition, but is it a sufficient one? Olson has argued that literacy introduces a new awareness of the precise verbal form, a fixed, stable version which can then be contrasted with the conveyed meaning. It brings certain distinctions into consciousness, among them, the notion of literal meaning (Dascal 1987; Olson 1996, 2003). The same questions apply regarding the development of the linguistic and textual expectations required for making sense of non-realistic poetry introduced in the later high school years. Let us take, for instance, the expectation that there may be multiple possible meanings in poetry. A model of metacognitive development which describes this developing flexibility is that of Kuhn (1999, 2001). She proposes a shift during adolescence from an Absolutist level, where children consider what they read in texts as objective facts, to a Multiplist

On reading poetry 

level, where students begin to often view what they read as opinions, freely chosen by their holders; they understand that knowledge from texts is subjective and uncertain, and that there may be multiple meanings. As we can see from some of the extracts from Study 2 (Section 2.2), such as (17) “I’m thinking that maybe this is an analogy for something else and I’m assuming that I am supposed to interpret for myself who the drug addict was,” 17- to 18-year-olds understand that poetry can involve multiple meanings, analogies, and symbolic content. But is this developing flexibility dependent on cognitive development, mainly or only, or on students’ experiences with literacy at school? From a “cognitive-developmental” perspective (Newkirk 1991: 336), the later language development required for analyzing literature depends upon the emergence of Piaget’s stage of formal operations where the “literalism” of the pre-adolescent child gives way to thinking about abstraction, symbols and all manner of possibilities (Applebee 1978; Inhelder & Piaget 1958). However, the nature of Piagetian formal operations is such that not all people show evidence of such thinking in all areas. In the highly complex, linguistically unique domain of poetry reading in particular, fewer than half the 18-yearolds in Svenssen’s (1987) study showed evidence of symbolic interpretations, and it is likely that even most adults are unconfident about interpreting highly symbolic poetry. Formal operations reveal pronounced “horizontal decalage” and, indeed, in Piaget’s later writings, he himself accepted that people reach the stage of formal thinking “in different areas according to their aptitudes and their professional specializations” including what he termed “advanced studies” (Piaget 1972: 10). While Piaget merely acknowledged the role of “advanced studies,” research on the development of expertise emphasizes advanced domain knowledge as the most important advantage that experts share. Ericsson and Charness (1994) characterize expert performance in terms of incremental increases in knowledge that result from the extended effects of experience and practice over a lengthy period. The resultant vast body of knowledge acquired is highly structured and organized in memory as schemata, and allows the expert to see large and meaningful patterns. For example, in the first study reported here (Section 2.1), the PhD students were able to use these patterns to create a contextual framework and anticipate the content: The expert, who in example (9) says she is “seeing words like Manna, so it’s going to have a religious meaning,” uses her knowledge to contextualize the poem as Metaphysical and anticipate what is to come. As this expert later expressed it, “I now have a portable context. I’m seeing an imaginary transparency which has Donne’s name at the bottom with notes in the margin.” While the poem was actually by Donne’s contemporary,

 Joan Peskin and David Olson

Marvell, this advanced level student had a frame of reference, a structured and integrated representation that systematically cued her knowledge. One implication of Study 1 for later language development is that an understanding of more difficult poems requires long and specialized instruction and practice in reading poetry. Indeed, a satisfying interpretation of Period poetry may require studies at the graduate school level. However, it was also clear (from both studies) that by the last two years of high school, students had had sufficient exposure to poetry reading to be familiar with many of the conventions and stylistic devices that are part of poetic discourse. When given a text in the shape of a poem, they showed consciousness of the text as a poem. They could set the text aside from ordinary discourse and treat it as an object to which various conventions apply. From a “cultural-developmental” perspective (Newkirk 1991: 336) later language development is tied up with the acquisition of societal practices through enculturation (Fish 1980; Gee 2001). As students progress through educational institutions, they increasingly become part of an “interpretive community” (Fish 1980) with its own particular interpretive strategies for reading. Students learn to become culturally literate. The identification of a poem activates certain assumptions and concomitant interpretive strategies. As Bruner (1996) notes, education aids young humans in learning by transmitting the “toolkit” the culture has developed, that is, the canonical ways of constructing meaning from texts. . Conclusion In conclusion, through cognitive development as well as the knowledge of and experience with poetry that comes from many years of education, students gradually acquire the culturally accepted ways of reading poetry: They expect to engage in an active search for coherence; imagining and elaborating; interpreting symbolic content, ambiguity and novel connections; and discussing how the poetic form helps to construct and amplify the meaning. In short, the reader understands that the writer means more than what he or she says. As in other studies on the development of expertise, the process is very lengthy. Symbolic interpretations, for instance, continue to be difficult for many adults not trained in poetry interpretation, even for otherwise well-educated adults. Acquiring an extensive, sophisticated vocabulary and the ability to construct well-organized expository prose texts are shown in other chapters in this volume to have a long developmental history. This seems even more so in the case of the specialized linguistic and textual understanding required to interpret poetic texts.

On reading poetry

Appendix A On a Drop of Dew by Andrew Marvell On a Drop of Dew See how the Orient Dew, Shed from the Bosom of the Morn Into the blowing Roses, Yet careless of its Mansion new, For the clear Region where ‘twas born, Round in its self incloses, And in its little Globes Extent, Frames as it can its native Element, How it the purple flow’r does slight, Scarce touching where it lyes, But gazing back upon the Skies, Shines with a mournful Light; Like its own Tear, Because so long divided from the Sphear. Restless it roules and unsecure, Trembling lest it grow impure: Till the warm Sun pitty it’s Pain, And to the Skies exhale it back again. So the Soul, that Drop, that Ray Of the clear Fountain of Eternal Day, Could it within the humane flow’r be seen, Remembring still its former height, Shuns the sweet leaves and blossoms green; And, recollecting its own Light, Does, in its pure and circling thoughts, express The greater Heaven in an Heaven less. In how coy a Figure wound, Every way it turns away: So the World excluding round, Yet receiving in the Day. Dark beneath, but bright above: Here disdaining, there in Love, How loose and easie hence to go: How girt and ready to ascend.

1

5

10

15

20

25

30



 Joan Peskin and David Olson

Moving but on a point below, It all about does upwards bend. Such did the Manna’s sacred Dew destil; White, and intire, though congeal’d and chill. Congeal’d on Earth: but does, dissolving, run Into the Glories of th’ Almighty Sun.

35

40

Appendix B Drug Addict prose Yesterday a drug addict was found in the dark shadows behind the beach homes. Drug Addict poem * Yesterday a drug addict was found in the dark shadows behind the beach homes. * Soldier prose This morning the injured soldier fled from the soaked field. Soldier poem * This morning the injured soldier fled from the soaked field.

The nature and scope of later language development* Liliana Tolchinsky

.

Introduction and overview

Few changes in the development of individuals are as remarkable as the emergence of language. It is not surprising, then, that the earliest stages of language acquisition – the appearance of the first words and the first combinations of words – have been of central interest to language researchers.1 This interest grew considerably following Chomsky’s proposal of the innate nature of language acquisition. Although Chomsky’s central claim – that language learning is possible due to a pre-wired Universal Grammar – is not an empirical statement but a logical requisite, it led psycholinguists to design empirical tests of this assertion. The bulk of such psycholinguistic research focuses on the speed with which children acquire language (e.g., McNeil 1970), the amount of innate knowledge they possess, and the similarity of their grammar to that of adults (e.g., Crain & Lillo-Martin 1999). Ironically, it was Carol Chomsky who, in the late sixties, suggested that children’s grammar may differ in some ways from that of adults even after age five. This idea gave rise to the domain of ‘later language acquisition’, the study of what happens to children’s language after age five, defined as “a frontier age psycholinguistically” (Karmiloff-Smith 1986b: 455). After age five, children produce longer sentences, acquire a larger repertoire of syntactic constructions, and are increasingly able to understand and produce constructions that were previously obscure to them. It would be wrong, however, to regard the age of five as a cut-off point between the use of basic and complex structures. What has puzzled researchers is “protracted acquisition beside formal similarities” (Johnston 1985), the fact that structures of simi-

 Liliana Tolchinsky

lar formal complexity but differing functions are sometimes acquired over a period of several years. There are also qualitative changes in the types of words children use, the ways they expand sentence constituents, the links they establish between sentences, and the strategies they use to interact with their interlocutors. Their communicative acts are embodied in new means of expression. Development does not consist of accumulating new linguistic forms; rather, previously acquired forms evolve to acquire new functions, and conversely, old functions come to be expressed by an increasing diversity of linguistic forms (Berman & Slobin 1994; Slobin 1973).2 This insight is one of the most potent ideas in the study of later language development. Numerous explanations have been advanced for the developmental changes in the relation of forms to functions. Some scholars have suggested that these changes have a cognitive underpinning. Researchers from various schools of thought – including those who hold that language development is guided by an independent language acquisition device that constrains the hypotheses that child learners make about language structure (Chomsky 1981) – suggest that such a language module may interact with other cognitive systems. Such interaction could take the form of general structural constraints on learning or it might reflect the limits that children’s available knowledge may set on their ability to discover linguistic patterns (Johnston 1985; Piatelli-Palmarini 1980; Slobin 1985). Other scholars explain the observed changes by children’s increasing experience with language per se (Karmiloff-Smith 1979, 1992; López-Ornat 1999). In this view, progress is seen as a move from ‘contextualized’ to ‘decontextualized’ mastery of language performance. That is, as children’s ability to reflect on language grows, they come to rely less and less on extralinguistic contexts for understanding linguistic expressions, they learn to consider such expressions in themselves, and are increasingly able to use language as a tool for understanding language. Children also become aware that a single linguistic expression may have multiple functions and, conversely, that a single function may be performed by several different expressions. All these are aspects of a process of ‘decontextualization’, not in the sense that language becomes dissociated from the contexts of its use but, rather, that the ability to separate language from context makes possible a profounder understanding of both, rendering the relation between them more flexible and dynamic. A third position holds that the crucial motor for change is discourse, which affects the internal organization of sentences and the meaning of sentence elements. It is not that discourse structure emerges only after age five, but that

The nature and scope of later language development 

the links children are capable of establishing between discourse components change, bringing about a reorganization of all linguistic systems. “Discourse provides children with a developmental mechanism for the acquisition of linguistic devices” (Hickmann 2003: 335). Discourse does not exist in a void, but within the specific constraints of genre and modality. Like Molière’s bourgeois, we speak prose, poetry, narrative, or whatever discourse genre the circumstances demand. There is no such thing as neutral use of language: People constantly attune their speech to specific intentions, purposes, and interlocutors. Thus, development is viewed in terms of the acquisition of different discourse genres, and the way that the cultural conventions of genres constrain the use of linguistic forms. Finally, theories of later language development are influenced by the idea of a strong linkage to literacy. In this view, literacy underlies numerous developments, including growth and diversification of the lexicon; assigning new functions to forms previously associated with a single function; establishing inter-sential connectivity; and understanding of non-literal uses of language, such as metaphor and irony; and far more (Fletcher & Garman 1986; Lee et al. 2001; Olson 1996). The goal of this concluding chapter is to show that these different approaches to explaining later language development – those that emphasize cognitive underpinning, experience with language per se, the effects of genre, and literacy – are inter-related, and their findings can be interpreted in a unified way. I will suggest that later language development is geared for two apparently opposing needs: appropriateness and divergence. By appropriateness, I mean the tendency of language use to become increasingly attuned to a variety of communicative circumstances; by divergence, its tendency to become increasingly individuated and heterogeneous. An utterance is appropriate when it takes into account previous and succeeding utterances, the interlocutor, and the communicative setting in which it is produced. This yields an acceptable choice not merely in grammar and sense, but also relating to the circumstances and interlocutor needs and expectations. Divergence implies that as children grow older, the discrepancies rather than the similarities in their use of language become more evident. This is because those features of language that develop with age are precisely those that are most sensitive to social and cultural experiences, such as advanced vocabulary or low-frequency syntactic structures. As children grow, their sources of language input become increasingly varied. Young children’s experience with language is mainly spoken, while schoolchildren and adolescents are also exposed to written input. As demonstrated across the chapters in this book,

 Liliana Tolchinsky

vocabulary becomes more specialized due to differences in schooling and the semantic specificity of the lexicon changes dramatically with age; children learn to distinguish linguistic registers to make appropriate distinctions between spoken and written usage and more colloquial compared with everyday usage; while the capacity to reflect on language increases, heightening children’s sensibility to figurative expression. I suggest that developing appropriateness and divergence involves two factors, one psychological and the other cultural. The psychological factor is possession of a ‘theory of mind’, the ability to understand and reason about the beliefs of others (Perner 1991; Siegal & Varley 2002). The cultural factor is literacy – not just in the narrow sense of learning to read and write, but in the sense of participating in the communicative activities of a literate community (Ravid & Tolchinsky 2002).

. Some features of later language development By age five, children have acquired the sentence structure of their language in keeping with target language typological constraints. This includes understanding how to group words within clauses, how to move them inside a sentence to create questions and passive constructions, and how to combine clauses in complex sentences. In languages with rich morphology, children early on master inflectional marking of agreement for such categories as number, gender, and person and they learn to apply principles of productivity and conventionality in the domain of derivation (Berman 2003a; Clark 1993; Clark & Berman 1984; Clark & Hecht 1982; Ravid, this volume). In lexical development, by age five children show less of a tendency to under- and over-generalize the meanings of words while concurrently being able to form better semantic representations of words in terms of their defining features or core meanings (Dockrell & Messer, this volume; Keil & Batterman 1984).3 That is, conventional meaning is attained. Regarding conversation, babies as early as age three months already use communicative gestures unique to our species. Mother and baby begin to talk to each other in turns long before the child is able to utter a word. Young children are good at turn-taking, both with other children and with adults. By the time they produce their first words, children can typically sustain long bouts of well-timed turn alternations (Kaye & Charney 1981; Snow 1977). By the age of three, children can produce and understand all sorts of greetings, prohibi-

The nature and scope of later language development 

tions, demands, requests for shared attention and promises. And they clearly possess a full repertoire of speech acts (Blum-Kulka, this volume; Ninio et al. 1994; Rollins et al. in press). What aspects of language use, then, develop later? Four main features characterize later language development: texture; meaning beyond conventional senses; dialogue beyond interchange; and the semantics of alternative worlds. . Texture With age, children’s mastery of linguistic texture grows in the sense that they are increasingly able to establish links between linguistic expressions (Both the words text and texture come from the same Latin word textus ‘tissue’.) For example, they learn to use anaphoric pronouns and to employ explicit markers of cross-utterance relations, such as for example or all the same. Crosslinguistic studies of narrative skills in children between the ages of four and ten, in English, German, French, and Mandarin Chinese show that the use of discourse-internal linguistic devices to introduce and maintain the characters in a story and to anchor them in space and time is a protracted development, which continues to evolve until age ten and probably afterwards (Hickmann 2003). For example, in languages that use indefinite markers for new information – the girl for a known girl, a girl for an unknown one – children begin to use this device to mark newness only at around age seven. In languages like Chinese, that mark newness by position and where new information is located post-verbally, this device appears even later, around age ten. Further, children in these studies were unable to provide adequate spatial anchoring for the location and displacements of the characters in the narrative, something only adult subjects did. And in marking temporal-aspectual distinctions, children began to use verbal morphology and particles for the grounding of such information in discourse only around age seven, a development that continued until age ten or later. The use of these discourse-internal devices implies that children learn to plan their discourse so as to reflect the interlocutor’s need to determine the newness of information, the locations of characters in the story, and the difference between discourse foreground and background (Hickman 2003). Older children create syntactically denser structures and more hierarchically organized texts. In both spoken and written language, they include more clauses in their sentences and connect sentences using an ampler range of devices. Studies in Spanish and Catalan on narratives and expository texts produced by nine- to sixteen-year-olds show that children in different age groups use different types of subordination, coordination, and juxtaposition for inter-

 Liliana Tolchinsky

clause combining (Aparici et al. 2002).4 In particular, use of lexically unmarked juxtaposed clauses and of additive coordination decreases with age, whereas relative and complement clauses become more common. The selection of connective devices, however, differs as a function of genre and modality. Coordination in general, and temporally marked adverbial clauses, are significantly more frequent in narratives than in expository texts, and in written than in spoken texts, whereas relative clauses are more frequent in expository than in narrative texts, particularly in subject position. In this sense, then, both amount and type of subordination can be said to function as a yardstick of syntactic development (Berman 1998). . Non-conventional meaning Later language development also involves a move beyond conventions, both at the lexical and the textual level. If early lexical development culminates in the understanding of conventional meaning, then the mastery of figurative language is a landmark in later language development. Figurative speech – metaphors, proverbs, idioms, and the like – is far from a marginal phenomenon, and understanding it requires important skills.5 These include distinguishing two or more meanings of a word; constructing a coherent representation of a text by integrating lexical and syntactic information with situational and linguistic context; and the awareness that what is said and what is meant do not always coincide (Levorato & Cacciari 2002; Peskin & Olson, this volume). Although some researchers attribute to very young children the capacity to understand and produce metaphors and irony (e.g., Winner 1988), when their behavior is carefully examined, it turns out that prior to age seven, children process language literally, even when it does not make sense in a given context (Cacciari & Levorato 1989, for idioms; Vosniadou 1987, for metaphors). It takes until age ten to twelve for speakers’ internal states and intentions to be taken fully into account. The ability to understand figurative language is strongly affected by the semantic transparency of expressions and by children’s familiarity with the domains involved. Thus expressions like ‘to cry over spilt milk’, in which the link between literal and idiomatic meaning is semantically motivated, are successfully interpreted by ten- and eleven-year-old children (Cacciari & Levorato 1989). Similarly for metaphors: attributive metaphors, in which there is a direct physical connection between the elements compared – e.g., ‘white as snow’ – are understood much earlier than relational metaphors like ‘smooth mood’, which involve a less obvious link between the terms (Gentner 1988). Further,

The nature and scope of later language development 

there is a considerable developmental gap between the ability to comprehend and to produce figurative language: only adolescents seem to use figurative language creatively. The ability to distinguish what is said from what is meant is also affected by genre. Lee et al. (2001) asked children to distinguish between a verbatim repetition and a close paraphrase of an utterance. The success rate was higher in nursery rhymes, which highlight surface form, than in narratives, where content takes precedence over form. With age, the effect of genre diminished and success became more consistent across tasks. Younger children’s sensitivity to genre constraints prevented them from forming a general distinction between verbatim repetitions and paraphrases. The authors interpret this as requiring the ability to reflect on texts, as a consequence of literacy. In fact, this genre sensitivity on the part of young children is remarkable: They seem to realize that paraphrasing is quite acceptable in narratives, but inappropriate in the case of poetry. The humor of high schoolers and college students in riddles and jokes is often linguistic, but ambiguity is a constant in all types of language use. The capacity to deal with ambiguous expressions like those as in (1) is considered part of the linguistic competence of every speaker (Pollock 1997). (1) a. The woman hit the man with a cane. b. The dog is ready to eat.

Yet developmental research shows that this is only true after around age eleven (Shultz & Pilon 1973). Younger children are able to deal with phonological and lexical ambiguity, but not with surface (1a) or deep (1b) ambiguity. In normal circumstances, children under age eleven use extralinguistic cues to select a suitable interpretation. However, the metalinguistic approach required to resolve sentence-level ambiguity in decontextualized, isolated sentences is a protracted development. Finally, in spite of the wide range and sophistication of young children’s speech acts, it is only from age five if not later that children use language to deceive (e.g., Sodian 1991; Sotillo & Riviere 2001). All types of figurative usage – metaphor, irony, idioms, and jokes – involve a distinction between literal and intended meaning, but the communicative goal is attained when the interlocutor perceives the intended meaning. In deception, the literal meaning coincides with the intended meaning, but – unbeknown to the interlocutor – this meaning is false. Such use of words to purposefully manipulate the mental state of others is a later development.

 Liliana Tolchinsky

Studies on young children’s use of genre (Hudson & Shapiro 1991; Pontecorvo & Morani 1996; Sandbank 2002) have found that they employ distinctly different forms to reflect different communicative purposes. That is, the problem of casting information into a given genre is solved very early. With age, children’s observation of genre constraints tends to relax as they gain increasing competence in the written modality. The boundaries between text types become more flexible, so that adolescents and adults can include reflective analyses in their narratives and personal anecdotes in their expository texts (Berman & Katzenberg 2004; Berman & Nir-Sagiv 2004; Reilly et al. in press). Text types become less canonical and increasingly divergent. . The development of dialogue As noted, turn-taking, non–verbal and verbal interchange, is a very precocious skill. During the two firs years of life, children in Western cultures are able to select appropriate answers to different speech acts based on real-world context and intonation patterns of the interlocutor. However, when interacting with peers, they fail to maintain a topic and rely instead on imitative devices – repetitions or ritualized variations of each other’s utterances – during these exchanges (Blank & Franklin 1980). Only after age four can children take into account not only the form and function of the other’s utterance but also the general structure of conversation, so that they start to produce responses that go beyond answering the interlocutor, and serve to keep the conversational interchange going (Blum-Kulka, this volume). For example, the use of utterance-final tags like ‘isn’t it’ and other types of conversational discourse markers serves this double function of reacting to the interlocutor and maintaining longer stretches of discourse. Similarly, use of auxiliary ellipsis (answering ‘Yes, I do’ instead of ‘Yes, I want to have some orange juice’) meets the dual function of strengthening cohesion and of promoting further talk (Karmiloff & Karmiloff-Smith 2001). With age, children increasingly attempt to create links within their own discourse and between their discourse and that of the interlocutor, a progression which inter alia leads to a decrease in unrelated conversational turns and increasing use of ‘perspective-related’ turns that take into account the perspective of the other person (Dorval & Eckerman 1984). In sum, although children from very early on are capable of sustaining rich and complex interchange, it takes until late preschool age, around five or six years, for them to move beyond interchange to dialogue and conversation.

The nature and scope of later language development 

. Alternative worlds In any situation in which people write or speak, they convey not only content, but also their relation to that content – their personal and social involvement – as well as their relation to their audience. This ‘discourse stance’ toward content, message, and audience has been studied from various perspectives (e.g., Berman et al. 2002; Ochs 1990). One component of this stance is the expression of epistemic and affective attitudes toward a message: its modality. These attitudes are expressed differently in different cultures, but every language has linguistic elements that function as modality markers to convey such information as the speaker’s certainty about an event or the desirability of its occurrence. Modality comprises a range of meanings, including intent, desire, doubt, possibility, and probability, but linguists have been mainly concerned with the difference between epistemic and deontic modality (Lyons 1977; Reilly et al. 2002). Semantically, the epistemic mode “characterizes the actuality of an event in terms of alternative possible situations, or worlds” (Chung & Timberlake 1985: 242), that is, it expresses a hypothetical or counterfactual possibility rather than a real situation. The deontic mode, on the other hand, characterizes an event as being allowable, necessary, or desirable in a given situation. The deontic mode requires a single, unique response to a situation; the epistemic mode opens the way to many alternative possibilities. In the cross-linguistic study on children’s developing skills in text construction (see Note 4) one notable age-related change in children’s attitudes towards the topic conflicts at school was the transition from a deontic to an epistemic mode (Reilly et al. 2002). Across different languages, children in the younger, grade school age-group considered that the conflicts they were asked to discuss were bad – as in (2a) from the oral talk of a grade-school boy; older children were more open to alternatives, possibilities, and uncertainty – as in (2b), from the talk of a high school boy. (2) a.

Que está muy mal ‘that it is very bad’ no se tiene que hacer en clase it should not be done in class’ b. Si no hubiera conflictos no seríamos buenos amigos ‘If there were no conflicts we wouldn’t be good friends’

The adverbs, modal operators, and modal auxiliaries that serve to express deontic or epistemic modes in different languages – like English may, could or

 Liliana Tolchinsky

maybe, possibly for epistemic and must, should, necessarily for deontic – are all part of children’s repertoire of linguistic forms from early on. But the use of these devices differs when discussing a topic and when telling a personal story about that topic. In the latter case, children use fewer modals and different types of adverbs. A similar move from the deontic to the epistemic mode, and another example of protracted development beyond formal similarity, is seen in choice of constructions that vary according to mood (conditional, imperative, subjunctive compared with indicative), especially in a language like Spanish, where this is marked inflectionally. Appropriate distinctions of mood present specific cognitive challenges, since they require children to form mental representations of events that are independent of, or even incompatible with reality. Spanish children show early mastery of inflectional mood paradigms (PérezLeroux 1998), and many produce subjunctive forms by age two (López-Ornat et al. 1994). However, these forms are used only with their deontic value in expressing commands and with complements of deontic verbs, as in (3) – from Hernandez-Pina (1984). (3) No bebas sufa [Rafael, 2;0] not drink-subj dirty ‘Don’t drink, it’s dirty’

Slightly older children use subjunctives successfully in narratives (Naharro 1996). But these children cannot transfer such uses to express non-actual entities by means of subjunctive relative clauses, as in (4), even well after relative clause constructions have entered their grammar. (4) Un cielo que tenga estrellas rojas A sky which has-subj red stars

The reason children fail to do so is because they cannot interpret the entities referred to – in this case, the sky – as non-actual. Even before age three, children are able to express two predications in a single utterance: a complement clause proceeded by a clause that expresses a mental attitude about its content, as in (5). (5) a.

Mira què he fet Look what I’ve done b. Quiero que juegues I want you to play

The nature and scope of later language development 

This occurs at an early age with verbs of perception or desire – as in the English equivalents of (5a) and (5b) respectively. Only after age five, however, is the same formal construction generalized to verbs of mental state, although, again, mental verbs like know are part of young children’s repertoire. Moreover, early uses do not accommodate the semantics of complementation – the possible discrepancy between the truth value of the complement clause and that of the sentence as a whole. Children under three cannot correctly answer the question ‘What did Juan say that he drank?’ – if Juan drank something different from what he said he did (de Villiers & de Villiers 2000). The semantics of complementation thus provide a window on other possible worlds, mental and communicative, in which propositions can be true even though their complements are not. Somehow, perhaps aided by the formal features of the sentence, children must come to realize that the truth value of the proposition needs to be computed with respect to these alternative worlds and not to the real world. What is the common denominator of the ability to participate relevantly in a conversation, to produce and understand non-literal uses of language, to use complex sentences, and to construct more cohesive texts? They all imply appropriateness: a more adequate adaptation to communicative circumstances. This includes consideration of context and expression; coordination of similarities and differences; and above all, awareness of the mental state of the interlocutor. At the same time, all these abilities involve an uncoupling – of intention from expression, of one possible meaning from another, of the intended from the shown – which enables divergence from convention, from canonical genre, and from literal meaning. Theory of mind (ToM) provides the psychological requisite for the development of these characteristic features of later language development; literacy provides the linguistic means for realizing this psychological possibility.

