input processing and grammar instruction

97
1 1 1 .J 1-56750-237-7 P53 V35 CL- 238$

Upload: alejandro

Post on 22-Jan-2016

202 views

Category:

Documents


0 download

DESCRIPTION

This book provides an alternative to the grammar debate in second language acquisition theory and teaching. Accepting that language acquisition is at least partially input dependent, the author asks how grammatical form is processed in the input by second language learners and is it possible to assist this in ways that help the learner to create richer grammatical intake. He answers these questions and explains why traditional paradigms are not psycholinguistically motivated. Drawing on research from both first and second language acquisition, he outlines a model for input processing in second language acquisition that helps to account for how learners construct grammatical systems. He then uses this model to motivate processing instruction, a type of grammar instruction in which learners are engaged in making form-meaning connections during particular input activities.

TRANSCRIPT

Page 1: Input Processing and Grammar Instruction

1 1 1

·-----~-------~---------------------.J

1-56750-237-7 P53 V35

CL- 238$

Page 2: Input Processing and Grammar Instruction

INPUT PROCESSING ANO GRAMMAR INSTRUCTION IN

SECOND LANGUAGE ACQUISITION

Page 3: Input Processing and Grammar Instruction

INPUT PROCESSING ANO GRAMMAR INSTRUCTION IN

SECOND LANGUAGE ACQUISITION

Bill VanPatten The University of Illinois at Urbana-Champaign

r UOIUOO A LA BIBLIOTECA

1 '.HPllEH A. BISTl'H" POR,

.b-<. l( t) a,.,< ~

11'\\ ABLEX PUBLISIDNG CORPORATION \j-\) NORWOOD, NEW JERSEY

Page 4: Input Processing and Grammar Instruction

SECOND LANGUAGE LEARNING SERIES A Monograph Series Dedicated to Studies in Acquisition

and Principled Language lnstruction

Robert J. Di Pietro, editor

TM. Catalan, Immersicn Progrom: A European lbint ofVU!w Josep Maria Artiga]

A Deuelopmental Psydwlinguistic Approoch to Seoond Language Teaching Traute Taeschner

Reading Deuelcpment in a Second úmguage: Theoretú:al, Empirical, and Classroom Perspectwes ·

Elizabeth B. Bernhardt

Vygotski.im ApproocM.s to Seoond Language Resea.rch James P. Lantolf and Gabriela Appel, editors

Bilingualism and Testing: A Special Case of Bias Guadalupe Valdes and Richard Figueroa

Elizabeth B. Bernhardt, editor

Input Processing and Grammar Instructicn in Seoond Language Acquisiticn Bill VanPatten

In preparation: Listen to the Silences: Mexican American Interacti-On in the ComposiJ.icn Classroom ami CommunU:y

Kay Losey

Page 5: Input Processing and Grammar Instruction

~·o. U.e <,

li';'1 ~+ ·'

í ..

-~:- ' • ... ~ .. . ' ..

• • • M i1A•11tN

Copyncht O 1996 by Ablex PuNisbing Ccwpontioo

Ali n¡¡hta--i. No pert olúuo puN;orion .,,_,.be rtprOducod, Mored U1 o retnevaJ .,.......,., or tnrwn.u.d in any rorm ... by any - olecu-onic. medwuc:al, phccocopyin& mlao6hrung. reoordJni, or otherwu., without per­mi...on ol IN pubboho<

Pnnl«I in lho Unll«I Statoo of ~

Ubrary of Coner- Cataloging-ln-Publlcatlon Data

VanPattm, Bill. Input p~ and gnunmar instniction in 9'lCOlld langua¡¡e

ecquialtion I Bill VanPatteo. p. an - tSecood language leaming)

ISBN 1-66760-237-7 - ISBN l-56750-23S-ó tpbkJ 1. l..en¡uage and language&-Study and teedung. 2. Second 1an¡uage

ecquisition. l Title. 11. Senes. P63V364 1996 418'.007-dc20 96-3502

Ablex Publiahing Corporation 356 Chcetnut Stre<!t Norwood, NJ 07648

CIP

DEDICATION

To Jim, with !ove and gratitudc for thc last 15 years and for encoureging me to spend time in Santa Fe to work on this book.

To Lucy and Ginger (arf-ar( woof-woof-woof, añ, woof-woot; pant, pant, pant) and to Murphy (meow).

Page 6: Input Processing and Grammar Instruction

TABLE OF CONTENTS

Series F.clitor's Preface ··-··-·····--··................................................................ xi Preface....... .. . .. .... -...................................................... xü

CHAPTER l : INTRODUCTION .............................................................. 1 Wbat is Processing lnstruction? ..................................... .... ........•..•......... 2 Motivation for Proocssmg lnstruction .... _ ........ . .... .. ........................... 2

Traditional Approachcs to Grammar Instruction............................ 2 The Role of Input................ ................... ............................................ 4 Input Processing and lntake ............................................................... 6

The ÜTganization of This Book ......... ................... .. .. .. ....................... ...... 8 The Intended Audicnce for This Book.................................................... 9 Terms Used in This Book......................................................................... 9

CHAPTER 2: INPUT PROCESSING IN SECOND LANGUAGE ACQUISITION ............................................................................................... 13

Introduction................................................................................................ 13 Attention and Input Processing............................................................... 14

The Principies........................................................................................ 14 Attention ................................................................................................ 15 The Push to Gct Mcaning.............. ............... ... ........ ............................ 1 7 Meaning and Fbnn: Lexical and Grommatical Encoding ............... 21 Meaning and Fbnn: Rclative Communicative Value ....................... 23 Processing Grammatical Fbrm of Little Communicative Value ..... 27 Summary ................................................................................................ 30

Assigning Grammatical and Scmantic Roles......................................... 32 The Principies........................................................................................ 32 Meaning Once Again ...... ............................................ .. .. .. .. ............ ...... 32 Summary ........................................... _..... ...................... ........... .......... .. . 42

Other Issues in Input Processing ............................................................ 43 Consciousness and Awanm-............................................................. 43 Fbrm Versus Mearung ........... - ... • .. --··--.. --····································· 47 Operating Principies............................................................................. 47 The Role of First Language Transfer ................................................ 50 The Competition Model... ... ·-·---.. ·---.............................................. 51

Conclusion................................................................ .. .. ....... ........... ............. 53

viii

CHAPTER 3: PROCESSING INSTRUCTION ......................................... 55 Introduction................................................................................................ 55 Motivation for Processing Instruction.................................................... 56

A General Model of Second Languagll Acqwsition .......................... 56 The Problcm with Tradttional lnstruction ....................................... 56

The Na tu re of Proccssing lnstruction .................................................... 60 Overview................................................................................................. 60 Explanation ....................... ..... .. . ............ ................................................. 60 lnforming Leamers of P1ocessing Strategies.................................... 62 Structured Input................... ......................................................... 63 GuidcliMs for Structured Input Activities........................................ 67 A Samplc l.etioion .......... ......................................................................... 71

Comparison with Other Approaches....................................................... 82 Comprehension-bo.sed Approaches..................................................... 82 Input Enhancement.............................................................................. 84 ConsciOUllllCSS Rrusing.......................................................................... 84

Conclusion. ...... ......................................................................................... 86

CHAPTER 4: RESEARCH ON PROCESSING INSTRUCTION .......... 87 Introduction................................................................................................ 87 Research on Syntruc: Objcct Pronouns and Word Order in Spanish.. 88

Ovcrvicw ........ ..... .... .. ...... ...... .. .. .... .... .......... ................ ............................ 88 Motivation for the Study ..... ....................... ......... .......... ..... .. ................ 88 Subject.s and Groups............................................................................. 89 lnstruction and Trcatment Matcrials................................................ 90 Pre- and Poottests .................... ........ .. ...... .. .... ...... .. ........ ...... ................. 91 Scoring .................................................................................................. .. 93 Summary of Research Design .... ........ ........ ............ .... ............. ............ 93 Resulte .................................................................................................... 94 Discussion of the Findings ... .. .................. .. . .................... ............. ........ 96

Rcsearch on Vcrb Inllcctions: The Spanish J>ast.tense ...... .. .............. .. 97 Overview .. . ...... .......... ........ .... .. .. ..... ............... .......................................... 97 Motivation for the Study ..... ...... ............. ......... ..... ................................ 97 The Study ................. .... .. ................................... ..................................... 98 Subjects and Groups............................................................................. 98 Instructional Treatment ...................................................................... 98 Asscssment Tasks and Sconng........................................................... 99 Results .................................................................................................... 99

Discussion of the Findings ··--················· .. ···························•·············· 102 Research on Communicative Output: Scntence-level Versus Discourse-levcl Tasks ................................................................................ 102

Overview ..... -·----------··············································· .. ·· ............... 102 Motivation for the Study ...................................................................... 103 The Study ······-.. -··········--······· .. ··························································· 104

ix

Page 7: Input Processing and Grammar Instruction

SubJcct.s and Groups ............................................................................ 104 lnstructional Treatment .................................................................... 104 Ailtic6sment Tasks and Scoring ......................................................... 104

Result.s ........ ···········································-···-··-··· ..................... 107 DIOCUssiOn of the Findings ......... ........................... .. .. .... ............. 111

Rc5carch on 1...exical·Aspectual ltems: ser and eal4r m Span.ish ......... 112 Ovt'rvicw ................... ..................................... ......•........ ....... .. .. ......... ..... 112 Motivation for the Study .......... ..... ........... .......... .. ....... ........ .... .. ...... ..... 112 Thc Study ............................................................................................... 113 Subjcct.s and Groups ............................................................................. 113 lnstructional Treatment ...................................................................... 114 ABscssment Tasks and Scoring ........................................................... 114 R<isulta................................................................. . ....... 116 Discussion of the Findings ................................................................... 120

Research on Monitoring: Explanation Versus Structured Input Versua Both.. .................................................. ............ . ........ ............ 121

Overvíew ................................................................................................. 121 Motivation for the Study .................................................................... 121

The Study ... ··········································-····-·--·························-············ 122 Sub.)CICIB and GT"OUps ....................................... - ................................. 122 Instructional Treatment .................................................................... 123 Aasc88ment Tasks and Scoring ........................................................... 124 Resulta .................................................................................................... 124 Discussion ofthe Findings ................................................................... 126

Conclusion ................................................................................................... 127

CHAPTER 5: CHALLENGES AND IMPLICATIONS ............................ 129 lntroduction ................................................................................................ 129 Universal Grammar, Syntax, and Input Procesaing ............................. 130

Universal Grammar and Language Acqui.sition ............................... 131 Fu"&t Uinguage Transfer .................................................................... 142 Summary ..... ...................... ............ ..........................•.•. .. ......................... 145

Cognlti•-c Aa:ounts versus Formal Languistics ...................................... 145 Proces&ng lnstruction and Syntax ...................................................... 148

Proces;ing lnstruction and Meaningless ltems: An Example from Verb Raising ................ . ................................... 148 Proocssing lnstruction and Functional Categonos ..•......... ............. 151

Rcmaining lssues Related to Processing lru;truction .................. .. ....... 152 Conclusion ................................................................................................... 154

CHAPTER6: EPILOGUE ............................................................................ 157

REFERENCES ............................................................................................... 161 Author Indcx ................................................................................................... 169 Subjcct Jndcx ................................................................................................... 171

X

SERIES EDITOR'S PREFACE

Second Language Learning was cstablished by &bert J. Di Pietro in 1987. At that time, he described thc series in the following way:

&cond Language Lec.rnmg L8 intendcd for the publication of research in to lleCOnd lang\IBg'e aa¡uj.sition and lll8tructíon. The series is distinguisbOO from others that addnl6& Wnilar toPIC8 by two fDllJOr featuree. First of all, the vol· umea in the series are f<>CUM!d in thci.r onentabon. They m.ay be authored by °"" or more penioos, but m ali C8IM. thar findingii come from theoretiallly coherent and unified reo;eerch pro¡ecla. Secondly, they are motivated by the dri,.., to"""°'"' prinaplesofacquwbonand INtnlction ratherthan by a wish to fulJill 80Dle immediate clwroom nood.

From 1987 until the time ofhis dcath, Professor Di Pietro sought out man­uscripts that exemplified that deacriplion. I am proud to say that he includ­ed my work among his seleetions.

ln 1993, I was cballenged to try to fil) the void left by the untimely death of Bob Di Pietro. That challenge meant carrying on the important work he had begun with the series and finding manul!Cripts that I believed Bob him­self would bave chosen. I am convinced that Bob would have included Bill VanPatten's book, Input Processing ond Grnmmar Instroction in Second Language AcquisitWn, for sure. Input Proce1lsíng ond Grnmmar Instro.ctwn in &ccnd Language AcquistJton indced foeuses on the specific issue of grammar and its role in contempora.ry language mstruction. Rather than providing us with a picture of grammar as yet another skilJ area, the VanPatten book helps us to see that grammar is the oognitive glue that holds the seoond language compreben.sion and use process togetheJ: Tbis book provides data generated on extenswe m-class work and places those data against the backdrop of the researcb literaturo in second language acqui.qj. tion. Inpt.d Processing and Gromm.ar Instr1u:tion in Second Language Acquisüion is the first contribution to the series Second Language Learning under my editorship. 1 am confident that readers wiU find it to be an out.­standing contribution to the ~arch literature in applied linguistica.

- Ellzabeth B. Bernhardt Series Editor

xi

Page 8: Input Processing and Grammar Instruction

PREFACE

The intent of this book is to reVlew a mO<lel of second language input pro­CC6Sing and examine one particular approach to grammar instruction dcrived from it-namely, proces,;ing instruction, an input-based focus on fonn that stands in direct contrast to more traditional fonns of fore1gn J.an. guage grammar instruction involving output practice (dril!, question­answer, and so on). This book also aims to swrunarize the research on processing instruction and to address potential challengee to and questions obout both the model on which it is based and the instruction ítselí

Of particular importance, 1 would like to thlnk. is that a research agenda on both input processing and processing instruction baa emerged. Whether or not one buys into processing instruction and the claim.; made in this book, my m06l minimal hope is that students of classroom 1anguage acqu.isition take note of the researcb agenda. Fbr example, in Chapter 4, the reader should focus not only on the research itseli; but also on how the research has evolved, what questions arise from each study that are subsequently addressed by additional rcsearch, and how the various studics build on each othcr to fonn a cohesive body of empírica! investigntion. The student of classroom language acquisition should also note that there is more research to be ~one and that this is not only a natural part of research-based language teeching but a clesirable one as well. Thr too oft.en, in my opinion, claims are made about language teaclung Cor 1anguage learning, for that matter) with too little empírica! evidence to bock them up. We have leamed too much and come too far since the dcmise of habit formation thcory and audio-lingual methodology to aocept the lawst thing to come down the pike without ask· ing for sorne evidence.

This book arises from 10 years of thinking about nnd reflecting on sccond language acquisition and the role of input in thc development of a grammatical system. 1 am indebted to a number of colleagues and graduate students who bave listened ro me or read my work. These col· leagucs and students have asked me ro clarify issue• or have challenged ~e on various aspects of both input processing and processing instruc· uon It has been a stimulating decade for me. l hope that l have cor· rectly remembered ali of you for the following list Calphabetically arranged, by the way)-and don't wony; no one will hold you responsi-

xil

ble for any of thc ideas, good or not-so·good, in this book.

Rodney Bransdorfer Teresa Cadiemo An Chung Cheng Rick de Graaf Donna Deans Binkowski Rod Ellis Wtlliam R Glass Jan Hulstl)n Peter Jonlens Stephen Krashen

James F. Lee Patsy Lightbown Barry McLaughlin Diano Musumeci Sollo Oikkcnon Isabel Pereira Cristina San2 Richard w Schmidt Bonrue Schwartz Tracy David Terrell (in memoriam) HelmutZobl

MY thanks al.so go to the Research Board ofthe University oflllinois for its constant support of much of my empirical resoarch since 1 arrived in 1985 and to the university for thc sabbatical !cave for the fall scmester 1994 during which 1 drafted this book. My thanks aJso go to 1 van Schulman for his support ~ ~ tenure as hoad of the departmenl lt l5 not essy being an applied linguist m a departmcnt J.argely dediait.ed to btcrary and c:u1tural studies, but lvan has always had a unique vision about where language departments should he heade<L 1 hope he is enjoying lus retircmenL l would aJso hke to thank the followmg colleagues wbo have supported me when the chips were down. You are the grcatest: Evelyn Garfield, Antonino Musumeci Sandy Savignon, Linde Brocato, Elena Delgado, Susan Gonzo, Numa Mar~.

Another round of t~ goes to my studcnts from the 1995 spring semester gradunte se11Ull81' who read the manuscript of this book and pro­vided feedback that helped to make the content more accessible: Louisc N~, ~ Córdova, Joe Barcroft., Renato Rodriguez, Nuria Sagarra. Karin Millard. María José Gonz.ález, Mariaol FernAndez.

A ~ thanks goes ro EJi1 abeth Bemhardt (series editor), the anony· mous reVlewers of the manuscript of this book, and to Anne Trowbndge plus ali the folks from Ablex Publishing for making this book a reality. Your efforts are very much appreciated.

My final thanks go to James F. Lee. His namc appoars on the dedication psge, but of ali pcople to whom 1 owe gratitude to hlm 1 owe the most. I wish everyone could have the support, encouragemcnt, ab$olute confidence, and !ove of somcone spocia1 that 1 do. Life can be fragile, and I don't want to miss any opportunities to say, "1 love you."

- Bill VanPatten Champaign, Dlinois

Febniary 19, 1996 (my birthday .•. )

XIII

Page 9: Input Processing and Grammar Instruction

!1 INTRODUCTION

Sorne nught argue that IK.'(X)lld language ac:quisition is an independent nonap­plied c:bsciphne (Ga:is & Schachter, 1989, p. 4), that its gool is to explain bow sec­ond languagl.'S aro lcruned. More partirularly, the gooJ. of second language ac:quisition rcsearch tS t.o undcrstand bow a linguistic system develops in a leamer's hcad Lil<e any numberoftheoretical domains, a researcberin second language acquisition identifies a phenomenon, examines the variables involvccl, oonducts ra;earch Cor theoriz.es about the phenomenon), and eventu­ally draW!I a oonclusion about the pbenomenon. In this view, second language ooqui.sition is divorood of pcdagogical interests. Tbis is not t.o say that issues in ec<>ond ILlngu4so toochin¡¡ nro not important; inatead, pedagogy must refied. ~n understanding of (a) theory of sccond language ooqui.sition and pedagogical principies and practiccs should be firmJy grounded in themy and research.

Others would argue that second language acquisition research is only as important as lhe pedagogjcal issues that motivate it. In this view, what should be researchod aro those matters that directly affect teachers and leamers in classrooms; resee.rch and theorizing should be at the service of pedagogy and not divorced from it.

The present book ia one at:tempt t.o link second language acquisition the­ory Md research with pedagogical issues. Its focus is grammar instruction. Although other books have been written about grammar instruction from a variety of thooret.acal vicwpoints Ce.g .• Ellis, 1990; Odlin, 1994; Rutherford, 1987), this book &llns t.o deecribe a particular approach t.o grammar instruc­tion callcd processing instruction. The book also aims to motivate this approoch via lhrory and to offer research to support it. In so doing, we hope to show that second language acquisition theory and research can make sorne very direct-and hopefully welcom~ns for second Jan. guage teaching.

Page 10: Input Processing and Grammar Instruction

2 VanPatten

WHAT IS PROCESSING INSTRUCTION?

J>rocesi¡,ng tnStruetion is a type of grammar mstructlon whooo purpooe is to affect the ways m which leamers attend to input data. lt is mput-based, as opposed to output-based. is consonant with both general seoond language aa¡uisition theory and communicative language teaching. It was first described in VonPatten (1991) and has since been described, discussed and rescarchcd in a number of publications (Cadierno, 1992. 1995; Cheng, 1995; Lee & VanPatten, 1995; VanPatten, 1993; VanPattcn & Cadiemo, 1993; VanPatten & Oikkenon, 1996; VanPatten & San~. 1995). Since we will describe procC$Sing instruction in detail in Chapt.er 3, we will not go into its specific characteristics nor the specifics of the theory that underlies it in this introductory chapter. lnstead we will begin by asking what the motivation for processing instruction is. Aft.er all, grammar instruction has been around for a long time. Fbr centuries leamers have received explanations about how the second IMguage works. followed by practice Wllng grammar rules and paradlgms. Why question this approach to grammar instruction?

MOTIVATION FOR PROCESSING INSTRUCTION

lraditional Approaches to Grammar lnstruction

Over the centuries, grammar instruction has t.akcn a variety of forros depcnding on the overall goal of instruction. Under grammar-translation approaches, the goal of foreign language leaming wos to learn a language in order to read texts (i.e., the great work.s of literatura) in that language. As Richards and Rodgers (1986) describe it,

lgrammar tranalation] approaches the 1anguagl! finll. throu¡h detailed analy-111 of ltl grammar rules. ro11o ... ec1 by application of t.lua knowledge to the task of tranalatm¡ eentence& and tezta mto and out of the t.arget ~. lt hence v\ev.'l lan¡ua¡e leammg aa consisting of tittle more than memonzing rules and facta morder to understand and manipulate the morphology and ayntaX of the forei¡n lanflU88'! (p. 3).

Although certain scholar-educators challenged the grammar-translation approach in the nmeteenth and early twentieth ccnturies man effort to place emphasis on the acquisition of oral IMguage skills (e.g., F. Gouin and the famous "Gouin series," the Direct Method), grammar-t.ranslation was a dom­inant languagll teaclúng paradigm well into tho middle part of the twentietb century. A major revolution in language teaching did not oocur until the advont of audiolingual methodology. Bom of the army's intcrest in develop-

lntroduction 3

ing oral language slrilJs during World War 11, audiolingual methodology was supported by American structuraJ linguistica and the theory ofbehaviorism. Explici.t grammar instruction was eschewed under th.is methodology. lnstead, the grammatical syst.em ofthe Janguage was to be "uncovered" by the learn­er through rote practic:e of sentence petterns, memorization of dialogues, and other "oral practice." Under audiolin¡¡ual methodology, the grammatical syi;­

t.em was viewed as a set ofhabits to be intemafued through consistent prac­tice and reinforcement. It was not neccssary for learners to know what rules they were learning, but it was neccssary for them to correctly repeat, trans­form, and perfonn other manipulations on sent.ences orally as a necessary first step toward communicative ability with tho languagll.

As behaviorism and Jinguistic structuralism lost favor in the sixtics, cog­nitive code leaming began to emerge. Essentially an evolution of audiolin­gualism, cognitive oode tbeory said that for learning to happen, learners must know what they are leaming and have sorne conscious mental repre­sentation of it before practicing. Those who advanced a cognitive-<Ode approach to classroom Janguage leanung ad'-ocated that grammar explana­tion and examples should procced sentence marupulation. By the l970s, moot da&room.s in the Unit.ed $tates in volved oome version or other of oudi· otingualism or cognitive-code theory.

Interestingly. not much was actually known about the processcs and products of second language aa¡uisition before 1970. Seen as the intersec­tion of psychology and linguistics, second 1anguage acquisition was thought to be not much different from more general leaming in which knowledge and practice were seen to go hand-in-hand. Syntactic rules, verbal and nom­ínal pamdigms, and other descriptions of Jinguistic features of the second languagll constituted the knowledge that lcamers needed to intemalize. Sentence dJills involving manipulations, t.rnnsfonnations, and the like, were the diet of classroom practice. To "intemalizc" language meant to memorire and regurgitat.e Jinguistic forms. Savignon (1983) summarizes the tenets underlying the Janguage program she taught in before she developed her ideas on communicati'-e competence and the language classroom. The fift.h tenet is perhaps the best description of what was happening in classrooms by someone who was an instructor at thc time:

The besic unit of practic:e shoold alW!\)11 be a complete atructure. Production ahoold precede from repetition to subetiwtion and continue until ""'JlOllS"S are automatic. Spontaneous expression ahould be delayed until the more advanced level& of instruction. Production erroni in ftruc:tural or phooological feetu.res m<'Wl that the patt.erns have not rooeived sufficient prior drilling (p. 20).

By the late 1960s and early 1970s, a particular approach to grammar leam­ing via practice was codified by Paulston (1972). She described a taxonomy

Page 11: Input Processing and Grammar Instruction

4 \lanPatten

of practice types and advocat.ed a sequential ordering of them. In th.is truc­onomy, Paulston claimcd that mccharucal practice sbould precede meaning­ful pract:ice. Meaningful practice should in turn precede communicative practice for any given linguistic structure or grammatical feature. This sequencing as wcll as chararterutics of each practice type are displayOO in Figure 1-1. Al. the time of the wnting oftlús book, litis sequential ordering of practice types and an emphasis on oral production was still the staple of most second language textbooka publi:;hed in the Uruted States, something that we will review m ChaptA?r 3.

The Role of Input

Smce 1970, we have gruned many insight.s mto sccond language aa¡uisition, owing to an ever increosmg number of reeearchers with special interest in the field. Although we nre far from complctely understanding how second language acquisition happens, 110me aspects of second language acquisition are clear to the research cornmuruty. One such aspect involves the role of input and the ooquisition of gnunmar. To illustrate, we will examine quota­tions from a number of wcll-known researchcrs.

Sequeneln¡

mochanical

¡ meamn¡¡ful

Figun 1 • I

I . Leomer OO.. not need to att.iu:h mooning to oenten<.,,. in order to complete the proctioe.

2. TheN w ono and onlY right comict response. Ex: trnn&fonnation dnll

l. .....,.,,.,. o...i. to ati.ch m°"""" to both áunulUll and -

2 There '* one &Dd onl,y n¡ht OOl'T'eC:t reoponse; lhe intendod --=>¡ ol t.he leomer is alnoedy known by t.he llllUud« (or fellow Jeanwl.

Ex. - queotlonl auch ... "\\'bol-·-thJI da. bectn?"

1 lAomer .-lo to oüach - to both llJmuluo and -

2 lntended -"'""' - ÍI oot lmown by the INtrud« <or fellow loomer>.

Ex - quoobons auch ... "Do you....., poo<en 111 your donn room?"

lntroduction 5

The Input llypothoaia clnims that humana acquire language in only one way-by understandmg mesaages, or by receiving "comprehensible input" !Kr!Uihen, 1985, p. 2).

Ali ao... of suOMISful fin!t and seoond langu11ge acquis¡tion are cbaract.erU.ed by the availabibty of comprehen&lble input. <Laraen-Freeman & Long, 1991, p. 142).

lt 1& adf~denl that L2 aoqul&ltion can only take place when the Jeemer has ~ to input in t.he L2 Thil input may come in written or spoken fonn. In the caoe of tpc>ken mpul. 1t may occur m the cont.ext of interaction U.e., the leemer's attemptt to converae Wlth a natlve si-)<er, a teachei; or another leamerJ or m th• context of non-rooprocal cli8ooune lfor example, listerung to the radio or waldlin¡ a film> lEllia. 1994, p. 26).

For the know~ l!)'lll4'm of a petticular ~ to gJOW; the acquirer must ha"e exposure to lftltancN or exrmplars of that pertirular ~. Withoot such expolUl'I' lanaull&'l'devolopment will ruit take place (Schwartz, 1993, p.148>.

The L2 lee.mer'a tMk bears a atrong reeemblance to that of the Ll 1eamer. L2 lerunera are at.o faced with the problem of making aense of input data, of oom­ing up with a oyatem which will aooounl for that data, and whcih will allow them to understand and produce st.ructures of the 1.2. Thus, their task can be co1>­ce1ved of aa foUows. eqwvalent to the Ll aa¡UJSit.ion task (\Vhite, 1989, p. 37).

L2 Input --4•~ 1 L2 Grammar 1

Although thc above cit.ed rcecarchers hold different perspectives on sec­ond limguage acquisition and may use different frameworks with which thcy invcstigate factors alTccting second language acquisition, ali concur in that meaning-bearing input is essential to second 1anguage acquisition. What this means is that lcarners mu.st be exposed to samples of language (and in great amount.s) that are uaed to communicate information. Sorne rcse.archers also suggest that the input must be modified or ailjusted during the course of interaction (c.g., Long, 1983) or be comprehensible to the leamer Whether moclified or not, without meaning-bearing input learners cannot bwld a mental representation of the grammar that must eventually underlie the1r use of language. This role of input is depicted in Figure 1-2. The term "dcveloping syotem" refers to the mental representation of the second language the learner is constructing over time. lt is tlús recognition of the estiential role of mput m second language acquisition that leads one to rethmk traditional approaches to grammar instruction. Tbe obvious ques­tiOn is whcther traditional grammar mstruction JS consonant with the idea that input is the be.sic building block for the construction of a mental repre.

Page 12: Input Processing and Grammar Instruction

6 Varl'atten

input • developing system

Fi¡ure 1·2 A mnplistic view of the role of 111put and aquisitlon.

sentation of the second languagc grammar. Recall that traditional grammar instruction involves cxplanation and

subeequent oral practice. This approoch preswnably resulta in a mental rep­resentabon of the grammar that will underlie commurucabve performance. However. if meaning-beanng input is one of the essential building blocks .ºr acquisition, where is the meaning·bearing input in output p~? Witb.rts emphas:is on output pracboe, a tniditional approach to grammar mstrucbon ignores the crucial role of input in second language acqws1bon-end the def­inition of input in second Janguagc acquisition does not include instructors' cxplanations about how the second language works. The dcfinition of input is limited to meaning-bearing input, language that the learncr hears or sees tbat is used to communicate a mcssage. Thus, in traditional instruction, Jcarners practioe a fonn or structure, but they are not getting the !nput that is nccded to construct t.he mental representation of the structure 1tsel( This

18 anaJogous to attempting to manipulate the exhaust fumes (output) of a car to make it run better. 1f we want to improve the performance of a car, we might first want to Jook into usmg a better grade of 1?88°linc .<input). .

In contrast to a tradibonal approach to grammar mstructlon, processmg instruction is an input-based approach to focus on fonn. Appropriate pro­ccssing instruction activities, 88 describcd in detail in Chapter 3, do not ask the lcamer to produce targeted grammatical items; matead: ~eru:ners ~ puahed to attend to properties of the language dunng actiu1ties m which they heor or see language that expresses so~ Tl'U!Olling. These activities con· trun "structured input" -purposefully manipulated sentences and discourse that cany meaning. During thcse activities, the leamer is pushed to a~d to particular features oflanguage while listening or readmg. Tbus, atte~l:Jon to features in input becomes an integral part of the grammar acqwsition process. Unlike traditional mstruction t.hen, Processing instruction is con­sistent with the inputrbased naturc of acquisition.

Input Processing and lntake

To be sure, processing instruction was not bom solely out of the acknowl· edged role of input in second language acquisition. lf input were ali that were nceded for successful language acquisition, t.hen t.here may be no need for 11 focus on form at ali; natural, free 00'.:W'ring input should do the trick! This is not the case, however. Sorne, if not many leamers, cvidence lack of

lntroductoon 7

acquisition or only partial acquisition of the grammar of a language even with exposure to input. Althougb we can suggest that this is due to a con­spiracy of factoni (Sharwood Smith, 1985) involving sociolinguistic, attitu· dinal, motivationnl, contextual, cognitive, and other factors known to affect second language noquisition, our focus in lhis book is a psycholinguistic one. From a purely )J6ycholinguistic perspectivo, we note that what lan­guage leamers hcar and see may not be what gct.s processed. lnput does not simply en ter the brain as the learner is exposed to it. lf this were true, thcn acquisition would be almost instantaneous. What is clear is that leamers filter input; they po6SCSS interna! proce&s0rs that nct on the input and only part of t.he input makeJi it.s way into the devclopmg aystem at any ¡pvcn time. The part of input that leamers procetiS is generally called intake, a term first coined by Corder (1967). What leamers do to input during com­prehension-that is, how intake is derived-is called input processing (see Figure 1·3). Although not the focus of this book, those prooesses involved in the incorporation of intake into the dcveloping system are referred to as accommodation and rcstructuring. We will touch upon t.hese proceases briefty in Chaptcr 5 88 we examine t.he rclationship between input process· ing and accommodation-restructuring 111& a vis Urúversal Grammar and 6rst language transfer. A third set of prooesses, !hose that are tl8ed by leamers to 11COOS11 the developing system to create output, are not repre­sented in the figure at t.his time.

Reoognizing that lcarners process input and that intake is ultimately the data made available to the developing system raises a number of questions: What is the nature of input proc:essing? What strategies or mechanisms are involved? Could input processing be responsible in sorne way for the partinl or incomplcte na tu re of the developing system? Chapter 2 of this book takes up the first two questions, and it is t.he BllSWCr to the third question that has more d.irectly motivated processing instruction Given that acquisition is intake-dependent and that intake is in tum mput-dependent., it is dear that input protes"1Ilg must play sorne role in the formation and development of · the leamer's lingwstic aystem. From the pei spective of a focus on form, then, we can ask t.he following question: Can input processing be manipu·

1 Il input---•• intake --•• developing

system

1 = input processing 11 • accommodation, restructuring

Page 13: Input Processing and Grammar Instruction

8 Vari'atten

lated, altered, or enhanced in order to make intake grammatically ncher? It 11 thía particular question that has led directJy to the devclopment of pro­oossmg instruction. As we will see in Chapter 3, processing instruction does not mean tbatjust any old input activity is viable and that learners wilJ sub­scquentJy make the com.'Cl targcted form-rneaning connlJ'1.ions. Processing instruction always considers lhe nature ofinput processing and 11tternpts to alter leamers' default processing strategies if they do not work to create optimum intake. Processing mstruction thus is beneficia! when it identifies an mcorrect or less tban optima! processing strategy and then fashions mput activities tbat belp to circumvent tbe straregy.

To surn up, processmg mstruction is an input-based. psycholinguistically motivated approach to focus on form. Un!ike traditional instruction with an emphasis on rule leern.ing and rule application during output activities, the purpose ofpmcessing instruction is to alter how learners procesa input ancl to encourage better form-meaning mapping that resulta in grammati<:ally richcr intake. This in turn should have a positive elfect on the nature of the developing systern.

TIE ORGANIZATION OF THIS 800K

The rem&nder of this book IS dcdicated to exploring procel8ing instruction m dctail. In Cbapter 2 wc will examine the nature of input processing. Rclying on experimental and descriptive evidence, we will outline a set of processes learners use to rnake form-meaning connections during input pro­CCSRÍng. Framed as processing principies, we will see how these processes result in filtered input data and how they help to explain the part.iaJ and somctimes incorrect nature of the grammatical systern evolving in the learn­er'e head.

ln Cbapter 3 we will examine prore;sing instruction in sorne detail reviewmg a set of guidelinca for developing what bave come to be caJJed structured input activities. We wilJ look closely at the types of octivities tbat leamers are engaged m and how they are derived frorn the insights about mput processing described in Chapter 2.

In Chapter 4 we will review empírica! research on t.he effocts of process­ing instruction. Alt.hough proccssing instruction is a relativcly ncw idea, a number of studies have been conducted that point to its beneñts. This rescarch exanúnes the effccts of proccssinginstruction using 11 variety oflin· guistic featurei: syntactic, morphological and inflect.ional, and lexico-seman­tic. The ñndings of these studics suggest that processmg instruction is an attractive altemative to tracbtional instruction for focu.'"1¡ on form in the communicative and input-nch language classroom.

In Cbapter 5 we explore a number of potential cnt.iosms of both input

lntroduction 9

processíng and processing instruction. Specifically we will examine the rela­tionship betwccn input processing and Uruversal Grammar (UG) as we explore sorne 188Ues in the acquisition of eyntax. \\\) will aJso see why input processing and processmg instruction do not succumb to the critici.sm! lev­eled at cognitive approaches to second language aoquisition and teach.ing. In this chapter, we sce that an important rC6CMCh agenda in the acquisition and teach.ing of syntax emerge.

TIE INTENDED AUDIENCE FOR THIS BOOK

This book is in tended for scholars and nonbcgmning students of second lan­guage acquisition and teaching. Although we have a~rnpted to Jimit the us:e of jargon and to explain relevant terms and concepta as we proceed, we nught not be able to cornpletely satisfy the needs of begirming studcnts of second language acquisition in this regard. As is t.he case with many books and research reports on specialized topics in &ee0nd 1anguage acquisition, the reader will most likely benefit from introductory readings on second lan· guage acquisition before reading the preeent book. Nonetheless, the ml\ic>r· ity of concepta ahould be """"""'ble to m06t readers including the beginning student of second language acquisition.

TERMS USED IN THIS BOOK

A number of terms are frequentJy used in this book. Since researchers may use these tenns differentially and be<:ause readcrs may imparta meaning to a particular t.erm be<:ause of the.ir own backgrounds, these terms are defined in the following. When used in the remainder of this book, these terms will imply the indtcated meanings.

Developmg system refers to the cornplex of mental representations that as an aggregat.e constitut.es the Jeamer's underlying lrnowiedge of the second language (phonolog)I syntax, morpholog)I et.e.). Sorne researchers refür to this as the "lcamer's granunar, n the Jeamer's "interlanguage, n or the leam­er's "(lingu.istic) competence" (see, e.g., Schwart2, 1993; Selinker, 1972).

Exp/icil instructwn refers to planned and organized teaching designed to inform learncrs of how the second language work:s. Explicit instruction gen· erally involves sorne kind of explanation (explicit information about the Jan­guagel and somc kind of practice, but not always. What makes explicit instruction explicit is t.he infonnation provided to the learner about how the language works.

Focus on form refers to any manner m wluch the Jearner's attention is sornehow drawn to the formal properties of the second language and how

Page 14: Input Processing and Grammar Instruction

10 Varl'atten

they work. This am bappen mor out of a classroom. However; the attention drawn to form happens witlun a commwúcam1e cont.ext so tbat clarification rcquestions (e.g., "Do you mean ?") and other interoctional devices may serve to draw a learners nttention to formal elcmcots during the expression of meaning. (soo Long, 1991).

Form (also callee! grcunmaJical form) refers to surfaai features of lnn­guage tbat include verbal nnd nominal morphology (i.e., inflcctions) and functional items such as prepositions, articles, pronouns, and other "noo­conteot" words.

Form-11U!Olli.ng mappmg (alao callee! form-mec.tUlllI oonn«Jwn) refers to the ooonection that the leamcr's mt.emal processors make between refer­ential reel-workl meaning and how that meaniog i.s encoded liogWstically. An example is tbe form-mearung coonection between the inflectional suffix /·V and "past time referenco" m English. Form-meaning mappings happen duriog input procossing and nre necessary for the building of mental rep~ sentations (see below).

lnpul is used to mean comprehensible, meaning-bcaring input. This is restrict.ed to samples of second language that leamers hear or see to which thcy atteod for its propositional cont.ent (message). This t.erm is synony­mous with Krashen's useofmputand Schwartz'sused of PLD (pnmary lin­gui'!tic data). It does not refer to explanatioos about the second 1aoguage nor expbcítly formulated irúormatioo IJlYen to learners about grammatical prop­erties of the second Ianguage (see Krashen, 1982; Scbwartz, 1993).

Input prottSSing refers to the derivation ofintake from inputduriogcom­prehension.

Intake is the subset of filtercd input that serves as thc data for accom­modation by the developing systcm. It is the input that has been proces.sed in some way by the learner during the act of comprehension. Intake i.s not eynonymous with intemalized lnnguage. Instead, intske are the data made available for further processing (e.g., int.emaliultion) once the input bas been prooessed

Menlal representalion refers to the largely unconsciOWI representatioo of any aspect ofthe liogWstic syst.em in the leamer's head. This representatioo mayor may not ex:ist as a rule, pnradigm, and so on, since the p¡ychological validity of rules and pnradigms is debated by linguists and psycholinguists (see, e.g., Bybee, 1991).

Processing strote¡¡y rcfers to the Oargely) unoonscious strategies that learners use to map meaning onto input sent.ences such as assigning agency to a noun m a sentence or detenrurung plurality versus non-plurality of a noun. These strategies are psycholiogWstic in nature and are MI syoony­mous with the ooocept of "lcarrung strategies" frcquently di&cussed in tbe hterature; these Iatter bemg debberat.e attempts to control one's leaming (e.g., usíng nemonic devices to remember words).

lntroductoon 11

Syntax refcrs to the structural propert1cs of senteoces and the rules that oonstrain the shape that sentences may take. Thi.s cootrasts withgrammar, which enoompasses syotax (cf. developing syi¡tcm) (see Chomsky. 1988; White, 1989).

Page 15: Input Processing and Grammar Instruction

!2 INPUT PROCESSING IN

SECOND LANGUAGE ACQUISITION

INTRODUCTION

As scen in Chapter 1, there is a consensus among second language researchcrs that input is an cssential component of second language acqui. sition. Lcn.rncrs use input in order to construct a mental representation of the grammar that they are acquiring. However, there is aJso consensus that not ali of input is attended to. Leamers filter the input, prooessing it so that only a reduood subeet of the inputr--<:alled intak&-is made available for aoomunodation by thc devcloping system. What do learncrs attend to in the input, and why? What strategies direct how leamers' rnake form-meaning connections? What. doce intake, as oppoeed to the input look like?

Sincc input is a critica) aspect of second language acquisition, the reader may think that there would be a plethora of studies describing input pro­cessing. Unfortunately, thi.s is not the case. Most empirical studies on input trcat issucs such as linguistic properties of input and modification of input (see, e.g., Elli.s, 1994 [Chapter 7]; the collection in Gass & Madden, 1985; Larsen-Freeman & Long, 1991 (Chapter 5)). Input processing only recently has become thc focus of research in second language circles and actual empirical studies are few Nonetheless, tbere is enough research, both theo­rotical and ompirical, to b<>gin d<>velopmg 9 mntlPI of input proces&rig. The point ofthe preecnt chapter is to summarize that researcb and develop a set of proccssing principies thal forro the nucleus of a model of how learners derive intake from input. These principies can, in tum, serve as points of departure for furthcr research .

13

Page 16: Input Processing and Grammar Instruction

14 VanPatten

The chaptcr is divided into three major sections. The first deals with oog­nitive aspects of input processing. In this section we will draw u pon various constructs from cognitive psychology, most notably attenlicn, and argue that second language learners are linuted capacity processors. As such, lhey can only atrend to so much lingWstic data ata time m the input dunng on-tine comprehension. \Ve will see that a cognitive approach to undcrstanding input processing yields important insights into thc acquisition of morpholo­gy. in particular, nominal and verbal inflections. We will see also that a oog· nitive approach offers insights into the acquisition of functional catcgories such as articles and prepositions.

The second major section deals with sentence-level aspect.s of input pro­cessing. Hcrc we are concemed with how leamers assign grammatical (suJ>. ject and object) and semantic (agent and patient) roles to nouns. Our focus will be on what is called tbe "first noun strategy." We will suggest that in the acquisition of a variable word order language likc Spanish, this particular processing strategy may have a subetantial effect on the developmg system. It is this particular strategy that informed thc first published study on pro­cessing instroction (VanPatten & Cadiemo, 1993).

The third major section focu.ses on a number of issues related to input processing gcncrated by the diecussion in the first two sections. Here we examine oollJ!CÍousness and awareness, language transfet; and other models rclated to input processing, m06tly from first language studies. Our point here will be !Al show that the pnnciples and model of second language input processing developed in this chapter are congruent with other aspects of sec­ond language theories. We wil1 show a1so that related models of language processing are complementar-y but not inclusive of the ideas prcscnted in this chapter.

ATTENTION ANO IM'UT PROCESSING

The Princlplu

Our aim in this section will be !Al motivate and offer evidence for the follow­ing principies m second language input processing. Pl and P2 represent two major principies with Pl(a)-{c) representing corollaries !Al Pl

Pl. Leamers process input for meaning before thcy process it for form. Pl(a). Leamers procesa oontent words in thc input before anything else. Pl(b). Leamers prefer processing lexical items to grammatical items

(c.g., morphological markings) for semantic information. Pl(c). Lcamers prefer processing "more meanmgful" morphology before

"less" or "nonmcaningful morphology."

Input Pro<:esStng 15

P2. For leamers to process form that is not mcaníngful, they must be able to procesa infonnational or communicative content at no (or little) coet to attention.

ln order to undcrstand the motavation for thesc pnnciples, we must first examme the cognitive psychology construct of attcntion

Attentíon

As 1 write at this moment, there are a number or stimuli that thrcaten to overwhelm me. Therc is the oonstant humming of the laser printcr, the steady rush of cars on the street, the faint sound or the TV coming rrom the other room, birds chirping, the clicking of the keyboard, and the sound of ncighbors next door. And these are just the auditory stimuli. Although ali stimuli are pen:cived {that is, the sounds are perceived initially), we rarely run into any kind of cognitive overload. lnstead wc function quite weU because our bnuns are equipped with processors that selectively attcnd to incoming data. Selective attention occurs because orientation of attention brings particular stimuli into focal attention rathcr than allowing thcm to be merely perccived (l .achman, Lachman, & Buttcrfield, 1979). Asan cxam­ple, when writing 1 orient my attention to tbe computer screen, wluch sub­eequently belps to filter out other competing (visual) stimuli

Attention is an unportant construct for learning. Researchers in cogni· tion generally beheve that learning takes place via attention. Peoplc simply don't leam unless they are attcnding to the stimulus to be learned. In tlrst language acquisition, for instancc, Slobin has argued that the dcgrce to which tinguistic information is acquired is detennined by whether or not the information is attended to.

... the only lin¡uiatic material t.hat can figure in languapmaking are stretch­ee of speech that atlrocl the chdd's alútntion ro a sufficrt1U degree ro be notu:.d wlá lu:W in ""'"'"? (Slubiu, 198ú, p. 1104, empha&io oddcd).

The belief that childrcn learn their mother tongue as wcll as second lan­guagl!S effortlessly is folklore at best.: Children must attend to incoming data IÍ they are to eventually incorporate them inoo a 1inguistic system (See Peters, 1985, for further discussion of the effortful nature of Ll acqui.sition.)

From the pcrspectivc of second-language acquisition, Schmidt has pro­vided the profession with an excellent synthesis and analysis of thc litera­ture on atrention, ooll8Ciousness, and lcarning. Arguing against any kind of suboonscious or sublinunal leaming, Schmidt concludcs that as far as mput processjng in second language acquisition is concerned, adult language

Page 17: Input Processing and Grammar Instruction

16 VanPatten

leanung reqwres attention to form in the input. <See aJso hís more recent ~on m Schmidt, 1994.)

The eXllUng data are oompatible with a very •trOng hypotheg¡s; You oon 't kom o fortrgn /(u¡gu(Jgt (or OJ1ytl11ng tl1t, for t/uJJ nl4lür) throogh subluninal pottp/IOll (Schnudt, 1990, p. 142, emphasis ll<ld«i).

Howcver, attention is effortful, and cognitivc ¡J8ychologists generally agree that attention involves a limited capacity to dcnl with stimuli: Only so much inoorning data can be att.ended to ata givcn time. Thia is particularly true of oompcting stimuli that engage the same modality, for example, the aura! modality (Wickens, 1984). Although it is the case that moat people can walk and chew gum at the same time, it is not the ca.se that moat people can watch Tv, listen to tlie radio, talk on the phone, and attcnd to aJ1 the aura! lingu.is­tic data of each activity at the same time. As a human cognitive activity, Jan. guage oomprehension consumes a great deal of att.entional resources, a notion that will become important when we d.iscw;¡; the push for leamers to get meaning froto the input. As Just and Carpent.er 0993) describe it, oom­prehen.ion involves oomputations in working memory. Although a number of t.he&e oomputations can pnxeed simultaneously (at least during native speaker oomprehension of the lirst langunge), Just and Carpenter note that

ü the number of ¡urn 1es [during on-line oomprt>henaiooJ ,.. large or, more prec:uiely, ü the amount of ad:iviation they try to propagate would exceed the capacity, then their attempta at propa¡¡ation are ecakd back to a leve! that keepe the total ad:ivatíon within the maximum bound.... When the task demnndl oxceed the available resouroos, thcn both storoge e.nd oomputation­al functiortS [ within working memory] are d<'llrt'graded (p. 4).