. How ToM relates to the central features of later language development Young children are aware that imagination and beliefs are mental phenomena, and that mental images are distinct from objects in the physical world. But they are unable to comprehend the indirect nature of others’ beliefs. This capacity seems to emerge around age five, and it has a neural basis: People with disorders such as autism often have a lingering impairment of ToM, and patients with brain lesions can develop such an impairment as a result of their injury (Siegal & Varley 2002).

 Liliana Tolchinsky

In tasks of ToM reasoning, most three-year-olds and some four-yearolds fail, but five-year-olds succeed.6 Younger children’s difficulties presumably stem from their regarding belief as a copy rather than a construed representation of reality (Wellman 1990). That is, they expect other people’s beliefs to be faithful representations of reality, without regard to how or when these beliefs were formed. Yet understanding mental states is essential for forming interpersonal relationships, for recognizing people’s motives for their behavior, and for anticipating reactions. It underpins social adaptation, sensibility to the needs of one’s audience, and the awareness of shared knowledge. Some studies (Pérez-Leroux 1998; de Villiers & de Villiers 2000) have established links between ToM and the emergence and generalization of certain linguistic elements or constructions. But even where such a link has not been experimentally tested, it is reasonable to assume such a relationship. Conversational interaction would be both a consequence and a prerequisite of ToM, since reading an interlocutor’s intentions is essential for carrying on a conversation guided by relevance principles. The same is true for semantics, particularly the lexical semantics of mental states. In this respect “conceptual and linguistic development may go hand and hand” (Gopnik & Meltzoff 1997). Without an adequate lexicon it is impossible to express mental states, and without some understanding of such states the lexicon cannot be acquired. A similar case is the difficulty in generalizing the use of subjunctives to describe counterfactual states as in (4) above. What is missing is the full capacity for mental representation of non-actuality, perhaps the same skill necessary for the understanding of false beliefs. From the perspective of formal semantics, both require the ability to consider more than one possible world. Finally, syntax is also involved, in the form of the syntactic process of complementation, which allows one propositional argument to be embedded in another. Language plays a role in the emergence of new ways of conceptualizing events, but at the same time, this cognitive development makes possible the kind of dual perspective involved in the skills discussed here: one’s own mental state versus another’s, true versus false beliefs. These new ways of thinking expand the repertoire of linguistic representations, and through them make possible new ways of reasoning. The language of complementation provides a new representational capacity for expressing propositional attitudes.

The nature and scope of later language development 

. How literacy affects later language development The term ‘literacy’ here refers not only to the acquisition of written language skills, but to social literacy, the process by which subjects growing up in a literate community become acquainted with the repertoire of discourse varieties that characterize that community. Literacy shapes later language development, as with age children are exposed to the increasingly wide range of linguistic forms associated with writing – not only through written texts, but also through spoken interaction (Blum-Kulka, this volume). Many of the linguistic features listed above as hallmarks of later language development are those associated with written registers. A widely acknowledged consequence of the introduction of writing in a speech community is that it makes available a greater diversity of linguistic forms for a variety of communicative circumstances. Even when written registers are first introduced, they differ linguistically from those of the preexisting spoken registers, to “include greater structural and lexical elaboration and higher informational integration, coupled with a lesser degree of structural reduction and interaction” (Biber 1995: 311). Literacy creates not only a more extensive repertoire of registers and expansion of the range of linguistic variation along several dimensions, but also broader patterns of linguistic variation. That is, literacy increases the heterogeneity of language uses, while concurrently facilitating the creation of normative standards. The evolving capacity to compare the functioning of linguistic elements in differing uses leads to a reorganization of those elements into coordinated systems (Karmiloff-Smith 1986b). Yet another way in which writing influences language development is by its specific nature as a mode of production. Writing frees language processing from the time pressures characteristic of oral language, so enabling the development of contemplative linguistic capacities – and consideration of sounds and words in isolation.

. Directions for future research This chapter has reviewed several facets of later, school-age language development, focusing on elements that are influenced by literacy, which can thus be partly attributed to the effects of schooling. To gain a more general understanding of later language development, a cross-cultural perspective is also needed. Although the psychological underpinning of the trends described above may

 Liliana Tolchinsky

be a universal human endowment, the ways these features are expressed are culturally determined. In other cultures, awareness of an interlocutor’s mental state may conceivably not be expressed by conversational turns, or by more cohesive texts, but by other linguistic or extra-linguistic means. The topic of figurative language, a crucial feature of later language development, also deserves further study, including production studies in contextualized elicitations (e.g., as part of a narrative), and by online comprehension studies (e.g., by measuring reaction time), and offline comprehension studies (e.g., by overt questioning, requesting paraphrases, etc.). And more careful control is needed of the kinds of linguistic materials that are used. Studying the developmental path of idioms, proverbs, sarcasm, metaphors, similes, jokes, riddles, and the many other non-literal uses of language can provide valuable insights into a range of questions. Among these would be how children establish links between linguistic expressions and contexts; how they distinguish alternative meanings of expressions at the word and sentence level; the role of metalinguistics in language use; the relation of lexicon to syntax, and so on. The entire domain of figurative language should be revisited from a psycholinguistic and cross-linguistic perspective, to identify the general cognitive and specific linguistic processes involved. It is generally agreed that by age five, children have access to the vast majority of linguistic structures of their language. However, as this volume has aimed to show, the way they deploy, recruit, organize, select, and modify specific linguistic structures in different communicative circumstances continues to develop well into later childhood. As children mature, these structures show increasing appropriateness and flexibility in adapting to different situations and speakers, as well as an increasing divergence toward more open-ended options of linguistic expression.

Notes * Writing of the chapter was partially supported by a grant of the Spanish Ministry of Education and Science DGES B-502000-076. I am indebted to Ruth Berman for her detailed comments and invaluable insights. I alone am responsible for misinterpretation or inadequacies that remain. . Examining the issues from two leading journals for the years 2000–2002, I found that out of 94 main articles only 7 had subjects over the age of six, plus two note-type articles concerning older speakers and a review of a book about later language development. Of the seven articles, two dealt with the evolution of figurative language and the rest with awareness

The nature and scope of later language development 

of linguistic units and with the relation of later language development to literacy. This sums up the scope of the field in a nutshell. . In this context, linguistic form is used for linguistic devices at all levels of language, from phonology to text including phonemes, syllables, free and bound morphemes, syntactic structures, and grammatical processes (Berman 1997a; Hickmann 2003). The term function refers to the way such forms are used to express referential content, the flow of information in extended discourse, and the many other goals of the use of language in context (Budwig 1991; Hickmann 2003) . Undergeneralizing is using a category word to mean only one particular object, as when a child says ‘cup’ to refer only to her own cup and not to others; overgeneralizing is using a word too broadly, such as saying ‘tree’ to refer to any kind of plant. . The study was conducted as part of the crosslinguistic project on Developing Literacy in Different Languages and Different Contexts (Ruth Berman, PI), referred to in the chapters by Jisa, by Khorounjaia & Tolchinsky, and by Wengelin & Strömqvist in this volume. . In American English about 5.9 non-literal expressions are uttered per speaking minute of free discourse (Pollio et al. 1977). Language addressed to children is full of figurative usages (Nippold 1991), and children’s stories contain various types of verbal comparisons (Broderick 1992). . A typical task for evaluating ToM is the “false belief task” (Wimmer & Perner 1983) concerning mistaken expectations. Children are shown a box used for some well-known product (say a familiar brand of candy), which turns out to contain something different, say pencils. They are then questioned about what they believed was in the box and what a friend who has not seen the unexpected content will believe is in the box. Other tasks are based on unseen displacement, following the same logic, aimed at exploring whether children can focus on and interpret different mental states, and what vocabulary they use to refer to them.

References

Abedi, J., Hofstetter, C., Baker, E. and Lord, C. 2001. “NAEP math performance and test accommodations: interactions with student language background.” CSE Technical Report 536. Center for the Study of Evaluation, University of California, Los Angeles. Adams, M. 1990. Beginning to Read: Thinking and Learning about Print. Cambridge, MA: M.I.T. Press. Aitchison, J. 2003. Words in the Mind: An Introduction to the Mental Lexicon, 3rd edition. Oxford: Blackwell Publishers. Allen, S. E. M. and Crago, M. B. 1996. “Early passive acquisition in Inuktitut.” Journal of Child Language 23: 129–155. Allen, S. E. M. and Schröder, H. 2003. “Preferred Argument Structure in early Inuktitut spontaneous speech data.” In Preferred Argument Structure: Grammar as Architecture for Function, J. W. Du Bois, L. E. Kumpf and W. J. Ashby (eds), 301–338. Amsterdam: John Benjamins. Alsina, A. 1996. The Role of Argument Structure in Grammar. Evidence from Romance. Stanford University: CSLI [Publication 62]. Alves, R. A. 2002. Limites e Processos Cognitivos na Composicao Escrita. MA thesis. Universidade do Porto: Faculdade de Psicologia e de Ciencias da Educacao. Alves, R. A., Castro, S. L., de Sousa, L. and Strömqvist, S. In Press. “The cognitive cost of execution in written composition.” In Recent Developments in Writing Process Research. Volume I: Basic Processes and Word-Level Effects, D. Galbraith, L. van Waes and M. Torrance (eds). Dordrecht: Kluwer. Andersen, E. S. 1992. Speaking with Style: The Sociolinguistic Skills of Children. London: Routledge. Anderson, J. R. 1993. Rules of Mind. Hillsdale NJ: Lawrence Erlbaum. Anderson, J. R. 1995. Cognitive Psychology and its Implications (4th ed.). New York: W. H. Freeman. Anderson, S. R. 1985. “Typological distinctions in word formation.” In Language Typology and Syntactic Description, Vol. III, Grammatical Categories and the Lexicon, T. Shopen (ed.), 3–56. Cambridge: Cambridge University Press. Anglin, J. M. 1993. “Vocabulary development: A morphological analysis.” Monographs for the Society for Research in Child Development 58 (10 Serial No. 238): 1–165. Aparici, M., Tolchinsky, L., Rosado, E. and Perera, J. 2002 – July. “A developmental analysis of connectivity in different genres and modalities.” Paper presented at the IXth International Congress for the Study of Child Language and the Symposium on Research on Child Language Disorders. Madison, WI.

 References

Appel-Mashraki, Y. 2001. Reading comprehension, prosodic reading and language abilities in 4th graders. MA thesis, Department of Communications Disorders, Tel Aviv University. [in Hebrew] Applebee, A. N. 1978. The Child’s Concept of Story: Ages Two to Seventeen. Chicago: University of Chicago Press. Ariel, M. 1990. Accessing Noun-Phrase Antecedents. London: Routledge. Aronoff, M. 1976. Word Formation in Generative Grammar. Cambridge, MA: MIT Press. Aronsson, K. and Thorell, M. 2002. “Adult-child talk and reaccentuation in children’s play.” In Talking with Adults: The Contribution of Multi-party Talk to Language Development, S. Blum-Kulka and C. Snow (eds), 277–295. Mahwah NJ: Lawrence Erlbaum. Ashby, W. J. 1992. “The variable use of on versus tu/vous.” Journal of French Language Studies 2: 135–57. Ashby, W. J. and Bentivoglio, P. 1993. “Preferred Argument Structure in spoken French and Spanish.” Language Variation and Change 5: 61–76. Ashby, W. J. & Bentivoglio, P. 2003. “Preferred Argument Structure across time and space: A comparative diachronic analysis of French and Spanish.” In Preferred Argument Structure: Grammar as Architecture for Function, J. Du Bois, L. E. Kumpf and W. J. Ashby (eds), 61–81. Amsterdam: John Benjamins. Assayag, N. 1999. Subject selection in discourse: Cognitive constraints and genre effects. Unpublished Tel Aviv University MA thesis. Astington, J. W., Harris, P. L. and Olson, D. R. (eds). 1988. Developing Theories of Mind. New York: Cambridge University Press. Atlani, F. 1984. “On l’illusionniste.“ In La langue au ras du texte, A. Grésillon and J.-L. Lebrave (eds), 13–29. Lille: Presses Universitaires de Lille. Auwarter, M. 1986. “Development of communicative skills: The construction of fictional reality in children’s play.” In Children’s Worlds and Children’s Language, J. CookGumperz, A. W. Corsaro and J. Streeck (eds), 206–230. Amsterdam: Mouton. Avivi-Ben Zvi, G. 2002. Morphological development in gradeschoolers: Nouns. MA thesis, School of Education, Tel Aviv University. [in Hebrew] Baayen, H. 1991. “Qualitative aspects of morphological productivity.” In Yearbook of Morphology, G. Booij and J. VanMarle (eds.), 109–149. Dordrecht: Kluwer. Baayen, H. 1994. “Productivity in language production.” Language and Cognitive Processes 9: 447–469. Baayen, H. and Renouf, A. 1996. “Chronicling The Times: Productive lexical innovations in an English newspaper.” Language 72: 69–96. Baddeley, A. and Wilson, B. A. 1994. “When implicit learning fails: Amnesia and the problem of error elimination.” Neuropsychologia 32: 53–68. Bakhtin, M. 1981. The Dialogic Imagination: Four Essays edited by M. Holquist. Austin: University of Texas Press. Baldwin, D. A. 1991. “Infants’ contribution to the achievement of joint reference.” Child Development 62 (5): 875–890. Baldwin, D. A. and Moses, L. J. 2001. “Links between social understanding and early word learning: Challenges to current accounts.” Social Development 10 (3): 309–329. Bamberg, M. 1987. The Acquisition of Narratives: Learning to use Language. Berlin: Mouton de Gruyter.

References

Bar-Ilan, L. and Berman, R. A. In Revision. Developing register differentiation: The Latinate-Germanic divide in English. Barthes, R. 1983. The Barthes Reader. New York: Hill and Wang. Bates, E. and Goodman, J. C. 2001. “On the inseparability of grammar and the lexicon: Evidence from acquisition.” In Language Development: The Essential Readings, M. Tomasello and E. Bates (eds), 134–162. Malden MA: Blackwell. Bates, E. and MacWhinney, B. 1987. “Competition, variation, and language learning.” In Mechanisms of Language Acquisition, B. MacWhinney (ed.), 157–194. Hillsdale NJ: Lawrence Erlbaum. Beal, C. R. 1990. “Development of knowledge about the role of inference in text comprehension.” Child Development 61: 1011–1023. Beard, R. 1993. “Simultaneous denominal derivation in word formation.” Language 69: 716– 741. Beck, I. L. and McKeown, M. G. 1991. “Conditions of vocabulary acquisition.” In The Handbook of Reading Research, P. D. Pearson (ed.), Vol. 2: 789–814. New York: Longman. Bedore, L. and Leonard, L. 1998. “Specific language impairment and grammatical morphology: A discriminant function analysis.” Journal of Speech, Language, and Hearing Research 41: 1185–1192. Bellaire, S., Plante, E. and Swisher, L. 1994. “Bound morpheme skills in the oral language of school-age, language-impaired children.” Journal of Communication Disorders 27: 265– 279. Benelli, B., Arcuri, L. and Marchesini, G. 1988. “Cognitive and linguistic factors in the development of word definitions.” Journal of Child Language 15: 619–635. Bentivoglio, Paola. 1994. “Spanish Preferred Argument Structure across time and space.” Revista de Documentaçao de Estudos em Lingüística Teórica e Aplicada (DELTA) 10: 227– 293. Bereiter, C. and Scardamalia, M. 1987. The Psychology of Written Composition. Hillsdale NJ: Lawrence Erlbaum. Berglund, E. and Sullivan, K. (eds). In Preparation. Computer Key-Stroke Logging and Writing: Methods and Applications. Dordrecht: Kluwer. Berman, R. A. 1976. “On derived and deriving nominals in Modern Hebrew.” In Studies in Modern Hebrew Syntax and Semantics, P. Cole (ed.), 57–98. Amsterdam: NorthHolland. Berman, R. A. 1978. Modern Hebrew Structure. Tel Aviv: University Publishing Projects. Berman, R. A. 1980. “The case of an (S)VO language: Subjectless constructions in Modern Hebrew.” Language 56: 759–776. Berman, R. A. 1986a. “A step-by-step model of language acquisition.” In Stage and Structure: Reopening the Debate, I. Levin (ed.), 191–219. Norwood NJ: Ablex. Berman, R. A. 1986b. “The acquisition of morphology/syntax: A crosslinguistic perspective.” In Language Acquisition, P. Fletcher and M. Garman (eds), 429–447. Cambridge: Cambridge University Press. Berman, R. A. 1987a. “A developmental route: Learning the form and function of complex nominals.” Linguistics 25: 1057–1085.



 References

Berman, R. A. 1987b. “Issues and problems in Modern Hebrew research.” Praqim 7: 84–96. [in Hebrew] Berman, R. A. 1987c. “Productivity in the lexicon: New-word formation in Modern Hebrew.” Folia Linguistica 21: 225–254. Berman, R. A. 1988a. “Word class distinctions in developing grammars.” In Categories and Processes in Language Acquisition, Y. Levy, I. M. Schlesinger and M. D. S. Braine (eds), 45–72. Hillsdale NJ: Lawrence Erlbaum. Berman, R. A. 1988b. “The ability to relate events in narrative.” Discourse Processes 11: 469– 197. Berman, R. A. 1988c. “Language knowledge and language use: Binomial constructions in Modern Hebrew.” General Linguistics 28: 261–285. Berman, R. A. 1990. “Acquiring an (S)VO language: Subjectless sentences in children’s Hebrew.” Linguistics 28: 1135–1166. Berman, R. A. 1993a. “The development of language use: Expressing perspectives on a scene.” In Language and Cognition: A Developmental Perspective, E. Dromi (ed.), 172– 201. Norwood NJ: Ablex. Berman, R. A. 1993b. “Developmental perspectives on transitivity: A confluence of cues.” In Other Children, Other Languages: Issues in the Theory of Language Acquisition, Y. Levy (ed.), 189–241. Hillsdale NJ: Lawrence Erlbaum. Berman, R. A. 1993c. “Crosslinguistic perspectives on native language acquisition.” In Progression and Regression in Language, K. Hyltenstam and A. Viberg (eds), 245–266. Cambridge: Cambridge University Press. Berman, R. A. 1994. “Formal, lexical, and semantic factors in the acquisition of Hebrew resultative participles.” In S. Gahl, A. Dolbey, and C. Johnson (eds), Proceedings of the Berkeleley Linguistic Society 20: 82–92. Berman, R. A. 1995a. “Narrative competence and storytelling performance: How children tell stories in different contexts.” Journal of Narrative and Life History 5: 285–314. Berman, R. A. 1995b. “Word formation as evidence.” In Proceedings of the 19th Annual Boston University Conference on Language Development, Volume 1, D. McLaughlin and S. McEwen (eds), 82–95. Somerville MA: Cascadilla. Berman, R. A. 1996. “Form and function in developing narrative abilities: The case of ‘and’.” In Social Interaction, Context, and Language: Essays in Honor of Susan Ervin-Tripp, D. Slobin, J. Gerhardt, A. Kyratzis, and J. Guo (eds), 243–268. Hillsdale NJ: Lawrence Erlbaum. Berman, R. A. 1997a. “Developing form/function relations in narrative texts.” Lenguas Modernas 24: 45–61. Berman, R. A. 1997b. “Narrative theory and narrative development: The Labovian impact.” Journal of Narrative and Life History 7, Special Issue: 235–244. Berman, R. A. 1997c. “Preschool knowledge of language: What 5-year-olds know about language structure and language use.” In Writing Development: an Interdisciplinary View, C. Pontecorvo (ed.), 61–76. Amsterdam: John Benjamins. Berman, R. A. 1998. “Typological perspectives on connectivity.” In Issues in the Theory of Language Acquisition, N. Dittmar and Z. Penner (eds), 203–224. Bern: Peter Lang.

References 

Berman, R. A. 1999. “Relevant features of spoken and written American English.” In Working Papers in Developing Literacy across Genres, Modalities and Languages, Vol. I, R. Aisenman (ed.), 4–19. Tel Aviv University: International Literacy Project. Berman, R. A. 2000. “Children’s innovative verbs versus nouns: Structured elicitations and spontaneous coinages.” In Methods for Studying Language Production, L. Menn and N. Bernstein-Ratner (eds), 69–93. Hillsdale NJ: Lawrence Erlbaum. Berman, R. A. 2001. “Crosslinguistic aspects of later language development.” In The Diversity of Languages and Language Learning, S. Strömqvist (ed.), 25–44. Lund University: Center for Languages and Literature. Berman, R. A. 2003a. “Children’s lexical innovations: developmental perspectives on Hebrew verb-structure.” In Language Processing and Language Acquisition in a RootBased Morphology, J. Shimron (ed.), 243–291. Amsterdam: John Benjamins. Berman, R. A. 2003b – February. “The role of register in sociolinguistic and psycholinguistic research.” Paper presented at Israel Association for the Study of Language and Society, Jerusalem. Berman, R. A. 2003c – July. “Tense/aspect shifting in narrative development across adolescence.” Paper presented at Child Language Seminar, Newcastle-Upon-Tyne. Berman, R. A. 2004. “The role of context in developing narrative abilities.” In Relating Events in Narrative. Volume Two: Typological and Contextual Perspectives, S. Strömqvist and L. Verhoeven (eds), 261–281. Mahwah NJ: Lawrence Erlbaum. Berman, R. A. This Volume. “Between emergence and mastery: The long developmental route of language acquisition.” Berman, R. A. In Press-a. “Genre and modality in developing discourse abilities. “ In Discourse across Languages and Cultures: Typological Studies in Language series, C. L. Moder and A. Martinovic-Ziv, (eds). Amsterdam: John Benjamins. Berman, R. A. In Press-b. “Developing discourse stance in different languages and in different genres.” Journal of Pragmatics Special Issue on Discourse Stance. Berman, R. A. and Katzenberger, I. 1998.”Cognitive and linguistic factors in development of picture-series narration.” In Organization of Learners’ Texts. [Special Issue of Studia Italiani I Linguistica Teorica e Applicata], A. G. Ramat and M. Chini (eds.), 27: 21–47. Berman, R. A. and Katzenberger, I. 2004. “Form and function in introducing narrative and expository texts: A developmental perspective.” Discourse Processes 38: 57–94. Berman, R. A. and Nir-Sagiv, B. 2004. “Linguistic indicators of inter-genre differentiation in later language development.” Journal of Child Language 31: 334–380. Berman, R. A., Ragnarsdóttir, H. and Strömqvist, S. 2002. “Discourse stance: Written and spoken language.” Written Language and Literacy 5: 255–290. Berman, R. A. and Ravid, D. 1999. The Oral Written Continuum: Developmental Perspectives. Final report submitted to the Israel Science Foundation, Jerusalem. Berman, R. A. and Ravid, D. Submitted. “Modeling information density in narrative texts across adolescence: Crosslinguistic and crossmodal perspectives.” Berman, R. A., and Slobin, D. I. 1994. Relating Events in Narrative: A Crosslinguistic Developmental Study. Hillsdale NJ: Lawrence Erlbaum. Berman, R. A. and Verhoeven, L. 2002a. “Developing text-production abilities across languages, genre, and modality.” Written Languages and Literacy 5(1): 1–44.

 References

Berman, R. A. and Verhoeven, L. (eds). 2002b. Written Language and Literacy, Volume 5, Parts 1 and 2: [Special Issue on Cross-Linguistic Perspectives on the Development of Text-Production Abilities in Speech and Writing]. Berrendonner, A. 1990. “Pour une macro-syntaxe.” Travaux de Linguistique 21: 25–36. Berrendonner, A. and Reichler-Béguelin, M.-J. 1997. “Left dislocation in French: Varieties, norm and usage.” In Taming the Vernacular: From Dialect to Written Standard Language, J. Cheshire and D. Stein (eds), 200–217. London: Longman. Bertram, R., Laine, M. and Virkkala, M. 2000. “The role of derivational morphology in vocabulary acquisition: Get by with a little help from my morpheme friends.” Scandinavian Journal of Psychology 41: 287–296. Bertram, R., Tønnessen, F. E., Strömqvist, S., Hyönä, J. and Niemi, P. In Preparation. Compound elicitation in Finnish: The Case of Writing. Technical Report. University of Turku: Department of Psychology. Best, R. 2003. Young children’s acquisition of novel science terms. Unpublished Doctoral Dissertation. South Bank University, London. Bialystok, E. 1991. “Metalinguistic dimensions of bilingual language proficiency.” In Language Processing in Bilingual Children, E. Bialystok (ed.), 113–140. Cambridge: Cambridge University Press. Biber, D. 1986. “Spoken and written textual dimensions in English: Resolving the contradictory findings.” Language 62: 384–414. Biber, D. 1995. Dimensions of Register Variation: A Cross-Linguistic Comparison. Cambridge: Cambridge University Press. Biber, D. 2001. “On the complexity of discourse complexity: A multi-dimensional analysis.” In Variation in English: Multi-Dimensional Studies, S. Conrad and D. Biber (eds), 215– 240. London: Longman. Biber, D., Reppen, R., and Conrad, S. 2002. “Developing linguistic literacy: Perspectives from corpus linguistics and multi-dimensional analysis.” Journal of Child Language 29: 458– 462. Biemiller, A. and Slonim, N. 2001. “Estimating root word vocabulary growth in normative and advantaged populations: Evidence for a common sequence of vocabulary acquisition.” Journal of Educational Psychology 93: 498–520. Bishop, D. V. M. 1997. Uncommon Understanding: Development and Disorders of Language Comprehension in Children. Hove: Psychology Press. Blanche-Benveniste, C. 1990. Le français parlé: Etudes grammaticales. Paris: Editions du CNRS. Blanche-Benveniste, C. 1994. “The construct of oral and written language.” In Functional Literacy: Theoretical Issues and Educational Implications, L. Verhoeven (ed.), 61–74. Amsterdam: John Benjamins. Blanche-Benveniste, C. 1995. “De la rareté de certains phénomènes syntaxiques en français parlé.” French Language Studies 5: 17–29. Blank, M. and Franklin, M. 1980. “Dialogue with preschoolers: A cognitively based system of assessment.” Applied Linguistics 1: 127–150. Bloom, P. 2000. How Children Learn the Meanings of Words. Cambridge, MA: MIT Press. Blum-Kulka, S. 1997. Dinner Talk: Cultural Patterns of Sociability and Socialization in Family Discourse. Mahwah NJ: Lawrence Erlbaum.