Onc of the processes (or oomputationa) thnt ia canied out during atten­tion is ~urtum. Detection is the "pro<:eM that sclocts, or engages, a partic­ular and specific bit of information" <Tomlin and Vtlla, 1994, p. 192). It is tbe procesa by which data are registered in workíng memory and is wbat makes a particular stimulus or piece of data available for further processing, that 1>, for aocommodation in the case of acquisition lt IS not enougb, then, that learnera simply att.end to data; particular data must be detect.ed if learning i.s to occur. ln this way, detection is a subprocess of attention and is the a.spcct of input processing most directJy relntcd to the derivation of intake.

Twn featul1lS of detection are relevant for the present discussion.

l Detcctcd information causes great interference with the proces•ing of othcr information.

2. Detccted infonnation exhausts more attcntional resources tban even ori­cntotion of att.ention (Tomlin and Villa, 1994, p. 192).

Input Processing 17

What the preceding discussion suggeo;us is lhat even ¡¡; say, a $pcech stream IS att.ended to by a second language learner, the interna! processors may not detect all of the linguistic data. 0..'l.ectmg one bit of information (pu!a! of lingu.istic data> may mt.erfere with the detection of others by oon­llWlllllg available resourcE6 in working memory The central issue for second language acquisition research is how lcamers' mtemal processors allocate attcntional resources during on-bne processing. In short, what causes cer· tain (linguistic) stimuli in thc mput to be detected and not otbers? (Note; from bere on processing will be a oovcr tcrm used for att.end.ing to and det.ecting lingu.istic data in the input.)

The Push to Get Meaning

Jt is important to recalJ that input rcfcrs to meaning-bearing input. This is language that the learner hears that enoodes a message. The oommunica­tive goal of the leamer is to oomprchend the message, to understand what the speaker said. (A)though input may be wntt.en, we will focus here on aura! input.> Fbr this reason, we posit PL

Pl. Leamers process input for meanin¡ before they process it for form.

S=ply put, Pl stat.es that leamera are driven to Jook for the message in the input ("What is this person saying to me?"} bcfore looking for how that message is enooded. The principie is oonaistent witb the observations of other researchers in botb first language acquisition and aecond language acquisi­tion (e.g., Faercb and Kaaper, 1986; !Oein, 1986; Peters, 1985; Sharwood Smith, 1986; $wain and Lapkin, 1989; Wong-Fillmore, 1976). In first Jan. guage acquisition, for example, Petera (1985) lisis a number of Operating Principies that guide children in prooessing input. Her first principle, how­ever, is a general one regarding the push to gct meaning.

EX: MEANING: Pay attention to uturanc:M that ha>,, a readily identifiable ,,_,.¡,,g. Extract and remember IDUnd eequenoea that have a clear oonnec­bon to a dear oonten. <p. 1034)

What this meaos is that first language acquisition is part oftbe act of oom­prebension. Leamers are constantly seeking some way to map whal they understand with how they heard it.

In serond language acqu1S1bon, l:lharwood SDllth (1986) states a similar position when he diatinguishe¡¡ betwcen processing for communication and processing for acquisition. He aays that

lt ia therefore appropriate to speak oflinguiatic input ea having dual relevance.

Page 18: Input Processing and Grammar Instruction

18 Vanl'atten

At moment X, the learner's main aim may be to extra<:t meaning and survive or succeed in a given interchange of m~. In thi.s ca:ie• only thooe as~ of the input wlúch will aid the learner ín t1ús commurucabve endea~or. Oin· guistic or otherwise) will be relevant ín this lirst aense: tM /.eo.mu will inter· pret for meaning (p. 243, emphasis original).

Altbough meaning and form are not necessarily mutually exclusive, a point talren up later in this chapter, Pl would Jead us to conclude that form and meaning may compete for processing resources. In _other words, w~en ali else is equal, fonn and meaning compete for detectíon-witb mearung generally winning out. Fonn here is defined as surface features oflanguage: verbal inftections nominal inllections, particles, functors, and so forth. (We will talte up ah~ syntactic rules in Chapter 5.) Again, our point of depar· ture is that comprehension is effortful for second language learners and con­sumes a great deal of on-Jine attentional resources.

Tbat Jearners process input for meaning before they Proc:e5S it for fonn Jeads us to ask what in the input Jeamers would attend to m order to get meaning. A Jogical place to begin would be with content words. Ali second tanguage Jearners come to the task of acquisition knowing what content words are. Intuitively they know that these words are the building blocks of mea.n.ing and the initial stages of Janguage acquisition ha ve been ~r­ized in many reports mainly as vocabulary getting. If comprebens1on JS

effortful for beginning and intennediate Jearners, it seems logi.cal, tben, that their attentional resources wiJJ be directed toward the detectJon of content words to belp them grasp the meaning of an utterance (see also Sharwood Smith, 1986, pp. 245-260). We thus posit Pl(a) as a subprinciple to Pl.

Pl(a). Leamers prooess content words in the input before anything else.

Pl(a) receives support in both 6rst language and second l~ research by the fact that learners in input-rich environments tend to p1ck out and start using single words or whole unanalyzed chunks of language (which they treat as content words) in the early stages_ ~d then combine these to form utterances. In child first Janguage acqws¡tiOn the early one· and two-word stages of acquisition are well-documented, with children pro­ducing utterances Jike tbose produced by Peter, a two-year-old, in the fol­

Jowing interchange:

Peter: (finding a car) Get more. Lois: You're gonna put more wheels on the dwnp truck? Peter: Dump truck. Wheels. Dump truck. (Lightbown and Spada, 1993, p. 3).

One could argue that Peter's and other children's procluction of one- and

Input Processing 19

two-word utterances as weU as unenalyzed chunks is tbe result of speech processing constraints and not input processing constraints. However, at some point, output also must offer some indication of what has gotten processed in the input (in addition to how it was stored), a point argued by Peters (1983, 1985). Her data on child fust language acquisition convinc­ingly show the child focusing on isolated words and unanalyzed chunks of language in the input, extracting these, and subsequently using them in his own speech. The child might process incorrectly, as the example that follows demonstrates, but it is clear that the child somebow is attempting to isolate content words and connect these to mea.n.ing.

Adult: That's an elephant, isn't it? What is it?

Adam: Init. (Peters, 1985, p. 1035)

Radford (1990), working within a generative approach, argues pretty much the same point. He notes tha(Í!i.e intial stages of first language acqu.i· sition involve developing an eleml!ntery vocabulary that. is acategorical (without grammar or grammatical properties) in natureJ Children pass through a pre-linguistic stage (e.g., babbling without words) anda one-word stage before they start to put words together and show evidence of having cateogorized them into nouns, verbs, and so on, or having assigned gram· matical categories to them such as subject and object (see Radford, 1990, Cbapter 2).

The primacy of content words in processing input is evidenced in second Janguage acquisition as well. Again, eerly stege leerners create output by stringing togetber content words and unanalyud chunks: Why test?, no drin.k b«r; me gusta las cl.as€s, are typical sentences for early stege learners. As in first language acqu.isition, this stage of output is suggestive of how tbe input has been processed. But Pl(a) receives support from several experi· mental studies as well. Klein (1986, Cbapter 5) reports on research con­ducted on adult second 1anguage acqu.irers of Gennan living in Gennany. Using a sentence repetition task, he asked ~ learners to repeat stim­ulus sentences immediately upon hearing tbeÍn. He found an overwhelming tendency for subjects to pick out and repeat oontent words, with the more advanced leamers being the only subjects able to pick out and repeat gram­matical items sucb as auxiliery verbs and erticles.

Mangubhai (1991) also found evidence for a primacy of lexical items in input processing. In his study, he gave five adults lessons in Hindi using Total Pbysical Response overa period of 10 weeks. He periodically examined what he called "tbeir processing behaviours," in our terrns, their processing strategies. He found that ali leamers focused on lexical words to a greater or lesser degree in order to get meaning from the TPR input. Mangubhai

Page 19: Input Processing and Grammar Instruction

20 VanPatten

shows that one leamer, Bany, was absolutely 00DS1Stent m this regard as he "reduced the load on his working memory by stnppmg the input to oontent wonls <cC. Sbapira, 1978)" (Mangubhai, 1991, p. 283).

Additional expenmental evidence oomcs from VanPatten (1990). In this expcnment different levels of subjects J.isWned to a short passage on infla. tion in therr seoond language, Span.ish. After listerung, sub¡ects were asked immediately to perform free written recalls in English, their 6rst language. They wcre instructed to write down anything and everything they oould remcmber. These recall protoools subsequcntly wcre soored for the number of correct idea units recalled, using Carrell's (1985) defuútion of idea unit:

Each unít consista of a single clause (ma.in or subordinate, ioclucling adverbial and relative claUBe8l. Each infinitival construction, gcrundive, oominalized verb phraae, and corijunct [is also) a separate id"" urut. In addition, opti.onal andlor b""vy prepo&itional prui..- [are aJsol id"" unita (p. 737).

As part of tbe experimental design, sub)'lCIB wcre assigned to one of four groupa: (1) listen to tbe passage only; (2) listen to the passage and note any

T..U

--<>-- S«ond·year ,.., .....

··<>·· Tblrd·y .. de•rnen

1'aak 1 • conttnt. only Tuk 2 • ron~t + IWcal item Tuk 3 • tw1'4.'nt + runctot Tuk 4 • oonk•nt + lnlltctlon

FICURE 2·1. R-11 aoo""' by wlt and leve! ofleemcn <bAlcd on Vanl'lltten, 1990>.

Input Processing 21

and aU occurrences of tbe oontenl word inf/acWn; (3) listen to tbe pru;sage and note any and aU occurrences of the defirute article /.a; (4) listen to tbc passage and note any and aU ocxurrences of tbe verbal inflection -n . "Noting" was operationalized by having sub)ElCt.s place a check mark, slash. or any otber kind of mark on a blank sheet of paper each time they noted the target ítem. lt should be pointed out that the passage was oonstructcd so that the oontent word, the defiruoo article and the verbal inllection oo:urred 11 times each and were evcnly distributed throughout the passage. Pl(a) would predict that if comprchension is effortful for leamers (uses up attentional resources) and if lcarncrs are processing content words in thc input before anytbing else, then those subjects in group (2) should perform bctter than those in groups (3) and (4). The results bore this out. Subjcct.s in the pa.ssage plus oontent word group recalled es many idea unit.s as those in tbe pa.ssage only group. Subjects in groups (3) and (4) who were asked to attend to a grammatical feature in addition to the overall oontent of the pas­sage, recalled significantJy fewer idea urut.s oompered with groups (1) and 12). lndeed, tbe soores for groups (3) and (4) dropped dramaticall); as illWl­trated in Figure 2-1. That tbe explicit processing of the oontent lexical iU!m did not mterfere with oomprehension is oongnient with the idea that con­tent lexical items are of primary unportance during secood language input processing (at least in the early stagcs of aoqwsition).

Meaning and Fonn: Lexical and Grammatical Encoding

Pl(a) has consequences in terms ofwhich surface grammatical features gct processed and which do nol. A great number of grammatical features enoodc sorne kind of semantic information. For cxample, the verbal inflection -ó in Spanish enoodes PAST as in hab/6, "he spoke." The English verbal inflection -s enoodes third person singular The plural inflection -i on nouns in Italian signals more than one as in ro¡f<JZZi, "boys" (cf, ra¡¡0=>, "boy"). Howe\'er, what is equally true is that many of tbese same semantic notions are expressed lexically in a language. In addition to being enooded in -ó, pastness is a semantic notion also oon...eyed by the words ayer, "yesterday," el año pasado, "last year," and hatt un mes, "a month ago" in Spanisb. Third per· son is a semantic notion underlying not only ·s in English but also the nouns "Maria," "cat," and "he." Also, plurality is not only enooded in the -i of r'O(Jazzi, but aJso in words such as due, "two" and molti, "many." Given that ICAmer& are initially driven to prooc:i& content words before anything ebe, we thus arrive at the following principie.

Pl(b). Learners prefer processing lcxical items to grammatical items (e.g., morphology) for semantic information.

Page 20: Input Processing and Grammar Instruction

22 VanPatten

This principie suggests, then, that learners first attend to and detect lexical markers of plurality before g:rammatical markers of plurality, that they attend to and detect lexical markers of temporal reference before verbal inflections of tense, and so on.

Tbere are two different lunds of evidcnce lll the second language litera­ture to aupport Pl(b). Thc first are those data that report how tense is first enroded in learner output. Typically, learners mark time early in the acqui­sition of verb morphology lhrough lexical items (e.g., yest.erday, last week) and only subsequently bcgin to add past-tense verb markings to their lin­guistic reperloire (Bardovi-Harlig, 1992). This appears to be the pattem for both clasaroom and nonclasm>om leamers. Agam, if one assumes tbat out­put is prutially (ifnot moetly) sbaped by the intake derived from input, then these data are consistent with the principie that lexical items are more like­ly to be attended to than verb morphology if both mark the same semantic feature. Clt is worth noting that sinúlar evidence exists for the acquisition of plurality with IEemers initially marlting plurality lexically, e.g., two dog, many problem. The same i.s true with third pcrson singular, e.g., she go, M sleep. For eome discussion, eee Pica, 1985, as well as the titerature on pid­gins and creoles, e.g., Todd, 1974.)

Recent cmpirical researcb provides evidcnce for Pl(b) by speci.fically focusing on tense. Cadiemo, Glass, VanPatten, and Lee (1991) report on the effect of providtng clas.sroom Jearners of Sparusb as a second language witb two different kmds of dJscourse leve! input, that is, a listeníng passage: one in wbich temporal adverbials are present and one in w!Uch they are absent. Pl (b) led the researcbers to predict that iflearners were attending to lexical cues for tense, then thoee in the "no adverbial group" would pcrform poor­ly on a test that asks them to identi.fy wbich eventa in the passage occurrecl in the past, wbich will oocur in the future, and wbich are currentJy in progresa. Tbe prediction was supported. Sub.)CCts in the "adverbial group" were bettA!r able to detemune tbe temporal rcference of an event compared with thosc in the "no adverbial (verb infiection only)" group. Glass (1994) reports on a follow-up study in wbich be idcntified severa! good and poor learners who were subjects in tbe preceding "adverbials versus no adver­bials" expcnmenl Glass reviewed tbeir post-listeníng tests witb them, replaying parts of tbe passage as tbey reviewed, and asked the sub.)CCts bow they detenruned that an cvent took place in the pastor not (e.g., probe: "You said that this event occurred in the past. How do you know tbat?") This introo;puclivu IJl'Obe revealed that the aubjccto wcre relying on the lexical information (adverbials) and not on vero inflcctions to assign tense (e.g., subject response: Because ltbe speaker) said el año pasado l"last year")). lmportant for tbis discussion i.s that unlike Englisb, the Spani:;b past-tcnse morphemes are sy!Jabic and receive strong stress; tbus the resulta of the research just described cannot be so easily dismissed on the grounds that the

Input Processong 23

grammatical items are "perceptually nonsalient." Musumeci (1989) conducted similar research at tbe sentencc leve! and

used classroom lcarnen; ofboth Italian and French in addition to leamers of Spanish. Musumcci asked her subjec:ts to assign tense to input sentences delivered under one of four conditions.

l. verbal inflections aocompanied by adverbials of time· 2. verbal inflections aocompanied by typical teacher pb.rsicaJ gesturcs (e.g.,

tbumb over the shoulder to indica te past) but no adverbial markcr of tem­poral reference;

3. verbal in&ctions aocomparued by both an adverbllll and a physical ¡es­ture; and

4. verbal inflections as the sole source of infonnation about the tense of the sentence.

Again, Pl(b) would predict that subjects rcceiving tbe lexical markers of tense in tbe input sentences would perform bettcr than those who did not. Musumea's resulta were similar to those reported in Cadiemo et al. ~ub~ in the verbal inflections plus adverb group and those in the verbal mflecbo~s plus ¡estures plus adverb group consistentJy outperformed the subjects m thc other two groups in which adverbs wcre absent in the input scntences. Her resulta clearly demonst:ra.te tbat the presence or abscnce of a temporal adverbial was the sigruficant factor determining co?Teet tense nssi~nt. Tlus findtng supports tbe primacy of lexical items over gram­matical markers during input processing.

In short, thcre is evidence from both second language processing studies as well as learner-language research on first and 8CCOnd Ianguage Jearners to support principies Pl, Pl(a) and Pl(b). The resulta oftbese studies sug­gest that leamcrs' focal attention during input processing is on meaning and pre~~mce is given to lexical oteros for tbe conveyance of that meaning'. When !eneal 1tems and grammatical features cont.aining tbe same informa­tion compete for processing time, lexical items again have precedence.

Meaning and Fonn: Relativa Comnuiicative Value

In both first-language and second lruiguage literature, a fairly consistent pie­~has devel~~ regarding the acquisition of verb morpbemes. ln English, •tn(f (progre.!!3•vc)"' alway$ acquircd before -s (thfrd fl<'1'501l pre..,nL) wul -ed (past tense) is acquired somewhere in between (soo VanPatten, i9s.sa, for more detaíl). Altbougb one could argue that -ing is syllabic and therefore pen:eptually more salient in the input, we are left with the question of why -ed is acqlrired before -s. Both are verb-final, consonantal, and nonsyllabic

Page 21: Input Processing and Grammar Instruction

l 1 1

24 vanPatten

(with the exoeption of one allomorph for each that oontains a schwaed vowel). Cleerly, something other than structural features is at work here. lf one loob at the acqlll&tion ofSpanish, one finds that person-number inftec. tions (-o, ·n, -mos, ·s, ·IS, ·0) generelly are acqu.ired before markings of a<ljec· tive ooncordanoe (-o, -a, -s, -0) (van Naerssen, 1981). Given thc overall lack of structural differcnces betwecn t.hese two sets of morphemes (all are word final morphemes, both sets oontai.n syllabic and nonsyllabic forma, none are stres.ed), something in addition to stnictural features is at work m acquis¡. tion. Although frequcncy may be a factor in detennining the rclative order of acquisition of a given grammatlcal feature (Larsen-Freeman, 1976), therc is another possibility, one that falls out of our di8CUSSion of attcntion, detec­tion, and meaning. N amely, it is lhe relative commw1u:atWe value of a gram­matical form that plays a major role ID detemurung the leamer's attention to it dunng input prooestjng and the likelihood of 1ts becoming detected and thus part of intake (VanPatten, 1984a, 1985b). Wc state this m tbe following principie.

Pl(cl. Leamers prcfer prooessing "more mearu.ngful" morphology before "less or nonmeaningful morphology."

In tbis principie, "more," '1ess," and "nonmeaningful" refcr to tbe com­municative value that a grammatical feature contributes to overall sentence mearung. Communicative value rcfers to the rclative oontnl>ubon a form makes to tbe referenbal meaning of an utteraiw:e and is based on the pres­ence or absence of two features: 1nherent semanbc value and redundancy within the sentene&-utterance. A form that can be classil\ed as having inherent semantic value and is not a redundant feature of language, will tend to ha'-e higb communicative value. For example, in English veroal mor· phology, -1ng tends to have bigh communicative value for the following two reasons. First, it has inherent semantic value because it enoodes progressive aspect, i.e., -ing = in progress. A sccond reason that -ing tends to be high in oommunicative value is that it is seldom redundant in naturally occurring discourse since, more often than not, no lexical information in the utterance CO-OOCUJ"S to provide cues to aspect.. When we calJ aomeone on the pbone we might say "bello" and tben "What are you doin¡¡?" Likewise, when we see someone lcaving we might ask "Where's she going?" In neither case do we add lexical information, for example "at tbis very moment," to the utter­w)Cle. The S>Je conveyor o( thc acrnontic notion ºin prosr-a" ia -in9., con­tnl>uting to íts relatively higb communicative value. We can swnmarize this disrussion by saying that in order to grasp the semantic notion of "in progress," the second language lcarner of English must process the veroal inflcction -ing in the input. (Although we are ooncemed with grammatical form here, note that content words by theír vcry nature can be considered

Input Process1ng 25

to be high in oommunicative value.) Other features of language will tend to have lower communicative value

if they 1ack inherent semantic value or are redundanL lnflections on a<ljec­tives in languages like Spanisb and ltalian tend to be low in commurucative value because tbey are botb redundant and lacking ID inherent semantic value. The -a in blanca does not carry any semantic meaning in the phrase ~ C080 blanca "t!'e whit.e house." BI= ends in-a because C080 is of a par­ticular grammatícal class that reqwres this. Note a1so that this -a is spread across three words in the ooun phrase, making it highly redundant.

Other grammatical fcatures may have inherent semantic value but are low in communicative value because they are almost always redundant. The English verbal inflection -s. for example, has inherent semantic value since it enoodcs the semantic notion ofthird person singular, but ít 18 aJso redun­dant because a sub)l!Ct noun-phrase is oJmost a!W11ys obligatory m Engl.ish syntax. Although oolloquial speech allows a certain amount of subject-omis­sion in English, for example, "Looks like rain," "Fcels funny, hch?" the vast majority of seotences in English have explicitly marked subjects in them. In naturally occurring discourse, one would most assunidly bear "He iol.-es me" and not "Loves me" asan answer to the question "How do you think he feels about you?" Given Pl(a) and Pl(b) and the primacy of lexical items in get­ting mcaning from input, the learncr is more likely to process the subject noun-phrase for person-number and not the verb form. Verbal ·8 thus is reJ. atively low in commurucative value, even though it can be said to have inher­ent semantic value.

Less clear cut cases are grammatical features such as tense markers. Clearly these have ínherent semantic value sinoe they encade temporal di.s­tinction. However, unlike third-person ·s and Spanish and Italian a<\jective inllections, tense markers are not oonsistently redundant in an utteraooe. At the same time, unlike progressive ·•ng, tbey are not consistently oonre­dundant. Sometimes they co-occur with lexical indicators of time in an utterancc, and sometimes they do not. In the foUowing ínterchange, a co­occurring adverbial (rcpresented in bold) is only mentioned in the first utterance.

Speaker A: iEbr qué n.o vinisle anoche? Speaker B: Trabqjé lodo el día y no tuve ganas aL salir. Speaker A: /Qld peno.! iLo pasarrws ton b~n! Speaker B: iSí? iQld hicieron?

Speaker A: Why didn't you come last night? Speaker B: 1 worked all day and I didn't feel likc ít after getting off. Speaker A: Too bad! We had such a good time! Speaker B: Yeah? What d'ya do?

Page 22: Input Processing and Grammar Instruction

1

26 VanPatten

At the present time, ali we can say is that grammatical forros with seman· tic value but inoonsistent redundancy will fall somcwhere between items of high and low (or no) oommurucative value.

A scale of oommunicative value might be clifficult to apply to ali features of language, but in a recent study Pl(c) was tested directly. Bransdorfer (1989) t.ested subjects' ability to procesa meaning and form simultaneously in Spanish when the form was cither the preposition de or the definite arti· ele la, neither of which can be oonsidered a oontent word Bransdorfer argued that these two forms are structurally similar (both are syllabic and occur in roughly the same syntactic position in a sentence, i.e., before noune), but he also argued that thc preposition would carry greater oom­municative value than would thc article. Acoording to Bransdorfer, the prepo51tion de bad mherent semantic value as a signa! of possession and, if absent, oould poee problems of interpretation for a leamer, Cor example, El libro es Juan, "The book is John" versus El libro es de Juan, "The book is John's." In the first case, the learner may not grasp wbether John is the poe­sessor of the book, thc goal of the book (the book is for him), the subject of the book, ami so on. On the other hand, the abecnce of the article la would not poee mterpretation problems, e.g.,Laplumaes uenk, "Thepen is green" versus Pluma es verde, "Pen is green." He thus classified la as having lower oommunicative value.

Bransdorfer used a simultanooua attention task 6imilar to that used in VanPatten (1990). Sub)CCts heard a brief passage and were tested on its oon· tent by means of an unmediate wntten recall in English. The sub.¡ects were divided into three groupe: (1) those who listened to the passage only; (2) tbose who listened to the passage and also not.cd any occurrcnce of tbc prepoeition de; (3) and those who listene<I to the passage and also noted any occurrenoe of the article la. The results of Bransdorfer's experimentation revealed that there mdeed was sorne kind of dilference between the two grammatical items. There was no significant dilTerence in recall seores betwecn tbe passagc only group and the passage plus de group. There was also no significant dilference in rocalJ seores bctween the passage plus de group and the passage plus la group. However, the seores of thc passage plus la group were significantly lower than those of tbe passage only group. Attencling to de while listening to thc passage affected oomprehension much less than attending to la. This suggests that as learncrs prooessed input for meaning, de was more likely to be attende<I to and detected tban la and sup· porta the oonstruct. oí relative oommunicative value.

In his 1991 dissertation, Bransdorfer replicaU!d his study using elements that carried strong stre5s: the lexical 1tem extínunu, "exams" and the OOJ>­ular verb está, "is." The oontent lexical item would obviously be oonsidered as having high oommunicative value since lexical items carry scmantic value and in an utterance are generally not redundant vis a vu each other.

Input Process1ng 27

Bransdorfcr classified the copular vcrb está as lacking inherent semantic value and thus lower in communicative value. Hia results revcaled that attencling ~ e:rámena while listening to a passage posed no problem for comprehension but attendmg to está did result in a drop in comprehension.

Processing Granvnat ical Form of Uttle Communícative Value

1_'he principies and supporting evidence presenU!d so far suggest that atten· ~on IS allocaU!d during on-line processing acoording to relative oommunica­tive value. Processing capacity limits what a learncr can attend to and detect when cn¡;aged in the ongoing and split-second process of deriving meaning from input. The intemal processor sceks to carry out efficienUy the task of getting information, and, in the early and intermediate stages of acquisition, ~e result IS a tendency not to process or bold m working memory those 1tems that do not contnbute to meaning. Howevcr, Ll learners eventually do acquirc most if not ali features of adult language, and som.e L2 leamers in input-nch environments do acquire manyofthe fcaturesoílanguage that do not oontribute to meaning (although in pidgjns and fossilized speech, it is p~ly thoee nonmeaningful featurcs that tend to be absent). Putting as1de SOC1oaffectiv~ n:iotivation for incre~ proflciency and focusing strict­ly on peycholinguistic aspects of learnmg, anothcr principie suggests the reason for wby ali grammatical forms eventually ma,y be acquired

P2. For lcarn~ to process form that 18 not meaningful, they must be able to prooess mformntional or oommunicative content at no or little oost to attentional resources.

Recalling that_ one of the features of detection is that it is a heavy oon­sum.er oíp~g resouroes, the preceding principie suggests that gram­matical form oflittle or no oommunicative value will be detecte<! only when the reso~ required b>'. detection to procesa meaning are not dcpleted Tbe prodictíon from th.is 18 that grammatical form of little or no oommu· nicative value will be prooessed much later in leamers' development and subsequcntly will be acquired later tban other grammatical form

There is as of this time no salid experimental evidence that directly SUJ>-1'.'.°rts this principie. In one ~dy, Leow (1993) hypothesized that simplifica· tion of mput would result m decreased attentional demanda on learners' pl"O<leMing of that input. Using a pre-tc.st....post-teat íormat, be mcaaured the gains on a test ofthe present perfect.and the presentsub]UllCtive in Spanish. Leow found that the subJects who read a simplified pessage had greater gain soores_ oompared with sulúects who read an unsimplified passagc. Jiowever, sampling errors suggcst that the results may have been owing to existing

Page 23: Input Processing and Grammar Instruction

1

28 VanPatten

differences in abilities between the símplified and unsimplified groups before receiving their passages. In any event, Leow's stimulus was written input, whereas our primary concem here is the processing of aural input.

In another attempt to investigate simplification and its posslble affects on attention, Beme (1989) replicated VanPatten (1990), but simplified the lis­t.ening passage by asking subjects in a pilot phase to tell her what sent.ences and parta of the pessage were difficult to comprehend. Based on their com­ments, she simplified the passage by making sorne sent.ences shorter, chang­ing sorne unlmown vocabulruy, and elíminating clauses. She then gave the list.ening passage two four groups as in the VanPatten study. focus on mean­ing only, focus on meaning plus a content word, focus on meaning plus a verb inflection, focus on meaning plus definit.e article. Her results, however, matched VanPatten's exactly: Leamers' comprehension dropped signifi­cantly when focused on form and meaning, but not when focused on a con­tent word and meaning. Símplification did not seem to help. Although surprising given that she had pilot feedback from subjects about the diffi. culty of the passage, Beme's findings may very well indicate that releasing attention to focus on nonmeaningful form may require a great e/rol of sím­plification ofthe input.

Another avenue to pursue regarding P2 would be to look at the effect of pausing. Blau (1990) found that of tbe three factors speed (slowing down the rate), complexity (reducing syntactic complexity and simplifying sentences), and pausing (adding brief pauses at natural breatb group boundaries). paus­ing significantly improved the comprehension of most of her Puerto Rican and Polish learners of English as a second language. Slowing down the rate of speech or simplifying the syntax did not have the same degree of facilita­tive effect. A number ofmethodological questions render Blau's study more suggestive than definitive, but nonetheless the study points to a line of research that can help to investigate P2 more directly; less or nonmeaning­ful grammatical features should be more easily detected when the input con­tains pauses that allow for processing time.

Although it is reasonable to conclude that simplification decreases demands on attentional capacity; attentional capacity does not interact sole­ly with communicative value to determine what is attended to in the input; attentional capacity is also affected by task demands of which processing time is a significant variable. Research needs to include studies in which pressure to perform the task is varied by amount of time available to do the tasi<. It could be that evon with simplified input, timP. pressure to compre­hend may place sufficient demands on the learner such that attentional resources cannot keep up with the demands of the task. While we await studies on this issue, it is worth stating that even if processing time proves to be an important variable in determining attentional load for L2 input pro­cessing, processing time alone does not ensure attention to formal features

Input Processing 29

of the input. In other words, prncessing time might be a necessary but not sufficient ingredient for attention to form in the input.

Loolting at P2 in another way, if we assume that output reflects to sorne degree what has been processed in the input, there is some indirect evidence for P2. Reca11 that leamers of English acquire verb morphemes in the fol­lowing order:

-ing

regular past

third person -s

with -ing being acquired fairly early on and ·s being acquired relatively late. As discussed earlier, -ing possesses high communicative value and -s pos­sesses little communicative value, whereas the past-tense marker falls some­where in-between. Thus, the order of acquisition matches the input processing preferences of leamers as the latt.er intersect with communica­tive value.

It is true that the example from Englisb verbal inflections is compound· ed by frequency in the input as well as structural differences. -ing is found to be much more frequent than -s, for example, in spoken English. -ing is also syllabic rendering it more peroeptually salient compared with nonsyl­labic -s. One could easily conclude that languages might simply bias salien· cy (based on structural features) and frequency toward items of higher communicative value, and that factors other than communicative value and processing capacity aocount for acquisition orders. Until we see substantial· ly more data from the acquisition of Janguages other than English, P2 wi1I continue to be theoretically motivated, but not empirically supported. Given the demands that detection places on processing resources, it seems logical that the detection of grammatical form of little communicative value is inhibited by the allocation of processing resouroeR tn getting meaning fmm the input. (See also Schmidt, 1990, for sorne discussion here.) Detection of these grammatical forms wi1I occur only if prooessing for meaning is rela­tively easy for the leamer.

Page 24: Input Processing and Grammar Instruction

30 VanPatten

Summary

In our model of i;econd language input processmg. form-meaning mapping is affcct.ed by the limited capacity to process incoming data Comprehension is seen asan efTortful phenomenon that consumes a great deal of attention· al capacity. Oriven to get meaning, leamers first allocate attentional capaci­ty to dct.ect conrent words in the input. Grammatical form conveying semantic information that also is encoded lexically wiU tend to not be detectr ed; thc learncr instead relies on the lexical items for thc semantic informa­tion. The leamers' interna! mechanisms will detcct grrunmatical form early on only if it is relativcly higb in communicativc valuc. Otherwi.se, grammat­ical form is detected over time only as the leamer's ability to get meaning from the input is increasingly automatized (beoomcs more efTort-free). This increasing autornatization of comprehension relea.ses attentional resources for the processmg of form that was prevíously skipped (undetected). In Figure 2-2 we sketch a model of how processmg based on limited resources works. Ar. data enter the processors, the latter first search the input for con­tent words. lf resources are depleted, then thc mtakc data delivered to the de.'elopmg system essentially are content leiocal iterns. lf resources are not

t'!GURE 2·2. A model of p"""°"""g grammabcol form ln tho input based on attention and communicative value.

Input Processing 31

depleted, then ~ processors rna,y make some form-meaning mappmgs usmg grammatical forms of high communicative value. If resources are ~epleted at this second stage of proct!ISlllg, then the intake will contain lex-1cal 1tems and sorne grammatical form. If the resources are not depleted, then further prooossing occurs. Tb.ís time, the processors make form-mean­mg connections using grammatical form of lesser or little communicative value. The resultant intake thus will contain a number offeatures that were no~ P~ previously. The diffcrent stages of processing, that is the pomts at which the prooessors determine whether or not resouroes aro depleted, can be viewed as developmental in nature; early stage learners suf· fer resource depletion early in processing so that lexicaJ items fonn the bullt of intake data. As the ability to ettend to lexical items and connect them to me~g begins ':°u~ fewer resources, then detection of grammatical form of higb commurucative value may occur. As attention to and detection of these grammatical forms becomes easier and the demands on processing resources are lessened, then the leamer can attend to and detect items of low communicative value. . Before moving on to our next duicus.;ion, we should acknow!edge that,

like any model, the one in Figure 2-2 is idealized and ignores individual char­acteristics of a given language. ln edclibon, this is a model of input process­mg and not. learning, A n;iodcl of leanung would attempt to capture accommodation of detected mput <intake) and how the developing system restructures. lssues such as frequency in the input and semantic complexi­ty of ~ form ~ter into e model of leaming, but not one of input processing. The ".JOb" of mput processing is to detect linguistic data and make initial form-meaning mappings, even if thcy are incomplete or partía! (i.e., the Ji:amer º?1Y maps one meaning onto a form). Thus, input processing pro­Vides the mtake for further cognitíve processing. Frequency is more relevant to the strength. of long term store of linguistic data, that is, the strength of the form-mearung connections as they ere stored in the developing system. Semantic ~mplexity-whether a single form has various meaningi; or funo­bons egam IS more relevant to ISSUe& of long term storage. Once a form­meaning mapping is detected and subsequenúy stored, it is the job of the developmg system to aocept or re,JCct e same form-different meaning map­ping made available to it. As we shaU see later in lhi.s chapte~ in our discus­sion about the assignment of grammetical and semantic roles, it is certamly the ~ that _the d~elopmg system may 1nfluence input procesaing itselfby allo~ or disallowmg cerlam form-meaning connections to be made.

Finally, our .model ignores issues of perceptual saliency for the time being. Perceptual saliency has been referred to earlier and involves whether or not a form has particular acoustic fcatures that render it more salient in the spcech stream than other grammetícal forms. Acoustic features that pr&­sumably affect perceptual saliency are syUabicity (SYUables are eas.ier to

Page 25: Input Processing and Grammar Instruction

32 VanPatten

detect than nonsyllables), stress (stressed syllables are easier to detect than weakly stressed or unstressed syllables). position in a word {beginnings and enda of words appear to be more salient than the middle parts of words except those carrying strong stress), and position in a sentence or utterance (utterance beginning and final po6itions are more salient that utterance internal positions, again except where strong stress is in volved). (See Hatch, 1983, for a discussion of acoustic-perceptual salieru:e.) It should be clear, then, tbat this model assumes that acoustic and perceptual factors are held equal. We will take up saliency again later.

ASSIGNING GRAMMATICAL ANO SEMANTIC ROLES

The Principies

Not ali ofinput processing is about attending to and detecting grammatical forms such as inflections and functors. When leamers hear a sentence, they must also process it in order to assign semantic or grammatical roles to the words they hear. Tlús means that they must assign roles such as agent and patient or subject and object to the nouns that they process. The present sec­tion will focus on this issue. The principies tbat we will examine are the fol­lowing.

P3. Leamers possess a default strategy that assigns the role of agent to the first noun(phrase) they encounter in a sentence. We call this the "first noun strategy."

P3(a). The first noun strategy can be ovenidden by lexical semantics and event probabilities.

P3(b). Leamers will adopt other processing strategies for grammatical role assignment only after their developing system has incorporated other cues (e.g., case marking, acoustic stress).

We will show that in sorne languages, such as Spanish, P3 has major con­sequences in terms ofthe type ofintake data tbat are fed to the developing gystem. We will suggest also that P3(b) may be developmentally late because of the intersection of P3 and the other principies examined in the previous section on cognitive aspects of input processing.

Meaning Once Again

In the previous section, we observed that learners are driven to get meaning from an utterance as part of comprehension. As we saw, tlús resulted in a series of principies in which the push to get meaning governed the process-

Input Processing 33

ing of discrete elements in input sentences. However, mearúng does not refer solely to the referential meaning encoded m words and grammatical form; meaning also refers to such things as who did what to whom. To grasp the meaning of an utterance, a leamer must also determine what semantic or grammatical role a noun has. For example, as fluent comprehenders of English we would assign the semantic role of agent to the lion and the role of patient to the hunter in the following sentence.

(1) The lion killed the hunter.

If we heard the following sentence, however, we would reverse our assign­ment of semantic roles, this time assigning the agent role to the hunter.

(2) The lion was killed by the hunter.

Our correct interpretations are based on our mature knowledge of English syntax. The syntactic rules in our mental representation of the lan­guage do not allow us to assign, for example, the role of agent to the hunter in the first sentence. What do second language learners do when the app~ priate mental representation has yet to be constructed? Somehow they must assign roles to the nouns they detect in a sentence. What principie or prin· ciples guide(s) them?

Research on second language leamers has revealed a strong tendency for them to rely on a strategy by which the first noun of a sentence is tagged as the agent. We can state tlús as the following principie.

P3. Learners possess a default strategy that assigns the role of agent to the first noun (phrase) they encounter in a sentence. % call tlús the "first noun strategy."

Simply put, this means that regardless of the actual syntactic configura­tion of the sentence, leamers interpret the first noun as the agent. Motivated by research in child first language acquisitíon {Bever 1970; Slobin, 1966), Nam (1975) showed how, for example, child and adult Korean learners of English as a second language incorrectly interpret passives. When presented with a sentence such as, "The lion was killed by the man," her subjects interpreted the sentence to mean that the lion did the killing and that the man was the entity that wound up dead In other words, they íncorrectly interpreted the sentence as "The lion killed the man."

Ervin-Tripp (1974) also reports on tlús processing strategy. She studied English speaking children attending a French-speaking school in Switzerland. She gave them passive sentences such as the following and asked them to act out the meaning of each sentence with toy animals.

Page 26: Input Processing and Grammar Instruction

..,

3-4 VanPatten

(3) La vache a dé chasstt par le cheuaL "The cow was chased by the horse. •

What Ervin-Tripp found was that her subjects tended to misinterpret the French passive sentences as actives, showing that the cow che.sed the horse ratbcr tban the other way around. She found this strategy in e.JI ages of chil­dren tbat she studied. Of note bere is tbe.t English and Frcnch form the pas­sive in e.lmost the exact se.me word-for-word wey. Thus, thc older children who clee.rly bad knowledge of pe.ssive structures in Engiish did not use this knowledge to interpret the French sentenoes. In other words, they did not rely on first language transfer to assign semantic roles.

A number of studies have been conducted in Spe.ni'lh (Bmkowski, 1992; Glil!an, 1985; Lee, 1987; LoCoco, 1987; VanPatten, 1984b) on how leamers of Spanish interpret sentences. Spanish has variable word order so that the following sentences are e.JI poesible, each meaning that the lion (león) killed (mató) the man (hombre).

(4) El león nwJé al hombre. (5) Al hombre IJJ nwJé el león. (6) Ma/I> el león al hombre. (7) Mall> al hombre el león

In addition, $panish obligatorily places clitic object pronouns in front of finite verbs. Thus the foUowing sentenoes are both poesible, each meaning that the ]ion (/eón) killed (malé) bim (/o).

(8) El león IJJ nwlé. (9) Lo nwJé el león.

Using various research instrumenta and different input scntences, these studies on leamers of Spanish show a strong tendency for them to misin­terpret sentenoes in which the first noun is not the agenl Thus, (5) Al hom­bre lo nwJé el león is misinterprctOO as, "The man killed the bon" instead of the correct "The lion killed the man." Sentenoes su ch as (9) Lo nwJé el Wm are misinterpreted as "He killcd the lion" rather than the correct "The lion killed him."

In her study cited ee.rlier, LoCoco also researched the sent.enoe interpre­totion of oooond Jansua¡;e loo.mora of German. Gennan mark.o nouns for case by inflec:tion of detenniners. Arnong her target sentenoes for compre­hension, LoCoco included sentenoes in which the dired. object preceded the verb 88 m the following example: Den LasbJx18m schubl da.s Auro, "The­ACC truck pushes the-SUBJ car" = "The car pushes the truck." There was a very strong tendency for lee.mera to ignore the case marking cues and use

l11>ut Proeess1ng 35

word order to a.;,;ign the grammatical-semantíc roles. Gass (1989) has also shown that English-spca.k:ing leamers of ltalian and

ltalian-speaking lcarners of English make use of thc first noun strategy to process sentences. This is an interesting finding vis a uis first language transfer since research on native ltalian-speaking adults suggests that othor cues (e.g., anirnacy) are more important than word order for assigning semantic-gramrnatical roles (see discu.ssion that follows on lexical seman­tics). MacDonald and Heilenman (1992) also provide evidence tbat adult English-speaking learners of French use a first noun strategy to assign semantic roles to nouns in sentenoes.

Research from various languages (both first and second) suggest.a that the first noun strategy may be a universal defaull strategy and not simply a matter of learners transferring knowledge of syntax from their first lan­guage during the act of comprehension. In one study of first language lcarn­ers of Hungarian, Pléh (1989) showed that her subjects relied on this strategy, even though Hungarian clearly marks case wlúch, in Bates and MacWlúnney'e (1989) terms, is a more reliablc cue to semantic roles. Pléh gave her subjccts sent.enoes in various word orders (SVO, OVS, etc.) that are possible in Hungarian. She found that thc Hungarian children do rely on sometlúng other than case marlting to assign scmantic roles, and concludcs that "as the data from 178 children [ J show there certainly is a strong bias toward using the first.noun-as-the-agent strategy " (p. 166). And whcn Pléh removed case markers from stimulus sentenoes, the first noun strategy waa even more prominent.

That the lirst noun strstegy may indeed be a universal default strategy in the beginning stages of acquisition is shown in the work of Bates and her colleagues (Bates & MacWhinney, 1989; Bates et al., 1984), even though they argue against it 88 a universal default strategy. In their Competition Model, word order is but one of a number of cues that languages may use to ensure proper assignstíon of semantic ro]eg by comprehenders. They clrum that no cue is uru,-ersaJ and that speakcrs of a language develop blB&eS toward one cue or the other based on the reliability, availability and C08t of the cue (see MacWhmney & Bates, 1989, Chapter 1 for detailed discussion). Based on data such as thoee presented in Bates et al. (1984), they arguc that adult speakers of Italian have a bias toward animacy as a cue and that English speakcrs have a bias toward word ordcr. They also provide data to show that children learn these biases early on. However, these biases only hold in thoee cases in wlúch tbe cues are put in to confüct and does not mean that the first noun strategy is not a universal or default strategy. In every testing situation that Bates et al. report, the ltalian speakcrs, children and adult, show evidence of be¡¡inning with a word order strategy and relying on animacy only in confhct situations. Thus, whcn g¡ven sentenoes such as the following in ltalian

Page 27: Input Processing and Grammar Instruction

36 Vanf>atten

(10) Thc rock bit the giraffe

speakers of lt.alian select giro/fe as the agent and not the first noun, rock. Animacy here is m conllict. wíth word order SU1ce inanimate things don't bite. But when given sentencea in ltalian m which animacy is held constant in the sentence, Italian speakers tend to selcct the lirst noun as the agent.

(11) The rock bit the Jea( (12) The monkey bit the girafl'c. (13) The monkey the giraJTe bit. (14) Bit the monkey the giraJTc.

In a certain scnse, then, thcse data cannot be used to argue against a uni­versal first noun strategy for a.ssigning agency to a noun; they can only be used to argue that speakers of aome language develop additional strategies or rely on additional cues in addition to word ordei: In short, everyone starts out wi:th a word order strategy but, dependmg on the language being learned, may wmd up with add.itional and, presumably; more reliable cues.

This leadii us to another principie for asaigrunent of semantic roles dur­ing input processing. We state it in the following way.

P3(a). The first noun strategy may be ovenidden by lexical semantics and event probabilities.

Lexical semantics refers to the constraints on a situation impo6(ld by the semantics ofthc verb involvcd. The verb kick, for example, requircs an ani­mate bcing with leg:s to do the kicking. Fbr this reason the sentence He lticked hu wUh his ear is anomalous as is the sentence The snake krdu!d the horse. An event probability refcrs to the likelihood that a given situation would eióst m the real world, ewn though lexical semantics allows it. The veril corret:t allows any anímate being capablc of instructional behavior to correct someone or something else. However, of the two situations that fol­low, one is more likely than thc othet:

(15) The parent corrected the child for bis behavior. (16) The child corrected the porent for bis behavior.

Most or ua would agree that tbe first sentl'DN' i• much more likely than the second. 1 t thu.s has a higher degree of event probability.

In first language acquisition and in adult native-speaker studies, lexical semantics and event probability alfect semantic role BSS1gnment. Bever (1970) showed that children who m.isinterpret passives such as The cow was kiclu!d by the horse do not regularly m.isinterpret sentences such as The pan-

Input Process1ng 37

cake was eaien by the man. The lcx:ical semantics of eot precludes an inani­mate object eating an animate obje<:t in the real world, and thus children do not use the word order of the pancake sentence in a.ssigning semantic-gram­matical roles. Bavin and Shopen (1989) show that Jexical semantics can ovenide thc first noun stratcgy for active sen\A>nces as well They gave Walpiri speaking children a senes of active sentcnces m which the lexical semantics allowed for either animate or inanimate nouns to pcrform or cause the action, for example, "trip" as in a pcrson can trip anothcr person or a log can trip a person. They also gave them sentences in which the lexi­cal semantics allowed for only arumate objec:U to perform the act.ion. They found that the first noun strategy was operati11e m the first type of sen­tences, but was m much less evidcnce with the second type. In nnother part of their study, they gave the children sentences m which event probabilities would favor one noun over anothcr as the agent The children showcd a very strong tendcncy in the earlier stages to use event probability rather than word order to 8bolgll the role of agent.

Both Gass ( 1989) and lssidondee and Hulstijn ( 1991) present data on the uútuence of lcxical semantics on the first noun strategy for second Ianguage leamers. Gasa gave Eng!ish·spcaking learners of ltalian and Italian-speak­ing learners of English sentences in which animacy conllict.ed with word order. That ia, the first noun was an inanimate thing but the Vtlrb required an animate being as the agent, for example, The rock bu the gira/fe. Her data were uneqwvocal in showing that both English- and ltalian-speaking learn­ers relied more on lexical semantics than word order to assign the semantic role of agent to the seoond arumate noun in thesc kmds of "conflict." situa­tíons. Issidorides and Hulstijn tcsted second language Jearncrs of Dutch with various first languages on thcir comprehension of adverb-verb-subject­object sentences. They found a significant drop in a.ssigning subject. status to the first noun when it was inanimate and the second noun was arumate.

Given the reeearch from first Ianguage aCQWSltion and the hmited rcsearch from second languagc acquisition, 1t is reasonable to posit P3(a) as a principie thot nlTect.s semantic role assignment. Although the first noun strategy may be a universal starting point for proccssing sentences, it seems that lexical semantics (and most likely event probabilities) will play a role in those situations in which the anunacy of the first noun is at odds with the lexical na tu.re of the veril.

In the previous scction on attention and input processing, we auggested that the formal features that !carnera could attcnd to and detect. in the input change over time as comprehension became leas effortful Concerning the assignment ofscmantic roles, what guides leamers' processing ofinputal.so changes over time. We posit the following general principie.

P3Cb). Learners will adopt other processing strutegies for grammatical-

Page 28: Input Processing and Grammar Instruction

38 VanPatten

semantic role assigrunent only after their developing system has incorporated other cues (e.g., case marking, acoustic stress, subject­verb agreement).

This principie is based largely on first language data, although there are a few studies from second language acquisition that are relevant. We will examme the data from first language acquisition first.

Languages vary as to whether they mark case or not and whether or not they have "free" word order. English is strongly SVO, whereas languages such as Spanish, Hungarian, Hebrew, Turkish, ancl Walpiri allow other orders, including SOV. VSO, OVS, and OSV. Significantly, most of the languages that allow other than SVO order mark nominal case in some way, usualJy marking nonsubjects. Spanish, for example, marks the accusative noun phrase with the preposition a when lexical semantics allows for the aooisative noun to be equal­ly capable of performing the action or event represented by the verb. The sub­ject remains unmarked. (Although languages like those listed above may also mark nouns other than the direct object [patient] of a verb, we will limit our· seives to discussion of subjects and direct objects.) In the following examples, the verb kül allows for either tM man or the lion to perform the act of killing. However, the verb reod does not permit the letter to perform the action. (The marker a contracts with the masculine singular definite article el ~formal.)

(17) El señor maJó al Ú!Ón. "The man killed ACC-the !ion."

(18) El león mató al señor. "The !ion killed ACC-the man."

(19) Al león lo maJó el señor. "ACC-the !ion him killed the man."

(20) Al señor lo maJó el Win. "ACC-the man him killed the !ion."

(21) El señor leyó la carta. "The man read the letter."

(22) •El señor leyó a la corta. (23) •A la carta la l.eyó el señor. (24) La carta la leyó el señor.

Hungarian, in contrast, inllects a noun with ·I (or -<!! and o{)t, depending nn thP. !dP.m of the noun) in order to mark accusative case. Like Spanish. the subject noun remains unmarked for case. The following sentences, taken from Pléh (1989), are ali possible in Hungarian.

(25) A macskcU átugorja "The cat-ACC jumps over

akutya. the dog."

(26)Amacska "The cat

(27) A fiú "The hoy

(28) A úinyt "The girl-ACC

átugorja jumpsover a úinyt the girl-ACC afiú the hoy

a kutyát. the dog-ACC." Mrgeti.. chases." Mrgeti.. chases."

Input Processing 39

Research in first language acquisition has shown that first language learners must become sensitive to these cues to role assigrunent during the couzse of development (see, e.g., Bates & MacWhinney, 1989, pp. 59-70). That is, the cues are not immediately apparent to children and it is possible that the cues in one language are acquired earlier than the cues in another. A frequently cited example involves Turkish. Turkish children seem to have acquired the case marking system for correct role assignment by the age of two years (Slobin & Bever, 1982). In contrast, children acquiring Hungarian show evidence of correct role assignment much later, after the age of five years (Pléh, 1989). Children of French do not acquire adult-like cues based on animacy until well into adolescence (Bates & MacWhinney, 1989). These cross-linguistic differences are attributed to a variecy of factors falling with­in the framework of the Competition Model (Bates & MacWh.inney, 1989; McDonald & MacWhinney, 1989), wh.ich we will discuss in a later section. These factórs largely are related to a=mmodation and restructuring of the child's developing system, and thus fall outside the scope of the present dis­cussion on input processing. What is relevant, however, is that input pro­oessing itself is not guided solely by universal and invariant strategies; as language acquisition progresses, input processing is also guided by informa­tion contained in the developing system itselt: That is, at sorne point, the knowledge stored in the developing system is utilized during on-line input processing, as learners either abandon the first noun strategy or utili.ze other granunatical information as well.

There is some evidence that the way(s) that second language learners assign grammatical and semantic roles during input processing develops over time like those of first language learners. McDonald & Heilenman (1992) report on second language learners of French and second language learners of English. In their study, hoth groups show gradual acquisition of sentence processing strategies that approximate those of native speakers with the second language learners of French showing a late development of the use of verb agreement as a cue to suQject-OQiect roles. In her report on learners of ltalian and English, Gass (1989) shows that ESL learners of var­ious first language backgrounds consistently rely on word order (the first noun strategy) for sentences in which lexical semantics plays no role. For sentences in which lex:ical semantics does play a role Gass found that the learners initially used both word order and the lexical semantics ofthe verb

Page 29: Input Processing and Grammar Instruction

40 VanPatten

equally when int.erpreting the Sl'nt.enoes. Overtime and quite gradually, the leamers increasmgly used word order only as the major cue to sentence pro-oessing.

The ability of second language leamers to rcly on cues other than word order may be slow whcn comparod wilh sucb development in first language leamers. The problem for seoond language input prooessing is lhe following. As we posiUd in the first sed.ion of this chapt.er, leamers are driven by the push to get meaning. Lexical 1t.ems are processed before grammatical cues wben both express the aame mearung. Gnunmatical and semantic roles oer­tainly are a part of mee.rung sinoe lhey are conoemed with the who and the whom of events. These roles are expreesed by nouns that are lexical items. As posited in the pre&enl section, leamers use word order cues to assign grammatical and semantic roles baacd on the serial processing of nouns Oex· ical items). Thus, we can al.so say that leamers prefer word order to gnun­matical cues Cor proccssing and w.signíng semantic roles to nouns in a sentenoe.

In the preceding ecenano, thc grammatical cues for role assignment are inelevant from the point of view of tbe leamer's intemal prooessing mech­anisms since the first noun strat.egy takes care of the bulk of tbe processing. Thus. these cues are initially low in commwúcative value. In a oertain sense, tbey function li.ke redundant markers (e.g., thtrd person -s, Spanisb a<ljec­tive agreement) for thc leamcr and we hypothesize here that they are ignored in the input for sorne time. Thcre is no empirical researeh to sup­port thts argument, however. Expcricnce with learners of Spanisb shows that the accusative markcr a is a late acquired item, often not showing up consistently in the output of graduating Spanish majors at the wúversity leve!. Expericnce also shows that thcsc somcwhat advanoed leam ers of Spanisb may still have difficulty with verba such as gustc.r, "to please" (Spanish has no equivalcnt of to l1ke), which requires an almost obligatory OVS word order in Spanish, e.g .. A Juon le gU$1c. el heladfJ, "ACC-Jobn pleases ice cream." Experienoo further shows that learners of Spanish have difficulty in setting thc null..subjcct parameter. In this latter case, many advanoed leamers of Spanish do not rcally have a null..subject system, but instead overtly mark sub)CCts of verbs far more than a native-speaker would. Another persistent problcm mvolves ol:ucct pronouns and rellexive pro­nouns that in Spanish obligatorily precede firute verbs, for example, Lo veo, "ACC-htm see-1," & cc~tc.. "REFI..- puts to bed-he." Errors such as *Lo viene for "He comoo" and *Mt hohlo mucho Cor "l tálk 8 lot" sugg<>St that leamers bave misinterpret.ed the grammatical and semantic roles of these pronouns, u.sin¡¡ them as subject pronouns instead. These problems in out, put may very weU be related to the first noun strat.egy that would mitigate ag:ainst the detection of oertain grammabcal forms and the wrong function •ssignment to certain pronouns. This in tum would have subEequent 1eam-

~

--------- ---- ., 1 1

l• 1 1 1 1 1

11

,.l., 1 1 1 1

: 1 1: • 11l1 : :::::~)

111! j 1~ 1

1 1 1 1 1 1 1 1 --------

_______ _,

~~

1· t~ •e '"' ~

l < ~

~~! -~ .. j i!. .s ~ Jl

J l .s 5

·t i ·.!!

l l e E. 's

i E <

~

~ o ¡;:

41

Page 30: Input Processing and Grammar Instruction

42 VanPatten

ing effec!$ far things such as the null·subject parameter. We leave these issues to future research.

Befare summarizing, one final point deserves mention. When we say that learners use a first noun strategy, there is an assumption here that learners know what a noun is, a verb is, and so on. When learners incorrectly assign the subject-agent role to macsluú, "cat" in an input string such asA macskát átugorja a kutya, they simultaneously show evidence of having parsed the sentenoe into noun phrases and verb phrases. Th.is would suggest that leamers have access to universal lexical categogies such as nouns, verbs, adjectives, and so on. We thus suggest that part of on-line input prooessing also involves first assigning words their referential meaning, e.g., macska = "cat," as well as their grammatical category, far example, macska = NOUN. In this way, input prooessing relies on oertain knowledge sources such as Universal Grammar (which contains the abstract grammatical categories) and the semantic network of referential meanings contained in the devel· oping system. Any model of input proces.si.ng would have to include an indi­cation of the processors making use of various knowledge sources.

Summary

In the previous section, we have examined the issue of grammatical·seman­tic role assigrunent during input processing. We have posited that word order is the starting point far second language leamers with a heavy reliance on a first noun strategy. This strategy may be overridden in those cases in which lexical semantics does not conform with the role assignment. We aJso have suggested that the word order strategy rnay be overridden with time as the leamers' developing system incorporates other cues to role assigrunent. These cues are utilized during on-line processing to help determine whether or not the first noun can indeed be assigned the subject case. Figure 2-3 summarizes the operation of these principies during on-Jine input process­ing. First, the processors assign grammatical categories (N, V, etc.) to the content words that are detected using information from UG (and, of course, the semantic meanings stored in the developing system). The processors then deliver this information plus the serial order in which the items occurred to that part of prooessing that assigns grammatical and semantic roles. In the early stages, the prooessors use the first noun strategy to assign grammatical roles. Later in second language acquisition, if the developing syntcm contnins othcr cues to grommatical and aemantic role MSignment, the processors will make use of these cues to assign grammatical roles and will override the first noun strategy if appropriate.

In Figure 2-4 we integrate role assigrunent into a more general model that includes the processing of grammatical forro as described in the first

·---------------------. ------. -.................................. . . ¡ i ¡ i ! !

Input Processing 43

FIGURE U An intA!grated modelo( aecond languaee input p"""'6Sing for gmrnmatical form and role assignment.

section of this chapter. The result is a more complete picture of on-line sec. ond language input processing.

OTHER ISSUES IN INPUT PROCESSING

There are a number of issues related to input processing that we have pur­posely ignored until now. Sorne of these issues relate to the conceptualization of attention as discussed in the first section. Others relate to issues of aooustic saliency and individual language characteristics as welJ as other models of language processing. Still others relate to so me possible objections or confu -sions about input processing that can be caused by uúsinterpretation ofthe ideas expressed in this chapter. We will take each of these issues in tum.

Consciousness and Awareness

In previous publications (e.g., VanPatten, 1985b, 1990, 1994, 1995) 1 had explicitly linked attention to consciousness and awareness. In agreement

Page 31: Input Processing and Grammar Instruction

44 VanPatten

with Schmídt (1990), 1 argued that processing grammat.kal form in the input required attention to form in the input. 1 viewed attention as having some degree of awareness or consaousness in it. Borrowing from Ceci and Howe (1983) and f'o5ner and Snyder 0975), 1 suggested that "an opera­tional defirution of attention B88umes aome kind of consciousness in that the Iearner is either aware of the process or the product of attention. Because attention involves sorne degree of consciousness, it is capacity robbing" CVanPatten, 1994, p. 28). It was this model of attention that informed my previous research.

More current research in cognitive pcychology. as described in Tomlin and Villa (1994), suggesta that awareness or consciousness is nota necessary component of eome aspects of attention. In reviewing work by Posner and bis coUeagues (e.g., Posner & PetenlCn, 1990; Posner & Rothbart, 1992), Tomlin and Villa conclude that

none of the oentral componcnta or attenbon-<llert.nea, orientation, ar detec­tion-requare awaren-. e1ther to operai. or as the result oí p,.,......;ng OeWction 11 the moot related to awaren-, but there is considerable evídence mdlcating that information can be oot¡ruttvely detected oven though the indi­vidual is not aware oí1t havm¡occurrod (p. 193).

In spite of these recent developmenta, it is not clear that this conclusion is readily applicable to second 1anguJl8ll acquisit.ion. Moet if not ali studies in attention and leaming use adulta, and none of these studies focuses on either first or second language acquisition, although a few involve language phenomena. In Jexical priming studics, such as Marcel (1983), as cited by Tomlin and Villa, adult native-spcaking subjects read a word such as nurse on a computer screen. Studies reveal that subjects read the word more rapidly wben appearing immediately after the semantically related word docwr as opposed to the more semantically unrelated word o.cwr. However, subjects cannot indiaite above chance that they have seen the word doctor prior to being asked te read the word nurse suggesting that they do not ha ve any awareness of the word doctor. Such studies are used te conclude that detection (in this CB11e, dcroction of a Jexical ítem) does not necessarily involve awareness.

Wbat is important here is that the rescarch uses adult native speakers. Tbese adult native speaken already have a fully devcloped semantic and lex­ical S}'8Wm 01 addition to weU-reheBl'llCd retrieval mechanisms that they use when proce;sng incunung l.ungua¡¡e. Tiie pou1icular lexicKI items they prooess in the above deecribed studiee are wonls (and not grammatical items, by the way) that thcy have a1roody learned and are well entrencbed in the long term store that comprises thCU' leiocal-&emantic network. In short, the subjed$ oren 't uwolU«i in the proca..ng of nooel fonn-nuxuting ma¡tpings;

Input Processing 45

they ore bemg askecl w perform with krwwledge that they already have. This is quite different from the mitial form-meaning mappings that a second lan­gw181? leamer must detect and lcarn. The lack of awareness of adult native­speeking suQ)ccts in studics such as lexical priming could be a result of klanung and exrended language use. We cannot conclude from these studies that initial pl'OaltiSing in the input thal is crucial to builcling the linguistic sys­tem in second 1anguage acquísition ocrurs withoul awareness.

In the same vem there are studies cited lo indiaite that leaming of linite state grammars and structural scquences happens without awareness. Firute state grammars contain e set ofletter nodcs (e.g., X. Y; Z. S. T. U) con­nected by one-way arrows. The direction of the arrows depict in what serial order the letters can be coloaited. These grammars can thus generate stnngs of lctters that function as "mput sentences" te adu1t subjects who subsequent to the mpul phase are gi\'en e surprise grammatiailityjudgment task with correct and incorrnct strings of letters (e.g., Carlson & Dulany, 1985; Carr & Curran, 1994; Dulany, Carlson, & Dewey, 1984; Reber, 1976). Subjects peñorm better than chance on surprise grammaticality judgment tasks ofletter stings but ofl.en are unable lo articulate why a string is gram­matical or noL 1 have roviewed the problems in applying these kinds of resulta lo 8CCOnd language acquisition elsewhere CVanPatten, 1994) and will briefty (re)state my argument here. lntemalizing the ordering ofletters in a string is not the samc as thc very complex processing and learning of new form-mcaning mappings in a second language. Finite state grammars do not encode any kind of mcaning (the lctters are simply letters and do not repre­sent any real world mcaning), they do not contain inflections, functors, and other suñace fcatures of language, they do not contrun movement rules and varying word ordcrs, and thc studies based on them involve purely visual processing during thc learning phase. Even Carr and Curran (1994, p . 207) &dmit that the application of such studics te d.iscussions on second language acquisition muet be viewed with some aiution and can only serve as a focal point for discussion nt this time.

In addition to these criticisms of finite state grammar, we are faced with the same problcm in applying the results ofthese studies te second language acquisition that wc faced carlicr. The subjects are not second language Jeam­ers acquinng new form-mcaning mappings; they are adults learning to mal.ch alread,y known fonns (letters) te serial patterns. However, second lan­guage leamers are proces&ng new forms and in some cases mapping these on to new mcarun¡:s such that thell' task cannot be deemed equivalent to th06e experienced by subjects in the experiments on cognitive psychology.

Because of these problems in interpreting cogniti\'e research on con­saousness and awarenCIS and applying them te second language acquísi­tion, 1 ha•-e opWd to eliminate references to consciousness and awareness altogether from the principies and the model of input prooessing developed

¡

l

Page 32: Input Processing and Grammar Instruction

46 VanPatten

in this chapter. Although my position is that awareness is probably a part of input processing at least mitially, it is not necessary for positing the strate­gies descn'bed m previous sectiona. What 1s necessary is the concept of lim­ited processing capacity Rccall that the model for attention to and detection of form m the mput developed in Uu.s chapter relies on the notion that detec­tion consumes attent.ional re&0uroes. In research on structural leaming, Carr and Curran (1994) report that, even though structural leaming seems to happen outside of awarenC88 with adulta m nonlanguage tasks, it does require focal attention-wlüch is synonymous with detection in their dis­cussion. When subject.s' must perform a concurrent task, their ability to leam the structural pattem&-and recall that these patterns are believed to be leamed without awarenC811 by the ~rity of researcher&-is impaired if not made impossible. Carr and Curran also cite the effect.s of a limited capac­ity on learnmg finite state grammars. Furthermore, Dienes, Broadbent, and Beny (1991) performed an expenment in which sub,JeCts were given the usual experimental exposure to input lctter-stnngs generated by the gram­mar but were ask.ed to perform a concurrent task_ The results revealed a marked decline in the subjccta' post-treatment ability to judge strings as grammatical or not suggesting that their implicit processing of the letter strings during the leaming pha.se was impaired. (See Whittlesea & Dorken, 1993, for additional related rescarch.)

We argued earlicr that the research on awareness based on finite state grammars is not neccssarily applicablc to second language acquisition because of the inherent complcxíty of natural languages and the fono-mean­ing mappings that must occur. Here, however, the resulta of studies such as those conducted by Dicnes, Broadbent, and Berry (1991) are relevant. If subjecta revea! a limited capacity to procees rather simple strings ofletters, logically it follows that to pl'OCCl!S ncw forms and perform ali the mental operations required to map form onto meaning in second language acquisi­tion must be at least as taxing if not more. As Carr and Curran (1994) state

The syntax of natural languagc p..-nts a very complícated hierarchic orga­niiation with many inten-elat.ed conditional rules on what elements can com­bine with one another. Thettfore, 1t aeema a plausible gueas that natural language leamin¡¡ taxea the 1)111.em'a abllity to keep track of context more eeverely than leerrung the model ¡nmmara used 1n structural leaming tasb ... ayntacbc leemmg ma,y well require, or et leest beoe6t from, focal altention (p. 224l.

Thus, while we sídestep the issue oí consciousness and awareness in the present chapter, we do so without !OM to our model of input pl'ON'S<Óng It seem.s that detection during proccssing consumes attent:ional resources and that we can test thís with or without ímbuing it with any degree oí con·

Input Processing 47

sciousnl!58 (cf., McLaughlin, 1990). We leave the issue of consciousness and awarenC88 to future rcscarch in both cognitive psychology and second lan­guage acqwsition.

Form Versus Meaning

Very often, d.tacu.ssiona of second language acquisition pit fono against meaning. This has led to discu.ssion in both second language acquisition cir­cles as well as language teaching circles of "attention to fono" and "atten­tion to meaning" (see, e.g., the discussion in Garrett, 1991). In posíting Pl, P l(a-<:) and P2, it IS cona!ivable that sorne would interpret these principies as dissociating meaning and fono as well Howevei; since input pl'ON'S<Óng IS concemed wíth how leamcrs make íonn-meaning connections when attending to input, the question cannot be whether leamers attend to mean­ing or to fono. The qucstion IS under what conditions they can alJmd ro both and how attcnt:ion to form and meaning develops over ti.me. As stated in VanPatten (1990)

What ia cnt1cal to keep m mind ia t.hat lhe lSBUe of attention to fono in the input ia only an iMue when t.he input is oommunicative in nature, i.e., it car· rico information to which lhe leamer is supposed to attend. This is the kind of input lhat ia typically found in nonclassroom settings and in certain kinds of classroom methodologfos, e.g., lhe Natural Approach. While humana may indeed dlrect rconacious] attention to form in and ofitself, the question is not whether they can do this; the qucstion is whdher or not tluiy can cUi this whi.h they prooess the input for moonrng (p. 288, emphasis original).

For this rea.son, thc principies sketched out in this chapter suggest con­ditiona undcr which form-meaning mappings can occur; they do not suggest that fono and meaning necessarily are separate. In adclition, they do not suggest that early stage lcarners should not have attention clirect.ed toward form via instruction; they only suggcst that certain kinds of fono are more difficult to detect than othcrs when leamers are left to their own devices. In the next. chaptcr, we will &ee just how Jeamers' attention can be clirect.ed toward form in the input without loss or meaning.

Operating Principies

For tho:;e familiar wíth thc work on Operatmg Principies in first language acquisition (e.g., Petera, 1985; Slobin, 1973, 1985), a number of questions may ari.se. To what extent are the prmciples described in the present chap­ter difTerent from those dcscribed in the child first language literature? Are thcre principies in first language aoquisition that are applicable to sec-

!.

Page 33: Input Processing and Grammar Instruction

48 VanPatten

ond language acquisition? To be oure, there is sorne relationship betwcen first language Operating

Principies and the second language input proce!ISing prinaples posited in this chapter .. In pnMous work (e.g., 'kn.P&tten. 1984b) 1 have acknowledged the role of first language acquisition research m sug¡¡csting research ques­tions for sea>nd language acquisition. Howcver, Opcrating Principies and the principies d~bed in the present chapter are not equivalent. Let us take, for example, Slobin's well known Opcrating Principie "Pay attention to the cnds of words." Slobin posits this Opcrating Principie to account for the fnct that nominal case in.fiections (noun-<?ndings) are acquired earlier by Turk:ísh children than case marking pronominal mnrkcrs are by Serbo­Croat.lan children. He suggests that childrcn are guided by a principie tbat directa their attention t.o the ends ofwords rather than t.o functors and pre­posed markers. Although his data certainly support this c1aim, he posited th.i.s Operating Principie t.o account for cross-linguistic differcnces and we certainly could apply the Opcrating Principie t.o acoond language acquisition as weU. We could hypothesize tbat English speakers learning Turk:ísh will acquire tia case marking system before English-speakers wilJ acquire the same for Serbo-Croatian. Such principies, then, could eaoily be inoorporated tnt.o thc present model and researched.

Whal is different about Slobin's Opcrating Pri11C1ples and the principies outlined in this chapter is that Slobin mixes input proccssmg strategies with both learrung and production strategies in his list of Opcrating Principies. "Pey attention t.o the ends ofwords," for example, is a principie tbat guides child input proces•ing. An example of a principie that guides production is "Avoid interruption or rearrangement of linguistic units." This principie acoounts for the fact that learners prefer canonical ordcrs in their output and is reflective of how the internal system has st.ored and organized the intake data. As such, the principie has little t.o do with actual on·line sen­tence proccsai.ng and instead refers t.o what leamen do with the data aft.er it has been initiaUy processed In our model of seoond language input pro­cessing, the principies are restricted t.o input pro<a&ing alone. We are not ooncemed with prinaples of learning, ie., principies dírectly related t.o the accommodation and restructuring of intake data Cace Figure 1-3 in Chapter U, although we will touch on this issue in Chapter 5. Likewise, we are not ooncemed with 1SSUes of on-line sentcnce production.

Petcrs (1985) also p00ts what she calls Extraction and Segmentation heuristics. &sed largely on phonological cues in the input, Extraction heuristics explain how children may initiaUy pcrccive and process units in the input larger than the word while treating thc unit as a word. For exam· ple, onc of Peters' Extract.ion heuristics is the foUowing.

EX: SlLENCE. An extractable unit is bound by a1lcnce.

Input Process1ng 49

This heuristic acoounts for how childrcn lflltiaUy perceive and attend t.o chunks of language that are bookended by pauses. Segmentation heuristics then aUow children to subsequcntly break off píeces of extracted units as they begin t.o make intemal oompansons. These a>mparisons result in novel data. One example of a segmentation heuristic ;,; the following.

SG: REPETITION. Segment otT subumta that are repeewd Cin tenns of seg· mental& or rhythm or intonationJ within on Extracted urut and store them separately.

Thus, a child who may have initially extracted Whensyour as a unit based on repeated input strings such as "Where's your nose?" "Where's your eye?" "Where's your truck?" and so on, may encounter new input oontaining sen· tences such as "Where's the spoon?" "Wherc's the dog?" In this case, acoord­ing t.o Peters, the child is equipped with aegmentation heuristics t.o compare the intemal unit Where8your with the new Wherest~ t.o determine that Wh.?res is an even smaUer unit. In J)06lting Extraction and Segmentation heuristics, Peters can aocount for why children 's units of speech may be dlf. ferent from thooe of adults and cert&n!y different from those of linguists.

Although Peters' extradion heunstics certainly seem t.o be ooncemed with input processing, her segmentation hcuristics often are related t.o leaming rather than ro proceesing. That 15, they are ooncemed with how the child accommodates new intake data and how the developing system subse­quently restructures itself. Her extraction principies dovetail nicely with the principies outlined in the present chaptcr, but are driven more by issues of perceptual (rcad "aooustic") salicnoo than thcy 11te by is.sues of attention and capacity. There can be no doubt that aooustic and perccptual saliency intcr­oct with the principies outlined in the present chapter. For example, let's take two forme of relatively equally low oommunicative value from the per­spective of the seoond language learner. One, howevez; happens t.o occur m the middle ofwords and generally reccivee weak stress; the other occurs at the ends of words and sentcnces and carnes strong stress. Our principies would prcdict that thEQJ forms would be processed and aa¡uired later than forms with oonsistently high oommurucative value. However, Peter's heurís­tica would predict that the form OOCW'Tlllg at the ends of utterances would get extracted and segmented before the one occurring in the m.iddle of words. Her heuristics, however, would not make any prcdiction based on rel· ative oommunicative value when perccptual saliency is held oonstant. Our principies oould not make any predlctlon about earlier or later processing when communicative value was held oonstant and perceptual saliency was varied Thus, the principies in the two different models are oomplemcntary and not mutually exclusive.

The oonclusion t.o be rcached is that a fully articulated model of input pro-

Page 34: Input Processing and Grammar Instruction

50 Vanl'atten

CC8Sing would need to include both principies based on attention and capac­ity as these mteract with communicative value and perceptual saliency. For the purpo11e5 of the present book, we have chosen to focus on the former­along with the principies involved m grammatical and semantic role assign­ment. Careful research needs to be conducted to mvestigate the intersection of cogrubve aspects of input processing in second language acqu.isition with acoustic-pen:eptual saliency.

The Role of Flrst Language Transfer

First languagc transfer has a long bistory of debate in seeond language acqui­sition theory and research. Most views of transfer focus on issues in leaming (ac:commodation and restructuring) or in speech processing (bow the first language inlluenoes the Wtrf in which lea.mera put utterances t.ogether). It is not unmccliately clear how first language transfer may inlluenoe input pro­oessing, but we can specidate by looking at the issue of word order.

First, 1t is widely accepted in second 1anguagc rseaicll that first language transfer is constrained by certain principies of leaming (e.g., Andersen, 1983; Hylt.enstam, 1987; Wbite, 1989). ln one view, transfer is triggered only when crucial similarity between the first and second language bec:omes apperent to the leamer or the leamer's developing system creates a rule that is similar to a rule in the first language. (Sce, e.g., Andersen's Transfer to Somcwhere in Andersen, 1983.) In tlús view, thc developing eystem is vuJ. nerable to "soopage" from the first languagc system as the internal mecha­nisms make comparisons.

For examplc, as posited in a previous section, learncrs tend to assign the first noun of an input string the role of grammatical subject or semantic agent. This means that they wiil assign the role of object to the second noun. Thc intake dats that they are feeding to thc dcveloping system would sug­gest thst the language they are leaming is mvariantly SVO. Thus, the accommodation and restructuring mechanisma involved in leaming would o-eate a rule that says, "This language is SVO." Now let's imagine that leamers with first languages like English that have mvariant or eystemati­caUy canoníeal SVO word order are leaming a language like Spanish that has variant word order: Tbe "natural rule" crcated in the developing eystem through the normal course of leaming has set up a &1tuabon ripe for first language transfer; the SVO rule of the first language strengthens the SVO rule or the developlng system.

In our model of input processing (see Figure 2-4) we have set up a devel· opmcnw.J situation in which learners' processing mochanisms begin to check thcir assigrunent of grammatical and scmanlic roles aga.inst the rules containcd in the developing system. ln this way, thc first nowi strategy may

Input Processing 51

still operate but the results of its operation are checked against the knowl­edge (rules) stored in the developing sysú!m. lf the developing system con­tams an SVO rule with no other cuesto as..igning grammatical and semantic roles to nouns, and this SVO rule has bcen strengthened by the first lan· guage, then we see that transfer operates m an mdirect way in second lan· guage input processing. mi will t.ouch upon first language transfer again in Chapter 5 as we explore the relationslup between input processing and Universal Grammar.

The Competition Model

Thc Competition Model has hecome a rooognizcd framework for examining both first language use (adult comprehension of sentenoes) and first Jangua.ge acquisitio.n. Only recently have scholars bcgun to explore the pot.ential ofthis framework for second language acquisition (e.g., Gass, 1989; McDonald & Heilenmann, 1992) and in the previous sections we bave referred to it on occasion The question that anses is whether or not input processing and the principies estabhshed in this chapter fall within tbe Competition Model, espe­cially saooe so much work in the Competition Model has been concemed with gnunmatical and semantic role assignmcnt. In order to answer this question, a brief overview of the Competition Model is in order.

The Competition Model is a functionalist model designed "to capture facts about comprehension, production and acquisition of language by real human beings, across a variety of qualitatively and quantitatively distinct language types" (Bates & MacWhinncy, 1989, p. 3). The principal idea ofthe Competition Model is that humans must dcvelop form-function mappings for language use, that each languagc possesoos "cu es" that are utilized dur­ing on-line comprehension and production. Adults use these cues, whereas children must leam what they are.

The Competition ModeJ has one major prcdictive consb"uct: cue validity. In sunple terms, cue validity refers to the degree to which a cue helps a com· prehender in making corTeCt int.erpretations during on-line sent.ence com· prehension- Cue validity itself cont.runs thrce components. The first is

availability. wluch simply means the degree to which a cue is available when )'OU need 1t. Subject.-verb agreement is nota read.ily available cue in English 6lllce verbs do not inllect for person-number, except for present-tense, third pcrson singular. Subject-verb agreement is a readily available cue in lan­guag~'ll like Spanish and !tallan. The t.eWnd wru.tru~i. ¡,, relliWifüy ttnd n:p­rcsents the degree to which a comprehender can count on a cue in making corro<.t interpretations. In English, word ordcr is about 100% reliable for assigning semantic and grammalical roles to nouns. Word order is less reli­able in Spanish. The tlúrd construct, confüct validity, refers to the degree to

Page 35: Input Processing and Grammar Instruction

52 VanPatten

which a cue reoders a oorrect interpretation when it is ID cooflict with anot.her cue. Cooftict situatioos are mfrequent m natural language situa­tiona but fono the &tapie of the research paradigm for t.he Competitioo Model. In languages like English, conJlict can never happeo with word order 6UlCe other cues (sub,ect..verb agreement, ca.se marking oo oouos) are gen­erally unavailable. Conftict can happen in languagcs likes Sparush as we saw ín a previous section of this chapter. Word ordcr ( "the first noun is the sul>­ject") may be in conftict with ca.se marking ("but the first ooun is ca.se marked with a") and so the comprehender must use one cue or the other to assign grommatical and semaotic roles.

Althougb the Competitioo Model accounts for adult on-line comprehen­sion issues, Bates and MacWhinney apply the modo! to first language acqui­sition. Within the Competitioo Model, childrcn's acquisitional problem is one oflcarning about cue validity. Over time, childrcn must intemalize what the best and most valid cues are for ali comprehension S1tuations in the lan­guage. According to the mode~ children are equipped with lcarning mecha­nisma that compute weights and strength.s of cues ID the de\-eloping system 80 that they wind up with the appropriate adult system <McDonald, 1989; McDonald & MacWhinney. 1989). In this way, the child develops an internal system that serves as a template or resource agrunst which the interna! proce&IOrs juclge mcoming cues.

From thia brief description, it should be clear that the Competition Model is an attempt to link knowledge sources with on-line prooessing. In terms of the principies presented in this chapter for sccond language input process­ing, the Competition Model can be seen as complcmcntary and u.seful for understanding how P3(b) develops; it cannot tell us how learners first <ktect and process gnunmatical fono in the input. Thc Competition Model simply assumes that grammatical form is prooessed by child first language acquir­ers. Recall that P3(b) says the following: Learncrs will adopt other process­ing strat.eg¡es for grammatical role assignment only aftcr thcir developing system has incorporated other cues (e.g., case marlong, acoustic stress). This principie saya that learners make use of the1r dcvcloping system (a k:nowl­edge IOU?'Ce) for on-line processing as they progresa through acquisition. Howcvei; the formal cues that this system might contain are ao indirect result of Pl(~) and P2. Thus, the Competition Modcl cannot tell us how the cues are mitially detectecl aod made available for th06C mechanisms tbat compute cue validity for the developing system. Tlus is not intended as a criticism, since the Competition Model is not meant to be a model of atteo­tion and detection.

Wc conclude this section, then, by saying that the Competition Model fits within our P3(b) and may prove to be a uscful framcwork for discussion of the eventual accommodation of new intake data and restructuring of the dcvcloping system in second language acquisition. However, because of the

Input Processing 53

relative incompleteness of second language ecquisltion for most leamers, we cannot assume that Jeamers attend to and detect cues in the input in the same way that this is assumed for cluld first language acqtrisition. For this, we necd to develop other frameworks for understaoding bow Jeamers iru­tially attend to and dete<:t grammat.ical fono in the mput, aod the principies aod moclel de-.-eloped in this chapter seek to do just that.

CONCLUSION

In this chapter, we have posited principies of second language input pro­cessing based on two major qucstions: What gu.ides the way(s) in which leamcrs process grammatical form in the input? How do Jeamers assign grammatical·semantic roles to nouns? Using the construct of attention and limited mpacity prooessing, wc havc posited a set of processing principies regarding how learoers prooess grammatical form. Using first and second language research on grammatical and semantic role •ssigoment, we have posited a set of principies related to how learners determine sub,Jeets and ob,Jeets of verbs. Along the way. wc have attempted to make links betwEen input processing and language aoqujsltion That is, mput processing shapes the 1Dtake data available for accommodatlon by tbe developing system. N; intake data are accommodated and re¡tructuring occurs, Jeamers' develop-1og systems take shape. We also have compared our model to others and found complementarity rather than opposition. We al.so ack:nowledged that pereeptual salience and its role in input processing must, at some point, be incorporated in a model of input proccssing.

The questioo we ask now concems the uscfulness of a model of input pro­cessing for second language instruction. Currículum developers and instroo­tors have tended to shy away from thoory since little direct connection seems to be made between thoory and actual practioe, at least in terms of grammar iostructioo. The Monitor Model, tbe Acculturation Model, Universal Grammai; and other model.s and lingu.istic theories may help JDStructors understand languagc acquisition in general but they offer few 1DS1ghts into how to best facilitate thc acquisibon of grammar in a day-to­day situation. In the next chapter, we will examine how knowledge of input processing can lead to very concrete applications to tbe classroom.

..

Page 36: Input Processing and Grammar Instruction

',

13 PROCESSING INSTRUCTION

INTRODUCTION

In the previous chapter we ouUined a set of principies for input prOCessÍJ\I: that inform second language acquisition theory about how leamers attend to input data. In thc present chapter we will examine processing instruc­tion, a type of explicit gramrnar instruction that initially was motivated by the crític:al role of input processing in acquisition. Our purpoee in doing so is to provide background on the instructional treatment researehed in the various studies summaríicd in Chapter 4.

In this chapter wc wilJ 6rst. review the motivation for processing instruc· tion. By examining traditional outputrbased grammar instruction, we will see that ll'aditional instruction is not informcd by general second langu~ acqui­sition theory. Our conclusion will be that traditional instruction is not pey­cholinguistically motivated. Noxt we will describe processing instruction in sorne detall, focusing mainly on the nature of "structured input" activities. We will see that underlying ali structured input activities is the push to get learn­ers to malee form-meaning mappings in order to create grammatic.ally richer int.ake. Subsequently, we will briefiy compare the nature of comprehension­baaed language teaching with processing instruction. Since there is potential for reductjorusm, our pur¡l06C here IS to poínt out that processing instruction cannot simply be l'QU8ted with oompreheosion·based language teaching.

55

Page 37: Input Processing and Grammar Instruction

... .

56 VanPatten

MOTIVATION FOR PROCESSING INSTRUCTION

A General Model of Second Language Acquisit ion

In Cbapter l we emphaslzcd thc unportance of oonununicativ"l! or meaning· i-ring input in second language aoquisition. Successful language acquisi· tion simply does not happen without it. Wc also saw that Jearners process the input to create intake, and that intake are tbe data that can be acrom· modated by tbe developing systcm. \'k sketched out a simple model of these sets of p1 o es and display it once agam in Fígure 3-1

In examming the model in Figure 3-1 we eee tbat the DU\jOr chain of events for acquiring a particular form go from Jet\ to right. First, learners get input oontaining tbal form. lf il IS prooe11sed, tbat form beoomes intake data As intake data, lhe form is a candidate for acconunodation by lhe developing sys­tem. lf acoommodated, lhen lhe form becomes pert of the developing system witb possíble restructunng in other parts of the systcm depending on the nature of lhe form and the (transitional) 8l8ie at wluch the system finds itselí lf acoommodatíon or n!6tructuring oocurs, lhe form can be tapped by the output processing mechanisms respollSlble for production.

What is relevant for the present discussion is lhat the acquisition offor· mal features of Janguage begina wilh input and input processing. For a leamer to be able to cvcntually occcss and use a form in output in an unmonitored way, that form must be part of the developing systcm. How did that form gct thcre? Again, formal foatures of language enter the system via lhe processes represented by l and 11 in Figure 3-1. Thls discussion sets the 8l8ie for an examinat.ion of the be.sic problem wilh traditional instruction.

The Problem w ith Traditlonal lnstruction

For most traditional approoches to grammar instruction, it is fair to say that most if not ali grammar mstruction is output,.oriented. This is particularly true of foreign language cla.'ISl'OOms in the Uruted States. Classroom Jeamers get explanations about rules and paradigma from instrucrors and materials,

input---lli intake ll !li developing oyatcm

1 • input procesaing 11 • accommodation, restructuring

FIGURE $-1. lnpu1 .,.._,., and uiW."' MXll>d longuop-IJOD.

t

Processing lnstruction 57

and then lhey practicc these rules and paradigms vía various output exercis­es and act.ivities. The ,_jer may recall from Chapter 1 that over 25 years ago, Paulston (1972) oodifuxl a particular sequencing of oral grammar practices. Sbe suggested that grammar-based oral practice sbould proceed from mechanlcal to meaningful to oommunicative exercises. Sbe also emphasized that mcchanical practice (pettern drills) were a neoessaJy 6rst step in the intemalization of tbe rules and forms of the second language. c.ouched with­in lhe dorrunantlanguage loomingtheoriesofthe 1950s and 1960s, Paulston's claim seemed sensible. BebaVJOrism, with its focus on obsei •able behavioi; hab1t formation, and immediate feedback, ser."l!d as the cornerstone for the devclopmenl of audJo.lmguabsm and its promulgation in language teaching. Oral production waa the observable behavioi; oorrect oral production and avoídance of grammatical error& were the habits to be formed, and on-truH;pot error oomictíon served as the immediate feedback. ~ of major theo­retical shi.fts in oognitÍOn, lingui.stics, and the vast research undertaken in child language acqUJ:jltion smce the early 1960s which clearly has dispeDed any appliclltion ofbehavionsm to first language acquisition, and in spite ofthe ret!C8l'Ch on ecoond 1anguage ooqw.sitíon since the early 19706, language instruction has oontínued to regard grammar leaming as largely oral. At the time of writing of lhis boolc, Paulston's hierarchy for grammar practice is alive and well, as evidenood by the type of grwnmar instruction and practice found in the vast majority of textbooks u.sed to teach foreign languages in this ooun· try. For example, onc of the m<Jllt popular Spanish college-Jevel texthooks on lhe market today follows this parti<.i.ilar approach to grammar instruction as evidenced by by thc following lesson on object pronouns.

Direct Object Pronouns • A direct o~ect noun is of\cn replaced by a direct object pronoun. The fo).

lowing chart shows thc forms of the direct object pronouns.

Blnplar

""'mo

" you lo you , ........ lum, ,, (mMC)

la you (fon>), her, 11 (fom)

Plunol

l10ll UI

.,. you (informal) '°' you <->. t.bem /& you (fom), t.bem

• Direct object pronouru agree in gender and number with the noun to which they refer.

Qu100 el libro. Quwo 1-04 bolaos. Uamo a TelU(l llaman a las chicas.

Loquiuo. Los qui.ero. La llamo. Las/laman.

Page 38: Input Processing and Grammar Instruction

l

58 Varl'atten

• Direct object pronouns are usually placed unmediately before the conju­gated verb.

iDóruk ves a Jorge y a Atkla? Los ueo en la clase. tQUW"r$ labJU$CJ, Mili? Si, la quiero.

(explanation continues)

Práctica

Where do you see Jorge and Adela? 1 seem in class. Do you want the blouse, Mili? Yes, 1 want it.

l. El (La)olvidazo (a). You and your roomat.e ha ve forgotten who is doing what. Modele: iQuién oo a comprar los sandwtcms? (Who is going to buy

the sandwiches?) Tú los vas a comprar. (You are going to buy

them. l. iQuién oo a llamar a las chicas? 2. iQuién oo a buscor el hiele? (3-6 continue in the same ve in 1

Práctica comunicativa 2. \.bnws a la pi.aya. With some frtcnds, organize a day at the beach.

Modelo: iQuiin lkvo la bolsa? CWho's t.alon¡ the bag?) Yo la 11_,. (I'm taking iL)

l. iQuién hact los sandw1c:Ms? 2. iQuién compra los refrescos? [3-6 continue in the same vein)

N. can be seen, a traditional structural explanation of the form and t.he rules that govem it is followed immediarely by mechanic.al output practice. (lnterestingly. the 8C>Ollled práctica comunlCOIWa is aJso a mecharuc.al prac­tice.) N. one more example of this traditional approach, the reader IS invit­ed to examine the following two parts of a "study tip" given to the learners rcgarding verb forms in Spanish.

• Practice coajugating severa! -ar verbe in writing first. Identify the st.em, then wrire the various verb fonns by adding the present rense endings list. ed on page xx_ Once you have done this, say the fonns you have written out loud severa! times.

• Next, you will nced to practice -ar verb conugations orally. Create two sets ofindex cards. On one, wrire down thc sub,Ject pronouns listed on page 36 (one per card). On the other set, write some of the -ar vcrbs you have leamed. Selcct one carel from each set and coajugat.e the vcrb with the selected pronoun.

Processing lnstructoon 59

The focus on output and mechanical practice as thc means by which the lm­guistic system develops should be evident in the cit.ed examples. Whnt is clear from this approach to grammar instruction is that the accommodation and rcstructuring of the dcvcloping eystem is seen to happen because of practioe, not becau.se of exposure to language sarnples in the input.

From the previous discussion the ml\JOr problem with traditional approaches to grammar mstruction emerges: There IS a mismatcli between the widely eccept.ed role of mpul in seoond languoge acquisition and the out,. pul-based and often mecharucal nature of grammar mstruction and practice in much of language t.eoching. Input is input and output is output o.nd the qucstion arises as to just whcre the communicativc meaning-bearing input is to be found in traditional grammar instruction. Output pract.ice is not the data upon which the developing syst.em relies for growth. Nor is the expla­nation about grammar provided by the instructor or the textbook. Explano.tions are information about language, but they are not the language data themselves.

A llU\ÍOr motivation for pl"O<let!Sing instruction, then, comes out of our crit,. ical cxamination of tradilional approaches to grammar instruction. Howevcr, there is anothcr motivation as well. As claimed in VanPatten (1993), classrooms are beooming increasingly communicative and input,. rich. lnstructors are encouraged to use the second language in a varicty of WBy!l to communicate information. They are encouraged to have leamers engage in tasks in wluch the leamers themselves must communicate infor­mation and u.se language mea.ningfully Ce.g., Lee & VanPatten, 1995; Nunan, 1989; Savignon, 1983). In addition, irultructors are encouraged to gel more comprehel\Slblc input into the cJass nnd to make comprehcnsion an esscntial component of tho classroom and thc curriculum (e.g., Krashen, 1982; Krashen & Terrell, 1983). The problcm in these directions for lan­guage teeching is that traditional grammar instruction just doesn 't seem to lit. Lockstep and largely mechanical in nature (especially in foreign Jan. guage classrooms in the Uruted States), tnditional grammar UIStruction conllicts with the more flwd, leamer--0entered, and cont.ent-based nature of communicative classrooms. And indeed, output drills just don't seem to go with input-based classrooms. Thus, a seoond motivation for rethinking grammar instruction conccma congruencc. Although not complet.ely ncces­sary, many ofus would like to have the various inslructional componen!$ of our clas..room blend !Dgelher m some hannonious whole.

The question to be asked is t.he following: lf a traditiooal output-bo..-;ed ~ approach to grammar instructlon Is incongruent wtth CUJTent theory about seoond language acquisition, then would an input-based approach to grnmmar instruction be better? Of no surprise, our answer is yes, but only if the instro.c­tion tak/!$ ir-.Jc consideration llu: nature of input proccssing. We tum our att.en­tion now to a description of thc components of pnxcssing instruction.

Page 39: Input Processing and Grammar Instruction

l

60 VaoPatten

TIE NATURE OF PROCESSING INSTRUCTION

Overview

The goal of proccssing instruction is to alter the processing strategies that leamers take to the task of comprehension and to enoourage them to make better form-meaning conncctions than they would if left to their own devices. To achíeve this, prooeaoing lllSlruction has tbree key components: (1) explanation ofthe relationslup between a given form and tbe meaning it can convey; (2) information about p1'()CNLS!ng strategies, sbowing learners bow natural processmg strategics may not work to tbeir benetit; and (3) "structured input" adlvit.ies m whích leamers are given tbe opportunity to process form in the input in a "controlled" situation so tbat better form­meaning connection.s might happen compered with wbat might happen in less controDed situations. Component ( 1) of\en may resemble some of the information provided to leamers in more traditional approacbes to gram­mar instruction, wbcreas components (2l and (3) are unique to processing instruction. Since these components fonn part of the treatment used in the research studies we wiJJ review in the next chapter, it i.s important to discu.ss them bere. We will take Cllch of tbese components in turn, spending the majority of our time in examining component (3), structured input.

Explanation

As in most approaches to focue on fonn, processing instruction provides leamers wilh somo informal.ion about grammatical form. Structural prop­erties of a feature are given and rules are dcscribed, but what is most impor­tant is that the explanat.ion attcmpts to link form and meaning. In the following example, takcn from tbe instructional materials used in VanPattcn and Cadiemo 0993), whích in turn were based on materials used in VanPatten, Lee, Glass, and Binkowski (1992), tbe explanation of object pronouns in Spanish infonns studcnts of how pronoWlB enrode meaning (semantic and gramrnatical roles) in addit.ion to the structural aspects of object pronoune.

Sample lesson An object of a verb is a grammabcal conoept different from a subject. An

object gcne....Uy i.a definccl aa a thing or peroon on wluch an action or prc>CC$S is performed. Thus, in the scntence "John writes lettcrs," "John" i.s the sub­ject and "letters" is the object (the action of wnting is performed on the Jet­ters). In the scntence "She has an idea," "She is tbe subject (pronoun) and "idea" is the object <lhc thíng on whlcb tbe prooess of "baving" is per-

Processing fnstruction 61

fonned). What 1s the subject and wbat is the object of the verb miran in the foUowing scntenoe?