References 

Blum-Kulka, S. This Volume. “The role of peer interaction in later language development: The case of speech representation” Blum-Kulka, S. In Press-a. “Modes of meaning making in young children’s conversational narratives.” In The Sociolinguistic of Narrative, J. Thornborrow and J. Coates (eds). Amsterdam: John Benjamins. Blum-Kulka, S. In Press-b. ”Rethinking genre: discourse genres as a social interactional phenomenon.” In Handbook of Language and Social Interaction, K. Fitch and R. Sanders (eds). Mahwah NJ: Lawrence Erlbaum. Blum-Kulka, S., Huck-Taglicht, D., and Avni, H. 2004. “The social and discursive spectrum of peer talk.” Discourse Studies 6(3): 307–329. Blum-Kulka, S. and Snow, C. (eds). 2002. Talking with Adults: The Contribution of Multiparty Talk to Language Development. Mahwah NJ: Lawrence Erlbaum. Blum-Kulka, S., and Snow, C. 2004. “The potential of peer talk.” In Discourse Studies: Thematic Issue: Peer Talk and Pragmatic Development 6(3): 291–307. Bock, J. K. 1986. “Syntactic persistence in language production.” Cognitive Psychology 18: 355–87. Bock, J. K. and Levelt, W. 1994. “Language production: Grammatical encoding.” In Handbook of Psycholinguistics, A. Gernsbacher (ed.), 945–984. San Diego: Academic Press. Bolozky, S. 1999. Measuring Productivity in Word Formation: The Case of Israeli Hebrew. Boston: Brill. Booth, A. E. and Waxman, S. R. 2002. “Word learning is ‘smart’: Evidence that conceptual information affects preschoolers’ extension of novel words.” Cognition 84: B11–B22. Booth, J. R. and Hall, W. S. 1995. “Development of the understanding of the polysemous meanings of the mental-state verb know.” Cognitive Development 10 (4): 529–549. Booth, J. R., MacWhinney, B. and Harasaki, Y. 2000. “Developmental differences in visual and auditory processing of complex sentences.” Child Development 71: 981–1003. Borer, H. and Wexler, K. 1987. “The maturation of syntax.” In Parameter Setting, T. Roeper and E. Williams (eds), 123–172. Dordrecht: Foris. Bosman, A. M. T. and van Orden, G. C. 1997. “Why spelling is more difficult than reading.” In Learning to Spell, C. A. Perfetti, L. Rieben, and M. Fayol (eds), 73–194. Hillsdale NJ: Lawrence Erlbaum. Bourdin, B. and Fayol, M. 1994. “Is written language production really more difficult than oral language production?” International Journal of Psychology 29: 591–620. Bourdin, B. and Fayol, M. 1996. “Mode effects in a sentence-production task.” Current Psychology of Cognition (CPC) 15: 245–264. Bourdin, B. and Fayol, M. 2000. “Is graphic activity cognitively costly? A developmental approach.” Reading and Writing 13: 183–196. Bowerman, M. 1987. “Commentary.” In Mechanisms of Language Acquisition, B. MacWhinney (ed.), 443–466. Hillsdale NJ: Lawrence Erlbaum. Bowerman, M. 1990. “Mapping thematic roles onto syntactic functions: Are children helped by innate linking rules? Linguistics 28(6): 1253–1290. Bowerman, M. 1996. “Learning how to structure space for language: A crosslinguistic perspective.” In Language and Space, P. Bloom, M. A. Peterson, L. Nadel, and M. F. Garrett (eds), 385–436. Cambridge, MA: MIT Press.

 References

Bowerman, M. and Choi, S. 2003. “Space under Construction: Language-Specific special categorization in first language acquisition.” In Language in Mind: Advanced in the Study of Language and Thought, D. Gentner and S. Goldin-Meadow (eds), 195–235. Cambridge, MA: MIT Press. Bowerman, M. and Levinson, C. S. (eds) 2001. Language Acquisition and Conceptual Development. Cambridge: Cambridge University Press. Bowey, J. A. 1996. “On the association between phonological memory and receptive vocabulary in five-year olds.” Journal of Experimental Child Psychology 63: 44–78. Bowey, J. A. 2001. “Nonword repetition and young children’s receptive vocabulary: A longitudinal study.” Applied Psycholinguistics 22 (3): 441–469. Brabham, E. G. and Lynch-Brown, C. 2002. “Effects of teachers’ reading aloud styles on vocabulary acquisition and comprehension of students in the early elementary grades.” Journal of Educational Psychology 94: 465–473. Braisby, N. and Dockrell, J. E. 1999. “Why is colour naming difficult?” Journal of Child Language 26: 23–47. Braisby, N., Dockrell, J. and Best, R. 2001. “Children’s acquisition of science terms: Does fast mapping work?” In Research on Child Language Acquisition: Proceedings of the 8th Conference of the International Association for the Study of Child Language, M. Almgren, A. Barreña, M. J. Ezeizabarrena, I. Idiazabal and B. MacWhinney (eds), 1066–1087. Somerville MA: Cascadilla Press. Branigan, H. P., Pickering, M. J. and Cleland, A. A. 2000. “Syntactic co-ordination in dialogue.” Cognition 75: 13–25. Bransford, J., Brown, A. L., and Cocking, R. R. 2000. How People Learn: Brain, Mind, Experience, and School. Washington DC: National Academy Press. Brinton, L. 1996. Pragmatic Markers in English: Grammaticalization and Discourse Function. New York: Mouton de Gruyter. Broderick, J. P. 1999. “Wallace Chafe’s light subject constraint in conversational discourse in the immediate mode of consciousness.” Word 50 (2): 143–155. Broderick, V. 1992. “Incidence of verbal comparisons in beginners’ books and in metaphor compression research: A search for ecological validity.” Journal of Child Language 19: 183–193. Brown, A. S. 1988. “Encountering misspellings and spelling performance: Why wrong isn’t right.” Journal of Educational Psychology 80: 488–494. Brown, G. D. A. and Ellis, N. 1994. Handbook of Spelling: Theory, Process and Intervention. Toronto: John Wiley and Sons. Brown, R. 1973. A First Language: The Early Years. Cambridge MA: Harvard University Press. Bruner, J. 1986. Actual Minds, Possible Worlds. Cambridge, MA: Harvard University Press. Bruner, J. 1996. The Culture of Education. Cambridge, MA: Harvard University Press. Bryant, P., Nunes, T., and Aidinis, A. 1999. “Different morphemes, same spelling problems: Crosslinguistic developmental studies.” In Learning to Read and Write: A Crosslinguistic Perspective, M. Harris, and G. Hatano (eds), 112–133. Cambridge: Cambridge University Press. Budwig, N. 1990. “The linguistic marking of prototypical agency: An exploration of children’s use of passives.” Linguistics 28: 1221–1252.

References 

Budwig, N. (ed.) 1991. “Functional approaches to child language.” First Language 11 [Special Issue]. Bybee, J. L. 1985. Morphology: A Study of the Relation between Meaning and Form. Amsterdam: John Benjamins. Cacciari, C. and Levorato, M. C. 1989. “How children understand idioms in discourse.” Journal of Child Language 16: 387–405. Campbell, R. 1985. “When children write nonwords to dictation.” Journal of Experimental Child Psychology 40: 133–151. Carey, S. 1978. “The child as word learner.” In Linguistic Theory and Psychological Reality, M. Halle, G. Miller and J. Bresnan (eds), 264–293. Cambridge, MA: MIT Press. Carlisle, J. F. 1995. “Morphological awareness and early reading achievement.” In Morphological Aspects of Language Processing, L. B. Feldman (ed.), 189–210. Hillsdale NJ: Erlbaum. Carlisle, J. F. 1996. “An exploratory study of morphological errors in children’s written stories.” Reading and Writing 8: 61–72. Carlisle, J. F. 2000. “Awareness of the structure and meaning of morphologically complex words: impact on reading.” Reading and Writing 12: 169–190. Carlisle, J. F., Fleming, J. E. and Gudbrandsen, B. 2000. “Incidental word learning in science classes.” Contemporary Educational Psychology 25 (2): 184–211. Carnine, D., Kameenui, E. J. and Coyle, G. 1984. “Utilization of contextual information in determining the meaning of unfamiliar words.” Reading Research Quarterly 19: 188– 204. Cassar, M. and Treiman, R. 1997. “The beginnings of orthographic knowledge: Children’s knowledge of double letters in words.” Journal of Educational Psychology 89: 631–644. Catts, H., Fey, M. E., Zhang, X. and Tomblin, J. B. 1999. “Language basis of reading disabilities: Evidence from a longitudinal investigation.” Scientific Studies of Reading 3: 331–361. Chafe, W. 1979. “The flow of thought and the flow of language.” In Discourse and Syntax. Syntax and Semantics 12, T. Givón (ed.), 159–181. New York: Academic Press. Chafe, W. 1987. “Cognitive constraints on information flow.” In Coherence and Grounding in Discourse, R. S. Tomlin (ed.), 21–52. Amsterdam: John Benjamins. Chafe, W. 1994. Discourse. Consciousness, and Time: The Flow and Displacement of Conscious Experience in Speaking and Writing. Chicago, IL: University of Chicago Press. Chafe, W. and Danielewicz, J. 1987. “Properties of spoken and written language.” In Comprehending Oral and Written Language, R. Horowitz and S. J. Samuels (eds), 83– 113. San Diego CA: Academic Press. Cheshire, J. 1991. “Variation in the use of ain’t in an urban British English dialect.” In Dialects of English: Studies in Grammatical Variation, P. Chambers and J. K. Trudgill (eds), 54–73. London: Longman. Choi, S. and Bowerman, M. 1991. “Learning to express motion events in English and Korean: The influence of language specific lexicalization patterns.” Cognition 41: 82– 121. Choi, S., McDonough, L., Bowerman, M. and Mandler, J. M. 1999. “Early sensitivity to language-specific spatial categories in English and Korean.” Cognitive Development 14, 241–268.

 References

Chomsky, C. 1969. The Acquisition of Syntax in Children from 5 to 10. Cambridge, MA: MIT Press. Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge MA: M.I.T. Press. Chomsky, N. 1970. “Remarks on nominalization.” In Readings in English Transformational Grammar, R. Jacobs and P. Rosenbaum (eds), 184–221. Waltham, MA: Blaisdell. Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris. Chung, S. and Timberlake, A. 1985. “Tense, mood, and aspect.” In Language Typology and Syntactic Description, Volume 3: Grammatical Categories and the Lexicon, T. Shopen (ed.), 202–258. Cambridge: Cambridge University Press. Clancy, P. M. 1986. “The acquisition of communicative style in Japanese.” In Language Socialization Across Cultures, B. B. Schieffelin and E. Ochs (eds), 213–251. Cambridge: Cambridge University Press. Clancy, P. M. 2003. “The lexicon in interaction: Developmental origins of Preferred Argument Structure in Korean.” In Preferred Argument Structure: Grammar as Architecture for Function, J. Du Bois, L. E. Kumpf, and W. J. Ashby (eds), 81–109. Amsterdam: John Benjamins. Clark, E. V. 1993. The Lexicon in Acquisition. Cambridge: Cambridge University Press. Clark, E. V. 1995. “Later lexical development and word formation.” In The Handbook of Child Language, P. Fletcher and B. MacWhinney (eds), 393–412. Cambridge: Cambridge University Press. Clark, E. V. 2003. First Language Acquisition. Cambridge: Cambridge University Press. Clark, E. V. and Berman, R. A. 1984. “Structure and use in the acquisition of wordformation.” Language 60: 542–590. Clark, E. V. and Berman, R. A. 1987. “Types of linguistic knowledge – Interpreting and producing compound nouns.” Journal of Child Language 14 (3): 547–567. Clark, E. V. and Clark, H. H. 1979. “When nouns surface as verbs.” Language 55: 767–811. Clark, E. V. and Hecht, B. F. 1982. “Learning to coin agent and instrument nouns.” Cognition 12: 1–24. Clark, E. V. and Wong, A. 2002. “Pragmatic directions about language use: Words and word meanings.” Language in Society 31: 181–212. Clark, H. H. 1996. Using Language. Cambridge: Cambridge University Press. Clark, H. H. and Gerrig, R., J. 1990. “Quotations as demonstrations.” Language 66: 764–805. Cohen-Gross, D. 2000. “The syllabic structure of Modern Hebrew.” In The Raphael Nir Jubilee Volume: Studies in Communication, Linguistics, and Language Instruction, O. Rodrigue Schwarzwald, S. Blum-Kulka, and E. Olshtain (eds), 359–369. Jerusalem: Carmel. [in Hebrew] Comrie, B. 1978. “Ergativity.” In Syntactic Typology, W. Lehmann (ed.), 329–394. Austin: University of Texas. Comrie, B. 1986. “Tense and time reference: From meaning to interpretation in the chronological structure of a text.” Journal of Literary Semantics 15: 12–22. Comrie, B. and Thompson, S. A. 1985. “Lexical nominalization.” Language Typology and Syntactic Description, Vol. III, Grammatical Categories and the Lexicon, T. Shopen (ed.), 349–398. Cambridge: Cambridge University Press.

References 

Cooper, D., Roth, F., Speece, D. and Schatschneider, C. 2002. “The contribution of oral language skills to the development of phonological awareness.” Applied Psycholinguistics 23: 399–416. Corley, M. M. B. and Scheepers, C. 2002. “Syntactic priming in English sentence production: Categorical and latency evidence from an internet-based study.” Psychonomic Bulletin and Review 9 (1): 126–31. Corsaro, B. and Rizzo, T. 1990. “Disputes in the peer culture of American and Italian nursery-school children.” In Conflict Talk, A. Grimshaw (ed.), 21–67. Cambridge: Cambridge University Press. Corsaro, W. A. 1985. Friendship and Peer Culture in The Early Years. Norwood NJ: Ablex. Corsaro, W. A. 1997. The Sociology of Childhood. Thousand Oaks CA: Pine Forge Press. Coulmas, F. 1986. “Reported speech: Some issues.” In Direct and Indirect Speech, F. Coulmas (ed.), 1–28. Berlin: de Gruyter. Cousin, M.-P., Largy, P. and Fayol, M. 2002. “Sometimes, early learned instances hinder the implementation of agreement rules. A study in written French.” Current Psychology Letters 8: 51–65. Crain, S. and Lillo–Martin, D. 1999. An Introduction to Linguistic Theory and Language Acquisition. Oxford: Blackwell Crain, S. and McKee, C. 1985. “Acquisition of structural restictions on anaphora.” In Proceedings of the 16th North Eastern Linguistics Society Meeting, [NELS 16], S. Berman, J.-W. Choe, and J. McDonough (eds), 94–110. Creissels, D. 1995. Eléments de syntaxe générale. Paris: Presses Universitaires de France. Croft, W. 1991. Syntactic Categories and Grammatical Relations. Chicago, IL: University of Chicago Press. Cronbach, L. J. 1942. “An analysis of techniques for systematic vocabulary testing.” Journal of Educational Research 36: 206–217. Culler, J. 1976. Structuralist Poetics: Structuralism, Linguistics and the Study of Literature. New York: Cornell University Press. Cunningham, A. E. and Stanovich, K. E. 1997. “Early reading acquisition and its relation to reading experience and ability ten years later.” Developmental Psychology 33 (6): 934– 945. Curtis, M. E. 1987. “Vocabulary testing and vocabulary instruction.” In The nature of vocabulary acquisition, M. G. McKeown and M. E. Curtis (eds), 37–51. Hillsdale, NJ. Erlbaum. Dapretto, M. and Bjork, E. L. 2000. “The development of word retrieval abilities in the second year and its relation to early vocabulary growth.” Child Development 71: 633– 646. Dascal, M. 1987. “Defending literal meaning.” Cognitive Science 11: 259–281. de Jong, P. F., Seveke, M. J. and van Veen, M. 2000. “Phonological sensitivity and the acquisition of new words in children.” Journal of Experimental Child Psychology 76: 275–301. de Villiers, J. G. and de Villiers, P. A. 1985. “The acquisition of English passives.” In The Crosslinguistic Study of Language Acquisition, Vol. 1, D.I. Slobin (ed.), 96–103. Hillsdale, NJ: Lawrence Erlbaum.

 References

de Villiers, J. G. and de Villiers, P. A. 2000. “Linguistic determinism and the understanding of false beliefs.” In Children’s Reasoning and the Mind, P. Mitchell and K. Riggs (eds), 191–228. Hove: Psychology Press. Demke, T., Graham, S. and Siakaluk, P. 2002. “The influence of exposure to phonological neighbours on preschoolers’ novel word production.” Journal of Child Language 29 (2): 379–392. Demuth, K. 1989. “Maturation and the acquisition of the Sesotho passive.” Language 65: 56-80. Di Sciullo, A. M. and Williams, E. 1987. On the Definition of Word. Cambridge, MA: MIT Press. Dickinson, D. K, and Tabors, O. P. 2001. Beginning Literacy with Language. Baltimore, MD: Brooks Publishing. Dickinson, E. 1890/1960. “I dwell in possibility.” In The Complete Poems of Emily Dickinson, T. H. Johnson (ed.), 657. Boston MA: Little, Brown and Company. Dixon, M. and Kaminska, Z. 1997. “Is it misspelled or is it mispelled? The influence of fresh orthographic information on spelling.” Reading and Writing 9: 483–498. Dixon, R. M. W. 1987. “Studies in ergativity.” Lingua 7 (1/1–4): 271–196. Dixon, R. M. W. 1994. Ergativity. Cambridge: Cambridge University Press. Dockrell, J. E. and Campbell, R. N. 1986. “Lexical acquisition strategies.” In The Development of Word Meaning, S. Kuczaj and M. Barrett (eds), 121–154. New York: Springer Verlag. Dockrell, J. E. and Messer, D. This Volume. “Lexical acquisition in the early school years.” Dockrell, J. E., Messer, D. and George, R. 2001. “Patterns of naming objects and actions in children with word finding difficulties.” Language and Cognitive Processes 16: 261–286. Dockrell, J. E., Messer, D., George, R. and Ralli, A. 2003. “Beyond naming patterns in children with WFDs – definitions for nouns and verbs.” Journal of Neurolinguistics 16 (2–3): 191–211. Dollaghan, C. 1987. “Fast mapping of normal and language impaired children.” Journal of Speech and Hearing Disorders 52 (3): 218–222. Dorval, B. and Eckerman, C. O. 1984. “Developmental trends in the quality of conversation achieved by small groups of acquainted peers.” Monographs of the Society for Research in Child Development 49(2), Serial #206. Du Bois, J. W. 1985. “Competing motivations.” In Iconicity in Syntax, J. Haiman (ed.), 343– 365. Amsterdam: John Benjamins. Du Bois, J. W. 1987. “The discourse basis of ergativity.” Language 63(4): 805–853. Dumesnil, J. E., and Dorval B. 1989. “The development of talk-activity frames that foster perspective-focused talk among peers.” Discourse Processes 12: 193–225. Durie, M. 1988. “Preferred Argument Structure in an active language: Arguments against the category ‘Intransitive Subject’.” Lingua 7 (4/1): 1–25. Durkin, K., Crowther R., Shire, B., Riem, R. and Nash, P. 1985. “Polysemy in mathematical and musical education.” Applied Linguistics 6: 147–161. Dutra, R. 1987. “The hybrid S-category in Brazilian Portuguese.” Studies in Language 11: 163–180. Eckert, P. 1989. Jocks and Burnouts: Social Categories and Identity in High School. New York: Teachers College Press. Eco, U. 1990. The Limits of Interpretation. Indianapolis, IN: Indiana State University.

References 

Elley, W. 1989. “Vocabulary acquisition from listening to stories.” Reading Research Quarterly 24 (2): 174–187. Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Parisi, D. and Plunkett, K. 1996. Rethinking Innateness: A Connectionist Perspective on Development. Cambridge, MA: Bradford Books. England, N. and Martin, L. 2003. “Issues in comparative Argument Structure analysis in Mayan narratives.” In Preferred Argument Structure: Grammar as Architecture for Function, J. Du Bois, L. E. Kumpf and W. J. Ashby (eds), 81–109. Amsterdam: John Benjamins. Ericsson, K. A. and Charness, N. 1994. “Expert performance: Its structure and acquisition.” American Psychologist 49: 725–747. Erskine, J. 1999. “Group and individual differences in the local level writing strategies of adult dyslexics attending higher education.” In The Process of Writing: A Progress Report, S. Strömqvist and E. Ahlsén (eds). Göteborg University: Gothenburg Papers in Theoretical Linguistics 83: 103–118. Ervin-Tripp, S. 1979. “Children’s verbal turn-taking.” In Developmental Pragmatics, E. Ochs, and B. Schieffelin (eds), 391–414. New York: Academic Press. Ervin-Tripp, S. and Gordon, D. P. 1986. “The development of children’s requests.” In Communicative Competence: Assessment and Intervention, R. L. Schiefelbusch (ed.), 61–96. San Diego, CA: College Hill Press. Evans, J. 1996. “SLI subgroups: Interaction between discourse constraints and morphosyntactic deficits.” Journal of Speech and Hearing Research 39: 655–660. Eyer, J. and Leonard, L. 1995. “Functional categories and specific language impairment: A case study.” Language Acquisition 4: 177–203. Fayol, M., Hupet, M. and Largy, P. 1999. “The acquisition of subject-verb agreement in written French: From novices’ to experts’ errors.” Reading and Writing 11: 153–174. Fayol, M., Largy, P. and Lemaire, P. 1994. “When cognitive overload enhances subjectverb agreement errors: A study in French written language.” The Quarterly Journal of Experimental Psychology 47A: 437–464. Fayol, M., Thévenin, M. G., Jarousse, J. P., and Totereau, C. 1999. “From learning to teaching to learning French written morphology.” In Learning to Read: An Integrated View from Research and Practice, T. Nunes, (ed.), 43–64. Dordrecht: Kluwer. Ferguson, C. A. 1994. “Dialect, register, and genre: Working assumptions and conventionalization.” In Sociolinguistic Perspectives on Register, D. Biber and E. Finegan (eds.), 15–30. New York: Oxford University Press, Fernández-Lagunilla, M. and Anula-Rebollo, A. 1995. Sintaxis y Cognición. Madrid: Síntesis. Fiser, J. and Aslin, R. N. 2002. “Statistical learning of higher-order temporal structure from visual shape-sequences.” Journal of Experimental Psychology: Learning, Memory, and Cognition 28: 458–467. Fish, S. 1980. Is there a Text in this Class? London: Harvard University Press. Flax, J., Realpe-Bonilla, T., Hirsch, L., Brzutowicz, L., Bartlett, C. and Tallal, P. 2003. “Specific language impairment in families: Evidence for co-occurrence with reading impairments.” Journal of Speech, Language, and Hearing Research 46: 530–543.

 References

Fletcher, P. and Garman, M. 1986. “Introduction.” In Language Acquisition: Studies in First Language Development, P. Fletcher and M. Garman (eds.), 3–8. Cambridge: Cambridge University Press. Foley, W. A. and Van Valin, R. D. 1984. Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. François, D. 1974. Français parlé: Analyse des unités phoniques et significatives d’un corpus recueilli dans la région parisienne. Volumes I and II. Paris: Société d’etudes linguistiques et anthropologiques de France. Freyd, P. and Baron, J. 1982. “Individual differences in acquisition of derivational morphology.” Journal of Verbal Learning and Verbal Behavior 21: 282–295. Frye, N. 1978. Anatomy of Criticism. NJ: Princeton University Press. Frye, N. 1994. “The critical path.” In Contexts for Criticism, D. Keesey (ed.), 271–279. London: Mayfield. Fukkink, R. G. and de Glopper K. 1998. “Effects of instruction in deriving word meaning from context: A meta-analysis.” Review of Educational Research 68 (4): 450–468. Funnell, E., Hughes, D. and Woodcock, J. Submitted. “Age of acquisition for naming and knowing: A new hypothesis.” Quarterly Journal of Experimental Psychology. Gadet, F. 1990. “Les outils grammaticaux au risque de l’oral non standard.” Travaux de Linguistique 21: 13–24. Gadet, F. 1997. Le français ordinaire. Paris: Armand Colin. Garvey, C. 1974. “Some properties of social play.” Merrill-Palmer Quarterly 20: 163–180. Garvey, C. and Berninger, G. 1981. “Timing and turn-taking in children’s conversations.” Discourse Processes 4: 27–57. Gathercole, S. E. and Adams, A. 1994. “Children’s phonological working memory: Contributions of long-term knowledge and rehearsal.” Journal of Memory and Language 33: 672–688. Gathercole, S. E. and Baddeley, A. D. 1989. “Evaluation of the role of phonological STM in the development of vocabulary in children: A longitudinal study.” Journal of Memory and Language 28: 200–213. Gayraud, F. 1998. “Distribution de sujets nominaux dans différents types de textes à l’oral et à l’écrit: Une étude développementale.” Paper presented at the 5ème Rencontres de l’Atelier de Linguistes de Paris VII. Gayraud, F. 2000. Le développement de la differenciation oral/écrit vu a travers le lexique. Unpublished Thèse de Doctorat, Universite de Lyon. Gayraud, F., Jisa, H. and Viguié, A. 1999. “The development of syntactic packaging in French children’s written and spoken Texts.” In Developing Literacy across Genres, Modalities, and Language, Volume I, R. A. Aisenman (ed.), 169–81. Tel Aviv: Tel Aviv University International Literacy Project. Gayraud, F, Jisa, H. and Viguié, A. 2001. “Utilisation des outils cohésifs comme indice de sensibilité au registre: Une étude développementale. ” Acquisition et Interaction en Langue Etrangère 12: 3–24. Gee, J. P. 2001. “Literacy, discourse, and linguistics: Introduction and what is literacy?” In Literacy: A Critical Sourcebook, E. Cushman, E. R. Kintgen, B. M. Kroll and M. Rose (eds), 525–544. Boston MA: Bedford/ St. Martin’s.