Lo$ podrt• nuran <> Ú>/I h(JOI

Right. Padres IS tbc subject (parents are thc ones doing the watching) and hyos is the obJec:t (tbe thmp being watched). Did you notioe tbat hijos is pre«ded by <>? Thii. a IS callcd "personal a" and is used to mark objects of a verb. You will learn more about ít later. Wbat is the subject pronoun that corresponds to padru? Ell"', él, or nosolros?

____ __ nuran <>loe htJOt.

Right agrun. Eli08. La8 padrra is tbe subjec:t noun and ellos is tbe subject pro­noun Subject pronouns are already familiar to you.

yo nOM>troslos tú oo.<10tros!os usl«i ustedes él/ella ellos/ellos

In Spanish (and English), not only are there subjed pronouns, but there are aL'iO object pronouns:

Tho parents wat.ch lhem (that i& tho kidal. Lo$ podres los nuran fu decir, a Ú>ll hvooJ.

Heni are the first set of subjoct and object pronoune that you wiJJ become familiar with.

<> ..... ~ Ob•ect .........-_·~~~~~~~·~~~~~~

Fim peroon 11nauJar yo lé> romprmd<J o nu htrnuuto. 11 underltand my brother.)

Socond - •incular tu 1ü rompm¡do o loo obr.dce. tYou undent&nd your pand_ ... ,

flnl por.>n plunJ _,... .\'<»OirwCOM~

o /os J1QJ11t11"11.

¡\\'e undenland OUT relat.veo.)

me M1 hmnono me comprvuk. tMy brolhe< ~ m•.J

le Loo~ le comprtwk1' CYour gnndperenta

wuler.Rand you.J

Page 40: Input Processing and Grammar Instruction

62 VanPatten

Me, t.e and nos are objects of the verb in the right hand column. In the first example, who is being understood? Me. In the second, who is being under­stood? You. In the tlúrd, who is being understood? Us.

There are two things to keep in mind about object pronouns:

l. They are placed in front of coajugated verbs. 2. They indicate on who or what the action-prooess is being performed, not

who or what performs the action-process.

We will not discuss explanation further and turn our attention to a unique aspect of prooossing instruction: irúorming learners of the potential­ly problematic outcomes ofprocessing strategies.

lnfonning Leamers of Processing Strategies

In addition to irúonnation on how a form or structure works, leamers receive clues or lúnts in processing instruction about paying attention to form in the input. Recall from Chapter 2 that leamers utilize a first-noun strategy to assign grammatical and semantic roles such that the first noun they encounter in the input string is assigned the role of subject or agent. This leads to wrong assignment in sentences in Spanish such as the following

(l) La Her-ACC

sigue el señor. follows the man.

"The man follows her."

Learners typically assign the pronoun la the subject role and misinterpret the sentence as "She follows the man."

In processing instruction, this tendency to rely on word order alone is pointr ed out to leamers, albeit in a nontechnical manner. Aft.er explanation about object pronouns they are then informed of sometlúng like the following:

What can get tricky in correctly uncleratanding a sentence is that often you wilJ see or hear a sentence in which the order is object-pronoun-verb-subject,just the opposite of what you would expectl

Nos inuitan las chica& a cenar. (The girls are inviting us to dinner)

No te compreruk el profe$()f: (The professor doesn't understand you.)

So be careful and don 't rnake the misteke of interpreting the object pronoun as the subject of a sentence! (from Van.Patten, Lee, & Ballman, 1996).

Sometimes this information is repeated in an activity so that learners are reminded of the problems in processing input. The following sample direc-

Processing lnstruction 63

tion lines come from an activity in a sequence of activitíes used in VanPatten and Cadiemo's (1993) study.

Actividad A Select the picture that best corresponds to the sentence. CKeep in mind that Spanish does not follow a rigid subject.-verb-object word order and that object pronouns may go before a ooajugated verb or at tite end of an infinitive.)

In her research on processing instruction and past tense in Spanish, Cadiemo (1992, 1995) informed students of bow to pay attention to tense cues in the input. Alter an explanation of how the past tense is formed, Cadiemo provided her subjects with the following:

Preterit (past tense) forms are often accompanied by temporal adverbs that indicate that the action of the verb occurred in the past.. Here you have sorne of these past temporal adverbe: {o/l!r {y<!sterday), anteayer (day before yester­day), on.oche (Jast night), la semana pasado (last week), el lun<JS, marWJ, etc., pasado (last Monday, Tuesday, etc.), IUJCe un mes (a month agoJ, etc.

Howevei; although these adverbials are a good clue to know that an action has occurred in the past, they are not always present in the sentences that you enoounter. Thís is the reason whY it will be ímportant for you to reoognize past tense verb forms. And remember, the best clue you bave is the spoken stress in the vowel of the endings of the verb (from Cadiemo, 1992, p. 318).

The last point is particularly important since the instructional explana­tion itself contrasted the stress pattems of tbe present tense with those of the past. Present tense verb fonns in Sparush almost always carry stress on the verb stem or root. Past tense fonns carry stress on the verb ending.

What Cediemo attempted to do in these instructional lines, then, is shift the focus of attention during input processing off of the temporal adverbials and direct it to the verb endings themselves. However, one thing is to tell leamers to do tlús; another is to provide them opportunities to do so. One thing is to tell leamers not to mistake object pronouns for subjects and another is to give them opportunities to make correct grammatical and semantic role assignments. We now tum our attention to activities used in processing instruction.

Structured Input

Activities in processing instruction utilize what is best termed "structured input." "Input" refers to the fact that during the activities, leamers do not produce the targeted grammatical form or structure. Instead, they are engaged in actively processing input sentences. The term "structured" refers to the fact that the input has been manipulated in particular ways; it

Page 41: Input Processing and Grammar Instruction

,

1

64 \larf>atten

is not free flowíng communic:ative disrourse, altbough it Í! mearung-bearing. The activities m processing instruction (al lcast as il has been used in the

emplncal st\ldies that we will examine in Chapter 4 l al.so use a mixture of referentially onented activities and alfectively onented activities. By refer­entially oricnted activities we mean activities dunng which the content ÍOC\18 of the input sentences is not on the learners but on sorne other tbird pereon(s). Thcse activities are used to check lhat lhe lcarners are actually procc68mg thc form. In these types of activitics, thcre is a right or wrong llll8wer that revcals whether or not the leamers have madc correct subject and object role assignment. In Figure 3-2 are two examples of referential activities used in Van.Patten and Cadicmo (1993) on thc t.eaching of object pronouns and word order.

As can be scen in the 6rst activity (Actividad 0) in Figure 3-2, leamers must chooec betwcen two English glossinp to show that they have assigned suJ:ucct and object roles correctly to tbe sentena?S that they reacL In the sec­ond activity (Adundad HJ, learners must choose between two gloesings to show tbat lhey have assigned subject and object roles correctly to t.he sen­tence& tbat they hear (Otber examples of referential activ:ities can be seen in t.he sample lesaon laid out 1aW in this chapter.) Referentially oriented activ­ities can be uaed al any time in a lesaon but are part.acu1arly aitical et tbe beginning of a sequence of activities. BecaUBe they entail eit.her right or wrong answers, they can serve as a sort of starting point in a sequence of activities to cnsure that learners are making comict form-meaning map­pinga. These particular activities ha ve a miJc of SVO and OVS type sentenoes, and thus pull)08cly push learners to rely on cues other than word order (the first noun stmtegy) to get the propositional meaning of each sentence.

Unlike referentially oriented activities, affcctively oriented activities have no wrong or right answer. Leamcrs indica te agreemcnt-disagrooment, true for me-not true for me, check boxes in surveys, in short, they provide ind.lcations of their opinions, beliefs, feelings, and personal circumstances. In our initial work on processing instruction CVanPattcn & Cadiemo, 1993), we mcluded alfect:íve structured input activ:ities because general commu­nicative 1anguage t.eaching incorporales affectively oriented activities on a regular basis. Thus, we were attempting to tie processing instruction to sorne of the more general tenets of commwúcatí~-e languagc teaching, whlch mclude an cmphasís on meaning and on learn(!)"-ccnteredness. Figure 3-3 contams two examples of affectively oriented activities used in VanPatten and Cadicmo !1993). In these activities. leamcrs provide information about thcmselvcs. Note that t.he senten;:es are simple and relatively easy to proccss. In addition, sentences do not follow SVO order so that leamers must attend to the meaning of the sentencc nnd makc correct form-mean­ing mappings in order to complete the activity successfully, as they did in lhe previously described referential activities.

l

1.

Actividad G: Pronombres Select the correct interpretation of the sen­tence. Keep in mind that Spanisb has flexíble word order and doesn't 111»

cssarily follow subject-verb-ob,Jeet order like English.

l. M1 hermana me llama {rec~tllln~nu. Who calls whom? a. I call my sister. b. My &ster calls me.

2. iTe escriben tus padres? Who writes to whom? a. Do you write to your parents? b. Do your parents write to you?

3. No nos escuchan los padres. Who doesn't listen to whom? a. Parents don't listen to us. b. We don't listen to parents.

4. Me conocen bim mui hertn0/IC8. Who k:nows whom well? a. My S1blings lmow me. b. 1 know my sibli.ngs.

Actividad tt ¿Objeto o s ujeto? Match each sentence you hear with one of the statements below. Remember that Spanish does not always follow subject-verb-object word order!

l. a. O A man is calling me. b. O 1 arn calling a man.

2. a. O My parents visit me. b.Q 1 visit my parents.

3. a. O 1 follow others. " b.O Others follow me.

4. a. O We are greeting a friend. b.OA friend greets us.

s. a.O Our relatives don't undenitand us. b.O We don't Wlderstand our relab\'es.

6. a.QA friend is inviting you to dinner. b. 0 You are inviting a mend to dinn(!)"

7 a. O The professor is wat.chmg us. b.OWe are watching the profcsaor.

ll R O MRriR i.~ lmking for you. b.O You are looking for Merla

9. a.O Juan believes us b.Q We believe Juan.

FIGURE S-2. Refenmtially on..nted acUviti• U80d in VonPal.tm and Cedi<>rno (1993).

65

1

Page 42: Input Processing and Grammar Instruction

66 VanPatten

The point of this chapter IS not to provide detailed inslructions in bow to pre­pare appropriate structurod input act.ivities for p~ inslruction. Tbis is done in significant detail in Lee and VnnPatten (1996, Chapter 5). Again, gjven tbat we are setting tbe 8ta@e for understanding tbe research described in the next chapter; it is important to lughlight a set of guidelines that bave begun to inform the construction of structured input llCl:ivit:ies used in p~

Actividad J: Los parientes. What are things tbat relatives do to us? Tbey can botber us, vi.sil us, cntioze us, love us, and so on.

Paso l. Reed eacb statemcnt and select the ones that you tbink are typical.

Los parienks . .• a. O nos molatan b. ::J nos cnhcan. c. O nos ayuden (belp). d. O nos 11isiJan. e. O nos quieren (qW!~r = to be fond oO. e Onos ______ _

Paso 2. Now scloct tbc altcmatives that you tlúnk make sense.

Los parientes ... a. Qpueden moleslarrtOll aunque (althougb) no deben. b. Q pueden criticamos aunque no deben. c. Qpueden ayudamos aunque no deben. d. O pueden uisilarrws aunque no deben. e. Qpueden querernos aunque no deben. ( Q pueden nos aunque no deben.

Actividad L: Mis padres. How do you interact with your parents? lndicate wbether or not eacb statcmcnt applies to you.

SÍ.SE ME APLICA

l. Los llamo con {rttuencia por teli(ono. o 2. Los uisilo los finea de semana. ::J 3. Los uisilo una vez al ma. o .._ ÚM abrato cuando loo ueo (abrcuar - to hug). o 5. Los comprendo muy bien. o 6. No me ímportan. ::J 7. Los aprecie (appreciatel mucho. o

NO, NO SE ME APLICA

o o o o o o o

Processing lnstruction 67

instruction. Thesc guidelines were fust reportro in VanPatten (1993) and are subeequenUy elaborated on in Lee and VanPatten (1995, Cbapter 5). For our purposes, we will review them bricfly. It should be noted tbat only tbe seoond, 6.fth. and SIXth gwdelines are related in any particular way to issues in input p~ and psycllolinguist as we bave ~ them in Cbapter 2. The otbers were dcvclopcd for proctical and experiential reesons as descn"bed in eecb case. It should also be noted tbat these are ll''idelines, not maxims. Variations in tbeir application from lesson to lesson may occur.

Guidelines for Structured Input Activit ies Tbe first gwdeline for developing structured input activities is the fol­

lowing.

l. Teoch only one thmg al a time. •

What tbis mcans is that paradigms and rules can and sbould be broken down into smaller parta. In traditional instruction of tbe past tense in Spanisb, for =ple, it is typical to give leamers ali the forros of regular verba for ali thz-ee verb coajugations in one large paradigmatic chart. After studying and mcmorizing this chart., learners practice ali tbe forms vía out­put drills. For processing instruction, we have advocated breaking up para­digma nnd actually "building" them during the course of tbe lesson. Tbus, lcarners might initially learn about third penion singular past tense verb forros in Spanish and work througb some structured input ac:tivities before going on to anothcr peraon in the vcrb paradigm. In VanPatten and Cadicrno (1993) and Cadicrno (1995) tbis is precisely what was done-and the length of tbc IC680n was not incrcased wben compared to time spent on explanation and practicc during traditional instruction; "full coverage" of the grammatical points was attained in both traditional and processing instruction in thc eame amount of time.

Tbe sccond guidelinc that we have developcd for structured input activi­ties is thc following.

2. Kttp mea111ng m focua.

Wbat this means is tbat the mput ~ must encocle sorne meaning tbat the leamer IS reqwred to attcnd to and respond to in some way. This particu­lar guidl'linf' wn• mobVllted by my mitial work with students of lruiguege teecbing. When mstruct.cd to <IC\-elop input-based grammar activities, I bave noticed tbat a nu.mber of 6tu.dents create activities in whicb it was possfüle to complete thc activity and not once connect meaning to fonn. Fbr example, one student crcated an activity in which the leamers bad to sean a sbort para­grapb and circlc alJ the past tense endings. Sbe was correct in tbat tbe leam-

Page 43: Input Processing and Grammar Instruction

68 Varl'atten

ers did oot ha>-e to produce anythmg. but this and other &etivities like it demonstrate that teachers may not understand the definítion of communica· tive meaning-bearing input or ot loo.st forget about it when developing ad.ivÍ· ties for prooessing instruction. Thcy have also forgotten that mput processjng involvca making forro-meaning mappings, not just noticing forro in and of il6elf. This guideline, then, servca as o reminder that the purpoee of structured input &etivities is to push leamers to ma.ke forro-meaning connections. If meaning is absent or if Jeamcrs do not have to pay attention to meaning to complete the activicy, then there ís no enhancement of input processing.

The third guideline for de.-elopmg structured input aclivibcs IS

3. úorners must do BOm~thing wllh /he input.

This gu¡deline means that leamers must be engaged activcly in processing lhe input sentences and must dernonstrate this by responding to the input sentcnccs in sorne way. They may et.ate agreernent or disagreement, say "Yes, that applies me," or "No, that doesn't apply to me," complete a survey. aclect elternatives (e.g., a, b, e, or d), offer a name, ma.ke an association, in short, indicate in sorne way that they have attended to the meaning con­tained in the input sent.ence(s). This giiideJine was motivated a¡¡aiD by my work with teachers-in-trainingwhocouldnotdistinguish between talkingat and talking with Jeamers. The guideline was formulated so tbat they might bettcr understand tbat input does no good ifthe leamer is not paymg atten· tion to 1t. Asking leamers to respond to the input in sorne way belps to insure tbat they are indeed paying att.ention.

A fourth guideline that wos dcveloped is

4.. Use boih oral and wrnten input.

This guídeline was established to 8'XOUnt for individual clifferences. Expericnce has sugge&OO tbat some leamers benefit to dilferent degrees from the mode of inpuL Sorne say they like to "see" the language, whereas others do not make this claim. A combination of oral and wnttcn input is a response to these claims and is not ticd direct.ly to the principies of input pro­ccssing outlined in Cbapter 2.

A fift.h guideline addresses sentential ve. discourse-lcvel input. It stotes

li Mntlt' fmm 1w1./J>nN>R lfl Mnrll't:t.ed discourse.

T!us gu¡deline meaos tbat early activities should invol'-e sentence leve! input such as those activities for objec:t pronouns reviewed in lhe preceding acctlon. Connected discourse (listening to conversations, monologues) should be reserved for lat.er m the lesson. The rationalc for this is tbat lean>-

Processing lnstructoon 69

ers ba\-e a better chance al focusing on and detecting the targeted form in sentences as oppc>E;ed to connect.ed discourse when 6rst engaged m &true· tured input act1V1tics. As mentioned in 'knl'atten (1993), connected dis­course may hindcr learners' initial processing of the targeted forro bccause of their limited capacity to process incoming data. Connected discourse may not give learncrs sufficient "processing time" as the sentences in lhc dis­eourse occur one all.er the olher to forma larger narrative or text. The result may be that much of the input is "noise" and the learners may bave diffi. culty in attcndmg to and detecting the relevanl gnmmatical item. In Chapter 2 we reV1ewed the di.fficulty expenenced by leamers in the expcn­ment report.ed in VanPatten (1990). Comprehension of content suf!'cred se.-erely when leamers also attempted to attend to forro. Terrell (1991) al:IO suggests the same, noting that learners may not be able to "bind" forms from discourse-level input and instead use strategies that allow them to sim­ply eomprehend (get the gist) and respond in appropriate fashion. Sincc we want leamers to att.end to content (mconing) and forro, starting a sequencc in processing instroction with discourse-levcl input does not seem optima!.

The sixt.h and final guideline is perhaps the most important sincc it makes explicit reference to input processmg. It says

6. Keep IM ~holmgulSlic procasing stratq¡ia m núnd.

This guideline is mtended to ensure tbat leamcrs' attention is appropnaw. ly guided dunng atructured input actiVltiCS. In the examples refcrred to in Figures 3-2 and 3·3, for example, the activities systematically push leamcrs to not rely on the firsl noun strategy. In both the initial referential activities and the subsequcnt afl'ective activities, leamcrs qulckly face the fact that the first noun may not be the subject, thus, dcveloping other forro-rneening mappings for scmantic and grammatical role assignment. These activities, then, keep in mind what the learners' natural processing strategy is and the input is structured to alter leamers' rebanee on the strategy.

Like severa! other guidelines, guidelinc 6 was borne out of my work with teachers-in-trairung. lt seems tbat sorne novice att.empts to create struc­tured input acttvities reflectan understanding that the activities should be input--based hul do not reflectan understanding of wbat Jearners are domg when processing input. For example, I have seen teachers-in-training creatc past-tense activities in which every single utterancc eontains a past tense adverbial. As pointed out in Cbapter 2, lhc presence of the adverbial actual­ly detracta from lhe role of the tense markcr on the verb since leamers pre. fer to process thc adverbial (lexical it.ern) as opposed to the grammatical form (inflection) when both enrode the same semantic information. J have also seen teachcr&-in·training who create actiVlties with object pronouno m Spanish but do no/ mclude activities in which the Jeamer must a.ssign the

Page 44: Input Processing and Grammar Instruction

70 VanPatten

grarnmatical role of object to the first (pro)noun in the sentence. In other words, they create activities in whicb the first noun is always the subject, thus reinforcing the first noun strategy. l have seen teachers and teachers­in-training create structured input activities in ESL for the continuous pre­sent (e.g., "1 am writing a book these days," "My spouse is teaching a course on reading research this semester") in which many of the input sentences contain adverbials such as "these days," "this semester," and so on. Thus, learners' attention is detracted from the mearúng of the continuous present since the same meaning is encoded in the adverbial phrases. And 1 ha ve seen teachers-in-training create activities in French for adjective agreement in which learners do not need to pay attention to adjective endings while mak· ing forrn-meaning connections.

In short., I ha ve seen well-intentioned teachers create interesting meaning· based input activities but these activities do not attempt to alter the procesa· ing strategies of learners. These activities do not systematically push learners to direct attention (detection) during input processing to targeted forrns.

To drive the point of guideline 6 home, we will compare two activities. Spanish requires that ali adjectives inflect for number and gender. Under many naturally occurring situations, adjective endings can go undetected since head nouns contain the same information. In the phrase la mujer espaiícla ("the Spanish woman"), mujer ("woman") is singular and femi· nine. The adjective marker -a is redundant and oflow communicative value; the learner does not have to attend to it to get the meaning of the phrase. Likewise, in dos hombres amerüxuws ("two American men"), the adjective marker -08 is redundant, since hombres is masouine and dos signals plural· ity. Hence, the adjective ending carries little communicative value.

Based on the preceding, we can determine that only one of the following two activities actually works at getting learners to attend to the adjective marker for possible forrn-meaning connections. The first one, A, was pro­duced by a teacher-student enrolled in a language teaching class. B, is adapt­ed from Lee and VanPatten (1995, p. 101). (Note: ella = she; es = is; the adjectives are purposefully cognates for the benefit ofthe reader who knows no Spanish. In addition, adjectives ending in -0 generally are masculine, whereas adjectives ending in -a are generally feminine.)

A. Listen as your instructor reads a sentence about Hillary Clinton. Decide whether what you hear is true or not.

st M

o o l ... . o o 2 .. . . o o 3 ... . o o 4 ... . o o 5 .... etc.

Processing lnstruction 71

(Instructor reads the following: l. Ella es dinámica.; 2. Ella es agresiva.; 3. Ella es egocéntrica.; 4. Ella es honesta.; 5. Ella es divorcie.da.; etc. )

B. Listen as your instructor reads a sentence. First, decide whether the sen­tence is about Bill Clinton or Hillary Rodham Clinton. Then, decide whether you agree with the statement or not.

agree disagree

o o l. O Bill o Hillary o o 2. O Bill OHillary o o 3. O Bill o Hillary o o 4. 0 Bill 0 Hillary o o 5. O Bill 0 Hillary

etc. (Instructor reads: l. Es dinámica.; 2. Es agresivo.; 3. Es egocéntri.ca..; 4. Es honesta.; 5. Es respef,ado.; etc.)

Of the two preceding activities, only B considers learners' allocation of attention during on·line processing. In activity B, learners first have to determine who is being talked about in each sentence. The only way to determine this is to attend to the endings of the adjectives. Thus, learners are pushed to connect forrn (a<ljective ending) with meaning (in this case, masculine vs. feminine). Second, learners must agree or disagree. In tlús way learners are not only attending to forrn and connecting it to its specific meaning, but they are attending to the meaning of the entire sentence as well. Activity A does not do tlús. Note how the activity does not encourage learners to attend to the a<ljective endings. Each sentence contains an explicit subject pronoun, ella, that enrodes feminine-singular. RecaJJ (again) that learners prefer to procesa lexical items as opposed to grammatical markers when both enrode the same semantic information. In short, the entire activity is set up so that learners do not have to attend to and detect IUijective endings. It does not encourage form·meaning mappings.

A Sample Lesson

Because the purpose of the present chapter is to describe processing instruc· tion so that the reader is better able to interpret the research presented in Chapter 4, we ofler bere part of the processing instruction that was used in VanPatten and Cadiemo (1993). This sequence focuses on third person (other forms of object-pronouns were presented in a previous part of the lesson-see guideline 1). lt was used as the set of materials for the second day of instruc­tion in a two-Oay long instructional treatment. (Note: the materials were not

l

Page 45: Input Processing and Grammar Instruction

72 VanPatten

actuaJly published in the VanPatten and Cadierno study. They do appear, however, in VanPatten, Lee, Glass, & Binkowski, 1992; Cadierno, 1992.)

Sample Lesson The most difficult object pronoun system for students of Sparúsh is the set

of third person object pronouns.

~~ect O~ect Ella besa a Juon. (She kisses John.)

El besa a Marf.a.. <He kisses Mary.)

Ellos observan a Marcos. (They observe Mark.)

Ellas observan a CarliJos. (They observe Charlie.)

Juan la besa. (John kisses her.)

Mari.a. lo besa. (Mary kisses him.)

Marcos los obseroa.. (Mark observes them.)

Garlitos las observa. (Charlie observes them.)

Keeping in mind that Sparúsh has flexible word order, what do you think the following sentence means?

Lo escucha TU>berto.

Rigbt.. Roberto listens to him.

Un Vistazo: El arte de besar In the following cartoon, who is kissing whom? __ Lo besa la mujer. La besa el hombre.

--.4-·-+-~t:.:~~=-:...·

Processing lnstn.iction 73

Actividad A. Select the picture that best corresponds to the sentence. (Keep in mind that Sparúah does not follow a rigjd subject-verb-object word order and that object pronouns may go before a coajugated verb or at the end of an infinitive.)

D a Ob l. Sus padres lo llaman por telé{oM.

Da O h 2. La8 invila. Monuel al eme.

D a O h 3. La abuela lo escucha.

Da Ob 4. Lo saluda lo. niña-

Da Ob 5. El chico la busca.

Page 46: Input Processing and Grammar Instruction

'

:1

7 4 VanPatten

Actividad B. lndicate whcther or not each statement about your parents applies to you. Share your responses with a classmate.

Sí, me No, no aplica me aplica.

- -- 1. Los llamo con (m:l.U!ncUJ por tclé(oTUJ.

-- -- ~ Los V1$iLo los fines de semana.

-- 3. Los 111$1/0 una vez al mes. ---- 4 Los abrazo cuantlo los ueo. --

(abraznr = to hug)

-- -- 5. Lo. comprendo muy b~n.

-- -- 6. Los ignoro compl~.

-- -- 7. Los aprecio mucho.

Did you notice that there are no exphot sub,Ject nouns or subject pronouns in each sentence' Because the yo form of lhe verb can only refer to yo, no ~ect pronoun is needed. Ali ofthc sentences are oflbesimple word order object pronoun-verb.

Actividad C. Select a female relative of yours (madre, hermana, tia, abl.U!/a, prima, etc.) and write her name bclow. Wbich of lbe statements descn'bes bow you fool about her?

Pnrienú: ..,,..-.,.....,,.--------- l. La admiro. __ 2. La respeto. __ 3. La quiero mucho. __ 4. TraJo de imitarla. __ 5. La detesto. __ 6. La ___ ? ___ _

Nombre: ________ _

Now selecta male relat.ivc and do the same! __ l. Lo admiro. _ 2. Lo respeto. __ 3. Lo qu~ro mucho. __ 4. Traro de im1tarllJ. - 5. Lo detaro. __ 6.Lo ___ ? ___ _

Compare wilb two olher people. Did you select lhe same relative(s)? Did you mark lbe same itcms?

Processing lnstruction 75

Actividad D. Listen to cach statement and select the appropriate picture.

1 l.J a Ob

2. ºª Ob

3. ºª O b

5. ºª Ob

(For activity D, lhe mstructor reads the following statements. l. Lo llama Juan por teléfono. 2. La t3Cucha ti señor. 3. La abro.za la mamá. 4 . Los saluda la mu¡tr. 5. El mño la miro.

Page 47: Input Processing and Grammar Instruction

76 VanPatten

Un vistazo : Una m a la relación Ar. article "El roso del hemuJlll) poses1110" appearcd in a magazine tar­

geted for t.eenagers. Read the excerpt for general meaning. Sorne '-ocabulary is provided to belp you out, but it's not necessary to read end understand every word. The actiVJtics that follow are based on thi.s reading.

seguir (to follow) aaustarse (to become frightened) ·oco~arse (to advise) soltar (to leave alone, release) jurar (to swcar) el colmo (the last straw) pai (peaoe)

DEBATE

Un drama familiar muy

común: ella se siente

dominada, perseguida por

su hermano mayor (que

puede ser menor, pero

con aires de grandote) y

no sabe cómo zafarse de

él. Alicia y Manuel so el

caso típico.

ALICIA CUENTA SU PARTE

"Manuel es muy posesivo. No me deja

respirar. Cada vez que VC1'I a salir, me pregunta con quién, a dónde voy, qué

vamos a hacer •.. A veces me sigue. Lo

juro. Cuando un chico viene a visitarme Manny lo interroga y él se asusta El

colmo: mis padres me de¡aron 1r con

unas amigas a un concierto de Bon Jovi... y Manuel les aconsejó que IR>

de ellos fue<a con nosotras, para

supeMsamos P0t poco lo mato. De veras, mi hermano es peor que mis padres. Por eso peleamos mucho. Le

he dicho más ele mil veces que él no es

mi papa y que ma deje en paz. Pero Manuel no me sueha."

Processing lnstruction 77

Actividad E. For each paso of t.his activity, work m pairs.

Pai;o l. Find the following in the rcad.inc me de¡aron ir me pregunta me sigue no me suelta viene a vistarme

In cach instance, Alicia is saying that somcone is doing something to her or for her. Can you identify the subject of eoch verb?

Paso 2. Find the following in the reading: lo juro lo mato lo munoga

Who is the subject of each verb? Who or what does each lo refer to?

Un vistazo: Mamy responde In the following selec:tion, Manuel responds to bis sister's claims. Read it

now for general meaníng. Then do thc actiVJtics that follow. no queda más remedio (no choice is left) había visto (had soon) oon{iar (to trust) mentir (to lie)

MANNYHACE UNA ACLARACION

"No querla decirlo. pero no me queda més remecio. Si viglo a mi hermana, es porque me ha dado mobvos para sospec:tw de ela En vanas ocasioi oes 1a sorprenc1r con un lal Se<gio, que es uno de 8905 1ebeldes SÍ"I causa oon la reputaaón por el suelo. Una vez le * 8 nis padres que lla al caie oon las amigas y ctesp oés un buOO amigo me contó que la h&­bla VIS!o en el cine... pero oon Sergio. ¿Cómo puedo confiar en mi hermana si miente a todos en la casa? Ella no conoce a los chicos. Ese tipo sólo busca una cosa Y yo no quiero que a mi hennana le suceda nada 1eo. •·

Page 48: Input Processing and Grammar Instruction

78 VanPatten

Actividad F. Fbr each paso in this activity. you should first work alone and then share your responses with somcone else.

Paso l. Find ali the UBeS of a third pcrson ob)l!Ct pronoun. Tu wbom or what do they oorrespond? What is the subject of eacli \'erb next to which you faund cach pronoun?

Paso 2. WlUch of the fallowmg best dcecnbes Manny' s feelíngs far his sister? __ La quiere mucho.

Laadmun. La tktttta.

Paso 3. Whicb oflhe foUowing doeti Manny probahly do on a Friday night if bis síster goes out? __ La s~ para ver lo que hacu y con quién. _La d<ja en paz porque es una adulia. __ La espera en caaa.

Recall that Spanish has tbe object marker a.

Lo8 podre• mm111 a k>8 hi¡o.. Uomo o mis podre•.

Tbis object marker has no equivalcnt in Englisb but is important in Spanisb since it providcs an extra cluc as to wbo did what to wbom. Since Spanish has flexible word order, the a rcmínds you that even though a noun appears befare the vcrb it docsn't have to be thc subject!

A Maria la 1'4ma Juan. A Maria Juan la llama. (John calls Mary.)

Note that when an object appears befare the verb, tbe oorresponding object pronouns must aleo be used. If you think that tbis is redundant, it is! But redundancy is a natural feature oflanguages, right? (Hint: Tbink about how we put tense endingi¡ on verbs whcn most of the tune we aleo say "yester­day," "last n.ight," and soon.) Whatdoes the fallowingsentence mean?Who is doing what to whom?

A la chica la buaca •l chico.

Right. The boy is looking far tbe girl.

Processing lnstruction 79

Actividad G. Select the English rendition of each sentence.

l. A mi mamá la besa nwclw mi papá. a My mother kissea my dad a lot. b. My father lo.sscs my mom a lot.

2. A mi papá. no lo comprendo yo. a 1 don 't understand my father. b. My father docsn't understand me.

3. A la señora la saluda el señot:. a The woman greets the man. b. The man greets the woman.

4. A los chicos los aorprende la profesora. a. The professor surprises the boys. b. The boys surpri.se the professor.

Page 49: Input Processing and Grammar Instruction

80 VanPatten

Actividad H. You wi1l hoor !iOme sent.ences in Spanish. Select the c:orrect p1cture for each.

ftt1 1 D a :Jb

2. ºª Ob

3. ºª Ob

4. ºª <For Actividad H. the instructor should read the following scntences.)

l. A la m'l}er rw la cree el hombre. 2. Al hombre la mujer rw lo cree. 3. Al chico lo sigue la chica. 4. El chico sigue a la chica.

Processing lnstruction 81

Actividad l. Un talenJo especial Paso l. Read the following passage to yoUr1!elf Then do the questions that follow.

Mis abuelos malCml)l ""1 ~y los qukro mue/u>. Vioen en San Josl y cuando v;,¡jo a California, siempre los vi.silo.

Mi abuela u llama Concepci/Jn y u Wl4 persona muy upecw.l. Tiene uno IUJbilidod psfqulCá (p!U!<k < <oer> > evenJoa de/. futuro y de/. ¡xu¡ado) pero no kJ usa con mucha {rttuttu:UJ. Du:t que es un rtgaJo tk DIOI y~ usarkJ con cuúl;,. <Ü> (care). 1bdOI tn la familU. la adrmramoa mudw.

Uno tltZ la pdria la IJam6 poro J'll(lu* """'°con wi avnen (wi -1. M1 abUIJa IOaS un ol¡¡do ptnONJl tk la vfdjma y tuvo uno < <""'6n> > dd hom'°"'4 1'1o muy clatO oJ.....,.. lsw QJOll. ro/u tk ptJo. de) y pro111o la poUda lo ropllU'd. M1 abUIJa ao conwtkl .,, una «kóndod tk ~a lo maliana (ovemigbtl

Paso 2. Select the litle that best fits the passage.

a. <<Mi abuela: victima de un crimen>> b. <<Por qué capturaron a mi abuela>> e. < <Un talento especial> >

Paso 3. Se!ect the best response based on what you read in the passage.

l . Mi abuela t$ una celebridad porque ... a la polu:fa la inuesti'6. b. un hombre la auu:ó pero ella pudo desarmarlo. c. ayudó a la policía.

2. Respet.o de su poder ps(quico ... a. lo usa pooo. b. M lo controla muy bien. c. rw lo roma en serio.

3. iQué describe me_¡or mUi sentimientos haCJa nu abuela? a La cnlico por su locuro (c:razineu). b. La quÚ'ro y la estime nuu:ho. c. No tengo rroo:ión porque nunca la veo ru la uUiilo.

Paso 4. Find the llCMln third person object pronouns tbat occur in the passagc and underline thcm. Tben tell to what they rcfer. Tbe first is done for you.

l ... .los quiero mucho. "Jos• refcrs to mis abuelos 2. 3. 4. 5. 6. 7.

Page 50: Input Processing and Grammar Instruction

82 VanPatten

COMPARISON WITH OTHER APPROACHES

The purpose of thc prcscnt scction is to make explicit the similarities and differences between processing in.struction and severa! other approaches. We will make three explicit compansons: one with general comprehension­besed approaches; anothcr with Sharwood Smith's "input enhancement;" anda third with Ruthcñord's "consciousness raising."

Comprehension-based Approaches

Sorne scholars of language wochmg and many in.structors might conclude that J>MC"SS'Ug ínstructíon is another type of comprehension·based instruc­tion. Because learners do not produce targeted grammat:i<:al items during structured input activibea, and indood bocaUBe the tenn "input" itself sug­gests comprehension, the conclusíon seems logical. However, there is a prob­lem in lumpmg processing mstrucbon together with comprehension-based approaches to language instruction.

The term "comprehension-based" refers to the general provi.sion of com· prehensible input by instructora and materia!Jl during the entire course of classroom acquisition. lnstructors use the second language as muchas pos­sible, modifying it for comprehensibility, and encouraging leamers to talk in the seoond language only whcn reacly. Krashen (1982) has the following to say about language classrooms that sums up the underlying position of most comprehension·based approachcs:

Quite simply, t.he role of t.hc S<.'OOnd or forcign language clMaroom is to bring a student to a point whcrc he can begin lA> use the outaide world for further second language acquisition. /vi exp..-ed in Chapwr U t.hia mea.ns we have to provide student.a wít.h cnough oomprehcn•ible input to bríng their seoond lan­guage oompcwnoo to the point whcre thcy can begin to understand language leard "on the out.sido," reod, and pertiopete in ooversations. Since they will be le611 than fully oompetcnt, we also need to provide t.hem wíth tools for encour· aging and re¡ulatiinr input (pp. 160-161).

Wbat Krashen means is that the provision of comprehensible input should be the BIM qua non of language teaching; other aspects of instruction are peripheral or only useful if they hclp to make input comprehensible. To be sure, comprehcnsion·based approaches t.o instruction may vary in the way in which instructors provide comprehensible input to classroom leamers and in the quality of the input. Total Physical Response CTPR) makes heavy use of commancb as 1nstnictors "ordcr" studcnts to peñonn actions. The Natural Approach uses aomc TPR techniques but ~ relies heavily on teacher.taJk. lnstructors use vío-uals and object.s around which they weave

Processing lnstruction 83

an oral text, involving studcnts in the co-<»nstruction of the discourse with simple answera. Content-based instruction is olten comprehension-based as well, particularly with low-lcvel leamers. In these types of classrooms, instructors teach about subject matter (geograph~ history, science, etc) usmg only the second language. lnstruct.ors modify the speech oftheír "lec­tures" and "discussions" t.o make the language comprehensl>le to their sec­ond language auchcnce. Since the focus is on content, like their first language counterparts these instructora make use of slides, pictures, graph.s, object.s, on·hands experimcntation, and other visual aids that assist leamers in comprehending the language.

In adchtion t.o the prei;ence ofinput, whatiscommon toall theseapproach­es is that w; long as leamers show evidenoe of understanding (comprehen­síon), thcn acquisltion is assumed to proceed. lndeed, Krashen (1982) states this when he saya that as long as the affective filter is low, romprehensible input muses acqwsibon (pp. 2<h32). This means that littJe or no attention should be pWd to formal clcmenta of the language by the instructors during classtime or in the evaluation of student peñonnanoe. The foci of compre­hension·based instruction are subsequently reflected in testing; content­based instructora test for content; Natural Approach instructors test for vocabulary, comprehension, and once beyond the earliest stages, for the cxpression of meaning. (For more detailed irúormation on comprehension­based approaches to language instruction, the reader is referred to Krashen & Terrell, 1983; Richards & Rodgers, 1986; Wuútz, 1981).

Although it is true that processing instruction is input-based, it is not equivalent to comprehension-based approaches. This is evident in that p~ cessing instruction's point of dcparture is to get leamers to process more fonn in the input (or to process it correctly). Recall the guidelines for struc. tured input activitics outlincd earlícr. Sevcral of them ha ve been established because of crroncous assumptions on the part of teachers-in-training (and sorne practicing teachera as well). These assumptions involve a simple equa­tion betwoon proc:cssing instruction and comprehension; any input activity that "embeddccl" the target form was seen to be prooessing instruction. As wc saw, this is not thc case. Prooessing instruction is guided by the insights from theory and research on input processing and attempts to inlluenoe input processing itself Comprehension-based instruction makes no such attempt. lts purpoee is to providc comprehensil>le input to language Jeam. ers and has not considered what learners do to the input when they "com­prehend" it. Whereas comprehension-based approaches ignore the psyholinguistics of mtake denvabon, processing in.struction actively seek.s to influence íntake denvation.

The point he re is that processing instruction is not just another kind df comprehen~ion-based approach to language instruction. Processing 111$/nu-twn 1$ o "P«ific opprooch to aplica grommar instructicn and thus

Page 51: Input Processing and Grammar Instruction

84 VanPatten

falls more clearly within the category of instructional treatments called "focus on fonn." lt is certainly more congruent with comprebension-based approoches to language t.eaching than the traditional approacbes to gram­mar instruction tbat we bave briefly reviewed, but to cquate tt witb com­prehcllSIOn·based approacbes is to ignore it.s intent and the th.eory and research tbat informs it.

Input Enhancement

Sharwood Smith (1993, and elsewhere) has coincd thc tenn "input enhance­ment" to dÚIC\lss focus on form. ln b.is framework, input enhancement is any extemal attempt (by instructors or materials) to make fcatures of the input more aalicnt to learners and could come in many fonns. He has discussed both positive and negative enhancement (Sharwood Smith, 1993, p. 177). Positive enbancement could be as simple as color codlng or boldfacing forros in reading texts. Negative enhancement involves providing leamers witb mformation that a given form is incorrect (e.g., pointing out an erroi; mak­ing a funny faoe, or offering a quizzical look when an error is produce:!).

ProceBsing instruction is consonant with Sharwood Smith's position but al9o goes beyond it in an impottant way. Note that Sharwood Smith is con­cerned with making fonns salient. tbat is, bnngmg them to leamers atten­tion in sorne way. Processing instruction does this but aJso attempts to providc opportunities for consistent fonn-meaning mappings in activities. Simp)y bringing a fonn to someone's attention is no guarantee tbat it gets processcd at ali or gets processed correctly (St.'C Cbapter 2). Fbr acquisition to happen, tbe intake must continually provide the dcvcloping system with examples of correct fonn-meaning conncctions that are the result of input proccssing.

In addition, processing instruction makes no claims about providing so­callcd negative enhancement regarding learners' errors. Since it is solely concemed with the processing of input data, explanation and practice are complete)y input-based. lt is true that processing instruction is concemed with erroneous input processing and attempta to on:umvent it; howevei; proc.&ing mstruc:tion does not address the role of output errors in second languagc de-'elopment.

Consciousness Raising

Rutherford (1987, and elsewhere) has discusscd "grammatical conscious­ness raising," which is, simply put, any "deliberate attempt to draw tbe lcarner's attention specifically to the formal propertics of the target lan­gunge" (Ruthcrford & Sharwood Smith, 1988, p. 107}. Wbat R:utheñord

Processing lnstruction 85

says is tbat language acquisition can be aidcd by sucb consciousness rBl:illlg, but that tbe actual form of consoousne:;,; 1'81:>ing can vary depending on first..second language contrasl.6 and the nature of tbe grammatical ttem or .tructure. Although the exact content and nature of consciousness raising ia amorphous, one thing is clear; for Ruthcñord, consciousness rai.sing cannot be equated ~th traditional grammar mstruction. The latteJ; according to Ruthcrford, IS an attempt to instill grammatical form in the leame~ eqwv­alcnt to writing upon the proverbial tabula rasa. Consciousness raising, on the otber band, does not instill anything; instead it assist.s the processes tbat underly acquisition of grammar and does not presume to contain the knowl­edge (or product) tbat must eventually be acquirod (Rutherford, 1987, 24).

In a real sense, processing instruction is a type of consciousness raising, although the term "consciowmess" is somewhat unfortunate. Since pro­ces.qing instruction attempts to inlluence the processes involved in the derivation of intake, it is nota product-oriented approach to grammar t.each­mg that Rutheñord appears to critique. That is, processing instruction does not seek to "pour knowledge" of any kind into leamers' beads; it assists cer­tain processes that can aid the growth of the developing system over time. For this reason, we conch.tde tbat Rutbeñord would not have mucb problem in co~dering p1?"5sing instruction as one manifestation of grammatical consoousness nusmg.

As noted eadiei; the term con.sc1ousness is somewbat unfortunate. The term carries with it the idea in many individual's beliefs that con­scious (explicit) knowledge must precede subconscious (implicit) knowl­edge, something that Rutbeñord surely would not prescribe. Thus a justi.fication for consciousness raising might lead these readers to justify trad1tional grammar instruction ifthcy do not grasp Rutheñord's mejor distinctions (see the preceding discussion). AB a probable type of con­sciousness raising, processing instruction does not claim tbat explicit knowledge leads to implicit knowledge and does not prescn'be any role for explicit knowledge. lndeed we cannot k.now what kind of grammati­cal knowledge is contained in the developing system. Again, at tbe risk of being repetitive, we state here lhal processing instruction is con­cemed with delivering form-mearung connections as intake to the devel· oping system. The ways in whicb the developing system accommodate• these data and eventually restructures is beyond the scope of both mput processing and processing inatruction Thus, whatever so-ealled explicit or consdous knowlcdge is containcd in processing instruction (i.e., the explanation part of processing instruction) is not what exists in the developing.system that is tapped for output processing. We would prefer, then, to thmk of our approach to instruction as not about raising learn­crs' consciousness about grammatical fonn but instead as enriching their subconscious intake.

Page 52: Input Processing and Grammar Instruction

86 VanPatten

CONCLUSION

We have descn"bed processing instruction m aome detall in this chapter, fOCUSUlg on t.he natw'e of structured input activitíes. We have seen that ~ro­oessing ínstruction is not just explanation about language and ~tical form. It mcludes infonnation to the leamer aboul what to attend to m the mput. Most unportantly, it includes structured mput activities that enrour­age Jeamcra to make form-meaning mappings thcy might not make when exposed to nonstructured or "spontaneous" input. An impor:tant ~ of structured input activities is that learnera attcnd to the mearung of an input sentcnce. It is not enough that leamers simply be directcd to t.he form; they must aJso use it to romprehend the meaning of tbc scntcnce. Thus, the activ­itiea ere formulated with the procesaing strategies of lcarners in mind.

A number of issues arise from the presentation in this chapter. Since input processing bere is related to form-meanlng mappings, is it possible to Will proc:essmg instruction for forros that do not carry meaning? A related issue is tbe relationsbip of processing instruction and input processing to tbe acquisition of syntactic rules. Many syntactic rules do not seem to be moti­vated by meaning and at 6rst glance it does appcar tbat procesgjng instruc­tion is useful in terms of belping leamera acquire these rules. These are important issues and we will address tbem in Chapter 5. First, howevet; we will examine t.he research tbat supports the eJTectiveness of processing instruction.

!4 RESEARCH ON PROCESSING

INSTRUCTION

INTRODUCTION

Suggestions about SECOnd 1a.nguage classroom methodology come and go. Richards and Rodgers (1986) and Lar&en-Freeman (1986), for example, examine met.hodologies such as Suggcstopedia, The Natural Approach, Total Physical Response, The Silent Way, Community Counse!1ing l.earning, among others. What is interesting about t.hcee met.hcxlological innovations ia that their purport.ed benefits are scldom researched. Empírica! studies about what learners gain in sucb prograrns as opposccl to otbersjust can't be found in the literature on SECOnd language Jearning and teaching. Only the out­romes of immersion approaches in Cenada seem to be investigated witb any consistency (e.g., Harley and Swain, 1984; Swain, 1985).

Research on explicit grammar instruction and focus on form, bowever, is much more plentiful as evidenced in revicws such as Chapter 8 of Le.rsen· Freeman and Long (1991), Chapter 14 of Ellis (1994), and Chapter 5 of Lightbown and Spada (1993). This plentifulness may be owing to tbe mther narrow focus of the research; re:;earch on grammar instruction and focu.s on form examines only one of the many romponents of any given metbodology and thus tbe research is easier to conduct than research that examines a mcthod in its entirety. The plentifulness may be owing to innovations in the­ory about language or innovations in rescarcb methcxlology that suggest ncw research pa.radigms. Or, t.he plentifulness of the research may be owing to the fact that grammar instructlon and focus on form have been hotly debated topics in SECOnd language studies for some time. People are simply intcrested in tbe issues related to thesc topics and thcy just don't go away.

The purpose of the present chaptcr is to examine the empírica) studies that have been conducted on procCMing instruction. Because it is a type of focus on form and not a methcxlolog)I processing instruction is readily

87

Page 53: Input Processing and Grammar Instruction

88 VanPatten

researchable and to date five major studies have been conducted on it. All five studies involve the leanúng of Spanish, although there is research in progresa that includes learners of English. The first study that we will review is VanPatten and Cadiemo (1993). This was the first empirical inves· tigation on processing instruction and its benefits. Since the study focused on object pronouns and word ordei; we classify it here as a study about the acquisition of syntax. The second study, by Cadiemo (1995), focuses on the past tense, whereas the third, VanPatten and Sanz (1995), is a partiaJ repli· cation ofVanPatten and Cadiemo and examines whether or not the effects of processing instruction are observable in a variety of meaning-based and communicative tasks. The fourth study, by Cheng (1995), focuses on the acquisition of lexical-aspectual items, namely, the copular verbs in Spanish. In ahort, we have three empirical studies reporting on different types of grammatical features in Spanish and one study that focuses on the assess­ment tasks used in research on processing instruction. The fifth and final study, VanPatten and Oikkenon (1996), reports on the role of explicit infor· mation in processing instruction.

RESEARCH ON SYNTAX: OBJECT PRONOUl\IS ANO WORD ORDER IN SPANISH

Overview

VanPatten and Cadiemo (1993) set out to investigate processing instruction based on P3, the first noun strategy, with learners of Spanish. Because the research methodology they used in.ftuenced subsequent studies, we will spend considerable time descnoing and reviewing it here. Their research questions were the following:

l. Does altering the way in which learners process input have an effect on their developing systems?

2. If there is an effect, is it limited solely to processing more input or does instruction in input processing aJso have an effect on output?

3. If there is an effect, is it the same effect that traditional instruction has (assuming an effect for the latter)?

Motivation for the Study

As we saw in Chapter 2, second language learners are k:nown to assign the first noun of an input sentence the role of subject even if it is something else, reflecting sorne universal principies of semantic role assignment. VanPatten

Research on Processing lnstruction 89

(1984b) shows that learners consistently misinterpret sentences such as La visita el chico as "She visits the hoy" rather than the correct "The hoy visits her." LoCoco (1987) has demonstrated that learners misinterpret sentences such as Hacia la madre empuja el chico a su podre as "The mother pushes the hoy to the father" rather than the correct "The hoy pushes bis father toward bis mother." (For additional research, see Binkowski, 1992; Gass, 1989; Lee, 1987.) VanPatten and Cadiemo argued that this processing strat­egy may cause learners of Spanish a number of problems in delivering intake to the developing SYStem. First, it may cause leamers to misinterpret object pronouns as subjects if the suhject is null or post-posed after the verb (see the preceding example). Altematively, if the subject is present and is aJso preverbal, leamers may not attend to the object pronoun once they assign the suhject its grammatical and semantic role. Leamers might sim­ply "fill·in" the object based on lexical semantics and context, not relying on the grammatical cue in the input sentence. Another problem is that leam­ers may skip over the object marker a in the input. If they are relying on word order, then the object marker is of relatively low communicative value for processing the meaning of the sentence. Finally, leamers may not rely on verb endings as cues to the subject when the suhject and object are of dif. fering number (i.e., one plural and the other singular). Given Pl(b), that leamers will process lexical items before grammatical items that enrode the same semantic information, learners may simply not process verb endings if the subject is present or ifthey misassign the object pronoun tbe grammat­ical role of suhject. The result of these possibilities is incorrect intake data delivered to the developing SYStem as evidenced by a number of learner errors: (1} misuse of object and reflexive pronouns for subjects, for example, Lo vieM for "He comes," Me hablo for "! talk;" (2) problems withgustar, for example, Yo gusto mi profe$0r for "l like my professor;" (3) a general over­reliance on subject pronouns (e.g., Manuel es mi compañero de cuarto. Lo es muy simpático. Él estudia arte."); (4) a complete lack of use of tbe direct object case marker a, for example, Tu no llamo mis padres; (5} an overre­liance on subject-verb-object word order and problems in the acquisition of object pronouns, for example, El/.Q8 llama mi, and (6) a delay in the acquisi­tion of person-number endings. In short, this one processing strategy may ha ve a significant effect on a number of areas of the learner's developing sys­tem of Spanish.

Subjects and Groups

VanPatten and Cadierno used the first noun strategy as the starting point for their study. They compared three groups ofleamers in the acquisition of object pronouns: (1) a control group that received no instruction: (2) a tra­ditional group that received instruction based on one of the most widely

Page 54: Input Processing and Grammar Instruction

90 VanPatten

adopted college-level textbooks and its accompanying workbook and lab manual; and (3) a processing group that received instruction developed by VanPatten and subsequently incorporated into \.án.Patten, Lee, Glass, and Binkowski (1992).

Six second-year c1asses of Spanisb from the University of Dlinois were randomly assigned to one of the three treatment groups, whole classes being used to avo.id problems inherent in voluntary self·selection of subjects. Alter pretesting and screening of subjects for a variety of background variables including first language, previous study of Spanish, knowledge of other Romance languages, and hearing and learning impairments, a total of 80 subjects participated in ali pha.ses of the study: 27 in the control group; 26 in the traditional group; and 27 in the processing group. Subjects were stu· dents in a 1anguage program with a communicative methodology based on the Natural Approach. Grammar instruction was absent from c1ass time, although subjects routinely completed some grammar focused exercises out­side of class as homework. Classes met four days a week with the majority of c1ass time spent on interaction, listening, and reading. At no time during the course of the study were object pronouns or word order a focus of study in the subjects' homework. All subjects' first language was English.

lnstruction and Treatment Materials

lnstruction for both the traditional and processing groups consisted of two days of class time with no outside homework; in each group the teaching materials were picked up at the end of the first class day as well as at the end of the second class day. Two days are about the normal amount of time that a language curriculum in Spanish would devote to the explicit learning of object pronouns. At no time were the subjects or their regular instructors aware that this was a comparative study and that learners in other c1asses were receiving different instruction.

The traditional materials consisted of explanation and exercises taken largely from Puntes de partida and its aocompanying workbook and 1ab manual. The researchers chose this textbook because at the time of the study it was the most widely sold and used college-level textbook in Spanish and probably represented typical grammar instruction for most college-level instructors. In these materials, students receive the full paradigm of direct object pronouns in Spanish, an explanation of what direct object pronouns are and where they are placed in a sentenoe. Students are subsequently led through a number of oral mechanical practices, oral meaningful practices, and finally oral communicative practices. (See the description of Paulston's hierarchy in Chapter 1 of the present book.) The workbook and laboratory materials consist largely of mechanical and meaningful oral and written practices. Emphasis is always on production albeit manipluated production.

Research on Processing lnstruction 91

Processing instruction consisted of the explanations and activity types described in detail in Chapter 3. Recall that the emphasis in these materials is on processing input and making correct form-meaning mappings. At no time did the subjects in the processing group produce an object pronoun dur­ing the instructional phase.

As they developed the materials, VanPatten and Cadiemo made acljust­ments in activities in order to balance the two sets of teaching materials and control for possible features that might influence the outcome of the study. Specifically, VanPatten and Cadierno controlled for the following features: total number of activities, total number of tokens (sentences produced in the traditional group versus sentences interpreted in the processing group), per· oentage of whole-class activities versus percentage of pair activities, peroent­age of oral (aural) versus written activities, and number of visuals. In addition, VanPatten and Cadierno a1tered vocabulary in the activities so that the vocabulary used in each packet was roughly the same and consisted of hlghly frequent and familiar vocabulary for students of second-year Spanish. At the same time, VanPatten and Cadiemo checked the vocabulary in the instructional packets agsinst the vocabulary used in the assessment tasks in order to avoid a vocabulary bias for one group or the other.

Since VanPatten and Cadierno were interested in the relative benefits of two ditrerent kinds of focus on form, they also controlled for a factor known to affect similar studies: the instructor. Previous comparison research has shown that instructora may bias toward one approach or another and affect the outcomes ofresearch (see, e.g., Spada & Lightbown, 1993). Rather than allow the regular classroom instructors to deliver the treatment, VanPatten and Cadiemo removed these instructors during the treatment and testing periods. Instead, Cadiemo taught both experimen· tal groups. Experienced in traditional instruction but not in processing instruction, she very carefully adhered to the tenets of the two approach­es and made sure that at no time during processing instruction did the subjects produce utterances with object pronouns. Cadiemo's hypotheses at the outset of the treatment period were that the traditional group would be better at producing object pronouns and that the processing group would be better at comprehending them. She kept these hypotheses until the first round of results appeared and as we wil1 see, her own bypotheses and past experience in teaching with traditional instruction did not influ­ence the outcome of the study.

Pre- and Posttests

A pretest/posttest design was used to measure gains due to instruction. Four versions (A, B, C, and D} of two different assessment tasks were developed. The first was an interpretation task. In this task, learners heard 10 target

Page 55: Input Processing and Grammar Instruction

92 VanPatten

sen ten ces along with five distractors. Five ta.rget sen ten ces were of the order object pronoun-verb-subject and the other five were of the order object marker + noun-object pronoun-verb-subject. The five distractors were sub­ject-verb-object sentences. Ali sentences were simple declarative sentences in the present-tense containing familiar vocabulary. Subjects demonstrated interpretation of each sentence by selecting one of two drawings projected on an overheed screen in front of the class. Thus, for the sen ten ce A la chica la abraza la mamá "The mother hugs the girl," subjects chose between a pic­ture of a woman hugging a child (who does not hug back) and a picture of a child hugginga woman (who does nothugback). This task hada time linút. Once the subjects heard a sentence, they had eight seconds to make their picture selection before the pictures were removed and they heard the next sentence.

The second assessment task was a sentence-level written production task consisting of five target items. Each item consisted of a twcrpart sentence that corresponded to a two-part drawing. The second part of each sentence was incomplete and was to be completed by the subject based on the visual cues. For example, one ítem was El chico pi.ensa en la chica y entonces =---= __ "The boy is thinking about the girl and so " The two-frame picture that accompanied this sentence shows the boy think­ing about the girl in frame one and then caJJing her in frame two. The cor­rect expected response would be to write y entonces lo llama "and so he caJJs her." Ali sentences were simple declarative sentences in the present-tense containing familiar vocabulary. This task aJso had a time linút. Subjects had 15 seoonds to complete eech sentence after they saw the pictures. (Time lim­its and vocabulary familiarity were detemúned by pilot testing of both the interpretation and production tasks.)

In order to break up the presentation of the two assessment tasks, VanPatten and Cadiemo administered a 10-minute diskactor task between the interpretation and production tasks. This distractor task consisted of 10 sentences to be translated from Spanish to English with none containing object pronouns.

We should point out here that the two assessment tasks used by VanPatten and Cadierno were designed purposefully to counter test bias. Note that the interpretation task would favor the processing group since it is similar to if not identical to some of the treatment activit.ies used in processing instruction (&ee Chapter 3). At the same time, the product.ion ta.sic would favor the traditional group. The design for thi• pArt.ir.ular task was identical to several of the product.ion tasks included in the treatment materials for the traditional group. Thus, VanPatten and Cadierno were able, if necessary, to account for any apparent task bias­es should they appear in the results. AJ3 we wil1 see, however, thls was not the case.

Research on Processing lnstruction 93

Scoring

Each ta.rget item in the interpretation task received a score of 1 or O, 1 for correct picture selection and O for incorrect selection or nonselection. The total points possible was 10.

Each target ítem in the production task received a score of 2, 1, or O points. If the subject produced a correct object pronoun and aJso placed it in the correct position in the sentence the ítem received a score of2. Ifthe sub­ject failed to produce any object pronoun even if the sentence was correct by any other standard, the item received a score of O. Inbetween cases (correct object pronoun produced but in the wrong place, incorrect object pronoun used but in the correct place, incorrect object pronoun in an incorrect posi­tion) received a score of l. VanPatten and Cadiemo adopted this liberal scor­ing procedure for inbetween cases in the event that the instructional treatments had sorne effect but subjects had yet to fully intemalized the tar­get linguistíc feature. Note, however, that in cases in which it was clear that the research subject was using the object pronoun as a subject pronoun, the researchers assigned the score of O, e.g., y entonces lo llama a la chica in which lo is incorrectly used for subject "he" and not object "him."

The pretest versions of the tests were used to establish baseline data and to elinúnate subjects from the p001. Van.Patten and Cadierno established an arbitracy 80% ceiling on the pretest for admission into the study; subjects who scored 80% and above on the pretest were eliminated sinoe they were already performing at levels in which they might not be affected by instruc­tion and could possibly skew the data. The group sizes given in the section above on subjects are the final group sizes after pretesting.

SUmmary of Research Design

The research design in Van.Patten and Cadierno, then, can be summarized in Figure 4-1. We add here that ali testing and i.nstruction was conducted in the subjects' regular classrooms and that the three posttests were administered immediately after the treatment on the second day of instruction, two weeks later, and one month later. The four versions ofthe tests (A, B, C, and D-see earlier on pre- and posttests), were used to create a split-block design to con­trol for test order. Thus, one group might receive test A first and then B, C, and Das the three posttests, whereas another might receive B as the pre-test with D, C, and A 8$ the three postl.ests, and so on. Raw ""°""' were subnútted to two separate one-way analyses of variance (ANOVA) with a repeated mea· sures design. One of the ANOVAs was conducted using the seores of the intepretation task, whereas the other was conducted based on the seores of the production task. Instruction was the Jabel used for the three treament groups, whereas t.ime was the label used for the pre- and two posttests.

Page 56: Input Processing and Grammar Instruction

1

1

94 VanPatten

Results

An ANOVA on both interepretation and production pretests revealved no sig­nificant clliferences between the three groupe prior to instruction. Thus, resull$ obtained on posttest measures are attributable to instructional effects.

Independent Variablea

lrult.ruction Time

~ ~ Control {none)

Traditional ProceMing Pre·in.struction Poct·inltruction

Oependent variables • ecores oo interpretation and production tas.ks

FIGURE 4.1, llA>search desi¡n in VanPatten and Cadierno (1993).

8

6

4

2

o._ I "• o ,' ...... 'O'-_____ _

I I

I I

I I

I I I

I ,' .......... o I <>•"""'" : º ................. ..

,/ ................. . ,,. ... ..

o ~~-.-~~~..,...~~~..,...~~~..,...~-'

~

j

--0- Control

....... <>....... Traditional

--0·· Processing

FIGURE 4-2. Reoults from Va.nPatten and Cadierno (1993) on the interpreU!tion task.

Research on Processing lnstruction 95

The resul1$ from VanPatten and Cadierno's study are interesting. On tbe intepretation task, there was a main effect for instruction, a main effect for time, and a significant interaction between instruction and time. Post-hoc tests revealed that tbe main effect for instruction was owing to the following:

8

6

4

2

o ............. .() ....... . p ·"- ........ .. ........ .. +- -o------·o ¡

I I

I I

I

Í ¡,• g:

..,

j

--0- Control

....... <>....... TTaditional

··O-- Processing

FIGURE 4-3. Rooulto from Vlml'atten and Cadierno (1993) on the production tasi<.

Table 4-1. ANOVASummaries for both lnterpretation and Production Tasks in Vanl'atten and Cadiemo (1993)

lntmpretation task Sourcé of vruiation

lnst.ruction Time Time x instruetion

Produ.ction ta.ele. Source ofvariation

lnslructioo Time Time x inst:ruction

·p < .01.

df SS MS F·value

2 3 6

2 3 6

1094.646 521.531 297.150

991.937 1361.453 244.392

&47.323 173.844 49.525

495.969 453.818 40.732

33.129'' 76.921" 21.913"

25.214" 97.016"

8.708"

Page 57: Input Processing and Grammar Instruction

96 VanPatten

the processing group madc significant gruns, whereas the traditional and con­trol groups did not. Thc main effect for time was due to the following: The posttest seores were significantJy groater than the pre-test seores. The inter­action was owing to the fact that the post.test seores for the prooessing group were greater than •ls pretest acores, but t.h.is was not true for the control and traditional groups. These finding:s an! summarized in Figure 4-2, which gives a sense of the tremendous gains made by the processing group.

Tbe results on the production t.a.sk were the same as for the interpreta­tion task, but for a different roason . The M<OVA yielded a main effect for instruction, for time, anda significant interaction between time and instruc­tion. However, the )>0$t-boc te6ts revealed that the effect for instruction was due to the fact that both the processing and traditional groups sbowed gains. whereas the control group did not. In addition, the processing and tradi­tional groups po&tletit acores were not sígnificantly different from each other. Tbe effect for time again was due to the pre-test/pcll3t-test differences, and the interaction between time and instruction was due to the two treatment groups' post-test seores being S1gnificantly higher than those of the control group. Figure 4-3 summarizes thcse findings.

Table 4-l displays the actual resulte of the two Analyses of Vilriance for those who wish to see the actual F and p values.

Discussion of the Findings

The resulls from VanPatten and Cadiemo's study suggest tbst processing instruction is more beneficial than traditional instruction. Subjects in th.is group not only gained in ability to interpret word order and object pronouns correctly, but apparentJy thc intake data made available to the developing system as a result of input procc;;sing hadan impact on the system such tbst learners were ablc to access thc ncw knowledge and use it during produc­tion. This is significant since cú ru> tinu: di.d .the subjccts in the processing group engoge in produ.clWn adiutiia during instroction. At the same time, the traditional group made gains only in production and not in interpreta­tion, suggesting that intruction did not affect thcir developing syst.em (a point we will explore in the next chapter). Thus the answers to VanPatten and Cadiemo's rcsearch questions an!: (l) yes, altering the way in which leamen process input data does have an efTect on thetr developing systems; (2) the effect is not limit.ed to input processing but is also ob5ervable in out­put; (~) "'""""""ing llM trA<11hnnAl 1ruitn1<1:ion "'""" diff PTPntial PJfect.<, with processing instruction be1ng superior overall.

Tbere are two limitations to the VanPatten and Cadiemo study that merit attention here. One is that the assessment tasks used were highly con­trolled sentence-level tasks. Although VanPatten and Cadiemo were not researching effects 8Cl"06S diffcrent t&ks but relative effects on one task

Research onProcessing lnstruction 97

attributable to differcnt instructional treatments, this limitation is impor­tant since we would like to know whether or not learners can access their systems on lcss controlled and more comrnurúcative tasks. We will take this up m a later section of this chapt.er wben we review VanPatten and Sanz (1995). The other limitation is thal only one aspect ofinput processing was ínvestigat.ed, P3, the first noun strat.egy. What of the other principies of input processing outlined in Chapter 2 and the effect.s that they have on acquisition? We tum our attention now to a study on "-ero inBections.

RESEARCH ON VERB l\Fl.ECTIONS: TIE SPANlSH PAST· TENSE

Overview

Cadiemo (1995) reported on a study in whicb she investigated the benefits of both trll<litional and processing instruction on the acquisition of the pretent tense (simple past) in Spanish. Her study was based on Pl(b), that leamel'!I procesa Jexical items as opposed to grammatical form when both encocle the samc aemantic infonnation. Cadiemo posed the following research questions al the outset of her study:

1. Will there be any difference in how leamers receiving no instruction, pro­cessing instruction, and Lraditional instruction interpret sentences in which post temporal referencc is only expressed by verb morphology?

2. Will thero be any differences in how leamers receiving no instruction, pro­cessing instruct.ion, and traditional instruc:tion produce correct past tense forma to expresa past tense meanings?

Motivation for the Study

The preterit tense is a morpbologically complex syst.em for learners of Spanish. Unlikc English, there are sorne 16 difTerent fonns for the regular pret.erit tense owing to the inflections for both tense and person-number and the type of vero bemg ínftected. A1so unlike English, the Spanish preterit undcrgoes a stress shift from the stem of the verb to the actual tense inftection with the exception of first person and sooond person plur­al In addition, varioua cl400Cll of,-crbo undcrgo a stem-vowcl alternation in certain form.s of the preterit . Tlus past tense system is compounded by the presence of a highly frcquent irregular subsystem in which the stress shift does not ocx:ur and in whích the st.ems of verbs undergo both \"Owel and final consonant alterations. The preterit system is notoriously difficult for leamers of Spanish.

Page 58: Input Processing and Grammar Instruction

98 VanPatteo

Cacbemo reasoned that one of the problems m the acquisition of the preteril system m Sparush is that Jearners may nol atlend to the verbal inilections in the input. Citing Pl(b), which she ca1ls "tbe lexical processing strategy," Qidiemo suggested that learners rely on lexical items su ch as adverbuils of temporal reference to interpret the tense of a sentence and ignore the verbal inflections. (See Chapter 2 for details on Pl(b) and a review ofthc relevant research.) She thus concludcd that prooos.sing instruc­tion, in wluch the adverbials of temporal reference were removed from the structurcd input sentences, would push learners to att.end to the oonnection bctween tenae and verb encbngs. Leamers would use the verbal inllections 88 indicators of tense since the lexical indicators of tense would be absent.

TheStudy

Given thal Cadiemo's study on the preterit tense W88 an almost exact repJi. cation ofVanPatten and Cadierno (1993) reviewed in the previous section, we will not examine the speci6cs ofher study mas much detail 88 we did for VanPatten and Cadiemo. We will insteacl make references to VanPatten and Cadiemo as we proceed, noting differences between the two studies as appropriate

Subjects and Groups Cadierno compared three groups: a control group {n•20), a group receiv­

ing traditional instruction (n = 19), and a group receiving proccssing instruc­tion (n•22). The subjects carne from the same population as that descnbed in VanPatten and Cadiemo but were not thc samc subjects. Cadiemo scrooned her subjects for background 88 in VanPatten and Cadiemo and aJso used a 60% cutoff score on pretests (more stringent !.han in the VanPatten and Cadierno study) to ensure a more homogenous pool of subjects. Only subjccta who completed ali phases of the experimenl remained in the final subject pool The n sizes given above were the final n sizes. Subjects in ali thtte groupe were unaware ofthe comparative nature ofthe study.

lnstructional batment As ID the VanPatten and Cadiemo study, Cadit>mo developed a set of

teaching materials for the traditional group based on Puntes de partida. She aJso developed a set of materials for processing instruc:tion and then baJanced tho two BCI.a of ln4toriala ÍOT thoee potential intcrvoning footu.ToG mentioned previou.sly: number of tokens, activity types, use of visuals, vocabulary, and so on. Proccssing instruction here focused on rcmoving temporal adverbs from the structured input activities so that leamcr attention could be direct· ed toward verb endings 88 indicators oftense. Instruction for both treatment groups consisted of a tw<rday treatment in class with no outside homework.

Research on Processing lnstruction 99

The control group continued with thcir regular activities and at no time dur-11\g the experimental treatment were the sub.)C(tS of any group scheduled to recewe ínstruction on the post tense as part of the regular ooursework. The regular classroom instructora were removed dunng ali testing and treatment penods with the researcher conducung ali testing and instruction herself

Assessment Tasks and Scoring Cadierno developed four versions CA, B, C, and D> of two assessment

tasks, an interpretation task and and production task. The interpretation task was a 20-item aura! test of simple scntences without adueroi.al indica­tions of remporal reference, e.g., A11to11io llamé a sus ¡xui.res por reléforw, "Anthony called h.is parents on the phone." Of these sentences, 10 itema were in the present tense (distractors) and 10 items in the past tense (tar­get items). Subjects had to listen to the sentence and then indicate whether thc sentence they heanl was in the prcsent or post tense or whether they could not tell. Tbe production task consisted of five written items with blanks in which the leamer had to complete the sentence with a coJTeCt '-erb cucd by an infinítive in parentheses, e.g., .l\)'er )'O (escuchar) ltJI nolirias, "Yesterday 1 (listen) to the news. • Cadiemo unposed time limits for both tasks as dcecribed for '41.nPatten and Cadiemo. She also administered an intervening chstractor task consisting of trnnsla­tion sentences unrelated to tense dist.inction.

Following VanPatten and Cadiemo, Cadiemo utilized a split block design and administered one pre-test and three post-tests varying the test versions among the groups. The post-tests wcre administered immediately af\er instruction, two weeks later, and one month later.

Scoring for the interpretation task consisted of a 1 versus O point systcm per item for a total possible of 10 points. A eubject received 1 point ifthe tar­get sentence was assigned oorrectly post tense reference and received O points if the tense assignment w88 wrong or the subject indicated an inabil· ity to determine the tense. Cadiemo scorcd the production task using a 2, 1, O point system for a pos51ble total of 10 po1Dts. A sub.)E!Ct received 2 points if the sentenal completion contained a verb ID the con-ect preterit fonn. lfthe verb was in the past tense but was the wrong person or if the verb was in the past tense but the subject had switched verb category endings, the sul>­Ject received a seore of l . Any other response received a seore of O.

Results

Cadiemo conducted separate onc-way ANOVAs on each assessment task using raw seores as the dependent variable. She used instruction (the vari­ous treatment groups) and time (pre· vs. posttests) as the labels for the indc­pendent variables. A preliminary set of ANOVAs on the pretest seores

Page 59: Input Processing and Grammar Instruction

100 VanPatten

8

9'----..

7 ' ' ' ' ' ' ' ' -0- Control 6 ' ' ' ' -~- Tradibonal ' ' ' 5 ' --<>-- Proces.stng ' ' ' '

FIGURE 4-4. Rewlta from CadJerno (1996) on t.he lnl<!rpn.tation task.

revealed no significant dill'ercnccs betwoon the three groups prior to tbe instructional treatments. Once again, any observable effects are attribut.ed to the instructional trcatment.s.

Cadierno's resulta were ident.ical to tbose ofVanPatten and Cadierno. For the interpretation task, she found a main ell'ect for instruction, a main ell'ect for time, and a significant interact.ion between instruction and time. Once aga.in thc post hoc tests rcvealed lhat the ell'ects were due to the following contrasta: Tbe processing group improved signifi­cantly from pre- to posttesta, whereas thc traditional and control groups did not. In addit.ion, therc waa no significant dill'erence between the con· trol and traditional groups. These findings are displayed grapbically in Figure44.

For the production task, the ANOVA yielded a main ell'ect for instruc­tion, a main efTect for time, and a &1gnificant interaction between instruc­tion and time. As with the resulta ofVanPatten and Cadiemo, these effects were cau.sed by the following contrasta as revealed through post hoc tests: Both the traditional and the processing group improved significanUy, whereas the control group did not There was no dilference between the performance of the tradlbonal and processing groups. The result.s of the

Research on Processing lnstruction 101

production task aro chsplaycd in Figure 4·5. Once aga.in, for lhosc wbo WlSh to see the F and p values of the analyscs,

these are present.ed m Table 4-2.

6

4

2

' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' '

~

·--<>·---0-

-·~-

--<>--

"'

J FIGURE 4-6. Retulta l\oom Cadlemo (1996) on the production task.

Control

Proces.stng

Table 4-2. ANOVA Summari,.. for both lnterpret.ation and Production Taska in Cadiemo (1995)

Interpreta.bon tul< Sourooot~ ),,.,,..,......

Tune Tune :oi lJ\lll'UCbon

.. p < 01

df

2 3 6

2 3 6

SS

228980 115.561 80730

394 624 508.399 141.203

MS

llt.490 38.250 13.456

197.312 169.466 23.534

f.vaJue

2'1.379'" 23.238"" 8.llT'

11526"' 58.442"" 8.116"'

l

Page 60: Input Processing and Grammar Instruction

102 VanPatten

Discussion of the Finding.s

Cadiemo's findings on the acquisition of the past ten:;e support the findings and oonclusions reported in VanPatt.en and Cadiemo; processing instruction sccms to be more beneficia! than traditional m:;truct.ion Subjed.s receiving Pl"OCC6!ilfl8 instruction make gains in both the ability to assign ten.se to input sentenccs when adverbials are absent and in the ability to produce past ten.se forms. Sub)Cd.s roceiving traditional instruction make gnins only in produc­tion of past ten.se forms. The answers to Cadiemo's rescarch questions, then, are: (1) there is a difference in the way in which thc three groups mterpret past ten.se scntenoos in which ten.se is encoded solely in the verb mo~holo­gy; only prooessing instruction improves the leamer'~ ability to_do this; (2) there is no dilference in the way that learners recei111ng processmg mstruc­tion and traditional instruction produce verb forms aftcr treatment; both result in improved performance. Aga.in we point out that the findings are espeaally remarkable oonsidering that the subjccts in the prooessing group never once produced past tense forms during the mstructional treatment.

We thus have evidence so far that proce551Dg lll5truction is indeed bene­ficia! for leamers and is superior to traditional instruction The evidence is basOO on two clistinct t.arget linguistic items, one synt.actic the other mor­phological. To be sure, Cadierno's resean:b sufTers from the same major lim­itation as that ofVanPatten and Cadierno's, that we have no evidence that the benefits of processing instruction are availablc during more oommu­nicativc task.,. We tum our attention now to this issue.

RESEARCH ON COMMUlllCATIVE OUTPUT: SENTENCE·LEVEL VERSUS DISCOURSE·LEVEL TASKS

Overview

ln their discussion of their results, VanPatten and Cadiemo (1993, pp. 23S-239) reoogn.ize that their results do not address the issue of ability to use grammar in oommurucath-e performance. Smce theU" aun was to compare the relative elfects of processing and traditional mstrucilon, they did not see the nced to incorporate a batteiy of taska to measure whether or not mstruc· tion results in more aa:urate performance during commurucative tasks.

VanPatten and Sanz (1995) set out to research tlus limitation in both the VanPatten and Cadiemo and the Cacliemo studJes. ln a pertial replication of VanPatten and Cadiemo's study utilizing only a oontrol group and a pro­ccssing group, they oompared the effects of instruction on three output mea­sures: (1) tho same sentence-level task as used in VanPntten & Cadiemo; (2) a structured question-answer interview; and (3) a video narration task. The

Research on Processing lnstruction 103

basic research question they asked is the following:

l. Do the observed effects for processmg in..m.tction on the sentence level task also obtain on other languagc production tasks?

Since both the VanPatten and Cadiemo study and the Cadiemo study uti­li1.Cd written production tasks, VanPntum and Sanz aJso wanted to know whether oral versus written production were equally or differentially alTectr cd by processing instruction. Thus a sccondaiy question is the foUowing:

2. Assuming the effects for processing instruction are observable in aU three assessment tasks, does the mode of task (written vs. oral) makc a difference?

Motivation for the Study

An examination of the production tasks u.sed in the previously re111ewed studies on prooessing instruction revea! that the tasks are essentially &en­tence-le--el tasks. Although a time linút was imposed for the completíon of cach item, it could be argued that the tasks invite monitoring since they appear to be form focused and quite oontrolled. ln VanPatten and Cadiemo, the production task was a sentcnce complction task in which subjects had to write a short phrase based on a visual cue. In Cadiemo, the production task was a fill·in-the-blank task in which subjects only had to provide a verb. In each case, the learners need only focus on thc grammar to be used since ali lexical material and oontext was supplied by the stimuli. These tasks have face validity for instructora, sincc thcy rcsemble the grammar testing that many instructors utilize. Tbus, anyone who wishes to argue that the testing was unfair to subjects in the traditional groups, cannot make the argument.

However, as pointed out in VanPntten (1988), one of the problems of research on the effects of instruction m geMrol is that assessment tasks nor­mally do not measure oommurucative ability. As lmowledge-based tasks, they supply the researcher with information about possible changes in the developmg system but they do not supply us with information about oom­municati~-e performance. Thcsc tasks include sentence-level compleaon as in the VanPatten and Cadiemo study. fill-m-the-blanks as in the Cadiemo •h1dy, as well as SPnlPnN> mmhimng, grAmmAticality judgments, error detection, cloze tests, elicited imitation, and other tasks.

Because of these assessment limitations, it is easy to dismiss the obscrved beneficia! results ofprocessing instruction described in the previous studies. VanPntten and Sanz thus decidcd to investig¡¡.te the effects of processing instruction independently of the clfects of traditional instruction by adding

Page 61: Input Processing and Grammar Instruction

104 VanPatten

two more -..smcnt tasks tbat involve langua¡e producbon beyond the sentence lcveL In t.hJs way, tbey could observe whether or not the mecha· nisms for malcing output could make use of the new knowledge in the devel· oping S)'lllcm Once again, since tbe study relies largely on \únPatten and Cadiemo's research design, we will be briefin our pre.entation, pointing out the DUIJOr differences between the two studies as eppropriote.

TheStudy

Subjects and Groups Subjecl8 were once again students ofSpanish at thc aecond-year college­

level culled from the same population as the previous studies but were not thc samc subjects. Sinoe VanPatten and Sanz were concerned witb assess­ment tasks and not traditional versus processing instruction, they utilized only two groups: a control group and a proce6Slng group. Alter background screerung, participation in all phases of the study, and performance at 60% or below on tbe pre-test (see the section that follows, ~ent Thsks and Sconng), tbeir final pool consisted of 44 subjects: 'l:T m the processing group and 17 in tbe control group. A1 no time dunng the expmment did subjects rece1ve regular instruction on the target 1teim.

lnstructional ñatment Subjccta in tbe prooessing group received the cXlld same two-day instruc·

tional treatment as tbe subjects in the \llmPattcn and Cadierno study. Tbe focus was on object pronouns and word order with thc materials designed to alter suQjoots' relianoe on the first noun strategy. Thc control group reoeived no instruction and continued with its regular classroom activities during the two-day period in wbich the prooessing group roceived its instruction. Tbe regular instructor for the processing group was removed and replaced by Sanz during the treatment period. As in the prcvious study, subjects in both groups wcre unaware of research being conducted in the other group.

Assessment Tasks and Scomg VanPatten and Sam used a split-block design with various versions of

each esacssment task and dilferent orders of presentation to control for test bias. Eru:h task was adm.ini.stered twice; once a os a pretcst and once as a posttcst immediately following instrucbon. The new tasks were developed by San>: a.. part of ht."1· dissertation researcl1 (SrutL·AlcWA. 1994).

The finrt task was an interpretation task. The only diJTerence between it and thc onc used in the \únl'atten and Cadicmo study was that the num· ber of target itema was doubled; tbe task containcd 20 targoted it.ems and 10 distrnctors. The l versus O scoring procedure from Van.Patten and Cadicmo was used once again. Since VanPatten and Sanz were conoemed

l ResearchonProcessing lnstruction 105

with peformance on various output tasks, the purpose of the interpretation task was to narrow tbe sub.)CCt pool. First., it served to eliminate subjects pnor to treatment; only sub.)ecia who scored 60% or below on the pretcst. remruned in tbe subject pool. Second, it Berved as a pretest/po6ttest measure of effects on the developing i;ystem; only processmg subjects who showed a grun after treatment remained in the subjcct pool for tbe e.nalyses ofthe out­put tasks. The reasoning here was that we were testing the effects of pro­cessing instruction on output and wanted to eee if those who benefited by prooessing instruction on the interpretation task where no output is required, could subsequently use thcir new knowledge on output tasks.

VanPatten and Sanz utilized threc diJTercnt production tasks. Thc first was a sentence-level completion task identical to that used in VanPatten and Cadiemo with the exoeption that 14 scntences were used rather than five: cight target itema and six distractors.

The second production task was a question·answer task based on a series of pictures that described a story. The series contained seven pic­tures and subjec:ts had to answer 11 questions based on what they saw in the pictures. Of tbese 11 questions, seven were target questions. A sample target question is iQué hace ~l chico con la banana? "What is the boy domg with the banana?" Tbese typeg of questions would require a direc:t ob¡ect pronoun and a conjugated verb witbin a simple sentenoe in natu· rally occurring discourse, for example, La saca de lo canasto, "He is tak· ing it out of the basket."

Tbe tbird production task consisted of a story telling task in which subjects reoounted the events they saw in a brief two-minute video clip. Each video story contained seven connccted events. Asan example, onc clip involved a man who comes home with bis groceries, pulls out a pota· to, washes it, peels it, cuts it up, fries it, and 80 on, until he eats it. Thc instructions specified that the subject should provide as much detail as possible, 80 that a student in another class could identify the video clip being descnl>ed from among a series of video clips. This was done not only to elicit as mucb information as possible, but also to provide a com· municative context to tbe task. Each video clip was shown twice and thc subJects were required to tell the story only after viewing it the second time. In this way, sub_¡ects created a short monologue of connected utter· anees in wbich object pronouns should be used to avoid unnatural repe­títion of object nouns.

Because VanPatten and Sanz were a1so interested in the effects of oral versus written mode on performance, they created botb a written and an oral version of each of the preceding tests. Thus, in the sentence-completion task, subjects had to either respond orally or in writing. In the question· answer task, subjects had to rcspond in writing to written questions about the pictures or orally to aura) questions about the pictures. In the video nar-

Page 62: Input Processing and Grammar Instruction

'I06 VanPatten

ration task, subJects either wrote out a narration or they produced an oral narration.

The use of three different production tasks wilh varymg numbers of items and types of responses, created a problem in comparability. To solve this, VanPatten and Sanz first transfonned ali scorcd mto ratios. To do this, they fonned a denominator using the number of of critica! items multiplicd by two. They then calculated the numerator by adding the amount of correct responses multiplied by two, the amount of incorrect attcmpts multiplicd by one and the amount of cases in which the item was not supplied at ali multiplied by O. Beca uso of thc nature of the video narration test, each subject could produce a ditTercnt number of sen· tences with preverbal object pronouns, which means thnt the denomina· tor in thc ratio could vary. To account for this, a special procedure was followcd One native speaker read the protocole and dccided on the num· her of obligatory occasions for the UBe of preverbal object pronouns gen­erated by thc subject. The number of occa.sions was multiplied by two and the result, representing the total possible po1.nts, was the denominator of a ratio. Two other native speakers independently scored a sample of the protocol~ and agreed on 100% of the cases.

Aft.er deriving the ratios for ali three tasb, 'hnPatten and Sanz tran.s­fonncd thc ratios into percentages and then transformed these percentages via a special function-y'=2 ara;in(sqrty)- in ordcr to satisfy the ANOVA nonnality assumption.

To summarize, VanPatten and Sanz wcre interestcd in researching the effocts of processing instruction across three ditTcrent production task.s con· ducted in both the oral and written mode. The research design of their study is summarized in Figure .4-6.

_.-T--_ 8fint.tnce ~ Video

i...v.I Amwer Narnuon

FJOUJU!: 4-8. ii-.rch Oeoi¡¡n in VanPatten and Sanz 0995).

1 Rcsearch on Processing lnstruction 107

Results

The researchers conducted a seri~ of /·tests on ali tasks prior to instruction. These tests revea1ed no significant dJJTerences between the control and the p~g groups. After instruction, VanPatten and Sanz conducted a /-test on the mterpretation task and found a sublitantial gain in the processing group's performance, similar to the substantial gains made in the previous stuclies. The control group's prctest/posttest seores were 5.71 and 5.74, respectively, whereas those of the processing group were 5.78 and 11.07. Thus, processing instruction again scemed to alter the strategies used by learners to interpret sentenccs. Wc tum our attention now to the results of the production data, examining first the qucstion of mode.

VanPatten and Sanz conducted scparat.e ANOVAs on each task using mode (written vs. oral) and time (pre- vs. posttestl as the independent vari­ables. They also conducted post hoc comparisons based on the results ofthe ANOVAs. Their findings revealed the followmgc

l. The processing subjects improved sígnificant]y from pre- to posttests on ali three tasks whereas the control group clid not. Tbe greatest gwns showed up in the sentence-level and video-narration t.asks.

2. Even though the processing sub)Cets improvcd, they perfonned better in the written mode than the oral mode on the sentence-level completion task and the video narration task, but no clilTerence was found Cor mode on the question-answer task.

In the course of this first set of analyscs, VenPatten and Sanz discovered that the question-answer task was yiclding problcmatic data. Even though the pro­ccssing subjects improved on this task and this improvement was signiñcantly diffcrent from pre- to posttcsting, the gains were slight. Overall, subjects seemcd to be producing few oQject pronouns in this tesk, opting instead to repeat full object nouns in their answers. VanPatten and Sanz therefore decid· ed to eliminate the question-answer task from further anaiyses, concentrating on clifferenoes between the sentence-level task and the video-narration t.ask.

G1ven the results on mode, VanPatten and Sanz conducta! another set of ANOVAs in which mode was removed as a variable. They ran two sets of analyses, one on the results in the written mode and another on the results of the oral mode with the indcpendent variables being instruction, time, and task. In tlus way, they could examine perfonnanoe across tasks within each moclP 'J'hp N>SU!tsofthe ANOVA1' and the aubsequent post-hoc oompariJloru can be summari2ed as follows:

l. The processing group made signiflcant gains in performance on both the scntence-level and the video narrution tasks in the written mode. The control subjects did not.

Page 63: Input Processing and Grammar Instruction

l Research on Processing lnstruction 109

:!l o ¡;; o ~ 2. The processing group made signjficant gains in performance on the sen· t1 ~ ¡¡¡ Cl o ~ "" tence-level task but not on the video narration task in the oral mode. The

~ control group made no significant gains on eíther task in the oral mode. to ,.

t ' t y The preceding resultl; are presented graphically in Figures 4-7 and 4-8. The • ff Pro-••" [ 9 • numerícal resulta of the statístical procedures are provided in Tables 4-3 ::J • • ¡; • • • and 4-4 . ¡¡ ::J • [ " •

f. §' ( l

<p • • ~ • 1.26 ª •

f •

1 ¡ ! , e. • " • t ¡; •

~ "' P<*-t.eet ó ¡ • ,.. g -o- ProceNi:ng Mn~IM ... r ·······<>- Pr'O(t'8Siof video-l'lamll.ion

0.75

j ¡:; o ¡:; ---0--- Conll'Ol 1efllenc&-~I ¡;: ~ o ~ ~ ~

º7 ---cr- ConlrOI v\dco.nlln"aUM !l o .. ····-·····-o

1 t • t y =:; • .........

i 9 "" 9 0.26

I • 1 J ' ::J

-g

1 ¡;>

f. ( ~ i ¡ ~ ~ l

.. a;

g o Tasks in the Written Mode ~ ~ 1

::J

i f ~ t6 • l Potl-\ellt

!i ,... g ... t -o- Proormit1g tentenoe-leveo!

~ e o ¡;; ~ ······<>··- Proa!ll:in¡ vldeo-narratlon o ... g. ---0--- C;.nt.Tel 1111ntenc:o-ie\·ei /,_.(> " 05

1 ---&-- Cont-rol VKloo-nan-aUon

t ' i y f _..,,,./ • 9 r Pfe.C.Nt

~ ¡¡ ,,.,. ... ___

' ~ 6:-- ----------g ~· ' 5·

f ~

1 .., :¡o o

t l ~ ] ~ t t. M

~ l • .. i l O' a ...

... !i ! ~ P<>tt>tc1l 3 i Ta.sks in t he Oral Mode ~

,.. FIGURE4-8. Resulu ofVanPatt.en and Sanz (1995) for both sentence-level and video-narra-tion wks wilh mode removed as a va.ria.ble.

108

Page 64: Input Processing and Grammar Instruction

Table4-3. Resulta of t.he Analyses ofVariance from VanPatten and Sanz (1994) for ali Output Tasks in Both Oral and Written Modes

df SS MS F-\"8.lue

Sente.nce complétion Souroe of variation

Instruction 1 9 .79 9.79 5.03'

Time 1 4 .56 4.56 7.25"

Mode 1 1.93 1.93 6.44" Ti.me x instru.dion 1 7.39 7.39 11.50" Mode x i.nstruction 1 .07 .07 .22 Timex mode 1 .54 .54 1.36 Instructlon x time x modé 1 .32 .32 .82

Quesllon1U18Wer Source of variation

lnstruction 1 1.14 1.14 4.36' Timé 1 .88 .88 3.98' Mode 1 .05 .05 1.34 Time x inst:Nction 1.08 1.08 4.90' Mode )( i.nstnlction 1 .u .11 2.81 TUne x mode 1 .04 .04 1.83 Instrw:tion x time x mode 1 .01 .01 .44

Video nanation Sou.rce of variation

lnstru<tion 1 5.61 5.61 4,83' Time 1 3.86 3.89 11.64" Mode 1 1.71 1.71 4.10'

Tiro& X instruct'°n 1 3.85 3.85 11.60" Mode x instruction 1 .12 .12 .30 Tunex mode 1 .37 .37 1.68 lnstNction x time x rnode .61 .61 2.76

'p < .05; ''p < .01.

110

Research on Processing lnstruction 111

Table 4-4. Results of t.he Analyse6 of Variance f'rom VanPatten and Sanz (1994) for t.he Written and Oral Versions of the Sentence Completion and

Video Narration Task.s

Writ.ten version Sou.rce of variation

lnstr'IJ<tion Time Task Time x instruction Tuk x instruction TimexWk lnstruction x tUne x task

Oral versions Sou.t'<» of variation

Instruction Time Task Time x irustruction Ta.sk X i.nst:Nction Time X laik Inatruction x time x task

'p < .05¡ "p < .01.

Discussion of the Findings

df SS

7.82 3.96 2.17

1 5.99 .01 .34 .09

1 7.29 1 4.46 1 1.94 1 4 .99

.47

.57 1.11

MS F·value

7.82 4.69' 3.95 6.12' 2.17 4.81' 5.99 9.2T'

.01 .01

.34 1.36

.09 .35

7.29 5.03· 4.46 9.78" 1.94 1.6T' 4.99 10.95"

.47 1.85

.57 2.58 1.11 5.02

The results from VanPatten and Sanz suggest that the effects of processing instruction are observable in the less oontrolled and more oommunicative output of learners. Subjects receiving processing instruction made gains on ali tasks in the written mode and on two of the three tasks in the oral mode. Only on the oral video narration task did the analyses fail to yield a signifi­cant difference between pre- and posttest performance. Nonetheless, as Figure 4-8 shows, processing subjects showed signs of sorne improvement on this task supporting the conclusion that processing instruction can have an effect on communicative performance.

To be sure, the results revea! that writt.en communicative performance is easier than oral performance and that sentence-level tasks are easier than story telling tasks. But these findings are not a function of instruction since the processing group oonsistently outperfonned the oontrol group. Instead, these findings can be attributed to the different task demands. In the story telling-video narration tasks, for example, not only did subjects have to access their developing systems for word order and object pronouns, they also had to put together entire sentences using oorrect vocabulary, tense,

J

Page 65: Input Processing and Grammar Instruction

112 VanPatten

and eo on. Recall that in the sentence-level tasks, subj(!cts only bad to pro­duce a short phrase and that vocabulary was proV1ded to them.

So far, then, we have seen evidence that processing instruction is effec· tivc with both syntax (object pronouns and word ordcr) nnd verb mor­phology. We have Biso seen that its effects are observable in more communicative and discoursc·Oriented types of tasks. Wc will examine one more study in which we sec the effects of processing instruction on a diffcrent kind of linguistic ítem.

RESEARCH ON l.EXICAL-ASPECTUAL ITEMS: SER ANO ESTAR IN SPANISH

Overview

Cheng (1995) conducted rcscarch on the effects of proccssing instruction and traditional instruction on the aoquistion of ser and estar, the two major copular verbs in Spanish. In her study, processing instruction is related to Pl(b) and Pl(c) in that the copular \'erbs are items of low communicati\-e value for second language learners and are redundant features of Spanish. She posed the following research questions as she began her study.

1. Do processing instroction and tradJtional instruction have the same effects on the acx¡uisition of ser and estar as measured by the following tasks?

a. interpretation b. sentence completion c. composition

2. lf there are effects, do theec hold over time?

Motivation for the Study

The copular verbs ser and estar comprise a complex system m Spanish. In many contexts, they are mutually exclusive. Fbr example, only ser can be uscd to connect nouns with noun predicates as in Maria es profesora, "Mary is a professor" and only estar is used in progressíve constructions, e.g., María está corrigiendo oomposiciones, "Mary is correcting papcrs." In the case of many a4jectives, ser and estar are both permissable but their selection is dctermined by the aspectual meaning intended by thc speaker. Durative aspect is represented by ser, e.g .. Juan es pobre, "John ia poor" meaning John belongs to the class of poor people. Punctual aspect is represented by estar, e.g., Juan está pebre "John IS poor," meaning that John bas recent1y become poor. In English, we might say he is broke or currenUy without

Research on Processing lnstruction 113

funds. However, with many a4iecti'"eS. only Mr or war can be used but not both. For example, aclJect;ives of nationahty appear only with ser, e.g., Soy argentino, "l'm Argentutian." Many a<lject:ives of cond.Jtion only appear with estar, e.g .. Est<ly bum, "l'm fine." In these and other cases, the underlying aspectual meaning of the a4iective (e.g., nationality tends to be viewed as durative, whereas condition tends to be viewed as punctual) dictatcs the use of ser or estar. ln a ccrtain sen.se, then, ser and e#/ar are redundant markers ofaspect.