References 

Geiger, V. 2000. Intervention program on promoting morphological awareness in 4th graders. MA thesis, Department of Communications Disorders, Tel Aviv University. [in Hebrew] Gentner, D. 1988. “Metaphor as structure mapping: The relational shift.” Child Development 59: 47–59. Gentner, D. 2003. “Why we’re so smart.” In Language in Mind: Advances in the Study of Language and Thought, D. Gentner and S. Goldin-Meadow (eds), 195–236. Cambridge, MA: MIT Press. German, D. J. 1984. “Diagnosis of word finding disorders in children with learning disabilities.” Journal of Learning Disabilities 17 (6): 353–359. Gillam, R. and Johnston, J. 1992. “Spoken and written language relationships in language/learning impaired and normally achieving school-age children.” Journal of Speech and Hearing Research 35: 1303–1315. Gillette, J., Gleitman, H., Gleitman, L. and Lederer, A. 1999. “Human simulations of vocabulary learning.” Cognition 73: 135–176. Gillis, S. and Ravid, D. 2001. “Language-specific effects on the development of written morphology.” In Morphology 2000, S. Bendjaballah, W. U. Dressler, O. Pfeiffer, and M. Voeikova (eds), 129–136. Amsterdam: John Benjamins. Ginsburg, H. P., and Opper, S. 1988. Piaget’s Theory of Intellectual Development. 3rd Edition. Englewood Cliffs NJ: Prentice Hall. Givón, T. 1975. “Focus and the scope of assertion: Some Bantu evidence.” Studies in African Linguistics (6): 185–205. Givón, T. 1983. “Introduction.” In Topic Continuity in Discourse: A Quantitative Cross Language Study, T. Givón (ed.), 1–41. Amsterdam: John Benjamins. Goel, V. 1991. “Notationality and the information processing mind.” Minds and Machines 1: 129–165. Goffman, E. 1981. “Footing.” In Forms of Talk, E. Goffman (ed.), 124–160. Philadelphia PA: University of Pennsylvania Press. Golinkoff, R. M., Mervis, C. V. and Hirsh-Pasek, K. 1994. “Early object labels: The case for a developmental lexical principles framework.” Journal of Child Language 21: 125–155. Goodwin, M. A. 1990. He-Said-She-Said: Talk as Social Organization among Black Children. Bloomington IN: Indiana University Press. Gopnik, A. and Meltzoff, A. N. 1997. Words, Thoughts, and Theories. Cambridge, MA: MIT Press. Gordon, P. and Chafetz, J. 1990. “Verb-based versus class-based accounts of actionality effects in children’s comprehension of passives.” Cognition 36: 227–254. Graves, M. 1986. “Vocabulary learning and instruction.” In Review of Research in Education, E. Z. Rothkopf (ed.), 13: 49–90. Washington DC: American Educational Research Association. Graves, M. 1987. “The roles of instruction in fostering vocabulary development.” In The Nature of Vocabulary Acquisition, M. G. McKeown and M. E. Curtis (eds), 165–184. Hillsdale, NJ: Lawrence Erlbaum. Green, J. 1988. “Spanish.” In The Romance Languages, M. Harris and N. Vincent (eds), 131– 170, London: Routledge.

 References

Grice, H. P. 1975. “Logic and conversation.” In Syntax and Semantics 3: Speech Acts, P. Cole and J. Morgan (eds), 41–58. New York: Academic Press. Guenthner, S. and Knoblauch, H. 1995. “Culturally patterned speaking practices – The analysis of communicative genres.” Pragmatics (5): 1–32. Gumperz, J. J. and Levinson, S. C. (eds) 1996. Rethinking Linguistic Relativity. Cambridge: Cambridge University Press. Gutierrez-Clellen, V. 1998. “Syntactic skills of Spanish-speaking children with low school achievement.” Language, Speech, and Hearing Services in Schools 29: 207–215. Habib, T., Huck-Taglicht, D., Kampf, Z. and Blum-Kulka, S. 2003 – July. “Conversational stories between children.” Paper presented at the Annual Conference of the Israel Society for Developing Literacy [SCRIPT], Zichron Yaakov, Israel. Halliday, M. A. K. 1967. “Notes on transitivity and theme in English.” Journal of Linguistics 3 (1): 37–87. Halliday, M. A. K. 1985. Spoken and Written Language. Oxford: Oxford University Press. Halliday, M. A. K. 1987. “Spoken and written modes of meaning.” In Comprehending Oral and Written Language, R. Horowitz and S. J. Samuels (eds), 55–82. New York: Academic Press. Halliday, M. A. K. 1988. “On the language of physical science.” In Registers of Written English: Structural Factors and Linguistic Features, M. Ghadessy (ed.), 162–178. London: Pinter. Halmari, H. and Smith, W. 1994. “Code-switching and register shift: Evidence from FinnishEnglish child bilingual conversation.” Journal of Pragmatics 21: 427–445. Hamo, M., Blum-Kulka, S. and Hacohen, G. 2004. “From observation to transcription: Theory, practice and interpretation in the analysis of children’s naturally occurring discourse.” Research on Language and Social Interaction 37(1): 71–92. Handel, G. 1988. Childhood Socialization. New York: Aldine de Gruyter. Harris, P. L. 2000. The Work of the Imagination. Malden, MA: Blackwell. Harris, M. and Hatano, G. 1999. “Introduction: A cross-linguistic perspective on learning to read and write.” In Learning to Read and Write: A Cross-linguistic Perspective, M. Harris and G. Hatano (eds), 1–9. Cambridge: Cambridge University Press. Hart, B. and Risley, T. 1992. “American parenting of language-learning children: Persisting differences in family-child interactions observed in natural home environments.” Developmental Psychology 28: 1096–1105. Hart, B. and Risley, T. 1995. Meaningful differences in everyday experience of young American children. Baltimore, MD: Brookes. Hastings, P. and Lytinen, S. 1994. “The ups and downs of lexical acquisition.” Proceedings of the 12th National Conference on Artificial Intelligence, 754–759. Menlo Park, CA: AAAI Press. Hawkins, J. 1983. Word Order Universals. New York: Academic Press. Heath, S. B. 1983. Ways with Words: Language, Life and Work in Communities and Classrooms. Cambridge: Cambridge University Press. Heibeck, T. H. and Markman, E. M. 1987. “Word learning in children: An examination of fast mapping.” Child Development 58: 1021–1034. Hemphill, L. 1989. “Topic development, syntax and social class.” Discourse Processes 12: 267– 286.

References 

Hernández Pina, F. 1984. Teorías Psicolingüísticas y su Aplicación a la Adquisición del Español como Lengua Materna. Madrid: Siglo XXI. Hickmann, M. 1993. “The boundaries of reported speech in narrative discourse: Some developmental aspects.” In Reflexive Language, J. A. Lucy (ed.), 63–88. Cambridge: Cambridge University Press. Hickmann, M. 2003. Children’s Discourse: Person, Space and Time across Languages. Cambridge: Cambridge University Press. Hickmann, M., Hendriks, H., Roland, F. and Liang, J. 1996. “The marking of new information in children’s narratives: A comparison of English, French, German and Mandarin Chinese.” Journal of Child Language 23: 591–619. Hicks, D. 1991.”Kinds of narrative: Genre skills among first graders from two communities.” In Developing Narrative Structure, A. McCabe and C. Peterson (eds), 55–88. Hillsdale NJ: Lawrence Erlbaum. Hidalgo, R. 1994. “The pragmatics of de-transitive voice in Spanish: From passive to inverse.” In Voice and Inversion, T. Givón (ed.), 168–185. Amsterdam: John Benjamins. Hoff, E. and Naigles, L. 2002. “How children use input to acquire a lexicon.” Child Development 73 (2): 418–433. Holes, C. 1995. Modern Arabic: Structures, Functions and Varieties. London: Longman. Holt, E. 1996. “Reporting on talk: The use of direct reported speech in conversation.” Research on Language and Social Interaction 29: 219–245. Hopper, P. and Thompson, S. A. 1980. “Transitivity in grammar and discourse.” Language 56: 251–299. Hoyle, M. S. 1998. “Register and footing in role play.” In Kids Talk, M. S. Hoyle. and T. C. Adger (eds), 47–68. Oxford: Oxford University Press. Hoyle, M. S. and Adger, T. C. (eds) 1998a. Kids Talk. Oxford: Oxford University Press. Hoyle, M. S. and Adger T. C. 1998b. “Introduction.” In Kids Talk, M. S. Hoyle. and T. C. Adger (eds), 3–23. Oxford: Oxford University Press. Hudson, J. A. and Shapiro, L. R. 1991. “From knowing to telling: Children’s scripts, stories, and personal narratives.” In Developing Narrative Structure, A. McCabe and C. Peterson, (eds), 89–136. Hillsdale, NJ: Lawrence Erlbaum. Hunt, K. 1965. “Grammatical structures written at three grade levels.” Research Report No. 3. Champaign IL: National Council of Teachers of English. Hupet, M., Fayol, M. and Schelstraete, M. A. 1998. “Effects of semantic variables on the subject-verb agreement processes in writing.” British Journal of Psychology 89: 59–75. Huttenlocher, J., Vasilyeva, M., Cymerman, E. and Levine, S. 2002. “Language input and child syntax.” Cognitive Psychology 45: 337–374. Indefrey, P, Brown, C., Hellwig, F., Amunts, K., Herzog, H., Rüdiger, S. and Hagoort, P. 2001. “A neural correlate of syntactic encoding during speech production.” PNAS 98: 5933–5936. Inhelder, B. and Piaget, J. 1958. The Growth of Logical Thinking from Childhood to Adolescence. New York: Basic Books. Iser, W. 1994. “Readers and the concept of the implied reader.” In Contexts for Criticism, D. Keesey (ed.), 137–144. London: Mayfield. Jackendoff, R. 2002. Foundations of Language. New York: Oxford University Press. Jakobson, R. 1987. Language in Literature. Cambridge, MA: Harvard University Press.

 References

James, W. 1890. The Principles of Psychology Volume 1. New York: Henry Holt. Jeanjean, C. 1980. “L’organisation des formes sujet en français de conversation: Etude quantitative et grammaticale de deux corpus.” Recherches sur le français parlé 3: 99–134. Jenkins, J. R., Stein, M. L. and Wysocki, K. 1984. “Learning vocabulary through reading.” American Educational Research Journal 21 (4): 767–787. Jerger, S., Martin, R. C. and Damian, M. F. 2002. “Semantic and phonological influences on picture naming by children and teenagers.” Journal of Memory and Language 47 (2): 229–249. Jisa, H. 2000. “Development of subject ellipsis in French narrative discourse: Reference in syntax.” Linguistics 38: 591–620. Jisa, H. This Volume. “Growing into academic French.” Jisa, H., Reilly, J. S., Verhoeven, L., Baruch, E. and Rosado, E. 2002. “Cross-linguistic perspectives on the use of passive constructions in written texts.” Journal of Written Language and Literacy 5: 163–181. Jisa, H. and Viguié, A. In Press. “A developmental perspective on the role of on in written and spoken expository texts.” Journal of Pragmatics, Special issue on Discourse Stance. Johansson, V. 2000. “Developing editing.” In Developing Literacy across Genres, Modalities, and Languages. Working Papers, Volume III. M. Aparici, J. Perera, E. Rosado., and L. Tolchinsky (eds), 235–249. University of Barcelona: International Project on Developing Literacy. Johnson, C. J. and Anglin, J. M. 1995. “Qualitative developments in the content and form of children’s definitions.” Journal of Speech and Hearing Research 38: 612–629. Johnson, M. H. and Morton, J. 1991. Biology and Cognitive Development: The Case of Face Recognition. Oxford: Blackwell. Johnston, J. 1985. “Cognitive prerequisites: The evidence from children’s learning of English.” In The Crosslinguistic Study of Language Acquisition, Volume 2, D. I. Slobin (ed.), 961–1004, Hillsdale, NJ: Lawrence Erlbaum. Jones, M. A. 1996. Foundations of French Syntax, Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press. Jucker, A. H. and Ziv, Y. (eds). 1998. Discourse Markers: Descriptions and Theory. Amsterdam: John Benjamins. Jusczyk, P. W. 1997. The Discovery of Spoken Language. Cambridge MA: Bradford Books. Kail, M. and Sánchez-López, I. 1997. “Referent introductions in Spanish narratives as a function of contextual constraints: A crosslinguistic perspective.” First Language 17: 103–130. Kail, R. and Leonard, L. B. 1986. “Word-finding abilities in language-impaired children.” ASHA Monographs: 25. Kameenui, E. J., Dixon, D. W. and Carnine, D. 1987. “Issues in the design of vocabulary instruction.” In The Nature of Vocabulary Acquisition, M. G. McKeown and M. E. Curtis (eds), 129–145. Hillsdale, NJ: Lawrence Erlbaum. Karmiloff, K. and Karmiloff-Smith, A. 2001. Pathways to Language: From Fetus to Adolescent. Cambridge: Cambridge University Press. Karmiloff-Smith, A. 1979. A Functional Approach to Child Language: A Study of Determiners and Reference. Cambridge: Cambridge University Press.

References 

Karmiloff-Smith, A. 1986a. “Stage/structure versus phase/process in modelling linguistic and cognitive development.” In Stage and Structure: Reopening the Debate, I. Levin (ed.), 164–190. Norwood, NJ: Ablex. Karmiloff-Smith, A. 1986b. “Some fundamental aspects of language development after age 5.” In Language Acquisition: Studies in First Language Development, 2nd ed. P. Fletcher and M. Garman (eds), 455–474. Cambridge: Cambridge University Press. Karmiloff-Smith, A. 1992. Beyond Modularity: A Developmental Perspective on Cognitive Science. Cambridge, MA: MIT Press/Bradford Books. Katriel, T. 1991. Communal Webs: Communication and Culture in Contemporary Israel. Albany, NY: State University of New York Press. Kaye, K. and Charney, R. 1981. “Conversational asymmetry between mothers and children.” Journal of Child Language 8: 35–50. Keenan, E. 1975. “Logical expressive power and syntactic variation in natural languages.” In Formal Semantics of Natural Languages, E. Keenan (ed.), 406–421. Cambridge: Cambridge University Press. Keenan, E. 1976. “Towards a universal definition of subject.” In Subject and Topic, C. I. Lee (ed.), 303–333. New York: Academic Press. Keenan, E. 1985. “Passives in the world’s languages.” In Language Typology and Syntactic Description, Volume I: Clause Structure, T. Shopen (ed.), 243–281. Cambridge: Cambridge University Press. Keil, F. C. 1983. “Semantic inferences and the acquisition of word meaning.” In Concept Development and the Development of Word Meaning, T. B. Seiler and W. Wannemacher (eds), 103–124. New York: Springer. Keil, F. C. and Batterman, N. 1984. “A characteristic-to-defining shift in the acquisition of word meaning.” Journal of Verbal Learning and Verbal Behaviour 23: 221–236. Kelly, D. J. 1997. “Patterns of verb use by pre-schoolers with normal language and specific language impairment.” Applied Psycholinguistics 18 (2): 199–218. Kemp, N. and Bryant, P. 2003. “Do beez buzz? Rule-based and frequency-based knowledge in learning to spell plural -s.” Child Development 74: 63–74. Khorounjaia, E. and Tolchinsky, L. This Volume. “Discourse constraints on the lexical realization of arguments in Spanish.” King, G. and Fletcher, P. 1993. “Grammatical problems in school-age children with specific language impairment.” Clinical Linguistics & Phonetics 7(4): 339–352. Kluwe, R. and Logan, G. D. (eds) 2000. “Executive control” [Special Issue]. Psychological Research 63 (4). Koenig, J.-P. 1999. “On a tué le président! The nature of passives and ultra-indefinites.” In Cognition and Function in Language, B. Fox, D. Jurafsky and L. A. Michaelis (eds.), 235–251. Stanford: CSLI Publications. Kothoff, H. 1998. “Irony, quotation, and other forms of staged intertextuality.” Inlist 5, http://www.ling.uni-konstanz.de/home/couperku/inlist-index.html. Koven, M. 2002. “An analysis of speaker role inhabitance in narratives of personal experience.” Journal of Pragmatics 34: 167–217. Kuhn, D. 1999. “A developmental model of critical thinking.” Educational Researcher 28(2): 16–25.

 References

Kuhn, D. 2000. “Theory of mind, metacognition and reasoning: A lifelong perspective.” In Children’s Reasoning and the Mind, P. Mitchell and K. J. Riggs (eds), 301–326. Hove: Psychology Press. Kuhn, D. 2001. “How do people know?” Psychological Science 12(1): 1–8. Kumpf, L. E. 1992. “Preferred Argument Structure in second language discourse: A preliminary study.” Studies in Language 16: 369–403. Kumpf, L. E. 2003. “Genre and Preferred Argument Structure: Sources of Argument Structure in classroom discourse.” In Preferred Argument Structure: Grammar as Architecture for Function, J. Du Bois, L. E. Kumpf, and W. J. Ashby (eds), 109–131. Amsterdam: John Benjamins. Kuno, S. 1972. “Functional sentence perspective: A case study from Japanese and English.” Linguistic Inquiry 3: 269–320. Kyratzis, A. 1999. “Narrative identity: Preschoolers’ self-construction through narratives in same-sex friendship group pragmatic play.” Narrative Inquiry 9: 427–457. Laberge, S. 1978. “The changing distribution of indeterminate pronouns in discourse.” In Language Use and the Use of Language, R. W. Shuy and W. Shnukal (eds), 147–162. Washington, DC: Georgetown University Press. Laberge, S. and Sankoff, G. 1980. “Anything you can do.” In The Social Life of Language, G. Sankoff (ed.), 271–93. Philadelphia, PA: University of Pennsylvania Press. Labov, W. 1972. Language in the Inner City: Studies in the Black English Vernacular. Philadelphia, PA: University of Pennsylvania Press. Labov, W. and Waletzky, J. 1967. “Narrative analysis.” In Essays on the Verbal and Visual Arts: Proceedings of the 1966 Meeting of the American Ethnological Society, J. Helm (ed.), 12–44. Seattle: American Ethnological Society. Lambrecht, K. 1981. Topic, Antitopic and Verb Agreement in Non-Standard French. Amsterdam: John Benjamins. Lambrecht, K. 1987. “On the status of SVO sentences in French discourse.” In Coherence and Grounding in Discourse, R. S. Tomlin (ed.), 217–261. Amsterdam: John Benjamins. Lambrecht, K. 1994. Information Structure and Sentence Form: Topic, Focus and Mental Representations of Discourse Referents. Cambridge: Cambridge University Press. Langer, J. A. 1995. Envisioning Literature: Literary Understanding and Literature Instruction. New York: Teachers College Press. Lapidot, S. 2003. Spoken and written morphological abilities in gradeschoolers with SLI. MA thesis, School of Education, Tel Aviv University. [in Hebrew] Largy, P. 2001. “La révision des accords nominal et verbal chez l’enfant.” L’Année Psychologique 101: 221–245. Largy, P., Fayol, M. and Lemaire, P. 1996. “On confounding verb/noun inflections. A study of subject-verb agreement errors in French.” Language and Cognitive Processes 11: 217– 255. Lee, E., Torrance, N. G. and Olson, D. R. 2001. “Young children and the say/mean distinction: Verbatim and paraphrase recognition in narrative and nursery rhyme contexts.” Journal of Child Language 28: 531–543. Leech, G. 1983. Principles of Pragmatics. London: Longman.

References 

Lennox, C. and Siegel, L. S. 1994. “The role of phonological and orthographic processes in learning to spell.” In Handbook of Spelling: Theory, Process and Intervention, G. D. A Brown and N. Ellis (eds), 93–109. Toronto: John Wiley. Leonard, L. 1998. Children with Specific Language Impairment. Cambridge, MA: MIT Press. Leonard, L., Deevy, P., Miller, C., Rauf, L., Charest, M. and Kurtz, R. 2003. “Surface forms and grammatical functions: Past tense and passive participle used by children with specific language impairment.” Journal of Speech, Language, and Hearing Research 46: 43–55. Leung, C. and Pikulski, J. J. 1990. “Incidental word learning of kindergarten and first grade children through repeated read aloud events.” In Literacy, theory and research: Analyses from multiple paradigms, J. Zutell and S. McCormick (eds). Chicago: National Reading Conference: 231–239. Levelt, W. J. M. 1989. Speaking: From Intention to Articulation. Boston, MA: MIT Press. Levie, R. 2002. Adjective formation in 1st and 6th grade children. MA thesis, the School of Education, Tel Aviv University. [in Hebrew] Levin, I., Ravid, D. and Rapaport, S. 2001. “Morphology and spelling among Hebrewspeaking children: From kindergarten to first grade.” Journal of Child Language 28: 741–769. Levorato, M. C. and Cacciari, C. 2002. “The creation of new figurative expressions: Psycholinguistics evidence in Italian children, adolescents and adults.” Journal of Child Language 20: 127–150. Levy, E. and Nelson, K. 1994. “Words in discourse; A dialectical approach to the acquisition of meaning and use.” Journal of Child Language 21: 367–390. Light, P. and Littleton, K. 1999. Social Processes in Children’s Learning. Cambridge: Cambridge University Press. Lindenmayer, D. B., Foster, D. R., Franklin, J. F., Hunter, M. L., Noss, R. F., Schmiegelow, F. A. and Perry D. 2004. “Salvage Harvesting Policies after Natural Disturbance”. Science 303, No. 5662: 1303. Linell, P. 1998. Approaching Dialogue: Talk, Interaction and Contexts in Dialogical Perspective. Amsterdam: John Benjamins. Loban, W. 1976. Language Development: Kindergarten through Grade Twelve. Champaign IL: National Council of Teachers of English, Research Report No. 18. Longacre, R. E. 1996. The Grammar of Discourse. New York: Plenun Press. López-Ornat, S. 1999. “La adquisición del lenguaje. Nuevas perspectivas.” In Estudios Psicolingüísticos en Lengua Castellana, F. Cuetos and M. de Vega (ed), 469–533. Madrid: Trotta. López-Ornat, S., Fernandez, A., Gallo, P., and Mariscal, S. 1994. La Adquisición de la Lengua Epañola. Madrid: Siglo XXI. Lucy, J. A. 1993. “Reflexive language and human discipline.” In Reflexive Language, J. A. Lucy (ed.), 9–32. Cambridge: Cambridge University Press. Lyons, C. 1995. “Voice, aspect and arbitrary arguments.” In Linguistic Theory and the Romance Languages, J. C. Smith and M. Maiden (eds), 77–114. Amsterdam: John Benjamins. Lyons, J. 1977. Semantics, Volumes 1 and 2. Cambridge: Cambridge University Press.

 References

MacWhinney, B. 1985. “Hungarian language acquisition as an exemplification of a general model of grammatical development.” In The Crosslinguistic Study of Language Acquisition, Volume 2: Theoretical Issues, D. I. Slobin (ed.), 1069–1156. Hillsdale, NJ: Lawrence Erlbaum. MacWhinney, B. (ed.) 1987. Mechanisms of Language Acquisition. Hillsdale, NJ: Lawrence Erlbaum. Mahony, D., Singson, M. and Mann, V. 2000. “Reading ability and sensitivity to morphological relations.” Reading and Writing 12: 191–218. Marchman, V., Wulfeck, B. and Weismer, S. 1999. “Morphological productivity in children with normal language and SLI: A study of the English past tense.” Journal of Speech, Language, and Hearing Research 42: 206–219. Markman, E. M. 1989. Categorization and Naming in Children: Problems of Induction. Cambridge, MA: MIT Press. Markman, E. M. and Hutchinson, J. E. 1984. “Children’s sensitivity to constraints on word meaning: Taxonomic versus thematic relations.” Cognitive Psychology 16: 1–27. Martin, J. R. and Plum, G. A. 1997. “Construing experience: Some story genres.” Journal of Narrative and Life History 7: 299–308. Matthews, R. C., Buss, R. R., Stanley, W. B., Blanchard-Fields, F. Cho, J.-R. and Druhan, B. 1989. “Role of implicit and explicit processes in learning from examples: A synergistic effect.” Journal of Experimental Psychology: Learning, Memory, and Cognition 15: 1083– 1100. McCabe, A. and Peterson, C. (eds). 1991. Developing Narrative Structure. Hillsdale, NJ: Lawrence Erlbaum. McCartney, K., Scarr, S., Phillips, D. and Grajek, S. 1985. “Day care as intervention: Comparisons of varying quality programs.” Journal of Applied Developmental Psychology 6 (2–3): 247–260. McCathren, R. B., Yoder, P. J. and Warren, S. F. 1995. “The role of directives in early language intervention.” Journal of Early Intervention 19: 91–101. McGregor, K. K. and Waxman, S. R. 1998. “Object naming at multiple hierarchical levels: A comparison of preschoolers with and without word-finding deficits.” Journal of Child Language 25: 419–430. McKeown, M. G., Beck, I. L., Omanson, R. C. and Pople, M. T. 1985. “Some effects of the nature and frequency of vocabulary instruction on the knowledge and use of words.” Reading Research Quarterly 20 (5): 522–535. McNeill, D. 1970. The Acquisition of Language: The Study of Developmental Psycholinguistics. New York: Harper and Row. McTear, M. 1985. Children’s Conversation. Oxford: Blackwell. Messer, D., Dockrell, J. E. and Murphy, N. In Press. “The relationship between naming and literacy in children with word-finding difficulties.” Journal of Educational Psychology. Metsala, J. L. 1999. “Young children’s phonological awareness and nonword repetition as a function of vocabulary development.” Journal of Educational Psychology 91: 3–19. Miall, D. and Kuiken, D. 1999. “What is literariness? Three components of literary reading.” Discourse Processes 28: 121–138. Michas, I. and Henry, L. A. 1994. “The link between phonological memory and vocabulary acquisition.” British Journal of Developmental Psychology 12: 147–163.