In VanPatten (l 985b, 1987a), m-e stages were documented for the acqw· sition ofthese verb. with stages ID-V quite protracted (see also Fienneman, 1990; Guntermann 1992 for supporting research>.

l. Omission of copular verbs Oearners Slmply string toget.her nouns and preclicatcs, e.g., Juan aJJo, "John tall."

Il. Acquisition ond overextension of ser. m. Acquisition of estar with progressive. rv. Acquisition of estar with locatives. V. Acquisition of estar with a<ljectives of condition and state.

VanPatten hypoth4l6ued that ser and estar are low m communicative value for the learner .. In and ofthemselves, i.hey have no inherentsemantic mcan­ing. The choioe of ser or estar is determined by the 111tended meaning of thc ac\jective or sentenai and very often sentence.. containing these verbs are interpretable vía the content lexical items and the context in whicb the scn­tence ~rs. For this reason, VanPatten argued, learners omit the copular verbs m the early stages; this is reflective of the fact that they may not be attending to lhcm at ali in the input. The subeequent acquisition and over­~neral.ization of aer may be owing to frequency in the input; ser is three times more frequent than estar.

Cheng reasoned that even thougb $(!T' and estar are devoid of semant.ic information themselves, processing instruction might be able to heJp leam· ers link the use of each copular 'wb with particular meanings of a<lject:ives and sentences.

TheStudy

Subjects and Groups Drawing on tho aruno student population na VanPnttcn and Cruiicmo (but

once again, not using the same subjc!cts), Chcng randomly assigned six class­es of second·year college.level Spanish to one of three groups: control, tradi­tional, and prooessing. Screening of subjects for background and the use of a~ cutofT score on the pretests yielded the final numbers of sub,Jects: con­tro.I (n = 33), trad.Jtional Cn = 36), process¡ng Cn m 36). As in the other &tud·

Page 66: Input Processing and Grammar Instruction

114 VanPatten

ies, subjects were unaware of the comparative nature of the study and were not scheduled to receive any k:ind of instruction on the copular verbs during the experimental period.

lnstructíonal li'eatment Cheng created two instructional packets for the traditional and process­

ing groups for a two-day instructional period with no homework. The tradi­tional packet was based on the leading second-year textbook Powjes, whereas Cheng developed the processing materials herself. The processing materials reminded students before begjnning the activities ro "pay atten· tion to the presence of ser and est,ar," because these verbs might help pro­vide clues as ro sentence meaning. Focusing exclusively on the use of ser and estar with adjectives and past participles, her referentially oriented activities included, among other kinds, picture matching activities. For example, stu­dents heard either Pablo es débil or Pablo está débil and had to select from one oftwo pictures, one in which Pablo is portrayed as a 00-pound weakling and another in which Pablo is s ick in bed and weak. Since ser seems ro be the default verb, Cheng made sure that estar was as frequent if not more fre. quent in the affectively oriented activities. These activities included those in which subjects checked boxes such as "often, sometimes, never" to indicate the frequency with which they find themselves in certain states and condi· tions, e.g., Estcy deprimi.®, "I'm depressed." Following Lee and Van.Patten (1995, Chapter 5) she created a variecy of structured input activities for the two-day instructional period.

As in previous studies, she carefully balanced the two packets for num· ber of tokens, vocabulary, activicy cypes, and so on, and checked the items of each packet against her testing materials to avoid a teaching-testing bias. The regular instructors were removed from the classrooms and replaced by a third-parcy neutral instruction who was not the researchei:

Assessment Tasks and Scoring Cheng developed various versions of three different assessment tests for

use in a split-block design. The first was an interpretation task that contained both aura! and written stimulus sen ten ces. There were 10 target items in this test, 4 with ser and 6 with est,ar. Cheng used a 1 (correct) versus O (incorrect) scoring procedure for this task for a total possible score of 10.

Cheng's second test was a written sentence-level production task in which students had to complete a sentence using ser or estar. Of the 10 tar· get items, 4 reguired ser and 6 required estar. Again, she used a l versus O scoring procedure based on correct use of copula and ignored any other errors in the scntence. The total possible score on this task was 10.

The third test was a guided composition task. Cheng provided subjects with a series of four drawings that narrated a story. Beside each drawing

Research onProcessing lnstruction 115

"'.en;, key words that subjects were instructed to use while writing their "sto­nes. Key words mcluded targeted adjectives as well as other helpful vocab­uJary to taJk about the people, objects, and events in the drawings. Subjects had to use total of 12 adjectives in writing their compositions, five requiring ser and seven requiring estar.

Cheng used ali three tests as both a pretest and two posttests adminis­tered immediately after instruction as well as three weeks late;, Using a split-block design with various versions of the tests, Cheng avoided an effect for test item familiaricy and test order.

To summarize, Cheng used a research design similar to that ofVanPatten and Cadierno. She compared three groups (control, traditional, and process­ing) in a pretest/posttest format (with only two posttests), but included three rather than two cypes of tasks: interpretation, sentence-level production, and guided composition. This design is summari.zed in Figure 4-9.

lndependent Variables

Control Traditional Prooessing

lnterpretation

Task

Production (Sentence-levelJ

Dependent variables = seores on tasks

FIGURE 4-9. Reoeorch desi¡;n in Cheng (19%).

Time

~ Pretest P06ttest

Compoeition

Page 67: Input Processing and Grammar Instruction

116 VanPatten

Results

An ANOVA conductcd on the pret.est seores revealed no dilTerences between i 1"1·110d

the groups. Separate ANOVAs and post·hoc comparisons on each of the _,¡ il .. l 5 three sets of pre- and posttc&t seores revealed the following.

~ª • 1 1 .2

j 31 ~ l l. On the interpretation task, both the processing and traditional groups 5 ~ i l 2M2·JIOd ., ""

made signilicant gnins but there were signilicant dilTerences in the gains -ª y i • §. .. • .... made; the processing group gwned even more, at least in the first post· 6 :I¡ t Q 5

• li test. On the serond posttest, the prooessing group's sa>re dropped to 8 • about the same level as the traditional group's but was still significantly -5 - J great.er than their pret.est seores. i 2. On the sen!Amce production task, both processing and traditional instJUc. tion improved signíficanUy and retruned thís performance aaoss the two tQ .. ll .. lll ! .. posttests. l

3. For the composition task, the results were idenlical to those of the sen- E tence production task. Both proce.;sing and traditional improved signifi- _,¡ 8

cantly from pre- to posttest.s; their performance stayed about the same '11191·110d ~ ] across the two posttcsts. e g

.2

f ] l 1J s g These results are displaycd grephically in Figure 4-10 and the stalistical ] tl • ! results summarized in Table 4-5. Even though the results do not appear to 1 ..,.,..,d p."" '3

1 be as dnunatic as those in the prcvious studics, Clwng's analyses ofthe data -a y i •

~~ • revealed coruiistent significant dilfcrences as described in the preceding Q 1 8 • numbers l~. •

Cheng pondered the rcsults on the interpretation task since they were _....., §

c f not identical to thosc of previous studies. Although the processing ~ instruction group made significant gains on the interpretation task, so .. • .. did the treditional group, although its gains were not as great as those ! of the proccssing group. Chcng decided to rcanalyze the data. She rea- .s soned that estar sccmed to be the problematic verb in tenns of acquisi- 8 tion. Recall that ser is ncquired early and overextended and that the real i problem that confronte learners 11 the acquisition of estar. If this is the g lltt-WOd ó

_,¡

~ case, perhaps perfomance on the ser items of Cheng's tasks was masking

~ª -~ 1 any effects for estar. Cheng thus conducted subsequent ANOVAs and J J post hoc comparisons on the estar items only and obtained the following ·ª lj d ! ] results. 1 '*1•*"°<1

~.,,

¿¡ e

y f ' ~ E " • J l. On tbe interpretation tasi<, only p~ ínstruction mude signilic:anl e- b ? ~ " ~ •

gains on both posttests. There was no dilTerence in performance between e !'! -

the tniditional and the control groups. -....i .. 2. 0n the sentence production task. the results were the same as for the ~

combined ser and estar data; both proce:;sing and tracütional instruc- ~ : .. .. .. ., • e • • co.

117

Page 68: Input Processing and Grammar Instruction

118 VanPatten

tion improved significantly and retaíned this performance across the two posttests.

3. For the composition task, ali groups seores improved significantly o •

~ ""'lll9d

from pre- to posttests, howevcr, the ga.ins made by the processing and • .;/ . I l traditional group were substantially greater than those made by the

• :a • g • ... ,., control group. There was no dJfference between the traditional and • i • c"2 e

• t3 processing groups. Q, ·- .2 o -a i;;; y ¿ • i • • •

• o lt seems, then tha1 by analyzmg only the es/ar data, Cheng's analyses yield- • • e .. i • 8 •

ed results similar to thoee obt.a.ined in the previous studJes. The results of • • • • 5

the reanaJyzed data with e:;tar only 8"' pn!Sented in Figure 4-11. Table 4-6 ....... ¡ contains a 6UJllm8lY of the stat.i8tical results. i .. .. .. .. § .. o

§. .;/ § ;¡ ...

z .-i·llllld

'"' ¡

e • ·f o

í 'il e • ] " ,., 1 l 1§ .. .. 1..,-.0d ll.. ...

y ¿ • 1 -.; ;¡ • Teble4-5. Resulta of the Ana.\y8'l8 of Varianoe from Cheng ( 1995) for

~ tl o ti ii l •

1 Combined ser and eattlr Date • e -- ~

df SS MS F·value e

1 c8 Interpretation taak .. • ;:¡ .. ;:¡ .. "' Source or vanat~ ..

lnstruction 2 30.6<1 1532 8.10 ..

Time 2 136.28 68.14 37.93"º ~ Time )( ínstruction 4 13.61 3 .40 1.89 ! Sentence procluction taak

p z 111t1·lll0d .. _,,,- l Source of variauon ' ' :a 1 1 ' lniUuctioo 2 6.l41 31.71 5.88º" ' ' ... b l 1 ó . Timo 2 26006 13002 47,03- .. § e: e • .g Time .. i.nstructaon 4 3032 758 2.74" d ::l o r .. .. Jlllt2·~ !s ¿ !I ..

y • ... • :! ~""" ...... .. e-~ o ........

Source of vanauon .. ., • .. - • lnslru<üon 2 6.15 307 6.11"· .. e ...¡ .. - -Timo 3 1693 8.47 26.BG"º - ... 1Uae ll UlllrUcUOn 4 341 .85 2.70'

~ • p < .05, - p < .01 .. .. .. ... CI . .. .. -...

119

Page 69: Input Processing and Grammar Instruction

120 VanPatten

Table~- Rmlalta o( the Analyses of \liriance &cm Cbeng (1996> for tttor Onzy Data dl SS MS F·value

ln~tmlt Soww or vonabon

l1>llr\KtlOll 2 14 84 742 4 .04"

Tuno 2 87.20 43 60 375¡-

'fune M U\ltrUdJOn 4 548 137 119

Sentenoe producüon Wk Source oí variataon

!Nt.Ncbon 2 20.28 10.14 2.17

Ti.me 2 121.96 60.98 31.04ºº

TUne x ÚlltructK>n 4 19 46 4.87 2.48"

Compo11uon taak Sowu or vanatíon

tn.uucbDn 2 12.19 610 3.88" Tuno 3 5872 29.36 51.54ºº Tame • inltNction 4 718 1.79 3.15·

p < 06, .. p < 01

Discusslon of the Findings

Cbeng's resulta provide further evidence for lhe benefits of processing instruction. Using the data on estar only, we see that prooessing instruction resulta in incrcascd gains on an interpretation task, whereas tradfüonal instruction does not. We see tbat botb processíng and traditional instruction result in increased gains on both a sentence-lcvcl writtcn production task anda guided composition task. Once again, the effects of pr«essing instruc­tion are obe<!rvable on output tasks but tbe effects of traditional instruction are not obe<!rvable on interpretation tasks.

In spitc of the consistent finding¡¡ regarchng processmg versus tradition­al or no instruction, there remains one question that Jl06CS potential prob­lems for thc previous researcb. Prooessing m.'<truction, like traditional mstruction, consísts of explanation plus sorne kínd of practioe, albeit input as opposed to output manipulation. One could conclude that, g¡ven the kinds oí tasks uaed in the previous studies, we cannot determine whether or not proccssmg mstruction is resulting in a change in the developing system or whether •u"1ACtA ..,.., simply monitoring on tbe ¡l06ttest6. That is. could the superior performance of tbe processing groups be due to conscious knowl­edge gaincd during the explanation part of lhe instructional t:reatment and not to any effect lhat the activities have in helping to restructure the devel­oping S)'lllem? Wc turn our attention to one final study that sets out to rescarch this question.

Research on Processing lnstruction 121

RESEARCH ON MONITORING: EXPLANATION VERSUS STRUCT~ED INPUT VERSUS BOTH

Overview

VanPatten and Oikkenon (1996), conducted a study to find out whether explanation or structured mput activtties were the causative variables Cor the previously reported research findings. Using the same materials and assessment tasks as uscd in VanPatten and Cadiemo (1993}, VanPatten and Oikkenon divided subjects mto thrce t.reatment groups. The first receivecl pr«essing instruction exactly as in VanPatten and Cadiemo's study. The scoond receivecl explanation only with no subsequent activities. Tbe third receivecl structured input activities on!y with no explanation. Theír research question was basically tbe following.

1. In terms of processing instruction, which of the following appears to be tbe mo5t significant vanable in acxounting for pretest/po6ttest gains.

a. explanation; b . structured input activibes; or c. explanation and structured input acbvities combined?

Motivation for the Study

Ali explicit instruction contains expllcit information about bow tbe sccond language works. Be it jargon-ladencd or not, this expllcit information tells lcamers about how the languagc works and, very often, about what ool to do in order to avoid errors. Processing instruct.ion is not different in this ~ ~ from Chapter 3 that processing instruction consists of expllc-1t informatíon plus structured input activities and tbat the explicit infonna­tion is oftwo types: ( l} bow the language works; (2) what natural processing strategies may be getting in tbe way of acquiring a particular form or struc­ture. As Krashen (1982) has argued. this explicit irúonnation can be stored on!y as concious knowleclge in the form oí a Monitor, a device that edita out.­pul when task demands allow for monitoring to happen .. For the Morutor to be engaged. Krashen has claimecl, the single most unportant feature oftask ~emands is time (Krashen, 1982, p. 89). He sta~ that it takes real process­mg time to remember and apply coll!!CÍOus knowledge or rules wben making uutpuL Fbr thls rew;on, it. U. wllkull. fur k ... rncllS to engage the Mon.itm du1-ing ongoing conversation and luned Wlitten tasks, whereas it is easy (but not guaranteecl) for learners to engngc the Monitor witb simple writtcn tasks (e.g., fi!ling in a blank) and untimed Wliting activities.

According to Krashen's viow, then, we must look at the interpretation

Page 70: Input Processing and Grammar Instruction

122 VanPatten

and sentence-level production tests u.sed in proc:ewng research as possible tasks in which leamers might easily Morutor Is the expliot and therefore conscious knowledge possessed by the leamers aftcr rcceiving explanation responsible for the gains observed in the research reported in previous sec­tions of tlus chapter? Since tbe task.s do not resemble "JX)ntaneous conver­sation in any way, many are wntten sentence-level tasks, it would seem reasonable the ask to what extent our results are due to Monitoring. What thia would mean in tenns of research is that if Monitoring were responsible for tho outcomes of previous research, then providing learners witb explicit information only should reveal similar pattems of gains from pre- to post­tests as in VanPatten and Cadiemo (1993). Howcver, if the use of explicit information in and of it.self is not responsible for observed gains, then two possibilities exist. The first is that the structured input activities tbemselves are responsible. As bonafule comprehensible 1Dput, if structured input alone is responsible for the results, tben we have furthcr support that acquisition is mdecd happening--Or beginning to happen-under processing instruc· tion. The other possibility is that the expliot ínformation and the structurecl input activítíes uigetber are responsible for the gains, ID which case we would have to argue for a weak interface posítion similar to that taken by Ellis (1994). He argues tbat conscious knowledge is used in a secondary manner by leamers to "notice" new forms and structures. The conscious knowledge enables leamers to engage bottom-up processing that he claims IS necca&ary for acquisition. <We, too, belicve that bottom-up processing is nooessary; that is what is meant by attention to mcaning and form during input processing. However, as we stated in Chapter 2, consciousness need not be linked to this bottom-up proc:ess.ing.)

To summarize, if conscious knowledge and Monitoring are responsible for the observed gains after proc:essing instruction, thcn we should see these gains in two groups: a group that rcceives explanat.ion only and one that rcceives regular processing instruction (explanation plus structurec! input activítics). Wc should see no gains in a group that rcceives structured input only 1( however, oonscious knowledge and Monitoring are not responsible for the ob6cr11ed gains, then we should see no ¡¡runa ID an explanation1mly group but inst.ead should see gains in a group re<:eivíng only structurecl mput as well as one receiving regular processing instruction.

THESTUDV

Subjects and Groups

Subjects from a local Champaign secondary school participated in the study. Ali were in their second year of Spanish courses in a somcwhat tra-

Research on Processing lnstruct1on 123

d1tional program in which mastery of granunar is emphasized and only minor importance is plaoed on skill developmenL As in VanPatten and Cadiemo (1993), a background questionnaire and pretests were adminis­tered to arrive at tbe final data pool. Ali subjects had to be native speak­ers of English with no outside contact with Spanish. Ali had to score less than 70% on the pre-tests to be included in the subject pool. The final total of subjects was 59, dividcd into the following groups: Explicit lnstruction Only (n = 22); Structured Input Activities Only (n = 20); Proccssing lnstruction (n = 17).

lnstructíonal Treatment

Processing lnstruction consisted of the materials u.sed in VanPatten and Cadiemo (1993) as well as VanPatten and Sanz (1995). No alterations were macle in tbe content of the material.s and no homework was allowed (ali packets were picked up at the end of each class day). Instruction was spread over three days-as opposed to only two-gíven the shorter class periods that generally mark secondary schools in the United States.

Explanation Only consisted of the explanatory sections of the materiaJs used in VanPatten and Cadiemo with no activities. No homework was allowed and ~ packets were picked up at the end of each da.it Subjecta read tbe explanation as the instructor reviewed it and provided examples. Subjects were encouraged to ask for claritication or repetition if nece&&ary. As in the Proc:ess.ing lnstruction treatment, the Explanation Only group received their explanation ovcr thc course of three days, each day's expla­nation and information corresponding to the explanation given in the Processing group on that day. Sincc no activities were used, the rest of class time was taken up with games nnd activíties unrelated to object pronouns nnd word order.

Structured Input Only consisted of the activities containcd in the materials used in VanPatten and Cadicmo with no explanation preceding them. The instructor merely bcgnn the activities and said whether or not students had made appropriate selcctions or not (during the referential activities). In a picture identification activíty, for example, the instructor would read a sentence and students would select picture a or b to match to what they had understood. lf students 1dentified the correct picture, the mstructor said "Good. Let's continuo with the next one." If students selected the incorrect picture. the instructor would merely say "No. The picture that goes with what you just hcard is . "At no time did the instructor inform students about rules or ohject pronouns in Spanish and if students asked explicitly about what they wcre to leam the instructor said, "Let's just see if at the cnd of thc week if you've gotten it." Moreover, the instructor did not call studcnts' attention to the object pro-

Page 71: Input Processing and Grammar Instruction

124 VanPatten

nouns or word ordcr and did not single them out in any way. As for the other two groups, the lreatment !asted for three days and the activities oonducted each day corresponded to thoee that were used in Prooessing Instruction on each day. No homework was given and ali packets were picked up at the end of each day.

The instructor was the same for ali three groups. She was an inexperi­enoed teecher-in-trauung who had not worked with processing instruction or struc:tured input previously

Assessment Tasks and Sco ring

The assessment tasks oonsistA:d of the same Ulterpretation and sentence­level production tasks as used m VanPatten and Cadierno (1993), but only one po&ttest was used. This po&ttest was administered the day after the instructional treatments were oompleted. Two versions of each task were used in a split-blocl< design (sorne leamers got version A as the pre-test and B as the poet-test and sorne got version B as the pre-test and A as the post. test. The scoring proccdure was the same as that in YanPatten and Cadiemo's study.

Results

An ANOVA oonducted on thc subjccts prctest results revealed no significant differences betwccn thc groups on either task. Once again, any observed effects are attributed to treatment.

The ANOVA oonductcd on thc interpretation task revealed a significant main effect for instruction (Explicit vs. Structured Input Activities vs. Prooessing), time (pre- va. postrest) and a significant interaction between instruction and time. A comparison of means revealed that the observed effects were caused by both proccssing and structured input only making significant gains from pre- to posttests on the interpretation task. The explicit only group made no gaina.

On the production task, the ANOVA yielded a significant main effect for instruction (Explicit vs. Structured Input Activities vs. Prooessing) and time (pre- vs. posttest) but no interaction .. Ali groups gained from pre- to posttest. However, the gains made by the structured input only group and the pro­cessing group were greater than tho6e made by the explicit only group. The precedlng result.s are summartzed in Fi¡,>ure 4-12 and TuW., 4-7.

6

• e 3 !! !:!

i 3

.5 e o

2

! ~

o

1

6

j

J Time

Time

-0- EltpJx7t Only

-<>- SUuciured Input Only

---<>-- ReguJu "'-"'

-0- Exptidt Only

··-.. <>--.. StTucturecl lnpul Only

---<>-- fteaul ... Proc:eoainc

PICURE +12. R.ul,. oC VanPatc... and ~ t19961 oo both toterprelallon and ...,. tenc.levd productlon taob

125

Page 72: Input Processing and Grammar Instruction

126 VanPatten

Tuble 4-7 ANOVA Summarie6 far Both lntapretation and Producuon Tasks 111

\Ml'ai:ten and Oililienon U9961

Tuno Ti.me )( WtrucllOll

Prodw:tion taak Souroo of •N"IBlion

lnstNCtion Timt Tune ,, 1nttructt0n

Discussion of the Fmdings

d! SS MS F-•'Blue

2 1 2

2 1 2

6:!896 159 947 88.785

48.M6 62.348 9.066

31948 159 947 44392

24 278 62348 4 533

9056" 50.322"" 13.96T"

3.220" 12.881" 0.937

The resulta from VanPatten and Oikkenon suggest lhal Monit.onng is not the causabvc or signficant variable for the resull8 observed in processing instruction F11st, the explicit only group did nol make s.gnilicant gains from pre- t.o poetletits oo the interpretation task. Jusl a.s unportant to note, how­ever, is that lhe structured input only group performcd as well as the regu­lar pl'OCC88ing group and made significant gains from pre- to posttests. Although ali groups improved on the production task, the gains made by the structured input only and the processing group werc greater than those of the explicit only group. These findings suggest that it is the actual struc­turcd input itself and the fonn-meaning connections being made during input pl'OCC88ing that are responsíble for the observcd cfTecta in the preseot as well a.s previous studies.

We must temper this conclusion, however, by noting lhal in the struc­lured input only group, learners were provided with ncgative feedback dur­ing reíerential actívities (i.e., "No. The correct picture is b. Let's try the next one. ") It is po551ble that leamers fonned conscious knowledge on their own as a resull of theo;e types of activities that interacted W1th subsequeot refer­ential a.s well a.s affectively oriented acbV1ties. If this is lhe case, then we could conclude that explicit knowledge does aid input processing and that is why the structured only and the regular pl"OCC$Sin¡¡ groups performed in sinúlar fashion across the experiment. What we can rule out, it seems, is the role of the Monitor on its own. That is, it is unlikely that Monit.oring is a sig­nificant factor in the research reported on in this chapter. If it were, the explicit only ¡,'l'Oup should have made more observable gains in the VanPatten and Oikkenon study.

ResearchonProcessing lnstruction 127

CONCLUSION

ln the fivc studies that we have reviewed in tlus chapter, the effects of pro­c:e.;síng instruction are consistently obeervable. Not only do learners receiv­ing processing instruction gain in the ability to procesa input better, but al.so thetr devcloping SYStem is afTected such that they can aCttSS the tar­geted linguistic features when rnaking output. This is the case with a vari­ety of linguistic items, and in lhe VanPatten and Sam study we saw that the efTects of processing instruction extcnd t.o a variety of output tasks. In addition, the findings of VanPatten and Oikkenon suggest that it is thc engagement in structured input activities within processing instruction that is the most significant variable; cxplicit information (explanation) does not appear to be critical.

On the other hand, traditional instruc:bon appears to have effects only on output t.asks. Subject.s who rcccived traditional instructi.on consistentJy gaincd in their ability to pnxluoe the targcted linguistic features, but they did nol gain in their ability to interprel the targcted linguistic ítem. This is a curi­ous finding t.hat we will explore along with other issues in the next chapter.

In spite of the fairly robust picture thal is emerging on the effects of pro­cessing instruction, one nuuor question has not '-n addressed: Are the cffects of prooessing instruction long-lasting? In two studies, VanPatten and Cadiemo (1993) and Cadiemo (1995), studied the efTecta for proe:essiog instruction held ovcr a one-month period. However, the researcbers did not conduct any testing ata later time. ln Cheng (1995), the effects for proces&­ing instruction held over a period of thrce wceks, with no furtber testing conducted. Other researcb, for cxample, White (1991) showed that the effects of traditional instruction may wcar off within one year, whereas in olher studies instruction has provcd t.o be more durable, for example, White et al. (1991) It would seem, then, that futurc researcb on prooessing instruc­tion should address this important issue.

Page 73: Input Processing and Grammar Instruction

!5 CHALLENGES ANO IMPLICATIONS

INTRODUCTION

Both input processing and processing instruction raise a number of issues for ru:qui.sition and instruction unaddressed in previous chapters. Tbese issues include possible challenges to processing instruction as well as impli­cations of input proce68Íng and prooessing instruction for theory building in second languagc ru:qui.sition reseorch. The present chapter presenta a di&­cussion of these issues.

The first issue concems input processing and problems in the ru:qwsi­tion of syntax as described by researchers using Universal Grammar (UG) as a framework. UG has gained increasing visibility in both first language and second language rescarch since the mid· 1980s (see, e.g., Eubank, 1991; Towell & Hawkins, 1994; Whlt.e, 1989). Perhaps of ali the linguistically ori· ented theorics that contain an innatist comp<>nent, UG has become the moet widely discussed and dcbat.ed (aee Eubank & Gregg, 1995; and the responsee by Scbumann, Jacobs, & Pulvermüller). Even in general reviews of second languagc acqwsition intended as textbooks (Ellis, 1994; Gass & Selinker, 1994) di8CUssions of UG bave come to be standard. Because ofita relative promincnce in the disoourse on second 1anguage acquisition, we have opt.ed to revicw the relationship of UG (to the exclusion of otber frameworks) to input processing in this chapter. Since tbeories often com· pete for explanatory p0wer, 1t ia not unreasonable to ask what input pro­~ 1...,, t.u ulTcr ... 'U.luJ lllugu..¡;e a.x¡u.isition tbeory that UG cannot aa:ount for. The argument will be that both mput processing and UG are compkrMnlary comp0ncnts of a la.rger more general model of second lan· guage acqui.sltion and use.

The second íssue is the role of first 1anguage transfer. Because input pro-

129

Page 74: Input Processing and Grammar Instruction

130 VanPatten

cessing does not incorporate (as of yct) a role for first language transfer but transfer is a recognized phenomenon in second language acquisition, it would aeem that input proccssing offers an incomplete account of second language acquisition. As in thc case of UG, we will show that both input pro­cessing and first language transfcr form part of a larger more general model of second language acqWSltion.

The third issue involves the tcachability of "meaningless" linguistic fea­tures. Sinoe input processing is largcly concemed witb form-meaning map­pings, how can prooessing instruction work effectivcly on those elements of language which Bn! devoid of meaníng? The argument will be that prooess­ing instruction can affect the eoqWSllJon of these features and that this is empirically t.estable.

The chapter will concludc with an examination of tbe differential out.­comes of traditional and prooessing instructional treatments reported in Cbapter 4. The issue here IS why Proce&illlg subJects unprove in both inter­pretation and production but the traditional subj«ts do not. Tbe resolution of this last issuc will come from the general mO<lel of language acqu.isition that forms the basis of the prescnt book.

U'.llVERSAL GRANMAR, SVNTAX, ~ 11\FUT PROCESSING

Research within the framework of UG and current government-bind.ing theory as applied to second language acquisition is concerned with abstract propertiea of syntax. Although both P3, the first noun strategy, and the researeh ofVanPatten and Cadicrno (1993) concern syntax since they deal with the processing of word order by second language Jearners, the syntax of the first noun strategy is not thc same kind of syntax .as that which conccrns specialista in UG and government-binding theory. Those working within UG are (principally) concerned with such matters as (1) how abstract principies conslrain the acquisition of grammar; and (2) paramcter setting versus (3) parameter resctting (i.e., first language transfer ofparameters and the subsequcnt rescttingofthc parameter to the second language value). Most researeh on UG and second Ianguage acquisition has concentrated on the lattcr two issues and the first ques­tion that arises is what input processing has to do with UG and para­meter resetting. A follow-up question is the extent to which a model of input processing instructaon can account for observed phenomena in sec­ond language acquisilion that UG cannot. Each of these questions is taken in tum Since the role of UG ii; taken up in considerable detall in a number of other publicat1ons (e.g., Eubank. 1991; Towell & Hawkins, 1994, White, 1989) the discussion bere will be brief for illustrative pur­poses only

1 Challenges and lmplications 131

Universal Granvnar and Language Acquisition

In order to understand the potenlial relationship between input processing and UG, understanding how UG is conceptualized is necessary. What is nec­essary a1so is to examine sorne of the issues addressed by UG-related research in second langua&e acquisition.

As currently fonnulated, UG is an innate knowledge SOUI"Ce. It consists of a set of abstract principies about language in addition to information on the ¡>06Sible syntactic vanations (parameters) that languages can follow. Because this knowledge is lM8te and uruversal in ali human beings, UG constrains first language acqwsition by allowing children to entertain only certam ¡>06Sibilities about the syntax of languages. These constraints are called principies. A classic example of a universal principie is X-bar Theory; which says that an Xº (also called a "bead") must project onto an X' (the head plus its complemcnt) and that the X' must in tum project onto an XP Cthe X' plus 1ts specilier). As an example, a N (noun) such as dog must pro­ject. onto an N' (the noun plus its complement) as in dog on a leash. Tbis in tum must pr<>Ject onto an NP (the N' plus a specifier) such as IM dog on a leash. (Note: "must project." IS a bit mislead.ing. As we ali know, a noun phrase can be as simple as the crog (See the dog?) In UG theory, the projec­tions within syntruc are thcre, they just might not have words that fil] them.) Since word order in languages is not a mere serial ordering of elements in that words group togethcr to make phrases, this principie provides for a hierarchical interna! structure of scntences. Thus children "know" that the languagc thcy are learning has phrase structure and ali tbey need to do is acquire lcxical itema and categorize them as nouns, verbs, prepositions and so on so that they can project them into the correct phrase structure (see, e.g., thc discussion in Radford, 1990).

Likewise, parameters within UG "inform" children about the possibili­ties for a given syntactic principie. Linguists bave cliscussed a number of parameters: thc head parameter, the null-subject paramet.er, and the verb raising parameter, to neme a few. These parameters pennit certain options and, since the parameters are part of UG, children possess tbe knowledge that the language thcy are learning will bave one of tbe options set forth by the parameter. In the case of verb raising, for example, children "know" that the language will eithcr allow verbs to raise or not. In oertain analyses (Pollock, 1989), a clause (sentence) contains an abstract representation such as that in Figure 5-1. Lexica1 verbs such as eot, dnnk, hit, and so or>-as opposed to awabancs and modals sucb as /uwe, do, and can- are generat­ed undcr the V' node, spectfica)Jy under VO. Agreement features for person and numbcr 8n! located under AV whereas tense features are located under T'. Accordíng to A>Uock, languages either allow lexical verbs to raise and land m AV to get a¡reement features while piclcing up tense features

Page 75: Input Processing and Grammar Instruction

132 VanPatten

along the way (as the arrows indicate in Figure 5-1), or they do not. French is one language that does. Movement ofthe verb rootboi- into T' and Agi-0 is indicated by the arrows. English, on the other hand, does not allow lexi­cal verbs to raise; hence the asterisk in front of the verb root dri.nk-. This one syntactic parameter accounts for a number of phenomena that differ between the two languages: placement of negators (e.g., "John drinks not coffee with milkJJean ne boit pas de café au lait.), placement of frequency and manner adverbs (' John drinks oft.en coffee with milkJJean boit souvent du cafee au lait.), requirement of do-support in English (John does not drink coffee with milk.r Jean ne fait pa.s boire du. café au lait.), among others. According to the theory of parameters within UG then, children's interna!

· processors should notice from the input data whether their language allows verb raising or not and they wiJl set the parameter accordingly. For example, children learning Sparush get consistent data about verb movement with simple yes-no questions that contain subject-verb inversions, e.g., Volvió tu

CP

/"--.. Spec C'

1 /"--.. ±wh e· Á

Sr A John/ Agr' TP

Jeon 1 /'-._ 3rd Spec T sing

1 /"--.. notJ T' yp

pas 1 /"--.. -past Spec

1 often/

souvetú

V'

/"--.. v· NP

I~ •drink·/ coffee with milk..

boi· du/de café au lait.

FIGURE l>-1. The strucwre of da'"""' ahowing dlJT....- between Engl.ish and Frencb with verb-rai.sing.

Challenges and lmplications '133

UG INPUT--+ ~

Principies Parameters

FIGURE 5-2.. The role ofUG in first ~ acquisition.

THECHILD'S ---+ DEVELOPING Ll

SYSTEM

hermane de la escuela? Did your brother get back from school? (Literally: Retumed your brother from the school?) Areording to generative approach­es to syntax, this inversion is dueto the verb moving and taking a spot high· er than the subject in the syntactic tree. On the other hand, children learning English never get data in whlch lexical verbs occupy a spot before subjects, e.g., Did your brother gel home?rDid gel home your brollu!r?rGot home your brother?

UG, then, is a knowledge source that restricts the range of possibilities for syntactic configuration in a language. Principies dictate particular aspects of syntax, whereas parameters allow for a narrow range of options for a given syntactic rule. These principies and parameters lie in wait for the relevant linguistic data that wiJl trigger their instantiation. In Figure 5-2 we summarize in simple fonn the manner in whlch UG works in first language acquisition (based on Wbite, 1989).

The exact role ofUG in adult second language acquisition is a matter of debate. Three possibilities exist. The first, that UG is no Jonger active in sec­ond language acquisition and that learners use more general (but still abstract) oognitive problem-solving strategies to develop a functioning sec­ond language aystem. The second possibility is the opposite; that second lan­guage learners have access to UG just like children acquiring their first language. The thlrd possibility is that second language Jearners have access to UG but that the parameters instantiated for the first language they pos­sess can interfere with this access. If we assume for the moment that seamd language learners have access to UG in sorne w¡zy (with or without first Jan· guage mediation), then just like first language learners they must possess "internally derived hypotheses" (Towell & Hawkins, 1994, p. 24 7) that interact with the input data. Thus, learners entertain particular hypotheses about the second language that are subsequently supported or rejected by

"L the input data. An English-speaker learning French as a second language, for example, núght come to the task of acquisition not expecting verbs to raise. The value of the parameter at the initial stage would be [-raising). That leamer will (re)set the verb-raising parameter to [ +raising] once the evidence from subject-verb inversion, adverb placement, placement of nega­tion, and so on, is encountered.

What the preceding accounts of both first and second language acquisi·

Page 76: Input Processing and Grammar Instruction

134 VanPatten

tion have in common is that UG interacts with input data ln Figure 5-2, for example, there are no mtervening prooesses that medial.e bc~-een UG and mput. As Wlute stated,

The runction or the input data in language acqUJ.>1tNln 18 to help to fue one of the po11>1ble lettmgB. Tlu.s IS called triggwing. In other wordo, the mput helps to make the choice between various &ettings <Wlute, 1969, p. 29).

Thwi, the instantiation of principies and parnmeters is dependent upon input. Therc is a problem in this conceptualization. lt assumes tbat data in the input are readily available for UG's use. It furthcr assumes that the¿ interna! etructurc of cvezy utteranoe is proccssed and is readily available to'. UG. The problcm is that these assumptions cannot be supported. Accorciini to thc model of input proeessing developed in Chapter 2, lcarners' process­ing of input results in a reduoed and sometirnes altercd subset of the input data. Thesc data, called intake data, are subjec:t to further proeessing (accommodation) that, when it occurs, can lead to l'Clrtructunng of the devel­oping syst.em. ln t.h.is view, the developing system uses mtake data and not input data for growth. IfUG fonns partofthe intemal mechanisms used in restructunn11 of the grammar, it follows that UG must make use of intake data; that is, UG dou not opera/e on inpuJ data for the constrw:twn of the developmg system; 111Steod, it operaJes on mtake dala.

In order to make our case, we must fust recall thc general model of sec­ond language ncquisition and use outlined enrlicr in this book. In this model, second language acquisition and use consists of distinct sets of processes as shown in Figure 5-3. The first set ofproccsscs filters the input data in particulnr ways resulting in intakc data. These data are then acrommodated or not by the developing system with subsequent restruc­turing of the system depending on the nature of thc intake data. If we combine Figure 2-3 from Chapter 2 with Figure 5-3, we can expand the general model. The result is Figure 5-4. At this point the only role for UG is in the identification and assignment of head lexical categories since every learner must be able to assign a word a category such as noun and verb during initial processing. If not, acquisition simply couldn't bappen.

I ll input __ _,.,. intake ---t.,• developing sy•tem

I = input processing 11 = accommodation, restructuring

FIGURE G-3. Set.t oíproc1S8eo m &e<Ond langu~ ocquilillon

1

1 1 l •

l ~

Challenges and lmplications 135

------------------··------·-----------·---------·-¡

; 1

"-··---·-----···-····-·-·-----------------; '--~~~~~~~~~~~~~~~~~--!~

FIGURE M. An expancled m<>del o( IK'<O<lCI ~ O<QUlllUOn and we llhowini the role of input p...-.ng.

(In Figure 5-4, and subsequent figures, XYZ represent content lexical items, whereas a, b, and e represcnt functors, morphology, and other sur­face grammatical features.)

It is now possible to examine a rather straightforwnrd case on the inter­action of UG with intake. Reca11 that the model of input processing devel­oped m Chapter 2 entails a push for meaning and that based on the limited capacity for processing input data learners will process lexical items in the mput bcfore anything else. The model also predicts that Iearners will process grsmmatical markers m accordance with the role they play in con­tributing to sentence meaning during comprehension. In Figure 5-4, if the processing resources are depleted in the first go round, that is, the process­mg of lexical items uses up alJ availnble resources, then a1I that is delivered M intake for further processing llJ"(' conwnt l<!Xical itema with sorne kind of serial ordering. This means that thc developing system would not receive sufficient, if any, data about most prepositions, auxiliary verbs, copular vcrbs, verb inflections, agreement markers, and so on. If UG is interacting almost exclusively with lexical iterns in the beginning, then the prediction would be that the leiácal categorics of syntax would develop before the func-

Page 77: Input Processing and Grammar Instruction

136 VanPatten

tional ones. Ail Z.Obl (in press) points out, tbis is precisely what happen.s. He states: ·

Lexical categories in t.hematic positions are in evidence long before functional categories appear in functional positions. In earlier stage& of L2 acquisition, t.he latter are either abaent, treated like lexical categories or are unanalyzed parts oí chunk-learned iterns. Moreover, t.he acquisition ofthe Iexical-themat· ic oomponent of L2 grammars seems virtuaJly assured while the same cannot be said of t.he functional oomponent (p. 24).

If UG interacted direcUy witb input, tben we would expect tbe functional elements of gyntax to appear much earlier than they do. Because they do not, our condusion is that the relevant data for UG are the reduced intake data and not the fuJJ input data.

There are more complicated scenarios tbat al.so suggest an interaction of UG with intake ll!ld not in_EI!_!. In Towell and Hawkins (1994, pp. 132-138) tbe authors reviewed UG and competing cognitive aoo>unts of second lan­guage acquisition to demonstrate the superiority of UG in accounting for certain developmental stages. Towell and Hawkins reviewed the following cases: (1) acquisition of obligatory subjects in English by speakers of null­subject languages; (2) acquisition of preposition stnmding by learners of English; (3) acquisition of object pronouns in French by speakers of English. In each case, although UG accounts for some very important acquisitional phenomena, there are other phenomena left. unexplained.

In the case of the acquisition of obligatory subjects, Tuwell and Hawkins cited research in which obligatory subject pronouns are acquired in the fol­lowing order:

referential --•.. quasi-argument ---1 .. • expletive.

This means that referential pronouns such as they and she are acquired before quasi-argument pronouns such as it, as in it is raining, wh.ich in turn are acquired before the expetive pronoun it of it is impossible to tell. Although one could argue that different pronoun types form part of UG (i.e., this knowledge is built into UG), Towell and Hawkins pointed out that UG cannot account for the order in which the pronoun.s are acquired. Although frequency in the input may be a factor, as noted by Towell and Hawkins, we could also claim that the pronouns are processed differen­ti..Uy in tbe input. Subject pronouna, auch as they, si.e, and others, are J.,,.. ical items of high oommunicative value since they encode referential meaning and since verbs in English do not encode for person-number. Learners must use them in processing input to account for subject sh.ifts in utterances they hear or even to establish a referent at the onset of oon­nected d.iscourse. For example, compare the foUowing sets of utterances

Challenges and lmplications 137

with and without referential pronouns and it becomes clear how learners would rely on them early on to establish reference.

W'llhoul re{trential prorwuns: Called Pedro ycsterday. Talked for at mast an hour. Said were oorning to dinner. W'úh referentWl. prrmoUlt8: 1 called Pedro ycsterday. We taJked for at least an hour. He said you were oorning to dinner.

Quasj-arguments and expletives are devoid of communicative value since they do not encode any referential meaning. Ail argued earlier, the principies in Chapter 2 would predict tbat referential pronouns would be detected and made available as intake before the other pronouns. This is supported by the fact that it as a referential pronoun in the example it is on the tabk is acquired before the other types of pronouns. When it functions as a refer­ential pronoun, it has clear communicative value and th.e form-meaning mappings produced by input processing are clear and unambiguous. In terms of the ordering of quasi-argument and expletive pronouns, frequency may indeed be a factor here because there is no theoretical reason why these two pronouns should not be acquired at the same time. (Towell and Hawkina do not provide information on tbe relative lag time of the acquisi­tion of these pronouns so that quasi-arguments and expletivas oould be acquired close together.)

Fbr preposition stranding, Towell and Hawkins cited Bardovi-Harlig's (1987) study. She found that learners first acquire structures without WH­movement such as She talked to whom? This is followed by a stage in which learners front a WH-element but no preposition is present, for example, Who did she tal.k? Towell and Hawkins oould account for this by suggest­ing that there is a problem with licen.sing, that learners have not learned that po (a preposition) can govem an empty category in English. Accordingly, learners delete the preposition to avoid violating tbe theory of licensing and empty categories. By deleting the preposition, the verb may govem the empty category left. behind by tbe moved WH-element. Although this fits within the theory of UG, Towell and Hawkins noted that it is problematic in tbat the process of deletion is not entirely understood. They then go on to suggest tbat perhaps learners have not correctly sub­categorized verbs, such as talk, asan indirectional transitive verb requiring a prepositional phrase. The question is, why wouldn't learners have sub­categorized this verb correctly? The answer may lie in tbat tbe prepositions do not form part of the lntake data in the early stages at all and tbus dele­tion is a misconceptualization of the facts. The relatively low communica­tive value of the preposition to may result in that the first round of processing of content leicical items in the input ignores it. Thus, learners may be providing intake data to UG that resembles l talk Jerry and We talk

1

Page 78: Input Processing and Grammar Instruction

138 VanPatten

him the phone last níght. Note that talk can also appear in the input with­out prepositional phrases: She talks a lot; He talks too ftult. In short, the developing system may be getting data in which ro is absent. In this way, we have an explanation for why learners m.ight have problems in initially determining the subcategorization of verbs like to. The question arises, then, as to why leamers' first stage includes the preposition, for example, She taJJwi, to whom? There are two possibilities here. The first is that the phrase to whom does not consist of two separate words but is actually one unit towhom. In spoken English, to whom would form a natural breath group and if any natural junctures are inserted in to an utterance, speakers would separate to whom from talk and not whom from talk to. In this way, towhom is initially encoded as an NP and not a PP during input process­ing. UG is thus given a consistent picture that the verb ta1k does not require a prepositional phrase. The only way we can con.firm that towhom forros a unit for the early stage learner is to show that whom is never used by learners without to in unmonitored speech during this stage of develop­ment. In any event, the developmental stages can be explained if we assume that UG is operating on intake data and not input data.

A second explanation lies in that the first stage of production is largely monitored and is notan accurate refiection of underlying competence. That is, when learners produce She talkd to whom? in the early stages, this very well could represent an example of monitored output. This receives some support if one considers that the naturally occuring input to learners is not likely to contain very many samples of ro whom. More likely is that learners would hear utterances such as She tallred to who? Their use of to whom, then, in the early stages could refiect something that they were taught and are attempting to use, but that natural stages of acquisition assert them­selves with increasing exposure to input.

A third case in which UG cannot explain ali of the observed phenomena in acquisition involves object pronouns in French. Towell and Hawkins reported that the first stage of object pronoun production in French by speakers of English involves placement of the pronoun at\er the verb: J'ai reconnu le, "l recognized him." This is followed by a stage akin to Bardovi­Harlig's data on preposition stranding in which the pronoun is m.issing; J'aí reconnu. In the third stage leamers incorrectly place the pronoun between the auxiliary verb and the past participle: J'ai le recon.nu. Towell and Hawkins could explain the overall developmental stages by arguing for the availability of null-0bjects in UG. That is, UG allows for null-0bjects and that perhaps learners of French have deternúned that Vo (a verb) can license pro (an empty category in which a pronoun would appear in a lan­guage like English). Assuming the validity of this argument, Towell and Hawkins point out that a UG account begs at least one question: "Why aren 't leamers able to recognise the preverbal location of object pronouns

Challenges and lfr4Jlications 139

as soon as they notice they are absent postverbally?" (p. 138). This question can be hanclled if it is assumed that the object pronouns are not processed in the input and therefore do not make their way into intake. Two points are relevant here. First, in French subject pronouns are obligatory and are always preverbal in declarative sentences. Thus, assigning the subject role to the first noun results in correct grammatical and semantic role assign­ment so that the problem learners of Spanish encounter does not enter here (see the discussion in Chapter 2). Second, full object nouns are invari­ably post-verbal as in J'ai reconnu Mo.ri.e, "l recognized Mary," as are object pronouns with affirmative commands: Faites-le, "Do it" Li.s-le, "Read it." Initial intake data m.ight very well be giving UG informal.ion that French is strictly VO (Verb-Object) regarclless of whether O is a full noun or a pro· noun. Now, recall that as leamers progress in acquisition, their developing system takes on a greater role in sentence processing. This is depicted in Figure 5-4 by the arrow leading to and from the developing system to var· ious aspects of input processing. If the developing system has determined that French is SVO, then this knowledge may constrain the processing of sentences at\er the earliest of stages. Note that the input data contain very weak and nonsalient object pronouns in preverbal position: The vowel is generally schwaed and tbe object pronoun elides with verbs that begin with a vowel: Je l• uois de temps en temps, "1 see him from time to time;" Je l'ai vu hi.er soir, "l saw him last nigbt." In spoken French, the vowel of the object pronoun may be weakened to the point that it is imperceptible. At the same time, French allows consonant clusters in initial position that are unheard of in English with the result in the cases of cliticized object pro­nouns being sequences such as /u for luoís. lf the learner cannot perceive the preverbal object pronoun in the earlier stages then the input string Je le vois may become intake resembling Je uoís since the developing system would push an SV assignment based on its interpretation that French is SVO. This intake datum is accompanied by the meaning "1 saw him" in which case the mechanisms in volved in tbe instantiation of principies and parameters of UG bave no choice but to consider that null-objects are per· m.issible in French. Here, then, our account involves the initial word order data that are delivered to UG as intake coupled with problems in acoustic perception of object pronouns in spoken French. That leamers perceive and produce object pronouns initially in post.verbal position is accounted for by input utterances containing commands with post-verbal object pronoun.s. (A second explanation of the first stage in which object pronouns are pro­duced post-verbally was also used to discuss the acquisition of preposition stranding. That is, the initial output oflearners could be heavily monitored, resulting in the use of French lexical items mapped onto English syntax. Leamers may be putting together surface features of language that they have consciously leamed, but since they are not generating the sentence

1

Page 79: Input Processing and Grammar Instruction

140 VanPatten

based on en acquired compctence, the syntax of tbe sentence ends up resembling the syntax of en Englisb sentence. More succinctly, leamers could be "thinking in Englisb" in lhe earlier stages and adding on French lexical items during output proccssing.)