References 

Miller, C. R. 1984. “Genre as social action” Quarterly Journal of Speech 70: 151–167. Miller, J. 2002. “Questions about constructions.” Journal of Child Language 29: 470–474. Miller, G. A. and Fellbaum, C. 1991. “Semantic Networks of English.” Cognition 41(1–3): 197–229. Miller, P., Potts, R., Fung, H., Hoogstra, L. and Mintz, J. 1990. “Narrative practices and the social construction of self in childhood.” American Ethnologist 17: 292–309. Moats, L. 1995. Spelling Development, Disability, and Instruction. Baltimore: York Press. Morris, N. and Crump, D. 1982. “Syntactic and vocabulary development in the written language of learning disabled and non-learning disabled students at four age levels.” Learning Disability Quarterly 5: 163–172. Nagy, W. E. and Anderson, R. C. 1984. “How many words are there in printed school English?” Reading Research Quarterly 20: 304–330. Nagy, W. E., Anderson, R. C., Scommer, M. Scott, J. A. and Stellmen, A. C. 1989. “Morphology and the mental lexicon.” Reading Research Quarterly 24: 262–283. Nagy, W. E., Diakidoy, I.-A. and Anderson, R. C. 1993. “The acquisition of morphology: learning the contribution of suffixes to the meanings of derivatives.” Journal of Reading Behavior 25: 155–170. Nagy, W. E. and Herman, P. A. 1987. “Breadth and depth of vocabulary knowledge: Implications for acquisition and instruction.” In The Nature of Vocabulary Acquisition, M. G. McKeown and M. E. Curtis (eds), 19–35: Hillsdale NJ: Lawrence Erlbaum. Nagy, W. E., Herman, P. A. and Anderson, R. C. 1985. “Learning words from context.” Reading Research Quarterly 20: 233–253. Naharro, M. A. 1996. “La adquisición del subjuntivo español en lengua materna.” In Estudios sobre la Adquisición del Castellano, Catalán y Eusquera y Gallego, M. Pérez Pereira (ed.), 117–130. Santiago de Compostela: Publicaciones de la Universidad de Santiago. Nation, K. and Hulme, C. 1996. “The automatic activation of sound-letter knowledge: An alternative interpretation of analogy and priming effects in early spelling development.” Journal of Experimental Child Psychology 63: 416–435. Nation, K. and Snowling, M. J. 1998. “Individual differences in contextual facilitation: Evidence from dyslexia and poor reading comprehension.” Child Development 69: 996– 1009. Nation, K. and Snowling, M. J. 1999. “Developmental differences in sensitivity to semantic relations among good and poor comprehenders: Evidence from semantic priming.” Cognition 70: B1–13. Nation, K., Marshall, C. M. and Snowling, M. J. 2001. “Phonological and semantic contributions to children’s picture-naming skill: Evidence from children with developmental disorders.” Language and Cognitive Processes 16: 241–259. Nelson, K. 1988. “Constraints on word learning?” Cognitive Development 3: 221–246. Nelson, K. 1996. Language in Cognitive Development: The Emergence of the Mediated Mind. Cambridge: Cambridge University Press. Newkirk, T. 1991. “The high school years.” In Handbook of Research on Teaching the English Language Arts, J. Flood, J. M. Jensen, D. Lapp, and J. R. Squire (eds), 331–342. New York: Macmillan. Nicholson, G. 1984. Seeing and Reading. Atlantic Highlands, NJ: Humanities Press.

 References

Nicolopoulou, A. 1997. “Children and narratives: Toward an interpretive and sociocultural approach.” In Narrative Development: Six Approaches, M. Bamberg (ed.), 179–213. Hillsdale NJ: Lawrence Erlbaum. Nicolopoulou, A. and. Richner, E. S. 2004. “When your powers combine: I am Captain Planet: The developmental significance of individually and collaboratively composed stories.” In Discourse Studies: Thematic issue on Peer Talk and Peer Development 6(3): 347–373. Ninio, A. and Snow, C. 1996. Pragmatic Development. Boulder, CO: Westview. Ninio A., Snow, C. E., Pan, B. A. and Rollins, P. 1994. “Classifying communicative acts in children’s interaction.” Journal of Communication Disorders 27: 158–187. Nippold, M. A. 1988. Later Language Development. Boston, MA: College-Hill. Nippold, M. A. 1991. “Evaluating and enhancing idiom comprehension in language disordered childen.” Speech and Hearing Services in School 22: 100–106. Nippold, M. A. 1998. Later Language Development: The School-Age and Adolescent Years (2nd ed.). Austin, TX: Pro-Ed. Nippold, M. A. 2003. “Expository discourse in school-age children, adolescents, and adults: A developmental investigation.” Paper presented at conference on Le langage oral de l’enfant scholarise: Acquisition, enseignement, remediation. Grenoble : Universite Stendhal. Nippold, M. A., Cuyler, J. S. and Braunbeck-Price, R. 1988. “Explanation of ambiguous advertisements: a developmental study with children and adolescents.” Journal of Speech and Hearing Research 31 (3): 466–474. Nippold, M. A., Hegel, S. L., Sohlberg, M. M. and Schwarz, I. E. 1999. “Defining abstract entities: Development in pre-adolescents, adolescents, and young adults.” Journal of Speech, Language, and Hearing Research 42: 473–481. Nippold, M. A., Ward-Lonergan, J. and Fanning, J. 2003 – July. “Persuasive writing in school-age children and adolescents: Syntactic, semantic, and pragmatic attainments.” Paper presented at the Child Language Seminar, Newcastle-upon-Tyne. Norbury, C., Bishop, D. and Briscoe, J. 2001. “Production of English finite verb morphology: A comparison of SLI and mild-moderate hearing impairment.” Journal of Speech, Language, and Hearing Research 44: 165–178. Nordqvist, Å. 2001. Speech about Speech: A Developmental Study on Form and Function of Direct and Indirect Speech. Doctoral dissertation. Gothenburg Monographs in Linguistics, 19. Nordqvist, Å, Leiwo, M. & Lyytinen, H. In Press. “From nonsense words to space adventure: Developing a test battery to study and diagnose writing.” Proceedings of the 13th European Conference on Reading. Tallin: Tallin Pedagogical University. Nunes, T. and Bryant, P. 2003. Handbook of Children’s Literacy. Dordrecht: Kluwer. Nunes, T., Bryant, P., and Bindman, M. 1997. “Morphological spelling strategies: Developmental stages and processes.” Developmental Psychology 33: 637–649. Ochs, E. 1979. “Planned and unplanned discourse.” In Syntax and Semantics, Vol. 12: Discourse and Syntax, T. Givón (ed.), 51–80. New York: Academic Press. Ochs, E. 1990. “Cultural universals: The acquisition of language.” Stanford University Papers and Reports on Child Language Development 29: 1–19. Ochs, E. and Capps, L. 2001. Living Narrative. Cambridge, MA: Harvard University Press.

References 

Ochs, E. and Taylor, C. 1992. “Family narrative as political activity.” Discourse and Society 3: 301–341. O’Dowd, E. M. 1990. “Discourse pressure, genre, and grammatical alignment – after Du Bois.” Studies in Language 14/2: 365–403. Olson, D. R. 1977. “From utterance to text: The bias of language in speech and writing.” Harvard Educational Review 47: 257–281. Olson, D. R. 1994. The World on Paper: The Conceptual and Cognitive Implications of Writing and Reading. Cambridge: Cambridge University Press. Olson, D. R. 1996. “Literate mentalities: Literacy consciousness of language, and modes of thought.” In Modes of Thought, D. R. Olson (ed.), 141–151. Cambridge: Cambridge University Press. Olson, D. R. 2003. “The cognitive consequences of literacy.” In Handbook of Children’s Literacy, T. Nunes. and P. Bryant (eds), 539–555. Dordrecht: Kluwer. Olson, D. R. and Astington, J. W. 1993. “Thinking about thinking: Learning how to take statements and hold beliefs.” Educational Psychologist 28 (1): 7–23. Pacton, S. 2001. “Children’s linguistic awareness and spelling: The case of derivational morphology.” ESCOP/BPS Conference posters, Annual Congress of the European Society for Cognitive Psychology, Edinburgh, Scotland Pacton, S. Submitted. “The role of homophony, frequency and syntactic position on agreement errors: Evidence from 12 year-old French children’s spellings.” Pacton, S. and Fayol, M. 1998. “La transcription de graphonèmes complexes en français. L’exemple de /o/.”Repères 18: 163–171. Pacton, S. and Fayol, M. 2003. “How do French children use morphosyntactic information when they spell adverbs and present participles?” Scientific Studies of Reading 7: 273– 287. Pacton, S and Fayol, M. This Volume. “Learning to spell in a deep orthography: The case of French.” Pacton, S., Fayol, M. and Perruchet, P. 2002. “The acquisition of untaught orthographic regularities in French.” In Precursors of Functional Lteracy, L. Verhoeven, C. Erlbro, d P. Reitsma (eds), 121–136. Dordrecht: Kluwer. Pacton, S., Fayol, M., and Perruchet, P. In Press. “Children’s implicit learning of graphotactic and morphological regularities in French.” Child Development. Pacton, S., Perruchet, P., Fayol, M. and Cleeremans, A. 2001. “Implicit learning out of the lab: The case of orthographic regularities.” Journal of Eperimental Psychology: General 30: 401–426. Paltridge, B. 1997. Genre, Frames and Writing in Research Settings. Amsterdam: John Benjamins. Parnas, D. L. and Clements, P. C. 1986. “A rational design process: How and why to fake it.” IEEE Transactions on Software Engineering 12: 251–257. Pascual-Leone, J. 1987. “Organismic processes for neo-Piagetian theories: A dialectical account of cognitive development.” International Journal of Psychology 22(5–6): 531– 570. Pearson, B. Z., Fernandez, S. C., Lewedeg, V. and Oller, D. K. 1997. “The relation of input factors to lexical learning by bilingual infants.” Applied Psycholinguistics 18 (1): 41–58.

 References

Pellegrini, A. D. and Galda, L. 1990. “The joint construction of stories by preschool children and an experimenter.” In Narrative Thought and Narrative Language, K. B. Britton and D. A. Pellegrini (eds), 113–131. Hillsdale, NJ: Lawrence Erlbaum. Pellegrini, A. D. and Galda, L. 1998. The Development of School-Based Literacy. London: Routledge. Pelli, T. and Sinimäki, S. 2004. ScriptLogtest för dubbelteckningsförmåga hos 9-åringar [ScriptLog-based testing of 9-year-olds’ ability to spell long consonants], Technical Report. Lund University: Department of Speech Therapy. Penno, J. F., Wilkinson, I. A. G. and Moore, D. W. 2002. “Vocabulary acquisition from teacher explanation and repeated listening to stories: Do they overcome the Matthew effect?” Journal of Educational Psychology 94 (1): 23–33. Perdue, C. and Bowerman, M. 1990. “The structure of the simple clause in language acquisition: Introduction.” Linguistics 28: 1131–1133. Perera, K. 1984. Children’s Writing and Reading: Analyzing Classroom Reading. London: Blackwell. Perera, K. 1986. “Grammatical differentiation between speech and writing in children aged 8 to 12.” In The Writing of Writing, A. Wilkinson (ed.) 90–108. London: The Falmer Press. Pérez-Leroux, A. T. 1998. “The acquisition of mood selection in relative clauses.” Journal of Child Language 25: 585–604. Perfetti, C. A., Rieben, L. and Fayol, M. 1997. Learning to Spell. Hillsdale, NJ: Lawrence Erlbaum. Perner, J. 1991. Understanding the Representational Mind. Cambridge, MA: MIT Press. Perruchet, P. and Pacton, S. 2004. “Qu’apportent à la pédagogie les travaux de laboratoire sur l’apprentissage implicite”? L’Année psychologique 14: 121–146. Perruchet, P. and Vinter, A. 2002a. “The self-organizing consciousness” (Target paper). Behavioral and Brain Sciences 25: 297–330. Perruchet, P. and Vinter, A. 2002b. “The self-organising consciousness: A framework for implicit learning.” In Implicit Learning and Consciousness: An Empirical, Philosophical, and Computational Consensus in the Making, R. M. French and A. Cleeremans (eds), 41–67 Hove: Psychology Press. Peskin, J. 1992. “Ruse and representations: On children’s ability to conceal information.” Developmental Psychology 28(1): 84–89. Peskin, J. 1996. “Guise and guile: Children’s understanding of narratives in which the purpose of pretense is deception.” Child Development 67: 1735–1751. Peskin, J. 1998. “Constructing meaning when reading poetry: An expert-novice study.” Cognition and Instruction 16: 235–263. Peskin, J. 2002. ”This one looks like it’s provoking thought: Higher-order thinking skills when high school students read poetry versus prose.” Paper presented at the annual meeting of the American Educational Research Association, New Orleans, LA. Peskin, J. and Olson, D. This Volume. “On reading poetry: Expert and novice knowledge.” Peterson, C. and McCabe, A. 1983. Developmental Psycholinguistics: Three Ways of Looking at a Child’s Narrative. New York: Plenum Press. Piaget, J. 1972. “Intellectual evolution from adolescence to adulthood.” Human Development 15: 1–12.

References 

Piatelli-Palmarini, M. 1980. Language and Learning: The Debate between Piaget and Chomsky. Cambridge: Cambridge University Press. Pickering, M. J. and Branigan, H. P. 1998. “The representation of verbs: Evidence from syntactic priming in language production.” Journal of Memory and Language 39: 633– 651. Pierce, A. E. 1992. “The acquisition of passives in Spanish and the question of A-chain maturation.” Language Acquisition 2: 55–81. Pine, K. and Messer, D. 1998. “The influence of social processes on children’s implicit and explicit representations.” Cognitive Development 13: 109–126. Pine, K. and Messer, D. 2003. “The development of representations as children learn about balancing.” British Journal of Developmental Psychology 21 (2): 285–301. Pinker, S. 1999. Words and Rules. London: Phoenix. Pinker, S. Lebeaux, D. S. and Frost, L. A. 1987. “Productivity and constrains in the acquisition of the passive.” Cognition 26: 195–267. Plunkett, K. and Wood, C. In Press. “The development of children’s understanding of grammar.” In Cognitive and Language Development in Children, J. Oates (ed.). Oxford: Blackwell. Pollio, H., Barlow, J., Fine, H. and Pollio, M. 1977. Psychology and the Poetics of Growth: Figurative Language in Psychology, Psychotherapy, and Education. Hillsdale, NJ: Lawrence Erlbaum. Pollock J. Y. 1997. Language et Cognition. lntroduction au Programme Minimaliste de la Grammaire Générative. Paris: Presses Universitaires de France. Pontecorvo, C. and Morani, R. M. 1996. “Looking for stylistic features in children’s composing stories: Products and processes.” In Early Text Construction in Children, C. Pontecorvo, M. Orsolini. and L. Resnick (eds.), 229–258. Hillsdale, NJ: Lawrence Erlbaum. Preece, A. 1987. “Narrative forms conversationally produced by young children.” Journal of Child Language 14: 353–373. Putter-Katz H., Banai, K. and Ahissar, M. In Press. “Speech perception in noise among learning disabled teenagers.” In Plasticity of the Central Auditory System and Processing of Complex Acoustic Signals, J. Syka and M. M. Merzenich (eds). Proceedings of the 6th IBRO World Congress of Neuroscience. Pye, C. and Quixtan Paz, P. 1988. “Precocious passive (and antipassives) in Quiché Mayan.” Papers and Reports on Child Language Development 27: 71–80. Rabukhin, E. 2003. Text production abilities in bilingual Hebrew-Russian children: A developmental study. MA thesis, Department of Communications Disorders, Tel Aviv University. [in Hebrew] Ragnarsdóttir, H., Aparici, M., Cahana Amitay, D., van Hell, J., and Viguié, A. 2002. “Verbal structure and content in written discourse: Expository and narrative texts.” Written Language and Literacy 5: 95–126. Ralli, A. 1999. Investigating lexical acquisition patterns: Context and cognition. Doctoral Dissertation. University of London. Ragnasdóttir, H. and Strömqvist, S. In Press. “The development of generic maður/man for the construction of discourse stance in Icelandic and Swedish.” Journal of Pragmatics. Special Issue on Discourse Stance.

 References

Ravid, D. 1990. “Internal structure constraints on new-word formation devices in Modern Hebrew.” Folia Linguistica 24: 289–346. Ravid, D. 1995a. Language Change in Child and Adult Hebrew: A Psycholinguistic Perspective. New York: Oxford University Press. Ravid, D. 1995b. “The acquisition of morphological junctions in Modern Hebrew.” In The Development of Morphological Systematicity: A Crosslinguistic Perspective, H. Pishwa and K. Maroldt (eds), 55–77. Tubingen: Gunter Narr. Ravid, D. 1997. “Between syntax and the lexicon: The parallel between N-N compounds and N-A strings in acquisition.” In Proceedings of the GALA 1997 Conference on Language Acquisition, A. Sorace, C. Heycock and R. Shillcock (eds), 138–141. Edinburgh: University of Edinburgh. Ravid, D. 1999. “On the essence of derived nominals in Modern Hebrew.” Hebrew Linguistics, 45: 61–78. [in Hebrew] Ravid, D. 2001. “Learning to spell in Hebrew: Phonological and morphological factors.” Reading and Writing 14: 459–485. Ravid, D. 2002. “A developmental perspective on root perception in Hebrew and Palestinian Arabic.” In Language Processing and Acquisition in Languages of Semitic, Root-Based Morphology, Y. Shimron (ed.), 293–319. Amsterdam: John Benjamins. Ravid, D. 2003. “Text-embedded lexical development: Noun types in later language acquisition.” Paper presented at the 9th Conference of the International Association for the Study of Child Language, Madison, WI. Ravid, D. 2004. “Cutting across inflection and derivation: A developmental perspective on linear formation in Hebrew nominals.” Paper presented at the 11th International Morphology Meeting, University of Vienna. Ravid, D. This Volume. “Derivational morphology revisited: Later lexical development in Hebrew.” Ravid, D. In Press. “Linguistic complexity in later language development: Evidence from expository text construction.” In Perspectives on Language and Language Development, D. Ravid and H. Bat-Zeev Shyldkrot (eds), Dordrecht: Kluwer. Ravid, D. and Avidor, A. 1998. “Acquisition of derived nominals in Hebrew: Developmental and linguistic principles.” Journal of Child Language 25: 229–266. Ravid, D. and Berman, R. A. 2003a. “The development of register perception in Hebrew text production.” Scientific Report to the Chief Scientist, Ministry of Education, Jerusalem, Israel. [in Hebrew] Ravid, D. and Berman, R. A. 2003b. “On modelling information density in narrative texts.” In Determination of Information and Tenor in Texts: Multidisciplinary Approaches to Discourse 2003, L. Lagerwerf, W. Spooren, and L. Degand (eds.), 17–37. Amsterdam: Stichting Neerlandistiek. Ravid, D. and Berman, R. A. In Press. “Developing information packaging in written and spoken narratives across adolescence”. In Language in the Context of Use: Cognitive Approaches to Language and Language Learning. A. Tyler, ed. The Hague: Mouton. Ravid, D. and Cahana-Amitay, D. 2004. “Verbal and nominal expressions in narrating conflict situations in Hebrew.” Journal of Pragmatics [Special Issue on Discourse stance].

References 

Ravid, D. and Nir, M. 2000. “On the development of the category of adjective in Hebrew.” In From Sound to Sentence: Studies on First Language Acquisition, M. Beers, B. van den Bogaerde, G. Bol, J. de Jong and C. Rooijmans (eds), 113–124. Groningen: Center for Language and Cognition. Ravid, D. and Shlesinger, Y. 1995. “Factors in the selection of compound-type in spoken and written Hebrew.” Language Sciences 17: 147–179. Ravid, D. and Shlesinger, Y. 1999. “Modern Hebrew adverbials: Between syntactic class and lexical category.” In Grammatical Categories, E. Contini-Morava and Y. Tobin (eds), 333–351. Amsterdam: John Benjamins. Ravid, D. and Tolchinksy, L. 2002. “Developing linguistic literacy: A comprehensive model.” Journal of Child Language 29: 417–447. Ravid, D. and Yagev, I. 2003 – July. “Language abilities in cleft-palate preschoolers and in their normally developing peers.” Paper presented at Child Language Seminar, University of Newcastle upon Tyne. Ravid, D. and Zilberbuch, S. 2003a. “Morpho-syntactic constructs in the development of spoken and written Hebrew text production.” Journal of Child Language 30: 395–418. Ravid, D. and Zilberbuch, S. 2003b. “The development of complex nominals in expert and non-expert writing: A comparative study.” Cognition and Pragmatics 11: 267–296. Ravid, D., Landau, H. and Lovetsky, M. 2003. “Passive morphology in the school years: An example of later language acquisition.” Script 5–6: 129–158. [in Hebrew] Ravid, D., Levie, R. and Avivi-Ben Zvi, G. 2003. “Morphological disorders.” In Classification of Developmental Language Disorders: Theoretical Issues and Clinical Implications, L. Verhoeven and H. van Balkom (eds), 235–260. Mahwah, NJ: Lawrence Erlbaum. Ravid, D., van Hell, J., Rosado, E. and Zamora, A. 2002. “Subject NP patterning in the development of text production: A crosslinguistic study of Dutch, English, Hebrew, and Spanish.” Written Language and Literacy 5: 69–94. Redington, M. and Chater, N. 2002. “Knowledge representation and transfer in artificial grammar learning (AGL).” In Implicit Learning and Consciousness: An Empirical, Philosophical, and Computational Consensus in the Making, R. M. French and A. Cleeremans (eds), 121–143. Hove: Psychology Press. Reilly, J. S., Harriet, J., Baruch, E., and Berman, R. 2002. “Propositional attitudes in written and spoken language.” Written Language and Literacy 52: 183–219. Reilly, J. S., Zamora, A., and McGivern, R. F. In Press. “Acquiring perspective in English: The development of stance” Journal of Pragmatics [Special Issue on Discourse Stance]. Reinhart, T. 1982. Pragmatics and Linguistics: An Analysis of Sentence Topics. Bloomington, IN: Indiana University Linguistics Club. Rice, M. 1990. “Preschoolers’ QUIL: Quick incidental learning of words.” In Children’s language, G. Conti-Ramsden and C. Snow (eds), 7: 171–194. Hillsdale, NJ: Lawrence Erlbaum. Rice, M. 2000. “Grammatical symptoms of specific language impairment.” In Speech and Language Impairments in Children, D. V. M. Bishop and L. Leonard (eds), 17–34. Philadelphia, PA: Taylor & Francis. Rice, M., Buhr, J. and Nemeth, M. 1990. “Fast mapping word-learning abilities of languagedelayed pre-schoolers.” Journal of Speech and Hearing Disorders 55 (1): 33–42.

 References

Rice, M., Oetting, J., Marquis, J., Bode, J. and Pae, S. Y. 1994. “Frequency of input effects on word comprehension of children with specific language impairment.” Journal of Speech and Hearing Research 37 (1): 106–122. Rice, M. and Wexler, K. 1996. “Toward tense as a clinical marker of specific language impairment in English-speaking children.” Journal of Speech and Hearing Research 39: 1239–1257. Rice, M., Wexler, K., and Hershberger, S. 1998. “Tense over time: The longitudinal course of tense acquisition in children with specific language impairment.” Journal of speech, Language, and Hearing Research 41: 1412–1431. Rice, M., Wexler, K. and Redmond, S. 1999. “Grammaticality judgments of an extended optional infinitive grammar: Evidence from English-speaking children with specific language impairment.” Journal of Speech, Language, and Hearing Research 42: 943–961. Rice, M. and Woodsmall, L. 1988. “Lessons from television: Children’s word learning when viewing.” Child Development 59: 420–429. Rispoli, M. and Hadley, P. 2001. “The leading edge: The significance of sentence disruptions in the development of grammar.” Journal of Speech, Language, and Hearing Research 44: 1131–1143. Robinson, E., Goelman, H. and Olson, D. R. 1983. “Children’s relationship between expressions (what was said) and intentions (what was meant).” British Journal of Developmental Psychology 1: 75–86. Rogoff, B. 1990. Apprenticeship in Thinking: Cognitive Development in Social Context. New York: Oxford University Press. Rogoff, B. 2003. The Cultural Nature of Human Development. Oxford: Oxford University Press. Rollins, P., Pan, B., Conti-Ramsden, B. A., and Snow, C. E. In Press. “Communicative skills in specific language impaired children: A comparison with their language matched siblings.” Journal of Communication Disorders. Romaine, S. 1984. The Language of Children and Adolescents: The Acquisition of Communicative Competence. Oxford: Basil Blackwell. Rosenblatt, L. M. 1989. “The transactional theory of the literary work: Implications for research.” In Researching Response to Literature and the Teaching of Literature, C. R. Cooper (ed.), 32–53. Norwood, NJ: Ablex. Roth, F. and Spekman, N. 1989. “The oral syntactic proficiency of learning disabled students: A spontaneous story sampling analysis.” Journal of Speech and Hearing Research 32: 67–77. Rubinstein, J. S., Meyer, D. E. and Evans, J. E. 2001. “Executive control of cognitive processes in task switching.” Journal of Experimental Psychology: Human Perception and Performance 27: 763–797. Rudel, R. G., Denckla, M. B., Broman, M. and Hirsch, S. 1980. “Word finding as a function of stimulus context: Children compared with aphasic adults.” Brain and Language 10: 111–119. Ruffman, T., Olson, D. R. and Torrance, N. 1990. “Young children’s understanding of the relation between verbal ambiguity, visual ambiguity and false belief.” Unpublished paper, Centre for Applied Cognitive Science, Ontario Institute for Studies in Education, Toronto.

References 

Rumelhart, D., and McClelland, J. (eds). 1986. Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Cambridge, MA: MIT Press. Saban, R. 1999. Developing syntactic and metasyntactic skills in Hebrew-speaking schoolchildren and adolescents. MA thesis, Department of Communications Disorders, Tel Aviv University. [in Hebrew] Sabbagh, M. A. and Baldwin, D. A. 2001. “Learning words from knowledgeable versus ignorant speakers: Links between preschoolers’ theory of mind and semantic development.” Child Development 72 (4): 1054–1070. Sacks, H., Schegloff, E. and Jefferson, G. 1974. “A simplest systematic for the organization of turn taking in conversation.” Language 50: 696–735. Sandbank, A. 2002. “The interplay between the conventions of the writing system and written language in writing texts.” In Developmental Aspects in Learning to Write, L. Tolchinsky (ed.), 55–76. Dordrecht: Kluwer Academic. Sandra, D., Frisson, S., and Daems, F. 1999. “Why simple verb forms can be so difficult to spell: The influence of homophone frequency and distance in Dutch.” Brain and Language 68: 277–283. Santrock, J. W. 1996. Adolescence (6th ed.). Madison, WI: Brown & Benchmark. Sawyer, K. 1997. Pretend Play as Improvisation: Conversation in the Preschool Classroom. Mahwah, NJ: Lawrence Erlbaum. Scarborough, H. S. 1989. “Prediction of reading disability from familial and individual differences.” Journal of Educational Psychology 81: 101–108. Scardamalia, M. and Bereiter, C. 1987. The Psychology of Written Composition. Hillsdale, NJ: Lawrence Erlbaum. Schecter, B. and Broughton, J. 1991. “Developmental relationships between psychological metaphors and concepts of life and consciousness.” Metaphor and Symbolic Activity 6 (2): 119–143. Scheepers, C. 2003. “Syntactic priming of relative clause attachments: Persistence of structural configuration in sentence production.” Cognition 89: 179–205. Schiffrin, D. 1995. Discourse Markers. Cambridge: Cambridge University Press. Schober-Peterson, D. and Johnson, C. 1991. “Non-dialogic speech during preschool interactions.” Journal of Child Language 18: 153–170. Schuele, C. M. and Nicholls, L. 2000. “Relative clauses: Evidence of continued linguistic vulnerability in children with specific language impairment.” Clinical Linguistics and Phonetics 14: 563–585. Schwarzwald, O. R. 1981. Grammar and Reality in Hebrew Verbs. Ramat Gan: Bar Ilan Press. [in Hebrew] Schwarzwald, O. R. 1998. “The formation of adjectival nouns and processes of derivation and inflection.” Helkat Lashon 27: 9–19. [in Hebrew] Schwarzwald, O. R. 2002. Studies in Hebrew Morphology. Tel Aviv: Open University of Israel Press. [in Hebrew] Scott, C. M. 1988. “Spoken and written syntax.” In Later Language Development: Ages 9 through 19, M. Nippold (ed.), 45–95. San Diego, CA: College Hill Press. Scott, C. M. 2002 – August. “Speaking and writing the same texts: Comparisons of school children with and without language learning disabilities.” Paper presented at the 12th annual meeting of the Society for Text and Discourse, Chicago IL.

 References

Scott, C. M. 2003 – July. “Literacy as variety: An analysis of clausal connectivity in spoken and written language of children with language learning disabilities.” Paper presented at the 24th Annual Symposium on Research in Child Language Disorders, Madison WI. Scott, C. M. 2004. “Syntactic contributions to literacy.” In Handbook of Language and Literacy. B. Shulman, K. Apel, B. Ehren, E. Silliman and A. Stone (eds), 340–362. New York, NY: Guilford. Scott, C. M. This Volume. “Syntactic ability in children and adolescents with language and learning disabilities.” Scott, C. M. and Windsor, J. 2000. “General language performance measures in spoken and written narrative and expository discourse of school-age children with language learning disabilities.” Journal of Speech, Language, and Hearing Research 43: 324–339. Scott, C. M., Windsor, J. and Gray, T. 1998. “A typology of grammatical error in older children with language learning disorders.” Paper presented at the annual meeting of the American Speech Language Hearing Association, San Antonio, TX. Scott, J. A. and Nagy, W. E. 1997. “Understanding the definitions of unfamiliar verbs.” Reading Research Quarterly 32: 184–200. Sénéchal, M. 2000. “Morphological effects in children’s spelling of French words.” Canadian Journal of Experimental Psychology 54: 76–85. Seroussi, B. 2003. Hebrew derived nominals in later language development. MA thesis, Department of Communications Disorders, Tel Aviv University. [in Hebrew] Seroussi, B. In Press. “Hebrew derived nouns in a sentential context: a developmental perspective.” Folia Phoniatrica et Logopaedica. Shalom, T. 2002. Linguistic structures and discourse stance in Hebrew L1 textbooks. Doctoral dissertation, Bar Ilan University. [in Hebrew] Shatz, M. 1984. “A song without music and other stories: How children’s cognitive process constraints influence children’s oral and written narratives.” In Meaning, Form, and Use in Context: Linguistic Applications, D. Shiffrin (ed.), 313–334. Washington, DC: Georgetown University Press. Shatz, M. 1987. “Bootstrapping operations in child language.” In Children’s Language, Volume 6, K. Nelson and A. van Kleeck (eds), 1–22. Hillsdale, NJ: Lawrence Erlbaum. Shibatani, M. 1985. “Passives and related constructions: A prototype analysis.” Language 61: 821–842. Shimron, J. (ed.). 2003. Language Processing and Language Acquisition in a Root-Based Morphology. Amsterdam: John Benjamins. Shlesinger, Y. and Ravid, D. 2003 – March. “The distribution of adjective classes in different text types.” Paper presented at the AAAL meeting, Virginia, MD. Shultz, T. R. and Pilon, R. 1973. “Development of the ability to detect linguistic ambiguity.” Child Development 44: 728–733. Shuman, A. 1986. Story Telling Rights: The Uses of Oral and Written Texts by Urban Adolescents. Cambridge: Cambridge University Press. Siegal, M. and Varley R. 2002. “Neural systems involved in ‘theory of mind’.” Nature 3: 463– 471. Simms, R. and Crump, W. 1983. “Syntactic development in the oral language of learning disabled and normal students at the intermediate and secondary level.” Learning Disability Quarterly 6: 155–165.

References 

Simonin, J. 1984. “Les répérages énonciatifs dans les textes de presse.” In La Langue au Ras du Texte, A. Grésillon and J.-L. Lebrave (eds), 133–203. Lille: Presses Universitaires de Lille. Singer, B. 1997. Parallels Between Spoken and Written Syntax in Children with LanguageLearning Disabilities. Unpublished doctoral dissertation. Emerson College, Boston, MA. Slobin, D. I. 1973. “Cognitive prerequisites for the development of grammar.” In Studies of Child Language Development, C. A. Ferguson and D. I. Slobin (eds), 175–208. New York: Holt Reinhart and Winston. Slobin, D. I. 1977. “Language change in childhood and in history.” In Language Learning and Thought, J. Macnamara (ed.), 185–215. New York: Academic Press. Slobin, D. I. 1985. “Crosslinguistic evidence for the language-making capacity.” In The Crosslingusitic Study of Language Acquisition, Volume 2: Theoretical Issues, D. I. Slobin (ed.), 1157–1256. Hillsdale, NJ: Lawrence Erlbaum. Slobin, D. I. 1988. “The development of clause chaining in Turkish child language.” In Studies on Turkish Linguistics, S. Koç (ed.), 27–54. Ankara: Middle East Technical University. Slobin, D. I. 1993. “Passives and alternatives in children’s narratives in English, Spanish, German, and Turkish.” In Voice: Form and Function, B. Fox and P. J. Hopper (eds), 341–64. Amsterdam: John Benjamins. Slobin, D. I. 1996. “From ‘thought and language’ to ‘thinking for speaking’.” In Rethinking Linguistic Relativity, J. J. Gumperz and S. C. Levinson (eds), 70–96. Cambridge: Cambridge University Press. Slobin, D. I. 1997. “Mind, code, and text.” In Essays on Language Function and Language Type: Dedicated to Talmy Givón, J. Bybee, J. Haiman, and S. A. Thompson (eds), 437– 467. Amsterdam: John Benjamins. Slobin, D. I. 2001. “Form-function relations: How do children find out what they are?” In Language Acquisition and Conceptual Development, M. Bowerman and S. C. Levinson (eds), 406–449. Cambridge: Cambridge University Press. Smith, M. 1998. Sense and sensitivity: An investigation into fifth grade children’s knowledge of English derivational morphology and its relationship to vocabulary and reading ability. Unpublished doctoral dissertation, Graduate School of Education, Harvard University. Smith, M., and Wheeldon, L. 2001. “Syntactic priming in spoken sentence production – An online study.” Cognition 78: 123–64. Snow, C. 1977. “The development of conversation between mothers and babies.” Journal of Child Language 11: 423–452. Snow, C. 1990. “The development of definitional skill.” Journal of Child Language 17(3): 697–710. Snow, C. 1995. “Issues in the study of input: Finetuning, universality, individual and developmental differences, and necessary causes.” In The Handbook of Child Language, P. Fletcher and B. MacWhinney (eds.), 180–193. Cambridge: Cambridge University Press.

 References

Snowling, M. J. 1994. “Towards a model of spelling acquisition: The development of some component skills.” In Handbook of Spelling: Theory, Process and Intervention, G. D. A. Brown, and N. C. Ellis (eds), 111–128. Chichester: John Wiley. Snowling, M. J., van Wagtendonk, B. and Stafford, C. 1988. “Object naming deficits in developmental dyslexia.” Journal of Research in Reading 11(2): 67–85. Sodian, N. B. 1991. “The development of deception in young children” British Journal of Developmental Psychology 9: 173–178. Soja, N. N., Carey, S. and Spelke, E. 1991. “Ontological categories guide young children’s inductions of word meaning: Object terms and substance terms.” Cognition 38: 179– 211. Solheim, O. J. and Uppstad, P. H. 2003. “Spelling quantity in Norwegian.” Technical Report. Stavanger University College: Center for Reading Research. Sotillo, M. and Riviere, A. 2001. “Cuando los niños usan las palabras para engañar: la mentira como instrumento al servicio del desarrollo” Infancia y Aprendizaje 24: 291– 306. Spelman-Miller, K. 2000. “Academic writers on-line: investigating pausing in the production of text.” Language Teaching Research, 4/2: 123–148. Stahl, S. A. and Fairbanks, M. M. 1986. “The effects of vocabulary instruction: A model based meta-analysis.” Review of Educational Research 56 (1): 72–110. Stanovich, K. E. and Cunningham, A. E. 1993. “Where does knowledge come from? Specific associations between print exposure and information acquisition.” Journal of Educational Psychology 85 (2): 211–229. Sternberg, R. J. 1987. “Most vocabulary is learned from context.” In The Nature of Vocabulary Acquisition, M. G. McKeown and M. E. Curtis (eds), 89–106. Hillsdale, NJ: Lawrence Erlbaum. Sternberg, R. J. and Powell, J. S. 1983. “Comprehending verbal comprehension.” American Psychologist 38: 878–893. Stiassnie, N. 2004. Language abilities in gradeschoolers with APD. MA thesis, Department of Communications Disorders, Tel Aviv University. [in Hebrew] Storck, P. A. and Looft, W. R. 1973. “Qualitative analysis of vocabulary responses from persons aged six-to sixty-six plus.” Journal of Educational Psychology 65: 192–197. Storkel, H. L. 2001. “Learning new words: Phonotactic probability in language development.” Journal of Speech Language and Hearing Research 44 (6): 1321–1337. Storkel, H. L. and Rogers, M. A. 2000. “The effect of probabilistic phonotactics on lexical acquisition.” Clinical Linguistics and Phonetics 14: 407–425. Stothard, S. E., Snowling, M., Bishop, D., Chipchase, B. and Kaplan, C. 1998. “Languageimpaired preschoolers: A follow-up into adolescence.” Journal of Speech, Language, and Hearing Research 40: 407–418. Straus, L. G. 2004. “As our world warmed”. Science 303: 1300–1302. Strauss, S. (ed.) 1982. U-shaped Behavioral Growth. New York: Academic Press. Strömqvist, S. 2003. “Language acquisition in early childhood.” In. Psycholinguistics – An International Handbook, G. Rickheit, Th. Herrmann & W. Deutsch, (eds.), 790–801. Berlin: Walter de Gruyter.

References 

Strömqvist, S. & Ahlsén, E. (eds.). 1999. “The process of writing – A progress report.” Gothenburg Papers in Theoretical Linguistics 83, Göteborg University: Department of Linguistics. Strömqvist, S., Ahlsén, E. and Wengelin, Å. 1999. “The production process in speech and writing.” In Developing Literacy across Genres, Modalities, and Languages. Working Papers, Volume 1. R. Aisenman (ed.), 1–20. Tel Aviv University: International Project on Developing Literacy. Strömqvist, S., Holmqvist, K., Johansson, V., Karlsson, H. and Wengelin, Å. In Press. “What key-logging can reveal about writing.” In Key Logging and Writing Research, K. Sullivan and E. Berglund (eds). Dordrecht: Kluwer. Strömqvist, S., Johansson, V., Kriz, S. Ragnarsdóttir, H., Aisenman, R. and Ravid, D. 2002. “Toward a cross-linguistic comparison of lexical quanta in speech and writing.” Written Language and Literacy 5(1): 45–68. Strömqvist, S. and Karlsson, H. 2002. ScriptLog for Windows – User’s Manual. Technical Report. University of Lund: Department of Linguistics, University College of Stavanger: Centre for Reading Research. Strömqvist, S. and Malmsten, L. 1998. Scriptlog Pro 1.04 – User’s Manual. Technical Report. Department of Linguistics, University of Göteborg. Strömqvist, S., Nordqvist, Å., and Wengelin, Å. 2004. “Writing the frog story: developmental and cross-modal perspectives.” In Relating Events in Narrative: Typological and Contextual Perspectives, S. Strömqvist and L. Verhoeven (eds.), 359–394. Mahwah NJ: Lawrence Erlbaum. Svensson, C. 1987. “The construction of poetic meaning: A developmental study of symbolic and non-symbolic strategies in the interpretation of contemporary poetry.” Poetics 16: 471–503. Swan, D. and Goswami, U. 1997. “Picture naming deficits in developmental dyslexia. The phonological representations hypothesis.” Brain and Language 56 (3): 334–353. Swanborn, M. and de Glopper, K. 1999. “Incidental word learning while reading: A metaanalysis.” Review of Educational Research 69 (3): 261–285. Tallal, P., Hirsch, L., Realpe-Bonilla,T., Miller, S., Brzustowicz, L., Bartlet, C. and Flax, J. F. 2001. “Family aggregation in specific language impairment.” Journal of Speech, Language, and Hearing Research 44: 1172–1182. Talmy, L. 2003. “Concept structuring systems in language.” In The New Psychology of Language: Cognitive and Functional Approaches to Language Structure, Vol. 2, M. Tomasello (ed.), 15–46. Mahwah, NJ: Lawrence Erlbaum. Tanaka, H. 1999. Turn-Taking in Japanese Conversation. Amsterdam: John Benjamins. Tannen, D. 1982. “Oral and literate strategies in spoken and written narratives.” Language 58: 1–21. Tannen, D. 1989. Talking Voices: Repetition, Dialogue and Imagery in Conversational Discourse. Cambridge: Cambridge University Press. Tannen, D. 1990. “Gender differences in conversational coherence: Physical alignments and topical cohesion.” In Conversational Organization and Development, B. Dorval (ed.), 167–207. Norwood, NJ: Ablex.

 References

Thévenin, M. G., Totereau, C., Fayol, M., and Jarousse, J. P. 1999. “L’apprentissage / enseignement de la morphologie écrite du nombre en français.” Revue Française de Pédagogie 126: 39–52. Thompson, S. A. 1978. “Modern English from a typological point of view: Some implication of the function of word order.” Linguistische Berichte 54: 19–36. Thompson, S. A. and Hopper, P. J. 2001. “Transitivity, clause structure, and argument structure: Evidence from conversation.” In Frequency and the Emergence of Linguistic Structure, J. Bybee and P. Hopper (eds.), 27–60. Amsterdam: John Benjamins. Tishman, S. and Perkins, D. 1997. “The language of thinking.” Phi Delta Kappa 78: 368–374. Tolchinsky, L. This Volume. “The nature and scope of later language development.” Tolchinsky, L., Johansson, V., and Zamora, A. 2002. “Text openings and closings in writing and speech: Autonomy and differentiation.” Written Language and Literacy 5: 219–254. Tolchinsky, L., Perera, J., Argerich, N. and Aparici, M. 1999. Rhetorical moves and noun phrase complexity in written expository texts”. In Working Papers in Developing Literacy across Genres, Modalities, and Languages, Volume 1, 183–219. Tel Aviv University: International Project on Developing Literacy. Tolchinsky, L. and Rosado, E. In Press “The effect of literacy, text type, and modality on the use of grammatical means for agency alternation in Spanish.” Journal of Pragmatics. Special Issue on Discourse Stance. Tomasello, M. 2001a. “General introduction.” In Language Development: The Essential Readings, M. Tomasello and E. Bates (eds), 1–12. Malden, MA: Blackwell. Tomasello, M. 2001b. “The item-based nature of children’s early syntactic development.” In Language Development: The Essential Readings, M. Tomasello and E. Bates (eds), 169–186. Malden, MA: Blackwell. Tomasello, M. and Kruger, A. C. 1992. “Joint attention on actions – Acquiring verbs in ostensive and non-ostensive contexts.” Journal of Child Language 19 (2): 311–333. Tomasello, M., Brooks, P. J. and Stern, E. 1988.” Learning to produce passive utterances through discourse.” First Language 18: 223–237. Tomblin, J. B. and Buckwalter, P. R. 1994. “Studies of genetics of specific language impairment.” In Specific Language Impairments in Children R. V. Watkins and M. L. Rice (eds.), Vol. 4, 17–34. Baltimore, MD: Brookes. Tomblin, J. B., Records, N., Buckwalter, P, Xhang, X., Smith, E. and O’Brien, M. 1997. “Prevalence of specific language impairment in kindergarten children.” Journal of Speech, Language, and Hearing Research 40: 1245–1260. Torrance, N. and Olson, D. R. 1985. “Oral and literate competencies in the early school years.” In Literacy, Language, and Learning: The Nature and Consequence of Reading and Writing, D. R. Olson, N. Torrrance and A. Hildyard (eds), 256–284. Cambridge: Cambridge University Press. Totereau, C., Barrouillet, P. and Fayol, M. 1998. “Overgeneralizations of number inflections in the learning of written French: The case of noun and verb.” British Journal of Developmental Psychology” 16: 447–464. Totereau, C., Thevenin, M. G., and Fayol, M. 1997. “The development of the understanding of number morphology in written French.” In Learning to Spell: Research, Theory, and Practice across Languages, C. A. Perfetti, L. Rieben, and M. Fayol (eds), 97–114. Hillsdale NJ: Lawrence Erlbaum.

References 

Treiman, R. 1993. Beginning to Spell: A Study of First-Grade Children. New York: Oxford University Press. Treiman, R. and Cassar, M. 1997. “Spelling acquisition in English.” In Learning to Spell: Research, Theory, and Practice across Languages, C. A. Perfetti, L. Rieben and M. Fayol (eds), 61–80. Hillsdale, NJ: Lawrence Erlbaum. U.S. Department of Education, Institute of Education Science, National Center for Education Statistics, The Nation’s Report Card: Writing 2002. NCES. 2003–529, by H. R. Persky, M. C. Daane, and Y. Jin. Washington, DC: 2003. van Hell, J.G., Verhoeven L., Tak, M. and Oosterhout, M. In Press. “To take a stance: A developmental study on the use of pronouns and passives in spoken and written texts in Dutch.” Journal of Pragmatics. Special Issue on Discourse Stance. Vedeler, L. 1997. “Dramatic play: A format for ’literate’ language.” British Journal of Educational Psychology 67: 153–167. Velmans, M. 1991. “Is human information processing conscious?” Behavioural and Brain Sciences 14: 651–726. Verhoeven, L., Aparici, M., Cahana-Amitay, D., van Hell, J., Kriz, S. and Viguié-Simon, A. 2002. “Clause packaging in writing and speech: A cross-linguistic developmental analysis.” Written Language & Literacy 5(2): 135–162. Véronis, J. 1988. “From sound to spelling in French: Simulation on a computer.” Cahiers de Psychologie Cognitive 8: 315–334. Vinay, J.-P. and J. Darbelnet. 1995. Comparative Stylistics of French and English: A Methodology for Translation. Amsterdam: John Benjamins. Vogel, S. 1977. “Morphological ability in normal and dyslexic children.” Journal of Learning Disabilities 10(1): 35–43. Vosniadou, S. 1987. “Children and metaphor.” Child Development 58: 870–885. Wade, E. and Clark, H. H. 1993. “Reproduction and demonstration in quoting.” Journal of Memory and Language 32: 805–819. Watson, R. 1995. “Relevance and definition.” Journal of Child Language 22 (1): 211–222. Waxman, S. R. 1994. “The development of an appreciation of specific linkages between linguistic and conceptual organization.” Lingua 92: 229–257. Weissenborn, J. and Höhle, B. 2000. Approaches to Bootstrapping: Phonological, Lexical, Syntactic, and Neurophysiological Aspects of Early Language Acquisition. Amsterdam: John Benjamins. Weizman, E. 1998. “Truth or fiction? Direct speech in the written press.” Hebrew Linguistics 43: 29–43. [in Hebrew] Weizman, Z. O. and Snow, C. E. 2001. “Lexical input as related to children’s vocabulary acquisition: effects of sophisticated exposure and support for meaning.” Developmental Psychology 37 (2): 265–279. Wellman, H. 1990. The Child’s Theory of Mind. Cambridge, MA: MIT Press. Wells, G. 1985. Language Development in the Pre-school Years. Cambridge: Cambridge University Press. Wengelin, Å. 2002. Text Production in Adults with Reading and Writing Difficulties. Göteborg University: Department of Linguistics.

 References

Wengelin, Å. In Press. “Examining pauses in writing – Theory, methods and empirical data.” In Key Logging and Writing Research, K. Sullivan and E. Berglund (eds). Dordrecht: Kluwer. Wengelin, Å. and Strömqvist, S. This Volume. “Text-writing development viewed through on-line pausing in Swedish” Werner, H. and Kaplan, B. 1963. Symbol Formation: An Organismic Developmental Approach to Language and the Expression of Thought. New York: John Wiley and Sons. Wexler, B. 2003, May 5. Opinion. Newsweek. Whittlesea, B. W. A. and Dorken, M. D. 1997. “Implicit learning: indirect, not unconscious.” Psychonomic Bulletin and Review 4: 63–67. Wiktorsson, M. 2000. “The production rate of prefabs – a pilot study.” In Developing Literacy across Genres, Modalities, and Languages. Working Papers, Volume 1II, M. Aparici, J. Perera, E. Rosado., and L. Tolchinsky (eds.), 211–224. University of Barcelona: International Project on Developing Literacy. Wiktorsson, M. 2003. “Learning Idiomaticity. A Corpus-Based Study of Idiomatic Expressions in Learners’ Written Production.” Lund University. Lund Studies in English 105. Williams, B. R., Ponesse, J. S., Schachar, R. J., Logan, G. D. and Tannock, R. 1999. “Development of inhibitory control across the life span.” Developmental Psychology 35: 205–213. Williams, E. 1994. “Remarks on lexical knowledge.” Lingua 92: 7–34. Wimmer, H. and Goswami, U. 1994. “The influence of orthographic consistency on reading development: Word recognition in English and German children.” Cognition 51: 91– 103. Wimmer, H. and Perner, J. 1983. “Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children’s understanding of deception.” Cognition 13: 103–128. Windfuhr, K. L., Faragher, B. and Conti-Ramsden, G. 2002. “Lexical learning skills in young children with specific language impairment (SLI).” International Journal of Communication Disorders 37: 415–432. Windsor, J., Scott, C. M. and Street, C. 2000. “Verb and noun morphology in the spoken and written language of children with language learning disabilities.” Journal of Speech, Language, and Hearing Research 43: 1322–1336. Winner, E. 1988. The Point of Words: Children’s Understanding of Metaphor and Irony, Cambridge, MA: Harvard University Press. Wolf, D. and Hicks, D. 1989. “The voices within narratives: The development of intertextuality in young children’s stories.” Discourse Processes 12: 329–351. Zilberman, L. 2003. The adjective category in narrative texts: A developmental study of later language acquisition. MA thesis, Cognitive Studies Program, Tel Aviv University. [in Hebrew] Zubizarreta, M. L. and van Haaften, T. 1988. “English -ing and Dutch -en nominal constructions: A case of simultaneous nominal and verbal projections.” In Morphology and Modularity, M. Everaert, A. Evers, R. Huybregts and M. Trommelen (eds), 361–393. Dordrecht: Foris Publications.

References 

Zucchi, A. 1993. The Language of Propositions and Events: Issues in the Syntax and Semantics of Nominalization. Dordrecht: Kluwer.