In each of tbe preceding three cases of acquisition (subject pronouns in English, preposition stranding in Engltsh, obJect pronouns in French), UG can account for various developmental phenomena successfully but cer· tain parts of lhe acquísitional p1eture are not aooounted for. Assuming that UG operates on mtake data rather than input data, then a more fully developed acoount of the observed phenomena emerges. To be sure, acoustic saliency end the phonetic realization of lhe mput speech stream became important in descnbmg the resulta of input processing in the sec· ond and third cases of acquisibon. Once agrun, a complete model of input processing must incorporate principies lhat involve acoustic features of input (see the brief discussion m Chapter 2 ) that is left to future research for the time being.

Although ToweU end Hawkine <lid not explicitly claim that UG operates on intake data, they hint at this in their model of second language acqui­sition. In their modcl, input data must pass through short-term memory before it can be compared to the intemaUy derived bypotheses-bypothe­ses that originate in eithcr UG or the first lenguage. Tbey state that short­term memory "is the mechenism which determines the information available to long-term memories" (p. 250). Later, they state that input data may not serve to confirm or deny the intemally derived hypotheses, citing that the developing system may be insufficient to aid in the pro­cessing of en utterancc or "becausc short-tcrm memory is limited in copacily [and) leamers are often unable to decipber tbe interna! structure of utterences whcn thesc are first hcard in context" (p. 253, emphasis added). What Towell and Hawkins suggested, then, is congruent witb the present proposal; that input proceesing may dcprive UG of tbe informa· tion it needs for the accurate conslruetion of the syntactic component of the second language.

Retuming to Figure 5-4, we are able to aituate UG in to our general model of aecond language acquisition and use. The result is Figure 5-5. In this model, UG's primary role lies in the second set of processes, accom.modation, end restructuring. M.er mput procea&ing creates mtake data, tbe intake is processed subsequenUy by those mechenisms that compare the data to tbe ¡n-inr.ip1"" Ancl pArAmPll'ir VRh.1"" in UG. Oepending on the state of the de-'el· oping system, the data may conñrm aome aspeet of syntax, force tbe restruc· turing of the system, or may be im?levant to a current hypothesi.s about the synblx. In short, whether or nota hypothesis generated by UG is confirmed or rejected dcpends on the intake data UG rem'-es end not the input that tbe Jeamer hears.

r····

i~ 1 -------------- -----:~-----

: J : . ' . ' . ' • • . . ' . ' ·--··-·

·----- -------------- -------------

r------------------------------ ·---------···--·-------

¡ •U,·---~ : : • : i : • : • • • • j : : • ! • : • • • • • • • • • • • :

r· ., : f ; • • i l ¡ , ____ ,

r·¡··¡ • • • • .... 1 ~ : 1 .......

' ~ . ' . . ' .. . ' ' 1 ~ 1 ' .. 1 iC 1 . .........

~---------------------·····----- - ------------------------.--·~··--'

g

H

.., ¡ l!'

l 1 .s '¡; !! i! " 5

f 1 ]

:J J .., e

i '¡;

1 l ~ o6 oh

§ ~ s:

141

j

Page 80: Input Processing and Grammar Instruction

142 VanPatten

First Language li'ansfer.

In the previous ~on, we focused on the relationslúp between UG and input proces&ng in the creation ofthe syntactic componcnt ofthe learner's developing syatem. However, lirst language transfer is n widcly reaignized phenomenon m second language aa¡uisition (e.g., Gass & Sclinker, 1983, 1992; Odlin, 1989; Wh.ite, 1989). What is its relat1onship to input process­ing? Given that thc first Jangua.ge is a knowledge source in the same way that UG is, it eeems logical that its role also would lie in dctennining the ini­tial hypotheJ!Cs that a leamer would entcrtain about tho syntax of the sec­ond language. For example, suppose that a speaker of English with a strict SVO word ordcr undertakes the acquisition of Spanish. The first language may generate an initial hypothesis that Spanish also is strictly SVO. Input processing, relying on the first noun strategy and unablo to attend to gram­matical markcrs related to case in the input, would dcbver intakc to the developing system m which subjects always seem to be preverbal and objects postverba). The result would be that the initial hypothesís u; confinned and indeed the in!A?rlanguage of English speakers of Spaniah is heavily if not exclu.sivcly SVO, even when pronouns are used (800 Andersen, 1983; VanPlltten, 1987b, for dascussion).

As another examplc, suppose that a speaker ofSparush learning English begins with thc hypothesis that negation is preverbal; more speciliailly, that a simple negator is placed in front of a finite verbas 1t is in Spanish (e.g., No hablo upañol, "I don't speak Spanish"). lf this is the case, then the struc­ture of the IP (lnflectional Phrase) is hypothesized to be something diJfer­ent than it is in English or that rules of movemcnt in which verbs and negators must travel together holds as it doos in English. As soon as the learner processes input containing the following sentences, the hypothesis is confinned-<>r, at least, not rejected. (Note: As will be discu.ssed latA?r, the leamer is probebly not attending to the support verb do in early stages of acquisition, henoe do ís placed in brackets.)

l. No talking' 2. !Dol not write on the papee 3. [Do) )'OU study or not (study) at night?

Thus, the prooessed data appear to confonn to the hypothcsized structure of negntion whieh in turn is based on the lirst language.

As onc final example, let's return to verb raismg. RccalJ that lexical (not auxiliary) verb roising is a parametric value in UG sueh that sorne languages allow it (are ( +raifilngl) and sorne do not (are [·raising)). Spnni¡¡h is [ +raising), whercas English is [·raising). Also recall that important rclcvnnt data for the interna! proocssors concerned with UG includes subjcct,verb inversion.

Challenges and lmplications 143

Languages that are [ +raising) allow sorne kind of subject verb inversion at loost in )-es/no questions. Thus Spe.nish allowi; iO:mueron les chicos ya? Did the boys a1ready eat? (Literally: Ale the boys aiready") with the verb preced­mg the sub_¡ect and indeed oo:upying sentence mitial position. Wbat is also of unportanoe here is that Spanish does not have true modals or auxiliaries like English. Thu.s, there are no lexical equjvalencies of ron, could, wül, would, might, do, and so on, since the work ofthesc modals is taken up in verb in&c­tions. Spanish does have an equjvalcnt of have, haber, wbich functions simi­larl,y to its English counterpart: Have you eaten? HC1$ comido? Spani.sh also has a verb peder that translates into English ns can but is notan auxiliary; in fact, the scmantic field of peder is mueh widcr thnn English's ron.

Now, what happens when Spaniah speaker& begin acquiring English? lf they begin with the interna! hypothcsis that English is [ +raising) just like Spanish, what kind of data do they get? lflearners inco!Tectly subcategorize can in English as a lexical verb rather thnn ns an auxiliary auxiliar)\ then sorne of the initial data they get would lead them to conclude that English (may be) [ +raising). Can you come?, Coukl you po&s IM saJJ? are sample data with subject-verb inversion.

Amending Figure S.S, the first ~ as a hypothesis generating knowledge souroe for the developing systcm is depicted in Figure 5-6. The t\rst Janguage is not seen to act directly during input prooessing. Instead, it mteracts with intake data to shape the dcveloping system.

Page 81: Input Processing and Grammar Instruction

i ; ~

l l o ~

l f l 5· = [ i

l i' t Q, [ ,.. e o B. ¡:

144

H

, ......... . ~ . : !!. 1

" 1 ~ :

t .. ~ .. J

........... . ~ ' .. ' : ~~ : • • . ' !.. .. t .. .!

:-¡--; . ~ . ¡ ~f ; ' . !.. .. .. J

Challenges and lmplications 145

Swnmary

It should be clear from the preceding discussion that input processing and UG are not placed in an either/or opposition as is often the case with UG and a competing theory. What Figure 5-5 suggests is that both input processing and UG play a role in the nature of the developing systems of second Jan. guage learners. By incorporating a role for input processing in delivering intake to UG, we can account for various phenomena in acquisition that at first glance seem to fall outside the scope of UG theory. In addition, the first language is also accounted for in that it is a lmowledge source that gener· ates hypotheses. As in the case ofhypotheses generated by UG, the first Jan. guage-generated hypotheses interact with the intake data for confumation or rejection (see Figure 5-6).

COGNITIVE ACCOUNTS VERSUS FORMAL LINGUISTICS

The previous discussion lays the groundwork for a number of other issues related to the role of input processing in developing a theory of second Jan. guage acquisition. It is true that the model of input processing developed in Chapter 2 is essentially cognitive. Utilizing constructs from cognitive psy· chology such as limited capacity. attention, detectíon, and so on, a model emerged that predicts the processing of linguistic features based on their role in providing cues to sentence meaning for second language comprehen· ders. In addition, the cognitive processing of sentences involving the first noun strategy, lexical semantics, and gramm.atical cues in the input forros the basis for the developmental nature of grammatical·semantic role assign· ment in second language input processing. ) Second language researchers working within the framework of genera·

tive linguistica have been critica! of cognitive accounts of second language acquisition. These criticisms claim that UG can account for the same phe­nomena that cognitive accounts attempt to explain. In addition, these criti­cisms show that UG can account for more phenomena. Fbllowing the argument of "parsimony in explanation," critics argue that UG is prefemid beca use it accounts for a wider range of phenomena. Finall)I these criticisms argue that cognitive approaches often propose ad hoc or poorly motivated principies that are untraceable to any theory. (Another criticism, not as important to the present discussion as the preceding but certainly important in the field of second language reaearch, ia that cognitive approaches very often do not bave a theory about the thing being acquired. That is, as theo­ries about learning, cognitive approaches do not first attempt to adequately describe that which is being learned: language.)

Towell and Hawkins (1994, Chapter 4) claim that common to all cogni·

Page 82: Input Processing and Grammar Instruction

146 VanPauen

uve approaches is that leamers initially decodc, analyze, store, and produce language using geneml cognitive strategies &milar to those proposed by Slobin (1973) and other first language researchers. To make their case that these ac<:ounts are inadequate, Towell and Hawkins examine two approach­es in detall. Tbe first is Pienemann's model (see, e.g., Pienemann, 1987) and the second is Wolfe Quintero's (Wolfe Quintero, 1992). Tbe first model is an attempt to aocount for the deowelopmental s~ of tbe aoquisition of Gennan word order In 1t, Pienemann made use of COl'l$e11JOásm m develop­mg the Canonical Order Strategy. Accordmgly, leamers attempt to conserve the dorninant word order orthe simple input strings they bave been cxposed to, for example, SVOAdv in the case of Gcnnan. To account for thc next stage, Pienemann used perceptual saliency in dcveloping the lnitialization and Finalization Strategy. At this stage, lce.rners can move sentence initial and final elements since they are more perceptually salient, for cxample, adverbs can be fronted now so tbat both SVOAdv and AdvSVO are poss.Ole. However; in German. fuute verbs appear in second position in simple declar­ativa sentences such that once adverbs can be fronted, learners should be producing AdvVSO or AdvAuxSOV colll!tructions that they do not. To account for tbis delay in ocquisition, espccial1y ofthe Aux and V separntion, Picnemann made use ofthe concept of continu1ty.

Wolfe Quintero (1992) attempted to eccount for the dcvelopmental stages ofWH-question formation. In her aocount, \\bife Quintero simílarly made use of conservatism to aocount for the Utitial invariant stage as well as continuity to aocount for nonseparation of prepositions from WH-elements. To aocount for subsequcnt stages of developmcnt, Wolfe Quintero posited an interaction among various cognitive mechanisms: uniqueness to account for one-to-one mappings, cumulative development to account for the building of one stage on anotber, generalization to account for the lack of eireeptions, and p~p/Wn to account for confinnation or denial of a hypothesis.

In their evalual.Jon of the preceding modcls, ToweD and Hawkins noted that the cognitive principies do not aocount for certain observed facts, most importanl, first langua¡¡e inftuence. Fbr example, continuity would not pre­dict that French speakers of English would consistently separatc verbs and objects by intervening adverbs from the earlicst sta¡¡es of developmcnt. That is, continuity would prcdict only '3ohn drinks coffee often" in thc earliest stage9 and not the attested ':John drinks olU:!n coffee." In a similar vein, the Canonical Order Stmtegy would not predict the out.come of Hulk's (1991) study in which first language speakers of Dutch leaming French as a serond Janguage initially utili.zc Germanic word order and not an SVO order. In both examples, thc general cognitive principies would not predict first lan­guage transfer and instead would predict that ali leamers regardless of first language would exhibit idcntical and invariont first s~ of ocquisition.

Thus there exists a situation in which cognitive models are pitted against

l Challenges and lmplications 147

UG and that the weaknesses of the cognitive models become salient as one begins to explore their predictions and compare these to the observed facts in a variety of acquisitional situations. Assuming that these cognitivo approaches are indeed inadequate (and by no means would evecyone agree), what is to prevent the model or input processing in the present book from succumbmg to the same criticisms? There are tbree re&l!Ons why the prescnt model of mput processing is not vulnerable to the criticisms leveled at othcr oognibve models. First and foremost, input processmg u not an acrounl o( karning; 1/ lS an acrount o( what JUnd o( intcrke data are made lWOilable for learni.ng. Rocal1 that in, Figure 5-3, ocquisition and UllC are separnted into tbree dii!tinct sets of processcs. lnput processing is only concerned with those processed depicted by I in the figure and is not intended to aa:ount for any other processes. ln their criticism of Pienemann's and Wolfe Quintero's oognifüe modela, Towell and Hawkins claimed that theo;e models are intend· ed to account for the deaxúng, onalysi.s, slorDlJe, and produdion of lan­guage. In other words, these models attempt to account for the sets of processes in Figure 5-3 in addition to how learneni create output. BecallllC input proocssing is concemed with a much more reduced set of phenomena, the present model does not makc predictions about how the developing sys­tem creates a syntactic componcnt nor about what the O\ltput of leameni may look like.

The aecond reason tbat our modcl of input processing is not subject to the same cnticism as more general oognitive aa:ounts of second language acquí­sition is related to the first rea.son; unlike cognitive models that are ali encompassing, input processing is not in oompetition with a theory based on UG. As Figure 5-5 illustrates, input processing and UG concern distinct) aspects of second language acquisition. The 6rst section of this chapter con-1 tains a discussion of how the oognitive aspects of input processing interface with the grammatical knowledge contained in UG By claiming that UG operates on intake data, the pnsent model incolllOrates a mutually com­patible if not conspiratoria! relationship between mput processing and UG.

The third reason that our model of input processing is not vulnerable to the criticislll! of ToweD and Hawkins is that first language transfer is allowed to play a role in the crcation of the developing system. As seen car· lier, both UG and the first language can play a role in generating the initial bypotheses that a learner entertains about the second language. This role of the first language is clearly depicted m Figure 5-ó.

To summarize, tbe principies oí input P""""'-«ing outlined in Chapter 2 do not att.empt to account for ali aspects of aoquisition; they do not oomprise a total model. lnstead, they form prut of a model that attempts to account solely for thc dcrivation of intakc data. Accommodation, restructuring, out­put generation, and other aspects of language acquisition and use must be accounted for by other theories or models. In this way, input processing does

. ,

Page 83: Input Processing and Grammar Instruction

148 VanPatten

not overgenerate predictions. When placed with UG and first language transfer in a larger more general model, a more complete aocount of acqui­sition emerges.

PROCESSING INSTRUCTION ANO SYNTAX

The previous discussion on the relationship between input processing and the development of syntax leads us to severa! questions about processing instruction.

l. Because processing instruction's aim is to push leamers to attend to ele­ments in the input that tbey might otherwise miss, and since input pro­cessing is principally concerned witb creating form-meaning relationships, how can processing instruction be used to teach items that are meaningless?

2. A related question concerns lexical versus functional categories. Zobl claims that the nature offunctional categories renders them unteachable. Thua explicit instruction cannot aíd in their development. Is this true for processing instruction as well?

Processing lnstruction and Meaningless ltems: An Example from Verb Raising

The first question is most easily observed in the problem of verb raísing in second langua¡¡e aoquisition. White (1992) and Trahey and White (1993) showed that first language speakers ofFrench have difficulty in eliminating the error of placing manner and frequency adverbs between verbs and objects when learning English. Because they assume that verbs can raise in English, these learners continue to make errors of the type noted in a pre­vious section ofthis chapter: •John drinJl.s ofüm coffee. 'Mary walches quiet­ly televi.súm. However, they do learn the permissible locations of adverbs in English that are disallowed in French. Thus, they can produce Joh.n often drinl<s!has o{ten drunk coffee. Even with explicit instruction and negative feedback, errors of the type John drinJ<s o{ten coffee are persistent. And input fiooding (i.e., providing the learners witb input over a short period of time that is saturated with correct positive evidence in the input) does not seem to help either (Trahey & Whlte, 1993). Asswning that this error is due to the differences between French and English regarding the verb raising parameter, we can see why traditional instruction and negative feedback would not remedy thesc learners' error. As Schwartz (1993) points out, UG can interact only witb data processed in the input; it is not a knowledge

Challenges and lmplications 149

source that can interact with explicit conscious knowledge about langw¡ge. Li.kewise, the input flooding would not work if the input did not contaín the right triggers. Note that adverb placement is a result of tbe verb raísing parameter and may not be a trigger for it. Thus one could question why adverb placement would be the correct instructional target if indeed the issue is that learners must restructure their grammars and reset tbe para­meter to (-raising].

The question is whether processing instruction can provide better intake for UG so that the correct verb parameter can be reset. Note that verb raí& ing and adverb placement cannot be considered meaningful aspects of lan­guage. Both John often drinks coffee and Joh.n drinks coffee often (as well as tbe erroneous John drinks o{ten co{fee) ali refer to the same referential set of circwn.stances. The different placement of the adverb, then, does not con­tri bu te to sentence meaning during comprehension and it is not clear how the form-meaning mapping¡¡ created by input processing are relevant here. Before examining this last issue, it is necessary to review sorne facts about verb raísing first.

Reca.11 that the clause has an abstract structure like that in Figure 5-1. Lexical verbs originate under Vo and move up to pick up tense and then land in Agr; to pick up agreement in French. This accounts for placement of adverbs (sentence 1 following) and negators in French to tbe right of finite verbs (sentence 2) as well as subject verb inversion that includes a further raísing of the verb into Co (sentence 3). (The use of Jean as a topic marker in sentence 3 is irrelevant here; note tbat the verb and subject pronoun are inverted.)

l. [¡p Jean[Acaboit; [yp souvent [y>e¡ du café au lait.]])] 2. [¡p Jean[AcRboit; ITP pas [yp souvent [y.e, du café au laít.]]]J) 3. Jean [C' boit¡ [¡p il [yp souvent (y>e¡ du café au laít.])])

In English, the features in Agr; and T> lower into the VP so tbat lexical verba do not move up. The negator not blocks the lowering of Agr; features so that do is forced. This accounts for the placement of adverbs and nega­tion to the left of verbs in English (sentences 4 and 5, respectively). Do can move into Co, which accounts for yes/no questions (sentence 6).

4 . [¡p John[AGR e; [yp often (yp drinks coffee with milk.]]]) 5. [¡p John(AGR does [TP not [yp often (yp drink coffee with milk.))]] 6. [cp Does; [¡p John[AGR e; [yp often (yp drink coffee witb milk.JJJJ

If the in.itial hypotbesis of learners of French is that English raíses ali verbs, tben the real clue to UG that English does not allow verb raísing is not the location of adverbs or negators but the presence of do. This analysis

Page 84: Input Processing and Grammar Instruction

150 VanPatten

reoeives support from Lightfoot (1993) who demon..<trates that the evolution from 1 +raÍSlllg) to l ·nusingl a.s Old English become Modem English is aocompanicd by the appearance and use of do as a dummy verb. In d:is­cussing the period of change from 1475 to 1550, Lightfoot says

Each iniiertion of a penphrastic dlJ to cany inllectional markers represents a en.., whcre the Y·to-1 (verb raisingl operation has not app~cd. ao a steady íncrm.e in the dlstríbution of dlJ entails fewer and fewer lnatances oí V-to-1· the two operotions are mutual\y incompatible (p. 207). '

In terms of second language acquisition, if leamcrs are attending to ali the lexical elcments in an input string then, the sequcnce cJc + NEG + VERB should tell the learner that Agr is structured difTercnUy in English and therefore verba cannot raise. In short, do is the triggcr to reset the parame­ter to ·raÍSlllg. The question then becomes whether or not learners are detectmgdo in the input.

Researcb on the aoquisition of negation m English suggests that learn· ers do not attend to do as a separate lexical item in the input in the early and mtermediate stages (for overviews of research on negation, see Ellis, 1994; Larsen-Freeman & Long, 1991; among othcrs). Stagc 1 of negation consista of NEG + phrase, for example, No go, No come tonljfht. Stagc 2 consists of the begiruúngs of clause structurc with NEG being sentence interna! and placed before the verb, for examplc, 1 no come tonight. At this aame time and also in Stage 3, we sce the use of don 't as an unana­lyzcd chunk that rcplaces no as the principie negator, for example, J cicn 't coni,e tonight; He don't study. At the same time, wc scc that in the acqui­sition of questions in Engtish, do is absent in thc first two stages. These data euggest that tbe intakc data delivercd to UG doce not contain a sep­arate do that would force an analysis of English Agt-0 and 'J'O that is more target-likc. The absence of do in the intake is prcd1c~ by our model of input proccssing. Ar. an item of essentially litUo or no communicative value (the root do carnes no semantic information), 1t is easily skippcd dunng input processing and its contraction with not-an 1tem of high communicative value because it is the only item in the utterance that wilJ indicate to a learncr that something is be1ng ncgated-m the forms of don 't and doesn 't easily could lead tbe mechanisms oí input processing to delivcr to UG the following information: no, not, don 't, and doesn 't are allomorpluc vanations of NEG.

If tlus analys1s IS corrcct and do is the key to forclng a ~on oí verb raising by Frcnch speakers learning English, then the instructional question is whethcr or not processing instruction can rud in attcntion paid to cJc in the input. lt would secm that this is possible. Sincc do carries both Agr fea­tures and T features, both of which are meaning·bascd, thcn these might be

Challenges and lmplicat1ons 151

exploi~ in the form of proc:essíng lllStruction to get leamers to actively attend to and detect do during structured input activities. As an example, structured input activities could encourage leamers to attend to do \"el'SIJS

does to get person andior do!does versus did to get tense (see tbe instruction m Cadiemo, 1995, in Cbapters 3 and 4). lt would then be up to UG to ana­lyze cJc correctly after it is irutially dctcctcd and the form-meaning ma~ are made. The outcome of tlús samano is, of course, an empirical question and one which depends on the proposed differcnces between Agr" and TO in Frcnch and English.

Prooessing lnstruction and Functional Categories

The question of lexical versus functional categories is related to tbe first question of teaching meaningless features of language. Zobl (m.s.) claims that functional categories are "semantically shallow or vacuous" observing the lack of semantic mcaning and purely grammatical nature of linguistic it~ms such as complementizers, conjunctions, and the aux­iliary do. Citing research from adult first language sentence and lexical processing, Zobl also clrums that the processing of functional categories is "automatic and encapsulated" and thus is separate from the process­mg of lexical categories. Based on th.is analysis, he argues tbat the acquisition of functional categories is not responsive to any kind of explicit knowledge derivcd from explicit instruction and that this is counter-theorctical to the weak interface positions (e.g., Ellis, 1993; Sharwood Smith, 1991).

At first glance, Zobl's claim appcars to be correct. However, there are a number of problems with thc argumentation. First, it is not clear that adult first language sentencc processing research is directly relevant to second language acquisition. Prcsumably, automatic processing is pre­ceded by rcpea_ted and consistent controlled processing (McLaughlin, 1987). From this one could argue that the automatic processing of func­tional elements in adult first language rcscarch might not have st.arted out as automatic processing m child first language acquisition. It would seem logical, then, that second language leamers have to build up the automatic processing of functional elements as well Second, it is also not clear that functional elements and tbe1r functional projections in syntax are consistenUy semantically vacuous. In the case of preposi­tinn•, thP. wP.11-knnwn Pnrl nf Linonln'• Gettysburg Addrcss is evidence against this: "A government OF the pcople, FOR the people, and BY the people shall not perish .... " Complcmcntizers can be signals of tense, agreement and/or mood features. In Spanfah, for example, the complc­mentizer que obligatorily heads tcnsed clauses with finite verbs. Contrast the following:

Page 85: Input Processing and Grammar Instruction

152 Vaof'atten

(la) QuU!ro lttr/c. 1 want t.o roed it.

Ubl Quuro que lo Iros. 1 want you t.o read it.

(2a) Creo ser mlelllfenle. 1 belicve l'm int.elligent.

(2b) Creo que eres 1ntel1gente. 1 belicve you are intclligent.

In cases such as theBC, que can serve as additional cuc during input process­ing that a t.enscd vcrb is coming and that there may be a su!Uect (person­number) shill

In the case of oo, it is important t.o recall that this support verb is a tense and ag:reement carrier. Although the verb itself is semantically empty, in a oertain senae there is semantic information encoded int.o this word.Does, for example, signala third·person singulai; whereas more unportant (from the Jeamers po111t of view) duJ signaJs past, whereas ooldoea signa) nonpast. Tbus, the tense contrast between Did you aJways pay attenhon? and Do .)OU

alway1 poy attenlwn? rests on processing the verb dui!oo 111 the 111put. Tbe problem with many functional elements as we saw m Chapter 2 is not

that they are semantically empty; it's that they often a><>ccUr with other lex­ical items or markers rendering them redundant, for example, Did you alwo¡ya poy aJtenlion in high echool? Thu.s, the relative lack of communica­tive value of functional elements is owing to built in language redundancy and not aey inherent semantic vacuousness. (For this rooson, we distinguish between ecmant.ic value and communicative valuc.) According t.o the model developed in Chapt.er 2, they are likely to be consist.ently att.ended to and detected ata later stage of development in input prooossing if at ali. Since the examples just ci!OO are no different in communicative value from such things as tense markers and copular verbs, and whereas there is evidence that pro­cessing instruction has a beneficial effect on these latter 1tems IChapter 4), it wouJd seem reai;onable t.o conclude that the aa¡uis:ition of many functional elementsoould bespeeded up by prooessing UIStruction. As in thecase ofverb raísing, the clTccts of proce;sjng instrudion on functional elements is an empirical que&tion.

REMAINING ISS~S RELATED TO PROCESSING INSTRUCTION

One of the more int.eresting findings from VanPattcn and Cadierno (1993), Cadiemo 1995), and Cheng (1995) is the differential clfcct that prooessing instruction and traditional instruction seem to have. In thcee studies the tra·

Challenges and lmplícations 153

ditional group's performance improvcd onJy on production tests. The procese­ing group's performance, however, impro\'ed both on interpretation and pro­duction. Why this difference? One obvio\is factor to examine is task familiarity caused by instructional treatmenl The production tasi< was one with which the tradibonal group was familiar, bccause there were l!e\-eral activities ha.sed on 1t 111 tlus group's instructional treatmenl The 111tcrpretation tasi< was not famíJ. iar to thcm 5Íllce it did not fonn a part of the instructional treatment. A logical conclusion would be that task familiarity is a factor in the outcome. However, the results of the prooessing group on both tasks leads us to question the role of task fümiliarity. RecalJ that the processing group had familiarity with the interpretation task but no familiarity with the production task. In processing instruction, leemers are engaged in structured input activities, not output octivities. Yet, the prooessing group was ablc to perfonn as well as ifnot better than the traditional group on a task with which the former had no fami.liarity but thelatt.erdid. Thu.s, taskúuniliaritycan accountforonesetofresults (thoee ofthe tradi.tional group) but not the other Ct.hoee of the processing grrup).

Tbese diJTerential outcomes can be explamed by the general model of soo­ond language acquisition and use that informa this book. In Figure 5-3, the 00..-eloping system is seen to be a result only of input prooes•ing and the acoommodation of intake and the restructunng of the system. In the expend­ed version of Figure 5-6, we have seen that input proa!S9ng, UG and the first language ali play roles in the formation and growth of the developing system. Processing instruction is a response to this rnodel. Since its aim is to alter or improve input prooessing, the result is bcttcr intake for the mechanisms that oocommodat.e data and restructure thc system. Our claim is that prooessing instruction is an approach that can directly affect acquisition (in Krashen's sonso) since it is input oriented; in short, the data are appropriate for the processes responsible for acquisition of granunar. Traditional instruction, on the other hand, does not provide input data to loomers. lt oonsists of explicit knowledge plus various kinds of output practice. ThW!, traditional instruction should not alfect acquisition.

lf we ei<pand the general model of language acquisition to include output proces:;ing, then we get something l.ike Figure 5-7. Hcre we see that the cre­ation of output in\'olves a set of prooesses that tap the developing sYStem in some way. One conclusion regarding the re&Ults of the research on process­rng versus traditional instruction is that the proce$S1ng subjects were per· fonrung with acquired knowledge, knowledge that has become part of their dcveloping system. For this reason, they were able to improve on both inter­pretation and production tasks. Sinoo trnditlonal instruction does not affect acquisition, our conclusion is that the tradit.ional subjects perfonned the product.ion task with leamed knowledge, knowledge that did not form part of their developing system. Hcro is whcre task fwniliarity may play some role. Given that the traditional greup pracliced producing the target items

Page 86: Input Processing and Grammar Instruction

154 VanPatten

1 II fil input---< .. ~ intake ---1 ... develop1ng system --..... ~ output

1 = input processing 11 = accommodation, rei;tructuring llI = access

FIGUREIVi. Throe .... olJ"• !1111....,..¡~ooqwauonanduse.

with their explicit knowledge, it is feasible that they gained sorne ability to use this knowledge on the fairly aunple and time-<entroUed tasks used in the studies. In elfect, they were Morutonng to borrow temunology from Krasben. However, whereas they had no pracbce with int.erpretation tasks, they could not effectively Morutor on the1;e timed mterpretation tasks.

CONCLUSION

This chapter has examined severa! challenges to input processing and pro­cessing instruction: the relatioMlúp of input processing to UG and to first language transfer; how our modcl of input processing differs from other cog­nitive framework.s that attempt to explain acqu.isition; whether or not "meaningless" features of language can be taught via processing instruc­tion; and the differential outcomes of traditional and processing instruction on interpretation and production ta.sks. Unlike other approaches to theory building, the proposed modo! of input proocssing is not offered in lieu of UG and first language transfcr as llCCOunta of seoond language acquisition. lnstead, it is offered as a companion aspcct of a general model of language acquisition and use. In this more general model, input processing is respon­sible for delivering intake that is thcn uscd to confirm or reject hypotheses generated by UG or the first language. In this way, the combined efforts of input processing, UG, and first language transfer account for more phe­nomena in seoond language acquisit.ion than any one could alone.

The discuSSJon m this chapter has led to a number of questions about the limits and effects of procesSlng lll8truction. Can meaningiess items be taugbt via processing instruction? Are the resulta of processing instruction attributable to Monitonne? Does proceeein¡ instruction actual]y affect the acquired system, what IS callee! in the general model the developing system? An ex•mination ofthese questions reveals that they are empirical questions and imminenUy researchable. Because processing mstruction is predicated upon Jl6)'Cholinguistac mechanisms and a theory of communicative value (that is, the relatiorish1p betwecn linguistic fonn and refemtial meaning

Challenges and lmplications 155

during the dcvelopmcnt of second language comprehension), and because it is not opposed to cun-ent linguistic tbeory but can draw on that theory to examine problcms in acquisition, processing instruction lends itself to exper­imental evaluat.ion. In both tbeory building and in practical application, experimental evaluation is a critica! and necessary process in the advanoe­ment of the f\eld.

Page 87: Input Processing and Grammar Instruction

!6 EPILOGUE

These are both interesting and exciting times for language instruction. In temis of theory. we ha ve come a long way in our understanding of the na tu re of second language acquisition sinoe the 1960s. We have moved from simple aocounts of how learners intemalize a grammatical system t.o an under­standing that acqui.sition is componential, multidimensional, and richly complex. We have a flourishing research field that explores everything from UG in second language acqui.sition t.o the nature ofreading comprehension t.o the role of anxiety in (un)sucoessful acqui.sition. Our more complet.e understanding of second 1anguage acquisition has led a number of us to rethink grammar instruction and t.o engage in active research on its elfects. For the first time we are beginning to make concrete suggestions for gram­mar instruction that are grounded in both theory and research.

It is the aim of this book t.o explore the theory. research, and implications of processing instruction. 'lb thi.s end we have examined in sorne detail the nature of input processing in second language acquisition. We ha ve reviewed an instructional approach that makes use of insights from input proeessing and we have examined in sorne detall 6.ve investigations on the effects of processing instruction. We ooncluded with an examination ofhow input pro­cessing 6.ts int.o the 1arger picture of second language acquisition and use, and thus, how it works in tandem with other factors such as UG and first language transfer. In so doing we ha ve addressed pot.ential criticisms of both input processing and processing instruction from a formalist perspective.

There is no doubt that input proces.sing and processing instruction will face additional chaJlenges as researchers tum more of their attention t.o how learners attend to input data to create intake. There is no doubt that a<ljust­ments in the model outlined in Chapter 2 and elaborated on in Chapter 5 wiJ1 occur. Perhaps a different model will replace it. As oommitted researchers and scholars, we should welcome these challenges, for they re¡>-

157

Page 88: Input Processing and Grammar Instruction

'158 VanPatten

resent progreiwi in the fields ofboth sooond languagc ncquisition and seoond language teaching. lndeed, at various points along the way. we ha•-e sug­gested what pat.hs continued researcb on input processmg and processing instruction might follow

Before concluding, there are two pomts that the readcr &lould oonsider. The present book has focusecl on one particular type of grammar mstruction and has nttemptcd to support it with theory and resean:h. The render should not conclude from this focus that we are advocating nn nbnndonment of the oommunicativc classroom. We are not advocating that processing instruction occupy ali of instructora' and leamera' time to the cxclU11ion of interaction, rending, and other oomponents of a communicative approach. Instead, as Lightbown and Spada (1993) state, "we nced to dcvelop a bctter under· standing of how form·focused instruction can be most cffcctively incorporat­ed into a oommunicative frarnework" (p. 106). To this cnd, not only do we nced to contmue research that explores the pSycholingWstic undcrpinnings ofprocessing lll8truction and its relath-e effects, but "''e Biso nced to ask our· selves questions of a much more practica) nature: Can and should processing instruction oocur outside the classroom. say, as homework? Does it neecl to be brought into thc classroom? Because processing instruction is mput-based, can computera deliver effective pu:assiug instruction? Pursuing questions such as thesc will belp instructors and CWTiculum deve.lopers maxi.mize oom· municative language W!e durúig the rather rninimum runount of time that language studenta spend in the classroom.

A sooond pomt wort.h bringing to the reader's attention concems the renction of instructora to processing instruction. Proccs.sing instruction admittedly is a radical alternative to traditional approaches to grammar instruction. FOC\1$ed on input and devoid of a mochanical stage, how does the expericnced instructor perceive processing instruction? 1 recall having been invited to a large state university to speak to a group of graduate teach· ing assistanta and experienced faculty about my resea.rch on processing instruction. Afu:!r a 45-minute presentation of sorne of the resea.rch pre­sented in Chapter <I of this book, 1 entertained questions from the audience. A graduate Waching assistant with sorne experience raised her hand and asked, "So,)'OUdon'tbelieve inexplicitinstructionora focw;on form then?" After overcoming my initial surprise, 1 responded that processing instruc­tion was a focw; on form. As my eyes swept the audience, 1 realized tbat she was not the only per50n who had ooncluded that procc&illlg mstruction was "implicit" instruction. At a reoeption held later 1 ficlded a number of relat­ed questions. lt seems that a number of peoplc in the audicnce had filtered my talk through thc schema that they associate(d) with Kmshen and com· prehensiblc input: no instruction in grrunmar. To talk of input processing, then, and proccssing instruction, was seen as to offer yet another version of the Natural Approach.

Epilogue 159

As resea.rchera, "''e must take caro as to how the nonresearching and pos.­sibly uninformed teacher interprets our l"CllCaJ'Ch. Tra<litional instruction is \'ery much alive, at least in foreign language teaching m the United States. Tcachers dutifully explain, drill, pass out worksheets, administer di.saete­point tests of gran:unar, ali in the name of helpmg learners internalize ("master") the granunar. And if we are absolutely honest, we would adrnit that m0&t of language teaching in the United States is not communicative at ali; it is grammar teaching with sorne communicative interaction added on. If rel!earch on grammar instruction such as that oontained in the pre­sent book eventually is to have an impact on language teaching, then we must be oognizant of the bclicfs and practices of teachera. lt would be in ter· csting indced to be alive in the ycar 2050 to scc what has happened to pro­ccssing instruction. Assuming it survives the theoretical and resea.rch challcnges that may come its way, will it surnve the challenge of existing teaching practices and beliefs? My hope is that it will.

Page 89: Input Processing and Grammar Instruction

REFERENCES

Andersen, R. (1983). Tramfer to somewhere. lo S. M. Gass & L. Selioker (Eds.), Longuoge lro11sfer in /anguage kornütg (pp. 177-201). Rowley, MA: Newbury House.

Bardovi-Harlig, K (1987). Markedn""8 and salieoce in secood language acquisition. Longuoge 1.-rning, 37, 386-407.

Bardovi-Harlig, K. (1992). The use of adverbial.s and natural order io the develop­ment oftemporal expression. Jnternatiornd Reu~w of Applied Linguislicll, 30, 299-820.

Bates, E., & MacWhinney, B. (1989). Functionalism and the competition model. In B. MacWhioney & E. Bates (Eds.), TM cross-ljnguistic study of ~pro­ctsSÍ"IJ (pp. 77-117). Cambridge: Cambridge University Prees.

Batee, E., MacWhinney, B., Caselli, C., Devesooví, A., Natale, F., & Venza, V. (1984). A cros&-linguistic study of the development of sentence interpretation strate­gies. Child Deuewpmern, 55, 341-854.

Bavin, E. L., & Shopen, T. (1989). Oues to senteoce ioterpretation in Warlpiri. In B. MacWlúnney & E. Bates (Eds.J, TM cross-linguistic study of sentence prow.s· ing (pp. 185-205). Cambridge: Cambridge University Prees.

Berne, J. (1989). The elfect of increesed comprehen.si'bility on leamer's abilityto con· scioosly atteod to cootent and form. Unpublished manuscript. Manuscript, University oflllinois at Urbana-Champaign..

Bever, T. G. (1970). The cognitiva basis for linguistic structures. Jo J. R. Hayos (Ed.J, Cognili.on and IM developmenl of languoge (pp. 279-362). New York: Wiley.

Binkowski, D. D. (1992). The effects of attenti.onlll focus, prtsentaJion mode, and /an. guage eJCP<!r~nce on second lall1Juoge Ú!<lrner's sentence processing. Unpublished doctoral theais, University of Illinois at Urbana-Champaign..

Blau, E. K. (1990). The elfect of syntax, speed and pauses on listening comprehen· sion. TESOL Quarlerly, 24, 746-753.

Bransdorfer, R. (1989). Processiog funct;ion words io input: Does meanin.g make a difference? Paper presentad at the annual meeting of the American As8ociation of Teachers of Spenish and Portugueee, San Antonio.

Bransdorfer, R. (1991). Chap. 2: Review of the literature. Communiootive ooiue and li.'lfU¡,,tu; kncwkd{;< in '5eC0'1d ~ oral input p~n¡f. Unpubl.i8bed. doctoral thesis, University of Illinois at Urbana-Champaign.

Bybee, J. (1991). Natural morphology: the organizaüon of peradigms and language acquisition. In T. Huebner and CA Ferguson (Eds.J, Crosscurrents in &cond Langu<J8e Acquisitions and Ü"IJUistic Theories (pp. 67-91). Amsterdam: John Benjamins.

161

Page 90: Input Processing and Grammar Instruction

162 VanPatten

Cadierno, T. (1992). Exp/u:tl in8trucl1on in grommar: o cornpori.8on of input based and oulput bo#d inslnldion 1n BtOOnd /(Jngu011e ocquuJ1twn. Unpublished doc· tmal thesis, Umvemty oí illinoia at Urbana-Champaign.

Cadiemo, T. (1995). Formal i!l4truction from a proceuing perspective: An inves· tigation mto the Spamah paat teru.c. Modern Uuiguagt Jounuú, 79, 179-193

Cadiemo, T., GIMs, W R., U!e, J F. & VanPatten. B U991l. Prooo;sing tense in sec:­ond language mput: Lexical wee Vl'l1IU.I grammallcal wea. The Vru''l!fsity of lllinoia at Urbana-Charnpmgn

Carlson, R. A., & Dulany, O 0965). Con.cioua attention and ahstraction m concept 1-.ung. Journal o{ Exptnmtntal ~~ Leornuy¡, MtmOT)', and Ccgruhon, 11. 45-M.

Carr, T. H_, & Curran, T. (199<1). CopubVe factora m leanung about 5tl'\lctUred eequeneee: Appbaitiona to syntax Stuáia 111 &rond Longu4(lt hqmsllwn, 16, 2()[).-230.

CarreU. P. (1985). Faalitating ESL .--lm¡ by ~ text structure. TESOL Qu¡uvrly. 19. 727-762

Ceci, S. J., & Howe, M. J. A. ( 1983>. Metamemory and effccta ofmtendmg, attending, and intending to atulnd In G Underwood CEd.l, Asp«:ta o{ conacÍO<l$nt:S$ (pp. 147-164). New York: Academ1c ~.

Cbeng, A. (1996). Grommar •n"1Uclion and UlpUl proceuing· TM ocquisllion o{ Sponi$h aer and .,141: Unpubliahed doctoral theeis. Universtty o( Illinoi.s at Urbana.Champrugn,

Chomsk;i\ N. (1988). ~ and probltma of know~: TM Managua kctures. Cambridge, MA: MJT.

Corder, S. Pit. (1967). The signillcanoo oí loo.rneni' errora. Republished in Corder (1981) Error OJUÚytlÍB and m~rlanguag<. Oxford: Oxford Uníversity Presa.

Dienee, Z., Broadbent, O., & Berry, O. (1991). lmplicit and explicit knowlodge bases in artillcial grammar leaming. Jounull of Experimenta/ P8ychol.ogy: Leaming. Memory, and Cognition. 17, 876-887.

Oulany, D., Carlson. R., & Dewey, G. U984). A.- of synt.actícal learning andjudg­ment: how conaciou.1 and how obstract? Joumal of &perimenta/ PByc/wlogy: GelwoJ. J 13. 541-665.

Ellis, R. (1990), /na/rud«I &cond languagt hquwtion. Oxfonl: Basil Blackwell. Ellis, R. (1993), The structural ayllabua and llCCOnd language aa¡ui.sitioo. TESOL

Qu¡uvrly, 27, 91 113. Ellis, R. (1994). His new book with Oxford. Ellis, R. (1995). ln1"'pl'l'tabon t8Ska ror grammar U'l01ruction. TESOL Quartuly. 29,

87-106 Ervin-Tripp, S. !.t 0974). la eecond 1an¡uage leanung really Wte the 6.rst? TESOL

Quartvty, 8. 11-128. Euba.nk., L. (1991). l\H1U10Ju1dc1JNu". U1uc..ynlUJ Grummar m tM Se.cond

langu"#f'. ~' Beqjami!l4. Eubenk, L., & Gregg, K 0995>. Et m amygdala ego? CG. <SJLA. and neurobiology.

Stucúts in 5-nd ÚUllfU48t hquu1tU>t1. 17. 35-68 Faen:h. C., & l<asper, G. ( 191!6). The role of romprehension in eecond language lo.aro·

ing. Applud únguisha. 7. 267-274.

References 163

Fmnemann, M. D. (1990). Markedneea and learner &trategy: Fonn-and meaning-Ori­entAld k>arnel'9 in the forcign language amtext. Mcxkm Uutlfllll8' Joumal. 74, 176-187

Garrett, N. (1991). Thooretical and ~ problems oI separating "grammar" from •communicabon • In B Freed !&!.), Forttgn Longu4(lt hquislli® RatCJTh and 1~ Claaroom (pp. 74-87). New York: D.C. Heath.

Gasa. S. 11989). Howdo leemenreeol'" lingwstic conl\icta? In S. Gass &J. Schacter (Eda.), 1...ui¡¡rmlu: Prr•pcdna on &rond ÚJn8U"8eAl:t¡uuiJúJn (pp. 183-199). Cambnd¡e: Cambrld6'e Unl\'~ty Preoa.

Gaoo, S. M, & Madden, C. G !Eda.l. U985l. Input 111 MIOlid /anguDgt acquisllton. Rawley, MA Newbury Houae.

0-, S. M, &: Sdulchter,J <Eda.l. 119891. ~<pa-spectWesonSftXJnd /anguDgt O()t¡UWU>tl. Cambrld6'e: Cambridge Univensity Pr-.

Gasa, S. M, & Selmker. L (Eda.l. U983l. langu"#f' lron8{u 111 /anguDgt ~­Rowley. MA Newbury Houae.

GM1, S. M., & Selmker, L. !Eda.l. ( 1992). ÚJll8UO#f' lron8{u in /anguDgt learmng. Amsterdrun 8el1J8mms.

Gasa. S. M., & Sehnker, L. <Eda.l. Sc<ond /(Jnguagt oa¡uisllwn: An Wroduttory rourw Hilladale, NJ: Erlbe.um

Gla.., W R. (1994 ). Paper delivered at the annu.al meeting oI the Ameriean Aaooaation for Apptied LmguJStics. Balbmore.

Gliaan, E. (1985). Thc efTcct ofword order on listeningcomprehensioo and pattem retention: Ar> experimcnt in Spruú&h as a foreign language. Longu4(lt Leaming. 35, 443-4 72.

Guntermonn, G. (1992). Ar> analy8ia of interlanguage development over time: Part 11, &r ondea/ar. Hi8panio, 75, 294-303.

Harley, B .. & Swoin, M. (1984). The ínterlanguagtl of immersion students and ita implícatio!l4 far 11000nd Jonguoge teoclúng. In A. Davies. C. Criper, & A P. R. Howatt (Eda.), lriterlorrguage (pp, 291-311). Edinburgh: Edinburgh Univcnlity P.-.

Hatch, E. (1983). Simplilled input and llOOOnd language acquisition. In R. W. Anderacn (Ed.), Pid1ginizal1on ornl Creoiizalion as Longu4(lt hquisiOOn (pp. 64-88). Rowley, MA: Newbury Houae.

Hulk, A. (1991), Parameter eetting and the aa¡uisition of word order in L2 French. &cond LanglJ.08'! Ra.orrh, 7, 1-34.

Hyltenstam, K. (1987), Markodncoa, Janguage univeraala, language typology, and sec­ond lan¡ua¡:o acqullition. In C. W. Pfaff (Ed.), Fint and 8tJC()nd /anguDgt O()t¡uullwn p~.- (pp. 65-78). Cambridge, MA: Newbury House.

lsmdoridea, D C .. & Hulatijn, J. H. 11992), Comprehension of grammatically modi· 6ed and nonmocWled 1entA!ncee by seoond language leamers. Applttd Ptyclwhnguutia. 13, 147- 171.