Author index

A Abedi, J. 57, 249 Adams, A. 46, 262 Adams, M. 163, 249 Adger, T. C. 197, 209, 265 Aitchison, J. 54, 58, 249 Allen, S. E. M 63, 91, 249 Alsina, A. 85, 249 Alves, R. A. 178, 185, 249 Andersen, E. S. 137, 249 Anderson, J. R. 54, 55, 166, 249 Anderson, R. C. 55, 271 Anderson, S. R. 54, 249 Anglin, J. M. 36, 39, 40, 41, 42, 48, 52, 55, 58, 249, 266 Aparici, M. 238, 249, 275, 285 Appel-Mashraki, Y. 65, 66, 75, 76, 250 Applebee, A. N. 229, 250 Ariel, M. 88, 108, 250 Aronoff, M. 60, 250 Aronsson, K. 196, 200, 201, 250 Ashby, W. J. 92, 109, 152, 154, 155, 250 Aslin, R. N. 174, 261 Assayag, N. 88, 90, 250 Astington, J. W. 220, 226, 250, 273 Atlani, F. 154, 250 Auwarter, M. 198, 200, 201, 250 Avidor, A. 56, 67, 68, 69, 70, 71, 78, 276 Avivi-Ben Zvi, G. 70, 71, 250, 277 B Baayen, H. 55, 56, 58, 60, 250 Baddeley, A. D. 46, 175, 250, 262

Bakhtin, M. 204, 250 Baldwin, D. A. 38, 42, 250, 279 Bamberg, M. 195, 251 Baron, J. 48, 262 Barthes, R. 212, 213, 251 Bates, E. 15, 34, 151, 251, 261 Batterman, N. 47, 236, 267 Beal, C. R. 226, 251 Beard, R. 34, 251 Beck, I. L. 40, 251, 270 Bedore, L. 114, 251 Bellaire, S. 117, 118, 251 Benelli, B. 41, 251 Bentivoglio, P. 91, 92, 109, 250, 251 Bereiter, C. 185, 222, 251, 279 Berglund, E. 178, 251 Berman, R. A. 1, 2, 3, 5, 6, 9, 10, 11, 12, 13, 15, 18, 20, 22, 24, 25, 26, 28, 29, 32, 33, 34, 48, 52, 54, 56, 58, 59, 60, 61, 62, 63, 67, 74, 77, 78, 79, 80, 81, 84, 88, 89, 90, 92, 96, 97, 98, 105, 106, 108, 109, 111, 112, 123, 124, 127, 128, 137, 143, 145, 146, 149, 151, 152, 153, 156, 160, 161, 176, 178, 179, 192, 195, 202, 212, 234, 236, 238, 240, 241, 246, 247, 251, 252, 253, 254, 258, 276, 277 Berrendonner, A. 142, 254 Bertram, R. 48, 183, 254 Best, R. 44, 52, 254, 256 Bialystok, E. 136, 254 Biber, D. 58, 89, 124, 128, 137, 245, 254 Biemiller, A. 45, 254 Bishop, D. V. M. 49, 254, 272, 282

 Author index

Bjork, E. L. 40, 259 Blanche-Benveniste, C. 137, 138, 139, 141, 142, 145, 254, 255 Blank, M. 240, 255 Bloom, P. 36, 38, 43, 255 Blum-Kulka, S. 7, 8, 191, 193, 194, 195, 196, 198, 200, 201, 207, 208, 209, 210, 214, 237, 240, 245, 255, 264 Bock, J. K. 150, 255 Bolozky, S. 60, 255 Booth, A. E. 37, 255 Booth, J. R. 48, 132, 255 Borer, H. 63, 255 Bosman, A. M. T. 176, 255 Bourdin, B. 171, 255, 256 Bowerman, M. 10, 13, 24, 32, 37, 256, 258, 274 Bowey, J. A. 46, 256 Brabham, E. G. 44, 256 Braisby, N. 40, 46, 256 Branigan, H. P. 132, 150, 256, 275 Bransford, J. 226, 256 Brinton, L. 59, 256 Broderick, V. 247, 256 Broughton, J. 48, 279 Brown, A. S. 175, 256 Brown, G. D. A. 163, 256 Brown, R. 112, 113, 256 Bruner, J. 195, 208, 213, 220, 230, 256, 257 Bryant, P. 163, 170, 175, 176, 257, 267, 272 Buckwalter, P. R 131, 284 Budwig, N. 63, 247, 257 Bybee, J. L. 34, 140, 257 C Cacciari, C. 238, 257, 269 Cahana-Amitay, D. 28, 56, 68, 79, 141, 275, 276, 285 Campbell, R. N. 36, 167, 257, 260 Capps, L. 207, 208, 272 Carey, S. 36, 257, 282 Carlisle, J. F. 44, 55, 58, 60, 118, 257

Carnine, D. 45, 257, 266 Cassar, M. 165, 168, 170, 257, 285 Catts, H. 113, 257 Chafe, W. 83, 88, 90, 142, 177, 187, 189, 212, 213, 257 Chafetz, J. 63, 263 Charness, N. 229, 261 Charney, R. 193, 236, 267 Chater, N. 169, 277 Cheshire, J. 197, 258 Choi, S. 37, 256, 258 Chomsky, C. 111, 112, 233, 258 Chomsky, N. 30, 34, 58, 67, 233, 234, 258 Chung, S. 241, 258 Clancy, P. 91, 99, 194, 258 Clark, E. V. 20, 21, 26, 32, 33, 34, 36, 37, 38, 39, 40, 48, 52, 55, 58, 62, 73, 79, 81, 236, 258 Clark, H. H. 73, 199, 202, 258, 285 Clements, P. C. 55, 273 Cohen-Gross, D. 59, 258 Comrie, B. 29, 56, 67, 108, 259 Cooper, D. 46, 259 Corley, M. M. B. 150, 259 Corsaro, B. 196, 259 Corsaro, W. A. 191, 196, 201, 259 Coulmas, F. 199, 202, 259 Cousin, M.-P. 173, 259 Crago, M. B. 63, 249 Crain, S. 10, 233, 259 Creissels, D. 154, 160, 259 Croft, W. 152, 259 Cronbach, L. J. 40, 259 Crump, D. 125, 271 Crump, W. 125, 281 Culler, J. 217, 218, 219, 224, 259 Cunningham, A. E. 35, 43, 259, 282 Curtis, M. E. 41, 259 D Danielewicz, J. 212, 213, 257 Dapretto, M. 40, 259 Darbelnet, J. 152, 153, 285 Dascal, M. 228, 260

Author index 

de Jong, P. F. 46, 260 de Villiers, J. G. 63, 243, 244, 260 de Villiers, P. A. 63, 243, 244, 260 Demke, T. 46, 260 Demuth, K. 63, 260 Di Sciullo, A. M. 59, 108, 260 Dickinson, D. K. 195, 260 Dickinson, E. 211, 260 Dixon, M. 175, 260 Dixon, R. M. W. 86, 260 Dockrell, J. E. 2, 5, 8, 35, 36, 40, 50, 56, 236, 256, 260, 270 Dollaghan, C. 48, 49, 260 Dorken, M. D. 166, 286 Dorval, B. 8, 192, 194, 196, 240, 260 Du Bois, J. W. 83, 86, 87, 88, 89, 90, 91, 92, 98, 104, 106, 260 Durie, M. 91, 260 Durkin, K. 47, 260 Dutra, R. 91, 100, 261 E Eckerman, C. O. 8, 192, 194, 196, 240, 260 Eckert, P. 197, 261 Eco, U. 224, 261 Elley, W. 43, 44, 261 Ellis, N. 163, 256 Elman, J. L. 34, 261 England, N. 92, 261 Ericsson, K. A. 229, 261 Erskine, J. 178, 261 Ervin-Tripp, S. 193, 196, 261 Evans, J. 115, 261, 278 Eyer, J. 120, 261 F Fairbanks, M. M. 44, 282 Fayol, M. 4, 6, 139, 163, 167, 168, 171, 172, 255, 256, 259, 261, 265, 268, 273, 274, 284 Fellbaum, C. 37, 271 Ferguson, C. A. 89, 137, 261 Fernández-Lagunilla, M. 94, 261

Fiser, J. 174, 261 Fish, S. 223, 230, 261 Flax, J. F. 131, 262, 283 Fletcher, P. 119, 120, 235, 262, 267 Foley, W. A. 145, 262 Francois, D. 141, 262 Franklin, M. 240, 255 Freyd, P. 48, 262 Frye, N. 213, 217, 220, 262 Fukkink, R. G. 45, 262 Funnell, E. 41, 47, 262

G Gadet, F. 141, 142, 143, 262 Galda, L. 192, 195, 199, 273, 274 Garman, M. 235, 262 Garvey, C. 193, 194, 262 Gathercole, S. E. 46, 262 Gayraud, F. 89, 141, 147, 262 Gee, J. P. 230, 263 Geiger, V. 65, 66, 70, 263 Gentner, D. 52, 238, 263 German, D. J. 50, 263 Gerrig, R., J. 199, 202, 258 Gillam, R. 122, 125, 263 Gillette, J. 38, 39, 263 Gillis, S. 163, 263 Ginsburg, H. P. 57, 263 Givón 83, 89, 90, 263 Goel, V. 57, 263 Goffman, E. 204, 263 Golinkoff, R. M. 36, 263 Goodman, J. C. 15, 251 Goodwin, C. 191, 195, 196, 197, 263 Gopnik, A. 244, 263 Gordon, D. P 196, 261 Gordon, P. 63, 263 Goswami, U. 49, 163, 283, 286 Graves, M. 43, 263 Green, J. 99, 264 Grice, H. P. 212, 264 Guenthner, S. 89, 264 Gumperz, J. J. 10, 264 Gutierrez-Clellen, V. 125, 264

 Author index

H Habib, T. 201, 264 Hadley, P. 132, 278 Hall, W. S. 48, 255 Halliday, M. A. K. 68, 88, 123, 124, 128, 212, 264 Halmari, H. 201, 264 Hamo, M. 210, 264 Handel, G. 191, 264 Harris, M. 163, 264 Harris, P. L. 210, 250, 264 Hart, B. 43, 264 Hastings, P. 54, 264 Hatano, G. 163, 264 Hawkins, J. 141, 264 Heath, S. B. 195, 264 Hecht, B. F. 236, 258 Heibeck, T. H. 36, 41, 265 Hemphill, L. 197, 265 Henry, L. A. 46, 270 Herman, P. A. 43, 45, 55, 271 Hernández Pina, F. 242, 265 Hickmann, M. 11, 12, 15, 32, 59, 96, 104, 107, 144, 200, 235, 237, 247, 265 Hicks, D. 84, 199, 200, 265, 286 Hidalgo, R. 93, 265 Hoff, E. 38, 43, 265 Holes, C. 80, 265 Holt, E. 204, 265 Hopper, P. J. 24, 78, 93, 265, 284 Hoyle, M. S. 197, 201, 209, 265 Hudson, J. A. 84, 240, 265 Hulme, C. 167, 271 Hunt, K. 133, 265 Hupet, M. 172, 261, 265 Hutchinson, J. E. 36, 270 Huttenlocher, J. 6, 131, 132, 265

I Indefrey, P. 132, 265 Inhelder, B. 229, 265 Iser, W. 213, 220, 266

J Jackendoff, R. 58, 266 Jakobson, R. 213, 220, 266 James, W. 187, 266 Jeanjean, C. 140, 141, 266 Jenkins, J. R. 41, 266 Jerger, S. 42, 266 Jisa, H. 3, 5, 7, 10, 25, 26, 34, 63, 90, 105, 124, 135, 145, 146, 152, 154, 160, 179, 214, 247, 262, 266, 277 Johansson, V. 178, 266, 283, 284 Johnson, C. J. 41, 194, 266, 279 Johnson, M. H. 34, 261, 266 Johnston, J. 122, 125, 233, 234, 263, 266 Jones, M. 153, 154, 155, 266 Jucker, A. H. 59, 266 Jusczyk, P. 32, 266

K Kail, M. 85, 266 Kail, R. 50, 266 Kameenui, E. J. 40, 257, 266 Kaminska, Z. 175, 260 Kaplan, B. 157, 286 Karlsson, H. 179, 283 Karmiloff-Smith, A. 12, 13, 41, 57, 85, 225, 233, 234, 240, 245, 261, 267 Katriel, T. 196, 267 Katzenberger, I. 10, 11, 143, 156, 253 Kaye, K. 193, 236, 267 Keenan, E. L. 25, 34, 63, 86, 267 Keil, F. C. 41, 47, 236, 267 Kelly, D. J. 49, 267 Kemp, N. 170, 175, 267 Khorounjaia, E. 3, 83, 119, 124, 156, 179, 214, 247, 267 King, G. 119, 120, 267 Kluwe, R. 57, 267 Knoblauch, H. 89, 264 Koenig, J-P. 152, 154, 155, 267 Koven, M. 204, 206, 267 Kruger, A. C. 38, 284

Author index 

Kuhn, D. 220, 228, 267, 268 Kuiken, D. 213, 220, 270 Kumpf, L. E. 90, 91, 268 Kuno, S. 153, 268 Kyratzis, A. 195, 268 L Laberge, S. 154, 268 Labov, W. 192, 197, 207, 268 Lambrecht, K. 137, 140, 141, 142, 143, 144, 153, 268 Langer, J. A. 222, 268 Lapidot, S. 65, 66, 75, 76, 268 Largy, P. 172, 259, 261, 268 Lee, E. 226, 235, 239, 268 Leech, G. 199, 202, 268 Lennox, C. 170, 269 Leonard, L. B. 50, 113, 114, 120, 133, 251, 261, 266, 269 Leung, C. 43, 269 Levelt, W. J. M. 143, 150, 255, 269 Levie, R. 75, 269, 277 Levin, I. 73, 74, 75, 269 Levinson, S. C. 10, 256, 264 Levorato, M. C. 238, 257, 269 Levy, E. 40, 269 Light, P. 55, 57, 269 Lillo-Martin, D. 233, 259 Lindenmayer, D. B. 3, 269 Littleton, K. 55, 57, 269 Loban, W. 3, 111, 125, 126, 132, 269 Logan, G. D. 57, 267, 286 Longacre, R. E. 90, 106, 269 Looft, W. R. 41, 282 López-Ornat, S. 234, 242, 269 Lucy, J. A. 199, 269 Lynch-Brown, C. 44, 256 Lyons, C. 152, 269 Lyons, J. 241, 269 Lytinen, S. 54, 264 M MacWhinney, B. 13, 27, 33, 34, 151, 251, 255, 270

Mahony, D. 55, 58, 270 Malmsten, L. 179, 283 Mann, V. 55, 270 Marchman, V. 117, 270 Markman, E. M. 36, 41, 76, 265, 270 Martin, J. R. 208, 270 Martin, L. 92, 261 Mathews, R. C. 176, 270 McCabe, A. 195, 274 McCartney, K. 43, 270 McCathren, R. B. 43, 270 McClelland, J. 57, 279 McGregor, K. K. 40, 270 McKeown, M. G. 40, 41, 251, 270 McNeill, D. 233, 270 McTear, M. 194, 196, 270 Meltzoff, A. N. 244, 263 Messer, D. 2, 5, 8, 35, 41, 50, 56, 118, 236, 260, 270, 275 Metsala, J. L. 46, 270 Miall, D. 213, 220, 270 Michas, I. 46, 270 Miller, C.R. 89, 269, 271 Miller, G. A. 37, 271 Miller, J. 124, 271 Miller, P 195, 270 Moats, L. 163, 271 Morani, R. M. 240, 275 Morris, N. 125, 271 Moses, L. J. 42, 250

N Nagy, W. E. 37, 41, 43, 45, 53, 55, 77, 271, 280 Naharro, M. A. 242, 271 Naigles, L. 38, 43, 265 Nation, K. 50, 51, 167, 271 Nelson, K. 36, 39, 40, 47, 269, 271 Newkirk, T. 229, 230, 271 Nicholls, L. 119, 120, 279 Nicholson, G. 225, 271 Nicolopoulou, A. 199, 271 Ninio, A. 191, 192, 195, 201, 237, 272

 Author index

Nippold, M. A. 1, 2, 3, 4, 5, 7, 10, 15, 37, 55, 56, 58, 77, 111, 247, 272 Nir, M. 17, 74, 277 Nir-Sagiv, B. 10, 26, 28, 33, 60, 84, 89, 90, 98, 106, 127, 240, 253 Norbury, C. 118, 272 Nordqvist, A. 178, 183, 190, 272, 283 Nunes, T. 163, 171, 257, 272 O Ochs, E. 58, 195, 207, 208, 241, 272 O’Dowd, E. M. 90, 272 Olson, D. R. 2, 3, 5, 7, 57, 201, 210, 211, 212, 220, 225, 228, 235, 238, 250, 268, 273, 274, 278, 279, 284 Opper, S. 57, 263 P Pacton, S. 4, 6, 139, 163, 165, 167, 168, 169, 170, 172, 174, 273, 274 Paltridge, B. 57, 273 Parnas, D. L. 55, 273 Pascual-Leone, J. 11, 273 Pearson, B. Z. 43, 273 Pellegrini, A. D. 192, 195, 199, 273, 274 Pelli, T. 178, 274 Penno, J. F. 44, 274 Perdue, C. 24, 274 Perera, K. 128, 274 Pérez-Leroux, A. T. 242, 274 Perfetti, C. 163, 274 Perkins, D. 220, 284 Perner, J. 178, 236, 247, 274, 286 Perruchet, P. 168, 169, 273, 274 Peskin, J. 2, 3, 5, 7, 211, 214, 219, 226, 238, 274 Peterson, C. 195, 274 Piaget, J. 7, 30, 229, 265, 274 Piatelli-Palmarini, M. 234, 275 Pickering, M. J. 150, 256, 275 Pierce, A. E. 63, 275 Pikulski, J. J. 43, 269 Pilon, R. 239, 280

Pine, K. 41, 275 Pinker, S. 56, 63, 275 Plum, G. A. 208, 209, 270 Plunkett, K. 35, 261, 275 Pollio, H. 247, 275 Pollock, J. Y. 239, 275 Pontecorvo, C. 240, 275 Powell, J. S. 45, 282 Preece, A. 195, 275 Putter-Katz, H. 71, 72, 275 Pye, C. 63, 275 Q Quixtan Paz, P. 63, 275 R Rabukhin, E. 60, 74, 79, 275 Ragnarsdóttir 89, 90, 254, 275, 283 Ralli, A. 43, 44, 52, 260, 275 Ravid, D. 2, 3, 4, 5, 7, 10, 12, 13, 16, 17, 19, 20, 22, 24, 28, 33, 41, 45, 53, 55, 56, 57, 58, 59, 60, 61, 63–65, 67–79, 88–90, 102, 106, 107, 112, 123, 124, 128, 136, 141, 142, 144, 149, 150, 160, 163, 164, 174, 178, 212, 236, 253, 263, 269, 275, 276, 277, 280, 283 Redington, M. 169, 277 Reichler-Béguelin, M-J. 142, 254 Reilly, J. S. 10, 89, 90, 106, 160, 240, 241, 266, 277 Reinhart, T. 153, 277 Renouf, A. 56, 58, 250 Rice, M. 36, 48, 49, 114, 118, 120, 121, 128, 131, 132, 277, 278 Richner, E. S. 199, 271 Risley, T. 43, 264 Rispoli, M. 132, 278 Riviere, A. 239, 282 Rizzo, T. 196, 259 Robinson, E. 225, 278 Rogers, M. A. 46, 282 Rogoff, B. 191, 278 Rollins, P. 237, 272, 278 Romaine, S. 197, 278

Author index 

Rosado, E. 25, 63, 105, 152, 160, 249, 266, 277, 284 Rosenblatt, L. M. 224, 278 Roth, F. 122, 125, 259, 278 Rubinstein, J. S. 57, 278 Rudel, R. G. 42, 278 Ruffman, T. 225, 279 Rumelhart, D. 57, 279

S Saban, R. 64, 65, 279 Sabbagh, M.A. 38, 279 Sacks, H. 192, 279 Sánchez-López, I. 85, 266 Sandbank, A. 84, 240, 279 Sandra, D. 172, 279 Sankoff, G. 154, 268 Santrock, J. W. 7, 279 Sawyer, K. 196, 199, 200, 201, 279 Scarborough, H. S. 49, 279 Scardamalia, M. 185, 222, 251, 279 Schecter, B. 48, 279 Scheepers, C. 150, 259, 279 Schiffrin, D. 59, 279 Schober-Peterson, D. 194, 279 Schuele, C. M. 119, 120, 279 Schwarzwald, O. R. 60, 65, 67, 279 Scott, C. M. 3, 4, 5, 6, 7, 8, 10, 56, 111, 112, 113, 116, 120, 121, 122, 123, 125, 126, 127, 128, 129, 130, 133, 279, 280, 286 Scott, J. A. 43, 271, 280 Sénéchal, M. 168, 280 Seroussi, B. 67, 68, 69, 71, 73, 78, 79, 280 Shalom, T. 57, 280 Shapiro, L. R. 84, 240, 265 Shatz, M. 11, 15, 27, 280 Shibatani, M. 63, 280 Shimron, J. 22, 33, 280 Shlesinger, Y. 58, 59, 74, 76, 79, 277, 280 Shultz, T. R. 239, 280 Shuman, A. 195, 196, 280

Siegal, M. 236, 243, 280 Siegel, L. S. 170, 269 Simms, R. 125, 281 Simonin, J. 154, 281 Singer, B. 125, 281 Singson, M. 55, 270 Sinimäki, S. 178, 274 Slobin, D. I. 9, 12, 26, 27, 28, 29, 32, 33, 56, 67, 96, 97, 107, 109, 112, 128, 145, 146, 150, 151, 157, 192, 195, 212, 234, 253, 281 Slonim, N. 45, 254 Smith, M. L. 58, 80, 150, 201, 281 Smith, W. 201, 264 Snow, C. E. 41, 43, 77, 191, 192, 193, 194, 195, 196, 198, 201, 236, 255, 272, 278, 281, 285 Snowling, M. J. 49, 50, 51, 170, 271, 282 Sodian, N. B. 239, 282 Soja, N. N. 37, 282 Solheim, O. J. 178, 282 Sotillo, M. 239, 282 Spekman, N. 122, 125, 278 Spelman-Miller, K. 185, 282 Stahl, S. A. 44, 282 Stanovich, K. E. 35, 43, 259, 282 Sternberg, R. J. 45, 282 Stiassnie, N. 65, 66, 71, 72, 75, 76, 282 Storck, P. A. 41, 282 Storkel, H. L. 46, 282 Stothard, S. E. 113, 282 Strömqvist, S. 4, 7, 10, 58, 90, 142, 149, 150, 156, 160, 177, 178, 179, 182, 188, 190, 212, 247, 249, 254, 282, 283, 286 Straus, L. G. 6, 282 Strauss, S. 15, 282 Sullivan, K. 178, 251 Svensson, C. 227, 283 Swan, D. 49, 283 Swanborn, M. 45, 283

 Author index

T Tabors, O. P. 195, 260 Tallal, P. 131, 262, 283 Talmy, L. 55, 283 Tanaka, H. 192, 283 Tannen, D. 197, 199, 203, 212, 283 Taylor, C. 195, 272 Thévenin, M. G. 172, 284 Thompson, S. A. 24, 29, 67, 78, 85, 93, 259, 265, 284 Thorell, M. 196, 200, 201, 250 Timberlake, A. 241, 258 Tishman, S. 220, 284 Tolchinksy, L. 3, 7, 12, 25, 33, 45, 55, 57, 63, 78, 83, 91, 97, 105, 108, 112, 119, 124, 128, 136, 142, 143, 144, 150, 152, 156, 160, 174, 178, 179, 212, 214, 233, 236, 247, 249, 267, 277,284 Tomasello, M. 13, 33, 38, 63, 284 Tomblin, J. B. 112, 131, 257, 284 Torrance, N. G. 225, 268, 279, 284 Totereau, C. 170, 171, 172, 261, 284 Treiman, R. 163, 165, 167, 168, 170, 257, 285 U Uppstad, P. H.

178, 282

V van Haaften, T. 67, 286 van Hell, J. 90, 160, 275, 277, 285 Van Orden, G. C. 176, 255 Van Valin, R. D. 145, 262 Vedeler, L. 199, 285 Velmans, M. 57, 285 Verhoeven, L. 10, 58, 80, 92, 97, 109, 111, 124, 127, 128, 156, 178, 179, 212, 254, 266, 285 Véronis, J. 164, 285 Viguié, A. 54, 262, 266, 275, 285 Vinay, J-P. 152, 153, 285 Vinter, A. 169, 274 Vogel, S. 117, 285 Vosniadou, S. 238, 285

W Wade, E. 202, 285 Waletzky, J. 192, 268 Watson, R. 41, 285 Waxman, S. R. 37, 40, 54, 255, 270, 285 Weissenborn, J. 10, 33, 285 Weizman, E. 199, 285 Weizman, Z. O. 43, 285 Wellman, H. 244, 285 Wells, G. 192, 195, 285 Wengelin, Å. 4, 7, 10, 156, 177, 178, 182, 183, 185, 247, 283, 285, 286 Werner, H. 157, 286 Wexler, B. 227, 286 Wexler, K. 63, 114, 255, 278 Wheeldon, L. 150, 281 Whittlesea, B. W. A. 166, 286 Wiktorsson, M. 58, 178, 286 Williams, B. R. 57, 286 Williams, E. 54, 59, 108, 260, 286 Wilson, B. A. 175, 250 Wimmer, H. 163, 178, 247, 286 Windfuhr, K. L. 49, 286 Windsor, J. 10, 56, 112, 114, 118, 119, 120, 121, 122, 125, 126, 280, 286 Winner, E. 238, 286 Wolf, D. 199, 200, 286 Wong, A. 38, 258 Wood, C. 35, 275 Woodsmall, L. 48, 278

Y Yagev, I.