Ju>t, M . A., & Carp.nt6. r. A. (1993). Acapaoty theoeyolcomprehension: lndlvid=I dilferencoe m worlúng memory Ptyclw/o¡¡=L Ret;.,.,,,, 99, 122-149.

Klein, w U91MH. The problem or anaJym. &rond /anguDgt acquisilwn. Cambridge: Cambnd¡e t.;ruvenity Pr-.

Knuihen, S 0982>. Stcond ÚUl8""1/t ocquú1hon and SftXJnd /tJngli.(J# ~­London· ~.

Page 91: Input Processing and Grammar Instruction

16-4 V&nPatten

Krashen, S. O 0985>. The mput hypotk$18. London: Longman, Krashen, S. D., & Terrell. T. O. U983>. The 114turol approach: ~ ocqwsr.bon

111 /Mclaaroom. San Francisco: Alemany Preos. Lechman, R., L.achman, J, & Butlerfield, E. U979J, Conoci<>lllll<* and attenbOD.

Ccpitiu. p.ycho/"f!Y and •nformatwn prott$1ing.· An 111lroductwn. Hilladale, NJ Erlbaum

l.anen-Frttman, O (1976). An explanatioo far the morpheme acquwtlon arder oC """""d lan¡¡uage leamers. ~ Ltanung, 26, 125-134

l.anen·Freeman, O. (1986). Techniques and principies in longu"8' tooch1f18. Oxford: Oxford Univeraity Preos.

Lanen-Frooman, O., & Long, M. H. (1991). An inlroduc1ion to treoond /anguage ocq1us1tion rt:1loorch. London: Longman.

Lee, J. F., & VanPattcn, B. (1995). Makin¡¡ communu:otiu. langua¡¡e koching happen. New York: McGraw-Hill.

Lee, J F. (1987). Morphological factora inJluencing prononunal reference assignment by IA>amers of Spanish; OedicaWcf to Joseph H. MaUuck. ln T. A Margan. J. F. Lee, & B. Vanl'atten (Eds.J, ~ and /an¡¡uagt l<M. Studia UI Spon18h (221-232}. IAnbam MD: Uoivt!f'Slty Preos of Amenca.

~ R. ( 1993). To IUilplify or not to simpbfy: A look at mt.ake. Stu~ ui &cl:Jnd ÚJn/fU4llt kq..-, 15, 333--355.

Ughtbown, P., & Speda. N. (1993). }{fJU} /anguD.gta orr ll'IJJ"Md. Oxford: Oxford UniYftlllty ~

Ught!oot, O. 0993). Why UG ._.is a k!emiog theory: tn¡gl'nn¡p-erb IDO\'e!Dellt. ln C. Jo- (Ed.), H181.or=i /1ngwstics: Prohkm& and f"'r8pttlwa (pp. 190-214). London: Longman.

LoCoco, V (1987). Le&rner oomprehen.sioo of oral and wnttcn ..,ntenoos in Gennan and Spani¡¡h: The importance of word order. In B. VanPntten, T. Dvorak, & J. F. Loo <Eda.), Fbrei¡¡n languogtleoming: A rtseorch f"'r8ptdtll<! (pp. 11~ 129). Cambridgll, MA: Newbwy Houae.

Long, M. H. (1983). Native speaker/noo·nat.ive speeker oonversotion and the negot.i· ation of oomprebensible input. Appi;,,d Li.n¡¡w.Btic1, 4, 126-141.

Long, M. H. 0991). Fbcua oo forro: a design feature in lan¡¡uage tooclu.ng methodol~ ID'· ln K. de Bot, D. Coste, R. Ginsberg, & C. Kramach <Eds.>. Forttgn úurgu.age ~ lll c:rootculJurolpmp«:twe(pp. 39-ó2). Amaterdam: John Beajamins.

Mac'WlunDe)I B , & Bala. E. <Eda.>. (1989). 1'llL c:rou-luiguuJ1c 111wJy o{ .,llJ#Jla proctM"Yf. Cambnd¡;e: Cambridge Uni'-ersity Preos.

Man¡ubhai, f' (1991). The processing behaviors of adult """"nd lan¡¡uage leamers and the1t relat.ionalup IO seoond langua¡e prolióency App/wd luigu181u:s, 12, 263-297

Maroel, A J. (1983). Conaaous and unoonscious perceptlon Experunenta on viaual tn88king and word reoognition. Co¡pu.tivt ~. 16, 197-237.

McDonald, J. (!~). Tho acqwsition of cue-cate¡ory m<1J>Plnp. In B. MaoWhinnoy & E Ba~ (Eda.J, The croas-1.uigu~ lludy o( ttnrenai pf'OCaBing (pp. 376-396). Cambridgll: Cambridge University Prcoe.

McDonald, J. l., & lleilenmao, L. K. (1992). Chan¡¡tlS in aentenoo pl'OCC8&ing as sec· ond language proflcicncy increases. ln R. J. Harria (Ed.), Cognitiue pl'OWJSing in bülllfluola (pp. 321h'J36). Elsevier.

References 165

McDonald, J, & MacWhinney, B. ( 1989). M8Xlmum likclihood moclels for sentence P""""'•mg. ln B. MacWhinney & E. Bata <Eda.J, The r:rou-1.uiguistJc study of «t1W1« procelWllf <pp. 397-421>. Cambridge: Cambridge Uni\-ersity Preos.

MdAughlin, B l1987l. Thtona o( «<Olid """"""' ocqrmitwn. London: Edward Amold.

McLaughlin. B. <l990J. "Consciou.s" vemaa "unoon.aou.s• learning. TESOL Quorterly. 24, 617~

Muswnea, O. (1989). The abillly o{ ucond longuag< leorrwa to assign tense cu IM aenttna! leve/: A CTOSS·linguiatic stwJy Unpubliahed doctoral thesis, Uruversity or Dlinoia at Urbana-Champrugn.

Nam, E. (1976). Child and adult pen:eptual strotcgie1 tn aecond language aequisition. Paper presented at the 1975 TESOL Convention, Los Angeles.

Nunan, O. (1989). Designing Ta.sks for tire Commwucotwe Cta..room. Cambridgll: Cambridge University Presa.

Odlin, T. (Ed.). (1994). hrsptelwts on Pe@¡Jogu:al Grarnmar. Cambridge: Cambndge University Pres&.

Paulatoo, C. B. (1972). Structural patt.em dnlla. ln ll B Allen & R. N. Campbell lEda.), Ttodúng Eng/uh G1 o otCOnd Úln/IUD#{• (pp. 129-138). New York; McCraw·HilL

~.... A 11983). The WIÚ$ o{ Ú1n/fWJG< GCqU1$r.bon. Cambridge: Cambridge Uoi,-enity Press. ~ A M. l1985l. Language segmentatJon ()peratm¡¡ pnoaple6 Cor the percep­

tion aod aoalysis of language. In O 1. Slobm ( Ed.l, The Cf'OS3·/ingwshc study o( Ú1n/fWJG< ocquisihon, ool. 2: Theomu:oJ wua (pp. 1029-1067). Hillsdale, NJ: Erlbaum.

Pica. T. (1985). Lingu.istic simplÍClty and lenrnability: lmplicat.ions for syllabus desi¡¡n. In K Hylteostatn & M. Piennemann CEd&.J, Mockling and asstsamg socond language acquisition (pp. 137 .. 162. Clevedon, UK: Multilingual Mattcrs.

P1enemann, M. !1987). Psychological con&t.rainta on the teachability of languagcs. ln C. PfaJf CEd.), First OJld second langua¡¡t ocqwalilon pro<:e88tB (pp. 143-168). Cambridge, MA: Newbwy Houae

Pléh, C. (1989). The development of eentence mtcrpretat.ioo io Hungariao. In B. MacWhinney & E. Bates <Eda.>, The crou./1ngu181u: sludy o{ IWJÚVll% proaJJ1· "Yf (pp. 158-184). Cambridge: Cambndgll Uruversity ""-·

Pollock, J. Y. (1989). Verb IDO\>ement. uru-1 ¡nunmar, and the suueture of lP. luigu18Uc lnquiry. 2JJ, 36&--424.

!Uner. M 1, & Petersen, S. E (1990). 1'he attcnbon aystem oí the bumao bram. Anmuil Rev_, o{Neu~na!. 13, ~2

Po.ner, M. l, & Rothbe.rt, M. K. (1992>. AttentJonal mechaníams and ooll8cious expe­rience. In A D. Miloer & M. O. Rugg (Ed.s.l, 1''oundalwll8 o{ Neurop6yChology Sm•• (pp 91-112) N•w Vnrk· Arftd~mir p,._

Posner, M., & Soyder, C. (1975). Attention and oogrulive oontrol. ln R. Solso <Ed.l, lnfomwtWn processing and CO((nilt<>n: Tlu Loyola S)lmposium (pp. 55-&). Hilladale, NJ: Erlbaum.

Radford, A (1990). SynJoc.tic theoryond theacqul8ilion of Engll8h 8'yntox: The noJure o( early child grommars of Engltsh. Oxford: Blacltwell.

Page 92: Input Processing and Grammar Instruction

166 VanPatten

Rebu, A S. l 19761. lmpliat leerning of synthetic la.nguagee; the role o( intltnlCOon­al set Joomal o( f:xpmnunJal l'tycJuJlbgy: Human Ll!anuflll ond Memory, 2, 88-94.

Richarda, J. C., & Rodgers. T. S. ll986l. Approoches and r11dJwcú m ÚJnKU08' t«u:h-111(/: A chscriptitm and analys11. Cambridge: Cambridge Uruveraity Press.

Rutlwñord, W (1987.) Second languo.¡¡e grammar: Leaming and l<'OChin¡¡. London: Longman.

Ruthcñord. W, & Sharwood Smith, M (1988). Conscious0088 raiaing nnd univel1l81 grammar. In W Rutheñonl & M. Shnrwood Srnith CE<ls.l, Grommar ond ...,. ond /~ ttodung lpp. 107-116). New York: Newbury HoU8<!

Senz.Alcalá, C. ll994l. T<2M, mO<k ond tJu. ef(.m o( inpuJ-baatd aplicll tnalruchon. Unpublished doct.oral the8ia. The Uniwrsity of lllinoía at Urbanll-Champaign.

SaVl¡nOn (1983). Comnuuucoh~ ron1/)Nnce: Theory ond duaroom proch«. Reeding. MA: Addison-W.aley

Schmidt, R. ( 19901. The role of ool\IClousncee in seoond 1anguage learnin¡¡. Applied Unguislll.'S, 11, 129-158.

Schmldt, R. (1994). Deoonstructin¡¡ oonsciousness in search of uscful definitions for opplied linguistica. A/LA Review, 11, 11-26.

Schwartz, B. (1993). On exphot and negative data effecting and affecting compe­tence and linguistic behaV10r. Sludin m Second Langua¡¡< Acquiallu>n, 15, 147-163

Sehnker, L. (1972.) lnterlanguage. lnttnUJlu>nal Rai<w o( Applied Úf18Uulu:a, 10, 209-231.

Sharwood-Srruth, M. (1985). From input to mtake: On argumentation m eeoond lan­¡¡uage acquisition. In S. M G&11 & C. G Madden CEds.>. Input m lltOOfld lan· ~ ooqui.sllron (3$4-400). Rowlcy. MA: Newbury House.

Sharwood Srnith, M. (1986). Comprehension versus acquisition: two waya of pro­"""8ing input. Applied ún¡p.tut1c1, 7, 239-274.

Shorwood Srnith, M. (1991). Speaking to many rninds: On the relevunce of dilTerent types of Janguage infonnation for the L2 learner. Second Langua.ge &reorch, 7, llS-132.

Sbarwood Srnith, M. (1993 > Input enhancement in instructed SI.A'. Thooretical i..-. SJUdus in &mnd ~ Atquwbcn, 15, 166-180

Slobm, D. 1, (1966). Grammatical tnllll(onnationr and -teoce comprehension in cluldhood and adulthood Joumal o{ V<rbal únnung & \Wbol &hm:ior, 5, 219-227

Slobm, D. (1973). Cognitive prere<¡Wtitlel for the development o( grrunmar. In C. Ferguson & D. Slobin (Eda.), Sr1tdin o( clu/.d langua¡¡e chut/op~nt (pp. l 7~208). New York: Holt, Rinchru-t, & Winston.

Slobin, D. (1985). Croos-linguistic evidcnoo fer the languQge-making copacity. In D. Slobin (Ed.), Th;, cross./1f1811islk sludy of langua¡¡e acquUi11km, llOI. 2: T/too,.tu;ol i$ru•• (pp. 1157-1249). Hillsdale, NJ: Lowrcn«> Erlboum A.-oc:iates.

Slobm. D., &: Bever, T. G. (1982>. Children U8<! canonical sentence IChernas; A Cl'068-

bn¡uiatic study of .,.ord order and mllections. Ccgrulu>n, 12, 229-265. Speda. N., & Lightbown, P ( 19931. lnauuction and the dl!'oelopment of questions in

L2 classrooms. Studla U1 Sccond Language Acquisrtwn, 15, 205-224

References 167

Swa.m. M. ( 19851. Commurucative oompeteooe. Sorne rolee of oompttbensible mput and oompre~ble output m its de-oelopmcnt In S. M Gasa & C. G. Madden (Eds.), Input"' BtC()nd language ooquu1twn !235-253>. Rowley, MA: Newbwy Hout1e.

Swain, M., & Lapkin, S. (1989). Canadian unmera1on and adult seoond lan¡¡uage teaching: What'e the oonnection? Mochrn f.,a.nlflUJKC J01muú, 75, 150-159.

Terrell. T. D. (1991). The role of grammar instruction m a communicative approoch. Mockrn Long1t0(/t Joumal, 75, 52-63.

Todd, L. (1974.) Ptdginr ond Crt.aks. Boeton: RouUC<l¡e. Tomlin, R., & Villa. V 11994). Artention m cogrutive 1C1ence and seoond lan¡uage

acqwsruon. Sludia m &mnd Langua¡¡< Acquuulwn. 16, 183-204. Towell, R., & Hawkina, D (1994). Approoclia 11> ll«Olld ~ ooquiallu>n

Cle\-OOOn. l,;K Multiungu81 Matters. Trabe)\ M., & Wlute, L. ( 1993). Positi'oe eviden<e and preemption in the seoond lan·

guage classroom Studia rn &mnd Langr10(/• Acquwhon, 15, 181-204. van Naerssen, M. (1981). Generalizing seoong languep hypotheses - lan·

gu&gl'8: A test C88ll in Spe.nish as a seoond longu•ll'l· Unpublished Ph.D. thc­sis, the Univeraity of Southem California.

Van.Patten, B. (1984a). Proceesing strategiea and morpheme aa¡uisition. In F. R Eckman, l. H Bell, &: D. Nelson fEds.). Uniwrsah ofsecond Jangua¡¡• acqu,. IUhon (pp. 88-$8). Rowley. MA: Newbury HoW18.

\'anP&tten, B 11984b>. Leemen' oomprehen.i.on of d1tic pronouns: More O'Vldence for a wonl onk·rotralegy. Hi3paruc únguu.Un, 1, 57-67.

VanPauen. B U986al. The acquisition of oer and eow by edult leamers ofSpruuah A prelmww:y invemtigation oftnnsitional stagut of oompetence. Hisponw, 68, 399-406.

VanPatten. B. (1985b). Communicative value and mfonnat.ion proce6Sing in seoond language aa¡uisition. In E. Judd, P. Nelson, & D. M.-n;chmittz (Eds.), On TESOL '84: A broue new wor/d. Washington, OC: TESOL, 88-99.

VanPatten, B. (1987a). Cl&IW'OOm learner's acquiaition of eer and estar: Aroountm¡¡ for developmental pntterns. In B. VanPatten, T Dvorak, & J. F. Lee (EdA. ), Fomgn langu"lf< ltomif18: A raeorch penpecl1~ (pp. 61-75). Cambndge, MA: Newbury HoU8<!.

Vanl'att.en. B U987bJ Claalroom and naturaliáJc: ~ acquisition: A oomp..-í· son of two ca&1> otudies m the ecquisttion of chtic pronouns in Spwúah; Decticated to Joeeph H. Matluck. In T. A Morp.n, J F. Lee, & B. VanPatten (Eds.), Úllllf""# ond /anguoge use: Slud~• m SpanWi (pp. 241-262). Lanham, MD: University l'l'e68 of Americe.

VanPatten, B. (1988). llow juriee get hung: problema with the evidence for a focur on form in t<.-achin¡¡. lAngrsage Learnuig, 38. 243-260.

Van.Pal.ten, B. (1990). Att.cnding to oontent and form in the input: An experiment in wrutcivwi11mb- &ucl~a "' &.YJild Le.u~ kq1,Uit1on, 12~ 287-301.

VanPatten, B U99U. Gnunmar instruction and input proces&in¡¡. Paper presented al the &pECial oolloqwum on the Role of Grammnr lnstruc:tion in Commurucnt.iw ~ -r-lun¡¡. Concordia Universty and McGill Unn"1'SÍt)I Montreal

VanPatten. B 09931. Gra.mmar teoching for the acqwatuon-rich classroom. Fomgn Langu'* Annah. 26, 435-450.

Page 93: Input Processing and Grammar Instruction

168 VanPatten

Van.Patten, B. ( 1994 ). Evaluating the role of oonacio1U1nees in seoond language aoqui· sition: Terma, linguaatic featurea and rceearch methodology. Al LA &vuw, 11, 27-36.

VanPatten, B. U995). Cognítive llllpeCta of input pl'OOOlllU'lg uueoond language aa¡ui­sitíon. In P Heshemipour, l . Maldonado, & ~t van Nael"8S<'n <Eda.), Futschrif! for Trocy DauJd Trrnll !pp. 170-183). New York: McGraw-Hill.

VanPatt.en. B. & Cadiemo, T (1993). Expliat inatrucllon and input proce$Sing. Studia 111 &cond Lon~ At:quuúwn, 15, 226-243.

VanPatt.en. B , Lee, J F,& Ballman, T. L. U9961. i&buJ11t/UI! .. i <2nd edl. New York: McGrawHill

VanPatten, B. Lee, J F. Glua, W R, & Binkowalci, O. O (1992). Manual ID AtaJmpany iSo.buu qUI! .• 1 New York; McG.,,w Hill.

VanPatt.en. B .. & Oikkerl<)n. S. 11996). 'The c:auatíve vanable. an p,........;ng itJslruc. tion: El<planaúon vs. otructured anput activitJe.. Stucúo rn &cond Longuagt ÑqUUÚIOll.

VanPatten, B., & Sanz, C 119961. From Input to Out.put: Procee8lng lnsuuction and Cammunicetive Taska. In F Eclunan, O. Hi¡hland, P. W Lee, J. Mileham. R R Web« (Eda.l. &cond LanguQ/lt At:quwllOn and Pwlagogy (pa¡¡l!S unavwl­able). Hillsdale, NJ Erlbaum

White, L. (1987). Agrunst oomprehenaible input.: The input hypothe&s and the devel· cpment of seoond-language oompctcnoe. Applitd LlnguUihc8, 8, 95-110.

White, L. (1989). Urnutnal J(1'Dmmarand ucond longulJ8e ocquisitwn. Amsterdam: Bel\Í8mll18.

White, L. (1991). Adverb pl-..ent in 8'lCOnd language acqui&t.ion: Sorne effect.s of positive and negat.ive evidence in the clMaroom. S=nd Language Research, 7, 133-161.

White, L. (1992). Long and short verb movement in 11000nd IBI18Uage aoquisition. Conadion Journol of ünguislics, 37, 27$-286.

White, L., Speda, N., Lightbown, P., & Rante, L. (1991). Input enhancement and L2 question fonnation. App/itd LinguialiC8, 12, 416-432.

Whittleeea, B. W. A, & Dorkcn, M. O. (1993). lncidcntally, thingii in general are par· ticularly dctcrmined: An epieodo-proceMing 8CCOUnt of implicit leaming. Journol of &p.n-nlaJ PlycMlogy: General, 12, 227-248.

W'ickeM, C. O. (1984). Proceeeinr reeoun:ee in attention. In R. Parasurman & O. R Oavi.,. (Eds.), ""1ttin of alttnlwn. New York: Acaclemic Pre.a.

Wmitz, H. <Ed.). 0981) Tk oompl't'htna1on oppl'OOéh ID {otTign /.anguage ooquis•· ti.on. Rowley, MA: Newbury llouoc.

Wolfe Quintero, K (1992). Leamability and tho 8"QUimt>on of extraction in relatn-e clauee. and wh- QUl!8tl<>lll Studia m S«ond langulJ8e Acqw.sitwn, 14, 39-71.

Wong-Fillmore, L. 09761. TM MOnd ¡,_ IJIT>Ulld. Unpubliohed Ph. O. thesis, t:C Berkeley

Zobl, H. Cm. s.l. Lexic:al-funeuonal modulanty 111 the lexíoon and the interface debate in L2 pec:lagogy. Unpubhahed manuacnpt.

Zobl, H. (m.s.). Critic:al rellectio111 on the .,...,¡, interface hypttb.i• Carleton UIUverllllY. Canada.

AUTHOR INDEX

A And<nen. R. 50. 142. 161

e llellman, T.L 168 BN'doYl·Harllr. K. 22. 131, 161 lla1'1t, E 35,39,51, 161, llifi llavui, E L 31, 161 llome,J 28.161 8ony. D 46, 162 lleveT, T C. 33, 36, 39, /6/, 166 Bmkowlki, D D 2-4. 60, 72,89, 90, 161, 168 Blau. E. K. 28, 161 BnlnOdoñer, R. 26, 161 B.-lbent, D 46, 161 Butterlleld, E 15, 164 Bybeo, J. 10, /6/

e Cad:lemo, T. 2, 14, 2'.l., 601 63, 64, 65. 661 67,

71, 72,83,94,96,9?,98, 100,101, 102, 121, 122, 123, 124, 121.130, 152, 162.168

CarUon, R. A. 45, /62 Carpenter, P. A 16, 163 CarT, '[ H 45, 46, 162 Carft!l. P. 20, 162 c-lli, c. 35, 166 e.a. s J. 44, /62 Oieng, A. 2, 88, 112. 115, 118, 119, 120. 127,

152, /62 O>omol<)I N 11, 162 Cord<r. s Pi~ 7. 162 Qinsn, '[ 45, 46, 162

D 0.-,A.35, 161 ~o 45,162 o.en... z 46, 162 Dorbn, M O 46, 168

DulanY. D. 45, /62

E r.u.., R. 1, 5, 13, 87, 122., 129, 150, 151, 162 Ervu>-Tnpp. s. M. 33, /62 Eubonk, L 129, 130, 162

F Foorch, C. 17, 162 FUUlel11MJl, M. O. 113, 163

o Garrett, N. 47, 163 º-s M. 1, 13. 35, 37, 39, 51. 89, 129, 142.

163 0111111. W,R. 22. 60, 72, 90, 162, 163, 168 Gllilan, E. 34, 163 Gt-eg, K. 129, /62 Guntermann, G. 113, 163

B Harloy, B 87, /63 Hat<h. E. 32, 163 HawkiruJ, D. 129, 130. 133, 136. 138. 140,

141>, 167 Heilenman, L. K 35, 39. 51.164 Howe. M J. A 44, 162 Hulk. A 163 Hulllvn. J. H. 37, 163 Hy\t.taut.am. K. 60, 163

1 1-londeo, o c. 37, 163

JUll. M. A 16, 163

K KMper. G 11, 1621 Kloui, w 17, 19, 163

169

------~!i. .

Page 94: Input Processing and Grammar Instruction

170 Author lndex

Krashen, S.D. 5, 10, 59, 82, 83, 121, 163, 164

L Lachman, J. 15, 164 Lachman, R 15, 164 Lapbn, s. 17, 167 Lareen-l"reernan, o 5, 13, 24, 87, 150, /64 Loe, J f'. 2, 22, 34, 69, 60, 66, 70, 72, 89. 90.

114, /62, 164, 168 Leo"\ R 27, 164 UChlhawn. p 87, 91, 127, 158, 164. 166. 168 ligjltfoot, D. 150, 164 l<>C<xa, V 34, 89, 164 Lcng. M. H 5, 10, 13, 87, 150. 164

M MacWlunney. B 35, 39, 51, 52, 161, 164. 165 ~e.o 13,163 l\laJi¡ublw. F 19, 20, 164 MaroeL A. J. 44, 164 Mclb>•ld, J. L. 35, 39, 51, 52, 164.165 McLau¡lilin, B. 47, 151, 165 Muswneci, O 23, 165

N Nam, E. 33, 165 Natalo, F. 35, 161 Nunan, d. 59, 165

o Odlin, T. 1, 142, 165 Oikkenon,•.2, 121, 126, 126,/68

p Paulst.on, C. B. 3, 5 7, 165 ""'-A. M. 15, 17, 19, 47, 43, 165 ~.s. E. 44, 165 Pica, 'J: 22, 165 Pienemann, M 146, 165 Pléh, C. 35, 38, 39, 165 fl>llock. J 'l 131, 165 Pooner. M. 1 44, 165

R Radford. A. 19, 131, 165 Ranta. L. 127, 168 Rebe.; A. s. 45, 166 Ricbanls. J c. 2, 83, 87, /66 Rod¡;ors, 'J:5. 2, 83, 87, 166

Rot.hbort, M K 44, 165 Rutherford, w l. 84, 85, 166

s Sani, c. 2, sa. 97, 1()2, 106. 109, 110, 111,

123, 168 Sala-Akalá, C. IC)4. J66 Sovi¡oon. s. 3, 69, 166 Schochter, J L 163 Sdumdt. R 16- 29, 44. 166 Schworu, 8 5. 9, 10, 148, /66 Sebnbr; L 9. 129, 142. 163, 166 Sharwood Srmth. U 7, 17, 18, 84, 151, /66 ~'t31.161 Slobon. o 1 15, 33. 39. 47, 146, 166 Snydor: c. 44. 165 S¡ieda. N 87, 91, 127, 158.164, 166. 168 Swoin ~t 17, 87, 163, 167

T T.....U. T O 59, 83, 164, 167 Todd, L. 22, 167 Tornlin, R. 16, 44, 167 TowelL R 129, 130, 133, 136, 138, 140, 145,

167 Trahey, M 148, 167

V van Nacn.en. M. 24,167 VanPatten, B. 2. 14, 20, 22, 23 ,24, 26, 28,

34, 43,44,45,57, 78,59,60,63,64, 66,66,67,69, 70, 71,72,88,90,94, 95, 97, 98, 102, 103, 106, 109, 110, 111, 113, 114, 121, 122, 123, 124, 125, 126, 127, 130, 142, 152, 162, 164, 167, 168

Vcrua, V 35, 161 Villa, V 16, 44, 167

w Wlu~. L. 5. 11, 50, 127, 129, 130, 133, 142,

148, 167, 168 WlutU-, B. \\1A.46-168 Wtdleoa C. O 16- 168 Wuuu, H 83. 168 Wolfo Qwnt«o. K 146- 168 Wonc·Nlmore, L. 17, /68

z l.obl. ll 136, 168

SUBJECT INDEX

A o, penonal "'Sp&ru•h. 61, 76 A«ommodalion, 133-134 AcculrurallOn Model, 53 Acoult>C-llOn, 139-140 AcwálC llhency Stt ~ual oohtncy AcwálC ltrwo, 26-27, 32. 4~. 62

and ..,.,.. an Sponuah, 97 -98 A<qu..,t.aon ord.,. oC En¡!Jth VOl'b mor·

ph•m•,29 Active ttnP11menL pnnciple. 68 Act1vill01. Stt oúo Proc.lan¡ anlltructlon

alToctiv•ly ONntod, 65, 114 referentW\lly ontnted, 85

A<l¡ectave *il'O"mon~ 40 Mjoctivo1 In Sponaoh

inftection oí, 25, 6~ 71 ,., and nlar with, 112-113

Adulr.a, bios toward Ll notaVHpeoken in att.enll(\nal rMMJ"Ch, 4;4-46

Advorbt,69 in French vo. En¡lilh. 148-149 temporol, 22-23, 98-99, 149

All'ec:üvely oriento<! octlvitiN, 65, 114 Stt olt0 Procettins inttruction

Agen<y/rolo au1gnment, 32-43 Stt olio Ro1.-.,.n1 ... 1gnment

Anunaey, 3W6, 37 ANOVA.o

ob}ocl pronouna o.nd word ord•r in Sponllh, 93-96

prec.mt ....... In Sp&ruah, 99-101 ................. , , .. dllCOW'll-i.--01 tuka.

107· l l l .. , and ,...,. "' Sponllh. 116-120 structund input vo. explanataon and

mon1tonnc. 124·126 ArudM ~l'WUI ~llON, 26 At<enW>n. 69

cl<cecbon aa part o( UH 7 eft'ortful natin o(, 16 tofonn. 47 Ulputp_,.and, IS-17 "' LI acqwsu-. 15-16 to meerung, 47 pousm¡ and, 27 eelective, 15 IUllpbñaWon oC input and, 27-28 aunultaneous, 2f>..27 !une p.......,., and, 27-28

Attentional capacity, 27-28, 30, 46-47 Audtolln¡w>I approech, 2-3 Awolwy·awúliazy verbl, 143 AwareneM, 43-47

B Bcl\aviorism, 3 a ...

instructor toward opproachea, 91 -.92 t.oward cu.., 35

Bottom·up P'°"""Ín& 122

e Canoni<al Order Stntegy, 146-147 ea.e marlang, 3S-40, 38-42, 52 Clanfimbon requeot.5, 10 Cocn>•" .. approechel, s. 9, 44, 145-148 eocruuve - oC inpul -ing.

14-32. Stt oúo Attention; lnpu< ~ oeumg; Meaning

Cogrutivo codo le&rru.ng, 3 Comm\lllQtlvedas&room COD<Jept. ~9. IM Commurucabve nlue

o( copular ...t.. 113 fonn oftittle ar no, 27-32 relaUve. 23-27

Competrtion ModeL 3S, 39, 51-63

171

Page 95: Input Processing and Grammar Instruction

172 Sltliect lndex

Complexity, 27, 31 Comprehen1ion, 16, 21, 30, 32-33, 69 Sr.

ahoMeeninc Comprehen11on baood app.-h .. va

.,._ uutruct1on, 112-86 ConJbct vabcht)l 51-.52 Conlciouso- 43-47 Consoousn.o rauOn¡ approoch, 84-85 Coateot-bued inaructi<in. 83 Coatent ........ 18-19, 24-2.5 Copular """'" , .. , and,.,.,, '· 112-126 Q-. 35, 39. 0-41 , 145 Cue valicbt)I 52 Cwnulallve dcvelopmont. 146

D ~30.37

at<MllOnal-.- and, 16 compeubon ol fonn a.nd meerun¡ íoi; 18 de6ned, 16

OeYelopmg lyllem, 7~. 31-J2, 39, 4 l, 60, 56,96,139, 147, 1153

defined. M, 9 Di.- Method (Gouin "'""">. 2-J Oiroct object, 38-42 Di.- objoct pronouna, 57-69. S.. aho

Objoct p""""1la Diocoune-level n. 11ntence-lavel taaka'

empirícal .-an:h, 102-112 Discourae \'l. oentent11ll level lnput. 68-69 Do as dummy vorl>, 160-152 Duteh l!UbjOCU, 146-147

E Empirical Jtudi.._ Stt Reaeatch Empty cate¡on .. , 137-138 Encoding, gnunmalleal and loxieal, 21·23 Englúlh, 25

.. lirst languagoe, 35, 60

role N1ignment "'· 33-12 .. oecond languagoe, 33-35, 37, 39-40 wrb raWnc and, 14iH50

EnNnmmen1

nopuve, 84 poSli..l'\"e,84

- <Sparud. oopular ....i.1, 112 130 &.nt pn>behútty. 36-37 Expla.nation va_ 1truttured anput e-mpln•

cal:ctudyofmon1tonn1o 121 .. 126 Expleavo ~. 136-137 Eipbot UIStnJ<tJon, 9 Ezpbot n. unpliot lcnowledce. 85

Extraction and Segmentat.ion heurutics, 48-49

F FemU1U1ow'muculm• forma, 71 f\ruta llata grammar otu<b .....

F\nt lan¡uage acquwuon. s.. Ll -·­

""º F\nt lan&uaee t.ranafer, 50-51 and Uru~raal Crammar, 129-130,

142-145 f\nt noun ltralAC)', 14, 41, 62-63. 69, 70,

89-90. S.. oáo Wonl order dofined, 32

111 rolo_ .. 33-42 u uru......i doíault, 35-37

Focuoon form, 9-10, 84, 91, 156 Focuo on m.mng. 67-68 torm

-tUlnpnllnUIOD,56 oommu.n.K:llbve value ancl. 23-27 definod. S.. Form foeu1 on, 9-10, 84, 91, 156 or btt&e or no commu.nitative vaJue.

27-32 venua mea.rung, 47 modo! of pro<e00mg, 30f precedenoe oí méBnln( over, 17

fonnal lingu.Lstict VI. cocnitive account6, 146-148

Fonn·meruun¡ rnapplng, 10, 30-31, 44-45, 46,86

French •• fint languap, 39, 148-149,

160-161 JexjcaJ verb moveme:nt in. 149 ob,iect pronOW111D, 138-140 u ...... d '-23, 34, 39-4-0,

132-133, 146-147 11\Jbiect P""'°""" in, 139 word order 111, 139-140

Fl/,ncllonaJ CltegonM, 151-153

o C..onhzauon. 146 c....n.u ...... ._ 146 C«mai>

u -..nd !an¡ua¡e, 19, 34-35 ....... order ocqwcboo m<Jdel ol

l'lenemonn, l46-l47

C-0 - ; Duect Methodi, 2-J eo-·...,,._,,_bmdmg theory. 136

Grammar finita ltata stuche1, 44 YI. synw, 11

Gramma.r UWb'ucltOn learrung va pnlctJoo <Paultt.on'• t.uon<>­

myl, 3-4, 57 lnlchti<Jnal app~ 111, ~ S.. aho

Tnidluonal oppr-i.. Crammar tranolauon. 2 CrammotJeai __ _

CrammatJeal eneochna, 21-23 Crammat>eal form S.. Fonn CrammatJeai role -nt, 32-43 Stt

aUo R<>lel...,,t-t GuMor. ~ with. 89

e ffierarducal modAtl ol P.ultton, 3-4. 57 Hindi .. aeoond ian¡uap, 1~20 Hunp.nan • fim lanclale. 36. 33-39

1 Idea untta, 20-21 lmptiat VI espltat ltMwladp, 85 Indirect1onal tl'&lllitlve vsrbo. 137-138 InRection,21,25, 26,29,34-35,38-J9 Informln¡ leemen about pr-.ng otrute-

g1ea, 62~ ·1ng endinp, 24-25, 2'9 Input

compered with intako, 7 deRned,6, 10 interactaon wtth Univeru.l Gro.minar, 183 moaning-boarinr. 6-8, 17, 68 oral and writr.en requll'41d, 68 role oí,~ Mnlenúal VI. dllOOUne leve!, 68 11mpblloaUon and au..nUon, 27-28 otructured, 6 VI. OUtpu~ 69

Input •nhanoem .. t ·~· 84 Input Roodtn& 148-149 Input HYJIO(h-. ~ lnput-

cocniu-.upecta cU.tontlon, 14 17 rorm of bttle eommuniaou.., varu.,

27-30 1o><al and pvamatxal anoochn¡.

21-23 ._,.Ul(, 17-21 relauve eommunac&UYf vaJue, 23-27

Subject lndex 173

oummary; 30-.12 for communication venua eoquilition,

17-18 oonduaont, 53 definad, 10 and 111take, ~. 56 -Compotibon Model, 51-.53

DOlllOOUlll- and ._ 43-17 fint lan¡¡uace tnmfer, 60-61 form Wl"IUI meen1ng 47 OperalUl( Prinaples, 37-60

... ...._leYel upecto and role --ment

at!Mtioa and ddec:tioo, 37-38 -~33-39 ch!T.....,_ ID 2L leamen, 3S-4C)

ewnt probability. 36-.'!7 fint·noun .uat.ec:r. 32-36 gramm•hMI dues. 39-42 pnnaplto, 32

IWlllnary. 42--43 lnotnicbon, expbcít, 9 lnlt.ructon. S.. Teaduiro lnta.l<e.~. 13,31

definecl 10 and Univenal Crammar; 133 VI. intemalizod lan-, 10

Interlangua¡e. S.. Developing l)'ltem lntemaliutlon, 3 lntemali.&ed langua¡e vs. intake, 10 lntemally derived bypoth-, 133. S.. alao

Univenal Grammar lnv.,..ion,l!Ubject,.veri), 131-132,14Z-143 !tallan, 25

u ftnt languagoe, 35 &I IOCODd language, 23, 35, 37

L Ll acquJStion

-lopmental models. 39, 146-147 ftnt DOUft áTatOfY in, 33 on• and two-word - 18-19

L2 aoquiabon Compebtion Model and, 51-.52 C\IOll m. 40 --41

lint-ianguoi¡e trmáer and, 50-61 input and. 4-.5 topul p-UI, 13-.53. s.. aUo Input

~""" """""'""' """'­.. mobvaboo for proceumg instru.ction, 56

Page 96: Input Processing and Grammar Instruction

17 4 Subiect lndex

reoearch bias toward adults, 44-45 role assigrunent in, 39-42 scarcity of research on, 3

Leerner'• grammar. S.. Oeveloping system Learning st.ratogieo vs. prooossing •traregy,

LO Leaming vs. anput proc ... ing, 48 Letters, string ordermg o~ 44-45 Lexica! encoding, 21-23 Lexieal priming &tudiea, 44 Limit.od p"""'6<ing capaaty, 45, 53 Llngui>tic oompetence. Su Developing sy&-

tem Long-term memory, 31

M Mspping. Set Fonn-meaning mapping Masculinwfeminine fonns, 71 Mooning

agency/role 8i8Ígnl1lent and, 34-42 oommunic:ative value and, 23-27 detéction of versus form. 18 focu. on. 67...Q! versua fonn, 47 as goel of inpul processing, l7-21 lexiceJ and grammaticeJ encoding fo<

21-23 prteedence OVér fonn, 17 push for, 135

Meaning-bearing input, 5-6, 17, 68 Mental representation dellned, 10 Monitoring, 133

empiricel study, 121-126 Monitor Model, 53

N Natural Approoch, 82-83

empiricel •tudy on objec:t pronouns and word order in Spanish, 88-97

Negationlnegators, 142, l49-150 Noun.s, assigrunent of role to. &e

Rol<188"nt assign.ment Null objeda, l38 Null oubject, 89

o Object of verb. Su also Object pronouns;

Role/agent osoignment

sample '""°""· ~2 Object pro.nouu

in French, 138-l40 sample l...,,n, 71-81

and word order in Spanish: empiricaJ studies, 88-97

One-thing·at·a·time concept, 67 Operating Principies, 17, 47-<IO Oral vs. written taSks, 107-l12 Organization of text, 8-9 Output

inta.ke as determining, 22 as proces&ing indicator, 19 vs. input, 59

Out.pul practice, 6

p Panunoter settinll"re&etting, 136-137. See

al.so Universal Grammar Panimony in uplanation a.rgument.. 145 Pass1ves, 33-34, 36-37 Potient. Set Oirect oqject; Role/agent

assjgnment Paul$ton'a taxonomy ofpractioe type&,

3-4,57 ro._49 Perception, 8Ubliminal, 15-16 Perceptual &eliency, 31-32, 49-60, 84, 140,

146 Per.onol a in Spanish, 61, 76 Pienemann '• model of Germao word ordér

""'!uísition, 146-147 PLO (primary linguú¡tic datsl. Set Input Polish as ñrst langua¡¡e, 27 Position in eeotence. Stt Word order Prnctice

out.put, 6 Paulston's taxonomy oftypeo, 3-4.57

Pre-emption. 146 Prepositiona, 26, 151 i'repo&ition stnu>ding, 137-140 Preterit tense: empiriceJ study. 97-102 Primacy linguistic dats (PLO). Set Input Proceosing. Set Input pro<e6Sing Processing in&truction

compered with other appr<>a<:he& comprehension·based, 82-84 consciou.snesa raising, 84-85 input enhanoement, 84 Lnu:litional, $

oomponents oí explanation with sample lesson, 60-62 infonnation about sb'ategy o.nd samplc

, ...... 6~ structured input and sample lesson,

63-82

1

de6ned, 2 input processing and int.ake, 6-8 motivation for, 2-8

direct object pronouns and. 67-69 problem..g 'N'ith traditional approeches,

56-57 second lan¡uaae acquisition in genera],

66 role of input in, 4-6 sumrnary/conclu~on, 86 tenninology, 9-ll

Procesaing stcategiee. iNonning learners about, 62-68

Proceosing •trategy defined, 10 Procesa in.st.ruction

and Universal Grammar, 147 empirical reoean:h on, 88-12S. See aJso

Researc:h: empiricel stumes issues and challengeo

oognitive ac:counts vs. forme! linguis­ti<>I. 145-148

dilTerentiel with traditionol in.struc-tion, 153-154

functioruil categories, 152-153 aynt.ax, 148-152 Umve.....i Grammar, 130-145

validity of concept, 147 ~n,24 Pl'lljp1!Slive tense, 112-l13 Pronouns

direct objec:t, 57-59 expletive, 136-137 object, 71-81, 88-97 oQject and refteicive in Speni&h, 40-41 quasi·argument, 136-137 reíerentiol, 136-137 oubje<t, 61, 139

Psycholínguistic approach, 7, 10 Puerto Rican rubjects, 27 Push for mea.ning, 135

Q Quasi-argument pronoo.n.s, 136-137 Queotio113, clarification re-. 10

R Recall prot.ocols. 20 Redundancy, 40, 69, 76, 152

and communic;ative vaJue. 24-27 Referentíally orientad a<tivities, 65. See

also P~g instruction Referential prono'""" 136-137

St.biect lndex 175

Requests, clariñeation, 10 Research

assessrnent limit.ations in, 1~104 empiriceJ studies

comrnunicative output.: aentence-leve1 vs. discourse-level tasks, 102-112

leicicel-aspectual items: s<r and estar. llS-121

mon.itoring: ~anation, structured input, combination, 121-126

synt.ax: objec:t pronoun$ and word order in Spanish, 88-97

verb inllections: Spanish past tense, 97-102

scarcity of L2 studi.., 13 Restructuring, 133-134 Role/agent ""8ignment, 60-51

ottention and detection, 37-38 case marking, 33-39

s

clifferenc:es in L2 leamers, 38-40 gnunmatic:al c:uea, 39-42 sumnwy, 42-43

Seliency, perceptual. 31-32, 49-ó-O, 84, 140, 146

Sécond langu88" ac:quU.ition. See L2 ac:quj· sition

Segmentation and Extraction heurlstics, 48-49

Seléctive attention, 15 Semanti< role ... ignment, 32-43. Set ol8o

Ro!e/agent assignment Semantk w.. communicative valué, 25 Sentence-1eve1 vs. disoourae-level ta.sks:

empiric;a! research, 102-112 Sentential vs. di~ level input, 68-69 Ser (Spanish copular verb), 112-120 Short~term memory, 140 Simplification and attention, 27-28 Simultsneous ottention task, 2&-27 4 8 inftection, 25 Spanish, 25

adjective infl.ection in, 69-71 copular verbo ser and u/ar in, 112-120 ompirical ctudio&

communicative output: '6etltenc:e-level vs. discours&-level ta.sks, l02-112

lexiceJ-nspectual items: ser and estar; 113-121

monit.oring: expl.a.nation, structured input, combination, 121-126

Page 97: Input Processing and Grammar Instruction

176 Subiect lndex

eyntu ob)OCt pn>nOllN and word arder U\ Spanllh. SS-97

veri> anllecUona. SpenlOh f'l$l ---. 97-102

u finl lonawiP. 27. 50 ob¡oct and reftun'9 pronouns in, 40-41 .,..._...¡. U\,61, 76 p-l - ID, 97-102 .. _,.,.¡ ~. 20, 22, 23, 26, 34,

142-143. Stt alto~ instnic­uon

word OrOOr Ín, 34 Speed ond allenhonal capacity; 27 Streee, oroullic, 26-27, 32, 49-60, 52, 97-98 Strin¡¡ ordenn¡¡ of lctters, 44-45 &ructural hn¡ulttic:o, 3 Struclurtd achv11.y v1 exph1.n•tion:

emp1r1cal 1tudy or mon1tor1ng. 121-126

Structwed U\pul, 6 ~ for U\IUUCt.on. 66-70 O•tcw .. •. 63-66 oamplo 1-ona, 71-81

SUbconlóouo/albhminal leenung -15-16

Suluect Ser """ Rolela¡ont ....,,,,., .. , nuU.89

Subject pronouna, 61 in l'reocb, 139

Subjoct-verb a¡¡recmcnt, 39-40, 61-ó2 Subjec:t-vcrb anvenion, 131-132, 142-143 Syllabicity, 81-32 Synt.ex, 143-162

deflned, ll and P""*"llli 1Ntnic:tion, 143-152

T Talkm¡ at vtrlUI with, 68 Tuk farrubanty, 162-15 T.chon

attJtU<loo -""' conooow. 1ul>cc>1•••JU1 knowledp,85

biu towvd _ _._ 91 em>,_., _,,.pllOOI o( 83

Tempor'1 odverl>I, 22-23 TM•. '1-?~. RQ, '7R

anddo, 152 ¡>mcnt In Span,.h: empincal study,

97-102 prosr-ivo, 112-113 verb ipRection1 m Sponilh past, 97-102

Tente cuu, 63

T..,... marun, :z.:>.26 Tenrunolof)', 9-11 T'"' orpru.z.i-. 5-9 Ti..ll>ltnlellOn 111-. 95-96 Time_.-., 27-26 Total Phy.-1 ~ approach, 19-20,

82-83 Tnidiuonal approkhet

c::ompared w1th erammatleal conteiou¡.­

neot ral.in¡, 85 comparilon w¡lh pl"OC8111ng 1nstruc:tion.

Ser RaMrch overvlew of, 2-8 problen11 with U\Cludi.na 1an1ple lcuon.

66-69 Trenaftt, flm.lan¡ua¡o, ~l. 129-1.10,

141-145

Trena1Wlnll """"· 56 Turlwh u 6nt lan¡w¡e, 39, 48

u U~146 Uon....J deí.Wt ..,._ 6nt"""" oo,

35-37 UnMnal Gnunmar. 9, 41, 53

bock¡¡round and pnnctpleo, 129 6nt 1&nguagw tranlfer and, 129-130,

141-145 and fim vo. aecond lan¡ua¡e acquliltlOn.

130-141 interection w1th input VI. intake, 133,

136, 140 Utteranc:o po11tions. 32

V Verb endinp, 76, 89 Verb morphemeo, 23-24

devolopmoni.I model U\ Englllh. 29 \'eri> ~ Stt o....ct obJOCt pcollOWll;

Ol!l«t or -1> ven."""""' 131. 133. 142-143. 149 v.n.

_ .... -- íor, 113 awahaey-awuliat)\ 143 oopu1.v (1<r and .. 1ar1, 112-126 dn 1111 dummy \lf"rh. lM-1~2 a.ndirect1onal tn.nltttve, 137-138 lezical v1. awuhary ond modal, 131- 132

Verb-subject o¡¡roernent, 811-40, 61-62

w Walpiri ea lirll languago, 37

'

Wolfe Qumttt0'• modo! of WIH¡ue1l1on formatloo, 145-147

Word endmp, 48 Wordorder

and lint """'- traosfer; 00-61 U\ Freoch, 1~140 U\ Spaniah: emporical studieo, SS-97

Sp.nW\ "" En¡b¡h. 64 Wor\:in¡¡ memory. 16 1,1.'ntt~ .. oral i.b, 107-112

X X-bar n-y, 131