63, 277

Z Zilberbuch, S. 20, 56, 60, 73, 74, 277 Zilberman, L. 74, 286 Ziv, Y. 59, 266 Zubizarreta, M. L. 67, 286 Zucchi, A. 67, 286

Subject index

A abstract vocabulary (see lexicon) academic achievement 55, 113, 131–132, 135–137 access (see lexical access) accessibility theory (see also pronouns) 83, 88, 94, 102 accusative (see case-marking) acquisition early 15, 21, 24, 36, 41, 233 late (see development) activation 55, 83, 88, 90, 106–107 active (see voice) adjectives (see also open class) 16–26, 60–63, 73–80, 172–173, 175 denominal, derived 17, 19–23, 73, 75–77, 79 adolescence (see age) adult (see age) adverbs, adverbial 3–4, 27–28, 125–127, 168, 175, 204, 241–242 adverbial clause (see clause) affect 59–60, 241 affix(ation) 22, 59–61, 63, 80, 117 33, 168, 175 prefix (prefixation) 60 suffix (suffixation) 17, 19–23, 29, 48, 60–62, 73–74, 117, 165, 168–169, 174–175 age, age-groups (see also grades) 91, 98, 107, 156–159, 178, 181–184, 187–189, 237 adolescents (high schoolers, 15to 17-year-olds) 20, 25, 57, 62, 64, 68, 70, 73, 76–80, 84,

90–92, 97, 100, 104, 109–112, 124–125, 135, 143–144, 147, 150, 156–158, 189, 194, 196, 209, 211, 214–230, 237–240 adult(hood) 10, 25, 28, 38, 54–56, 64–69, 70, 75–80, 84, 88, 92, 97, 100–104, 109, 112–113, 117–118, 135–138, 143, 146–151, 156–158, 167, 170–175, 178, 184–185, 188–189, 193–195, 198, 200, 207, 211, 214, 226–230, 237, 240–241 gradeschoolers (schoolchildren, 6- to 10-year-olds) 11, 16, 19–20, 23–29, 35, 42–44, 48, 49, 52, 53, 55–57, 62–63, 67–69, 74, 76, 80, 111–113, 118, 116–125, 130–133, 188, 193–194, 197, 199–201, 225–228, 233–238 preadolescents (junior high schoolers, 11- to 14-year-olds) 24–25, 42, 46, 92, 108, 109, 116–117, 121–122, 125, 127, 156–159, 169, 172, 189, 224, 227, 238–239 preschoolers (3-, 4-, 5-year-olds) 10–11, 15–16, 18–20, 22–25, 28–30, 32, 37, 42–47, 49, 55–56, 64, 68, 74–75, 80, 111–114, 119, 131, 188, 194, 196–198, 200, 207, 210,

 Subject index

225–227, 233–234, 236, 239–246 age-matched 51, 71–72, 75, 77, 117–118, 120, 122, 125 agent (agency, agentivity) 25–26, 70, 78, 90, 89, 152–155, 200 agent downgrading (demoting) 15, 25–26, 63, 152, 154–160 (see also voice) agent orientation 90, 106 agreement, grammatical 13, 19, 31, 86, 93, 114, 138–140, 171–173, 236 alternative worlds 220, 237, 241–243 ambiguity 211, 222, 225, 239 ambiguous constructions 143, 159, 239 lexical ambiguity 239 anaphoric pronouns (see pronouns) animacy 37, 89, 91–92, 94, 96, 99, 100 APD (see Auditory Perception Deficit) Arabic 80 argument (structure) (see also object, subject) 31, 83–92, 95–106, 108, 116, 119, 141, 154, 157 core arguments 85–87, 92, 95, 105 articles (see also definiteness) 115–116 definite 24, 95, 107, 120, 155–158, 237 indefinite 107 aspect (see tense-aspect) assessment (see lexicon) Auditory Perception Deficit (APD) 65, 66, 71, 72, 75, 76 auxiliary (auxiliaries) 63, 140–141 B binyan (Hebrew verb patterns) 17, 20, 56, 62–63, 65–68, 78 bootstrapping 15, 30 brain 132, 243

C Canadian French 154 case-marking 24, 29 accusative 23–25, 29, 78 dative 93, 94 nominative 86 oblique 92, 93, 154 genitive 29, 73 Catalan 237 causativity (see also inchoativity) 17, 24, 61–62 Chomsky (Chomskian paradigm) 10, 30, 58, 67, 112, 130, 233–234 clause adverbial 125–127, 130, 238 clause combining (see also connectivity) 15, 30, 108, 113, 123–128, 130–131, 143–147, 238 clause construction 24, 89, 242 clause length 160 complement (see complementation) coordinated (see coordination) relative clauses 107, 119, 124, 127, 138, 141, 238, 242 simple clauses 24, 29, 85, 128, 131 subordinate (see subordination) clitic (see also pronouns) 93–94, 102, 146, 152–153, 155 closed-class items 59, 113, 115, 117 coda (see narrative) cognition, cognitive abilities 57, 144 cost (demands, effort, load) 10–11, 41, 150, 178, 185, 188–189, 220–221, 228 development (see development) constraints 15, 178, 212 resources 83, 177, 181–182, 185, 188–189 system 48, 234

Subject index 

coherence 12, 58–59, 194, 199, 201, 206–207, 217–218, 238 cohesion 30, 59, 146, 192, 194, 197, 207, 241, 243 colloquial usage 12, 26, 28, 59, 154, 160, 202, 236 comprehension tasks 40, 49, 50 132 communication, communicative acts, activities 83 234 236 competence 135, 191 context, situation 14–15, 32, 57, 78, 83, 108, 135–136, 143, 154, 235, 243–246, goals, intentions 137, 151, 239, 56, 240 oral (see speech) written (see writing) complementation 95, 119–120, 126, 129, 238, 242–244 compounds (compounding) 41, 48, 68, 74, 79 conjugation (see binyan) conjunction (see connectives) connectives, connectivity (see also clause) 29–30, 115, 123–124, 126–128, 142–147, 159, 194, 200, 235, 238 content words (see open class) context, contextual 26, 31, 42, 45, 51, 55, 67, 97, 193, 208, 227–229, 246 communicative 9, 14, 32, 57, 83, 136, 243 context sensitivity 15, 31 decontextualization 192, 195, 234, 239 extra-linguistic 15, 38, 51, 234 linguistic 43, 67, 231, 238 conversation(al) (see also dialogue) 16, 30, 88, 90, 115, 119, 135, 141, 145, 192–197, 201–203, 206–209, 213, 236, 240, 243–246 coordination (coordinate clauses) 85, 117, 119, 120, 123, 126–129, 237–238

copula 90–92, 97, 98, 105, 106, 114 cross-linguistic perspective 10, 25, 97, 124, 156, 158, 195, 241, 246

D dative (see case-marking) definiteness (see articles) derivational (see morphology) development (see also acquisition) cognitive development 35, 54, 56–57, 80, 157, 159, 191, 229, 230, 244 developmental phases (stages, steps) 12–14, 30–33 developmental path (route) 31, 79, 193–195, 209, 246 developmental patterns (processes) 21, 136, 178, 195, 225 developmental language impairment (see language impairment) dialogue, dialogicity (see also conversation) 90, 192–193, 199–201, 208, 237, 240, 192, 207, 209 diminutives 4, 62, 165, 169, 174, 170 direct speech (see speech representation) disabilities (see language impairment) discourse (see also conversation, expository, narrative) 14, 26, 83–94, 99, 104, 108, 188–190, 192, 195–202, 207–209, 234 constraints 32, 83, 117 context 11, 112, 115, 151, 154 functions 9, 12, 15, 25, 26 discourse markers 59, 240 discourse stance 35, 63, 77–79, 151–153, 156–157, 204, 207, 209, 241 discourse style 55, 78, 150, 160, 178, 212

 Subject index

monologic 28, 97, 121–123, 128, 135, 137, 143, 151, 156, 159, 211 disorders (see disabilities) Dutch 25, 32, 163, 172 dyslexia (see language impairments) E education 43, 54, 108, 113, 132–135, 176, 230, 246 ellipsis 3, 94, 145, 146, 160, 194, 207, 240 English 16–19, 21–33, 56–59, 62–63, 67, 73–74, 80–81, 85–86, 107–109, 113–119, 124, 140, 149, 153–155, 163–164, 167, 170, 175, 177, 214, 219, 227, 237, 241, 243, 247 EOI (see infinitive) 114, 118, 133 epistemic mode 241–242 errors 14, 18, 20, 40, 49, 50, 65, 71, 113, 115–123, 164, 171–173, 175 ethnography 195–198, 209 existentials 98 experiments (see methodology) expository discourse 10, 16, 26–29, 33, 58, 68, 74, 77–79, 84, 88–92, 97–109, 114–116, 120–127, 133–137, 142–143, 147–149, 151–161, 179, 213, 230, 237–240 eye-tracking 188 F familiarity 47, 55, 176, 192, 199, 208, 238 figurative language 227, 236, 238–239, 246 finiteness (see tense) first grade (see grades) flow (see information) foreground 152 63, 152 196 form-function 9, 12, 15, 25–26, 41, 59, 78, 84, 86, 120, 123, 131, 138, 152–158, 195–200, 209, 234–235

form-meaning mappings 10, 14, 17, 19, 21–22 France 154, 165 French 25, 85, 107, 109, 179, 237 frequency 37, 42, 55, 58, 63, 74, 97–98, 100–101, 111, 115–118, 123, 125–127, 130, 142, 144, 147, 156–159, 165–171, 172–175, 185–189, 238 frog story 147, 150 function words (see closed class items) future (see tense) G gender 13, 19, 31, 34, 73, 93, 195, 197, 198, 236 61, 81, 109, 138–139 generic (see reference) genitive (see case-marking) genre 14, 57–58, 68, 78–79, 84, 88–108, 113, 121–130, 133, 135, 151, 211–213, 224–227, 238–240 German 28, 30, 50, 108, 163, 237 Germanic 17, 22, 29 gerund(ive) 27, 29, 33, 94, 168 27–29, 67 grade (see age) grammar 9, 12, 13, 15, 30–34, 53, 84–86, 90–92, 95, 101, 105–108, 111–116, 118–122, 128, 130–132, 137, 145–146, 150–154, 171–173, 176–178, 201, 217, 233, 235, 242, 247 generative 138, 202, 210 grammatical relations (see arguments, object, subject) grammaticization 14, 31 graphemes 49, 164, 167, 174 graphotactics 169–175 H Hebrew 12–17, 19–34, 53, 56, 58–63, 67–68, 73–74, 78–86, 90–91, 106–107, 109, 149, 152, 163, 202, 205

Subject index 

high school (see age) homophones 170–173 I idioms 48, 58, 238–239, 246 idiosyncracy 60, 67, 68, 79, 164 imageability 38, 59 imitation 14, 194, 240 impersonal 26, 78, 85, 90–92, 97, 105, 106, 153, 154 implicit (see learning) inchoative (see also causative) 17, 61, 67 indefiniteness 95, 102, 103, 107, 140, 154–156, 158, 237 infinitives (see verbs) 16, 27–29, 30–33, 94, 116, 140, 153 EOI (Extended Optional Infinitive) 114, 118, 133 inflection (see morphology) information, flow of 68, 83–89, 92, 142, 177, 189, 247 information packaging 142, 143, 149, 151 input 37, 54, 56 instruction (see teaching) intonation 130, 143, 160, 201, 208, 212, 240 intransitivity (see also transitivity) 17, 85–89, 92, 93, 97–102, 105, 108 Inuktitut 91 irony 235, 238, 239 Italian 85, 196 J jokes 239, 246 journalese 23, 26, 74, 124 juxtaposition 123, 237–238 K keyboarding 119, 178 kindergarten (see age) Korean 32, 91

L language acquisition (see acquisition) language development (see development) language impairment (difficulties, disorders) 19, 49, 73, 112–118, 120–128, 131–132, 133n, 203–205, 243 Language Learning Disabilities (LLD) 71–72, 76, 113–116, 243, 131–132 reading and writing disorders 113–114, 131 Specific Language Impairment (SLI) 49, 114, 128, 133 130 20, 49, 51, 65–66, 71–72, 75–77, 114–115, 117–120, 133 language-matched 71, 72, 77, 115, 118, 120, 125–128, 130 language-specific 23, 24, 37, 163 Latinate (see Romance) learning 14, 15, 30, 35–39, 42–46, 48, 51–55, 64, 68, 70–73, 79–80, 111–118, 122–125, 130–133, 136–137, 144, 149, 159–163, 169–178, 189–190, 193, 209, 230, 233–236 explicit 46, 165, 168, 173–176 implicit 165, 168–169, 173–176 learning disabilities (see language impairment) level of usage (see register) levels of schooling (see also grades) 80, 143, 147, 177 lexicon, lexical 11–12, 16–17, 35, 45, 48–49, 53–57, 62, 67, 80, 102, 107–108, 115, 132, 142, 160, 164–166, 178, 201–202, 230, 235, 236, 238, 244, 246–247 access 42 ambiguity (see ambiguity) assessment 35, 40–45, 49, 52 convention 18, 53, 73, 76, 166

 Subject index

core lexicon 56, 58 entries 41, 49 lexical-decision tasks 54 mental lexicon 53–59, 142, 239, 243–244, 246–247 regularities 164–165 semantics 35–41, 43–51, 63, 76, 89, 99–100, 236–238, 243–244 lexicalization 59, 74–78, 105–107, 130 LI (see language impairment) light subject constraint 88, 142 linearity 20, 22, 23, 61, 62, 73, 143, 149, 179, 180 linguistic universals 31–32 literacy 12, 14, 20, 26, 29, 33, 35, 49, 52–60, 67–68, 73, 80, 107, 108, 112, 123–124, 128, 131, 135–136, 143, 150, 160–161, 177, 178, 192, 195, 196, 202, 207, 209, 211, 228, 229–230, 235–236, 239, 243, 245, 247 discursive 192, 195, 196, 202, 207, 209 linguistic 12, 55–57, 112, 128 literate language 123–124 literate lexicon 53–58, 73, 80 literary prose 74, 212 LLD (Language Learning Disability) (see language impairment) longitudinal studies 16, 43, 74, 111, 114, 194, 197, 199 M Mandarin Chinese 237 markedness 17, 18, 20, 22, 24, 32, 39, 58, 68, 86, 93, 94, 104–106, 109, 112, 115, 116, 118, 126, 138, 139, 143, 149, 168, 173, 180, 182, 200, 202, 206, 210, 238, 242 masculine (see gender) matching (see age-matched, language-matched) maturation 12, 13, 56

maturity 9, 14, 15, 28, 37, 57–59, 73, 78, 142, 151, 246 mean length of utterance 49, 132 meaning, conventional 226, 236–238 figurative (see figurative language) literal 227, 228, 239, 243 memory 46, 50, 55, 57–58, 107, 147, 165, 168, 173, 212, 229 long-term 83, 90 phonological 46, 50 mental lexicon (see lexicon) metalinguistic awareness 41–42, 53, 57, 113, 132 metaphor (see also figurative) 214, 217–219, 227, 235, 238–239, 246 methodology 45, 62, 150, 164–166 structured elicitation 16, 18, 20, 24–28 middle (see voice) mind (see Theory of Mind) MLU (see mean length of utterance) modals (see verbs) modality (see also speech, writing) 14, 32, 58, 89, 106, 107, 113, 121–128, 130–138, 142–144, 146–147, 156–161, 235, 238, 240–241 monologue (see discourse) mood 34, 238, 242 morphemes 58–61, 112–117, 121, 183, 247 morpho-phonology 62, 73, 78 74 morpho-syntax (morphosyntax) 10, 14, 24, 31, 35, 52, 56, 58–59, 113, 115, 117–121, 123, 128, 164–168, 175 morphology 12–13, 16–19, 24–25, 29, 34, 41, 45, 48, 164, 169, 173–175 derivational 13, 16, 18, 41, 48, 53, 58, 60, 65, 80, 167, 169 inflectional 12, 13, 19, 31, 41, 48, 53, 73, 78–79, 85,

Subject index 

113–114, 117, 140–141, 165, 170–173, 175, 200, 202, 236, 242 N naming 11, 42, 47, 49–52 narrative (discourse) (see also genre) 9–11, 13, 26, 28, 32, 33, 44, 49, 58, 67, 78, 86, 88–92, 96–109, 112, 114–117, 121–122, 125–128, 133, 135, 141–156, 177, 180, 184–189, 195–196, 199–204, 206–210, 222, 235–238, 246–247 coda 143, 208 conversational 195, 202, 203 personal experience 16, 79, 84, 96, 160, 100, 105, 151 schema 188 naturalistic language 121, 132 neural activation 132 new information 55, 83, 86–89, 91, 95–100, 102–103, 105–108, 130, 237 nominalization 13–15, 27–32, 67–68, 70, 73, 123, 130, 141–142 nominative (see case) non-finite subordination (see subordination) non-literal uses of language 235, 243, 246 nonfinite 3, 6, 25, 27–29, 92, 97, 98 105, 116, 120, 125, 126, 150 nonfinite complement clause (see clause) nonfinite verb 116, 120 nouns (see also open class) 11, 13, 17, 19=22, 25, 34, 38, 41, 58–62, 67–68, 71–76, 70–81, 85–88, 91–96, 102–104, 107, 109, 114–115, 118, 120, 125, 127, 130–131, 138, 141–142, 144–145, 164–165, 170–175 action nominal 29, 67 bare nouns 95

derived nominals 28, 29, 61–62, 66–68, 70–73, 76–79 noun phrase 13, 24–25, 63, 68, 79, 85, 88–89, 93, 95–96, 102–103, 107, 109, 116, 120, 131 novel word 37, 46 49, 117 NP (see noun phrase) null subject (see subject) O object (see also argument) direct 24–27, 63, 78, 119, 138–140, 153 indirect 84, 119 oblique (see case-marking) online (see processing) online comprehension studies 246 open-class 58, 60 (see also adjectives, nouns, verbs) oral language (see speech, see also modality) orthography 60, 163–166, 172, 175 P packaging (see connectivity) participle, participial 16–25, 28, 63, 116, 120, 133, 138–141, 168, 175 passive participles 16–19, 22–23 past participle 17, 116, 120, 138–139, 141 particles 237 PAS (see Preferred Argument Structure) Passive (see voice) past tense (see also tense) 61, 81, 115–117, 121, 133, 141, 155, 170, 210 past perfect 3, 56, 112 passé composé 141, 155 passé simple 141, 146 patient (semantic role) 26, 78, 152, 154

 Subject index

pause 177–180, 185–189, 210 peer talk 191, 196–199 periphrasis 60, 63, 138, 140 personal pronouns (see pronouns) phases (see development) phonemes 49, 164, 174, 247 phonology 10, 35, 39–40, 42, 45–46, 48–52, 54–55, 71, 93, 115, 117–118, 136, 164–166, 170, 174, 239, 247 Piaget 30, 229 picture naming 42, 49, 50 plural 19, 93, 115, 116, 138, 139, 154, 155, 159, 164, 165, 170–175 poetry 148, 211–222, 224–225, 228–230, 235–239 possessive 29, 60, 95, 107, 114 post-verbal positions 85, 94, 96, 99, 100, 106–107, 144 pragmatic competence 192, 193–195, 207, 210 preadolescent (see age) pre-subject position 26 pre-verbal position 85, 94, 96, 99, 100, 104, 141, 146 predicate 26, 31, 84, 85, 101, 152, 155 31, 34, 84, 89, 92, 102, 105, 108 123, 242 preferred argument structure (PAS) 87–89, 90–92, 95, 104–105 prefix(ation) (see affixation) preposition 11, 23–25, 33, 95, 107, 114–116, 120, 153, 168 prepositional phrases 11, 107 preschool (see age) present (see tense) priming 54, 132, 150 pro-drop 85, 94 processing 11, 15, 23, 41, 55–57, 67, 83, 87, 112, 118, 128, 131–132, 171, 175–176, 212, 219, 245 on-line 11, 15, 42, 137, 142, 144, 178, 189, 212, 246

productivity 17–20, 22–26, 28, 33–34, 39, 48, 73, 77, 156, 236 proficiency 9–10, 56, 186 pronominalization (see also pronouns) 109, 141 pronominals (see also pronouns) 93, 116, 120, 141, 153 pronouns (see also accessibility theory) 11, 29, 85–91, 93–97, 102, 104, 106–109, 115–116, 119, 138, 140–146, 152, 153, 160, 194, 200, 237 anaphoric (see also reference) 145, 160, 194, 237 clitic pronouns (see also clitics) 96, 104, 138, 141, 144–145 ellipsis 145–146, 160 personal 87, 90, 93–94, 109 relative pronouns 94, 144–145 propositional content 128, 192 propositions 136, 143, 146, 151, 218, 243 prosody 142–143, 194, 200–201, 206 protagonist 11, 89–91, 100, 102, 105, 204, 206, 209 proverbs 238, 246 pseudo-word 165–167, 169, 170 psycholinguistics 30, 54, 80, 178, 195, 196, 212, 233, 246 punctuation 143, 177, 182, 188, 189

R reaction time 246 reading 33, 43–45, 49–50, 113, 114, 131–133, 135–136, 141, 147, 152, 159, 163, 173–174, 176, 179, 197, 211, 213–219, 220–225, 228–230, 244 reading achievement scores 132 reference (see also anaphoric pronouns) 11–13, 38, 85–90, 94, 101–102, 104, 106–107, 137, 144, 146, 153–155, 160

Subject index 

generic 59, 78, 85, 90, 102, 107, 109, 133, 151, 153–159, 203, 204 register 12, 29, 32, 57, 99, 135, 145, 201, 236, 245 12, 14, 22, 23, 26, 57, 61, 89, 135–138, 140, 145, 146, 160, 192, 201–203, 209–210, 236 formal 29, 146 99, 145 variation 12, 137, 140, 201 relative clause (see clause) repetition tasks reported speech 199–201 reporting (see reported speech) resultatives 16–18, 20–25, 60–61, 152, 157 riddles 239, 246 Romance 17, 29, 58, 73, 80 root 17, 23, 27, 33, 41, 45, 48, 59–62, 67 rote learning 14 34

S Sakapultec 86, 91 sarcasm 246 school-age (see age-groups) Scriptlog 179–181, 185 semantics (see lexicon) semantic relations 37, 51, 149 Semitic languages 22, 33, 61, 63, 86 SES (Socio-Economic Status) 15, 65–66, 75, 136 silent letters 168 similes 218, 246 slang 136 SLI, Specific Language Impairment (see language impairment) sociolinguistic 89, 137, 191, 195, 197, 198, 201, 209 sound (sequence) (see also phonology) 35, 46, 54, 55, 165–167, 169, 170, 194, 213, 226 sound-to-spelling correspondences 164, 167

Spanish 28–30, 83, 85–87, 91, 93, 96, 99, 101, 102, 104, 107–109, 156, 179, 237, 242, 246 speaking 9, 112, 115, 119, 123–125, 131, 133, 140, 143, 144, 149, 150, 177, 178, 197, 200, 201, 204, 212, 247 special education (see education) Specific Language Impairment, SLI (see language impairment) speech, spoken language (see also modality) 15–16, 18, 24, 26, 30, 32–34, 46, 114, 50, 56, 58, 74, 78–7, 91, 101, 106, 112, 115, 117, 119, 121, 124, 137, 139, 142–144, 146–147, 153, 156, 159, 170, 174, 177, 190194, 197–204, 206, 209, 212–213, 235, 237–240, 245 child-directed speech 26, 74, 78 speech acts 192, 194, 237, 239, 240 speech events 198–200, 209 spoken discourse 136, 142, 159 spoken French 138–139, 141–142, 144, 157–158 speech representation (see also reported speech, voicing) 191, 193, 199–203, 206, 209 spelling 118–119, 144, 139, 163–165, 168–173, 175–177, 183–189, 208 stages (see development) stance (see discourse stance) stem 17, 20, 23, 60, 62, 67, 68, 73, 75, 79, 117, 173, 244 story (see narrative) structural complexity 10, 20, 22, 23, 27, 29, 31, 48 structured elicitations (see methodology) style (see also discourse style, register) 14, 22, 26, 28, 34, 55, 77, 78, 131, 150, 160,

 Subject index

177–178, 189, 192, 197, 202, 210, 212, 222 sub-lexical regularities 164–165 subject (grammatical relation) 25–27, 29, 78, 84–101, 107, 108n, 109, 116, 119–121, 129–130, 141–147, 151–155 null subject 11, 13, 26, 85, 109, 129 subject position 83, 88, 96, 100, 107, 141, 145–46, 153, 238 subjectless constructions 24, 63, 78, 85 subjunctive 242, 244 subordination 28–30, 68, 120, 123, 125, 127, 129, 142, 145–148, 150, 160, 237, 238 nonfinite 145–150, 159–160 suffix (see affixation) SVO (see word order) Swedish 1, 4, 150, 156, 177–180, 227 syntax (see also morpho-syntax) 15, 24–27, 55, 58, 62–65, 73, 80, 83, 111–113, 124, 131, 132, 152, 160, 238, 244, 246 complex 6, 15, 73 syntactic architecture 41, 58, 61, 85 syntactic complexity 15, 25, 29, 30, 58, 73, 113, 125, 132 syntactic constructions 12, 13, 16, 32, 57–59, 79, 127, 147, 233, 247 syntactic packaging (see connectivity) T T-units 128–130, 133 target language typology (see typology) teaching, teachers 43–44, 52, 95, 135, 227 temporality 31, 33, 195 tense (see also past) 29, 31, 34, 61, 81, 113–118, 121, 133, 138,

141, 145, 155, 170, 202, 204, 210 finiteness 27, 29–30, 92, 94, 118, 120, 123–126, 129, 149, 180 future 138, 140, 202, 210 present 16, 34, 42, 47, 65, 79, 84, 86, 90–91, 97, 98, 104–105, 108–109, 113, 117, 133, 135–138, 142, 151, 155–156, 159, 163, 168, 175, 177–179, 181, 185, 188, 191, 204, 242 tense-aspect 28, 31, 113, 89, 116, 237 text (see discourse) texture 37, 58, 213, 237 Theory of Mind 178, 225–227, 243–244, 247 thinking for speaking 150 ToM (see Theory of Mind) topic 89–90, 160 topicalization 26 153–154 transcription 97, 164, 166–169, 171–175, 210, 214 transitive verbs (see verbs) transitivity (see also arguments, intransitivity) 13, 17, 24, 31–34, 62, 93, 78, 85–88, 91–93, 97–102, 105–106, 153 Turkish 28, 29 turn-taking 192–194, 207, 209, 236, 240 typology 10, 12, 15, 21, 22, 24, 29–33, 38, 49, 85, 120, 236 Tzeltal 32 U U-shaped curve 15, 32 unanalyzed amalgams 14, 17 university students (see age, adults) unmarkedness 86, 93, 96, 99, 100, 141, 210, 238 utterance 49, 83, 107, 111, 115, 117, 122, 131–132, 136, 150–151,

Subject index 

180, 200–201, 204, 210, 212, 226, 228, 235, 239–240, 242 length 111 V valence 23, 155 variation 11–15, 28, 31–32, 45, 99, 120, 124, 127, 137, 138, 140, 145, 146, 201, 210, 245 Venezuelan Spanish 91 verbs (see also open class) 17–22, 24–29, 31, 33, 38–41, 49–50, 56, 60–63, 67, 71–73, 78–81, 84–86, 92–94, 96, 98–99, 106, 108, 113–119, 121–1268, 141, 153–155, 164–167, 175, 180, 199, 202–206, 220, 242, 243 aspectual verbs 27 change-of-state verbs 67 deontic verbs 241, 242 finiteness (see tense) 114, 118 mental state 243 modals 111, 241 verb-argument structure (see also preferred argument structure) 31, 120 verb-pattern (see also binyan) 17, 22, 24, 25, 34, 56, 62 vocabulary (see lexicon) voice (see also agent downgrading) 13, 23, 25, 26, 34, 62–64, 78, 111, 116, 120, 152–153, 157, 159, 204–207, 212 active 19, 20, 23–24, 64, 68, 78, 152 adjectival passive 25, 63, 78, 152 middle 26, 152–153 syntactic (verbal) passive 15–19, 22–26, 32–33, 54, 56, 60–67, 75–78, 111, 116, 120, 133, 152–160, 236

voicing 199–200, 204, 206, 209 Vygotsky 30 W waterfall model 55 WFD (see word finding difficulties) word finding difficulties (WFD) 50–51 word frequency (see frequency) word order 29, 31–32, 85, 96, 141 word-formation (see morphology, derivational) writing (see also modality) 16–17, 20, 28, 58, 61, 79, 81, 92, 108–109, 112–113, 115, 118–119, 121–125, 127–129, 131–137, 141–144, 146–151, 153, 156, 159–160, 163, 172–175, 177–188, 226, 228, 236, 241 245, 246 notational system 144, 174, 178 written language (see also modality) 16–17, 22–26, 28, 33, 40, 45, 55–58, 68, 74, 77–80, 98, 104, 107, 111–112, 114–118, 118, 120–130, 136–144, 146–150, 155–161, 163–164, 168, 170, 174–180, 189–192, 195, 202, 212, 213, 226, 228, 235–238, 240, 245 Y young children (see also age) 11, 16, 21, 26, 30, 32, 36, 38, 48, 74, 78, 84, 91, 104, 112, 113, 193–195, 198, 201, 225, 227, 235–240, 243 Z zero marking 85, 87, 88, 93, 96, 101–104, 106, 115, 170, 171, 173

In the series Trends in Language Acquisition Research the following titles have been published thus far or are scheduled for publication: 1 2 3

CENOZ, Jasone and Fred GENESEE (eds.): Trends in Bilingual Acquisition. 2001. viii, 288 pp. MORGAN, Gary and Bencie WOLL (eds.): Directions in Sign Language Acquisition. 2002. xx, 339 pp. BERMAN, Ruth A. (ed.): Language Development across Childhood and Adolescence. 2004. xiv, 307 pp.