1991 crookes ssla

7/29/2019 1991 Crookes SSLA

1/19

SSU, 13, 113-132. Printed intheUnited States ofAmerica.

SECOND LANGUAGE SPEECHPRODUCTION RESEARCHA Methodologically Oriented Review

Graham CrookesUniversity ofHawaii, Manoa

RecentSLA theory development, supported by related developments incognitive psychology, has made the study of SL speech production,hitherto neglected, a promising area of work. Recent developments inL 1 pr oducti on studies have pr ovided a gradual ly strengtheningfoundation for investigations of L2 production with both use andacquisitional concerns. This article briefly sketches the current firstlanguageposition as a necessary preliminary to a critical discussion ofrecent SL production research with particular regard to methodology.

Research intospeech production lacks visibility within the overall psycholinguisticsliterature (Levell, 1989). Aspart of the language-processing field, whichcovers bothproduction and comprehension research, it "is inherently interdisciplinary, [ s o ~ e -search reports tend to be scatteredamong the various journalsserving psychology,linguistics, computerscience, and behavioral neurology" (Swinney & Fodor, 1989, p.1). Despite somerecent developments (ct. Jarvella& Deutsch, 1987), comprehensionand production are usually considered separately, with studiesof languagecomprehensionand linguistic competencegreatlyoutnumberingstudies of production. Forexample,in a representative recent collection of language-processing papers (Swin-ney & Fodor, 1989), only one-tenth of the studies were of production. Similarly,secondlanguage (SL/L2) production studiesare lesscommonthan studiesof comprehension, withina smalleroverall bodyof work.fnaddition, thisresearchsubfield hasI amgrateful toGabi Kasper, Eric Kellerman, and Riidiger Grotjahn for comments on anearlier version ofthisarticle, and toMary-Jane Henning for connectionist references. This article does not consider the writtenmodality. Inwhat follows, terms such as second language production, and so forth, should betaken torefer tooral language alone.

1991 Cambridge UniversityPress0272-2631/91 $5.00 + .00 113


2/19

114 GrahamCrookesnot found a safehavenin anysingleapplied linguistics journal.(Such institutionaliza-

tion facilitates cumulative work, and often characterizes the establishment of a productive suhfield in an academic area.) SL production studies seem more likety tooccur ascollections of conference presentations, which canbeslowtoappear inbookform. Increasingly variedmethods of investigation are beingemptoyed, yet usuallythe onty types of documentthat afford the spacenecessaryfor adequately detaitedpresentation, defense, or discussion of arguments concerningresearch methodotogyare the dissertation and the technical report.Neglect ofsecond language production is tikely to continueuntess there is steadypressureto assertits importance and tomakethe somewhat scatteredfindings generally accessible. The future looks promising, however, as theoretical discussions thatprovide a rote for performance in SL tearning are increasingly availabte (e.g., Ellis,1985, 1988; Mahle & Raupach, 1987; Sharwood Smith, 1981, 1983, 1986). At itssimplest, the reason for studying SL performance is that to knowa language is notsufficient to be able to usea language, and instudying the devetopment ofthe ahilityto speak an SL, wewishto knowwhat the changesare in the cognitive systems used

for production, aswell asin the underlying competencethat theydrawupon. tnorderto facilitate SL production research, attentionneedsto be givento the methodologyemptoyed in such investigations. This is the rationale for the present article. Toestablish the background for a consideration of methods of investigation, first tbriefly present a standard L1 production model as a basis for the L2 model whichbuilds directly on L1 workor largelytakes it for granted. t then considerthe roleofL2 production in SL learning. In the final sectionof the article t focus on the methodsused to investigate L2 production.

THE BASIC L1 PRODUCTION MODELMicrostructure

The languageproduction system is commonly conceptualized as a subcomponent ofan information-processing modelof humancognition (e.g., Butterworth, 1980; Carroll, 1986; Foss & Hakes, 1978; Levell, 1989; ct. O'Connell &Wiese, 1987), structuredwith a hierarchy of levels associated with utterance representations of increasingcloseness to actual speech. Speech errors (Fromkin, 1971; Garrett, 1975) are theprimary data for the construction of such models, having succeeded in this respectdescriptions of pausal phenomena(Goldman-Eisler, 1958; Grosjean, 1980). Errorsofproduction lead to statementsabout the variousmodules or levels of processing thatmust exist giventhe classes of errors observedand other classes that fail to appear(ct. Cutler, 1982). The classic work of Fromkin (1971), for example, referred to sixstagesinthe generation andoutputofEnglish utterances: (1) specification ofmeaningto be expressed; (2) selection of a syntactic structure; (3) generation of an intonationcontour; (4) retrieval oflexical itemsand their insertion into"slots" setup inStage2;(5) additionof affixes to itemsestablished inStage4, and insertion of function wordsat appropriatepositions; and (6) phonological realization.Models differ in the number of levels they posit and their degree of interactive-


3/19

Second Language Speech Production Research 115

ness. Butterworth (1980). for example. expanded Stage 1 to contain a semanticsystem. a prosodic system. anda pragmatic system (ct. Bock. 1989). Models alsovaryin the extentto which additional controlmodules. suchas editorsand monitors. areincorporated. Important recentworkincludes initial attemptsto specify the processesintervening between the levels of structural representation (Kempen & Hoenkarnp,1982; Lapointe & Dell. 1989). as well as the development ofmodels (e.g., Dell, 1985.1986. 1989) that utilize associationist or connectionist theory (McGelland, Rumelhart,& Hinton. 1986).MacrostructureControl over the speechproduction system cantake at leasttwoforms: manipulationofitsproducts while the utterance isbeingformulated. ormanipulation in the processof articulation. The former may be said to be a kind of planning; the latter is oftenreferred to as monitoring.' Bothhavebeen classified as executive controlprocesses(Calfee, 1981; ct. Kluwe, 1987).

Planning. fn speech production a plan is the "representation interven[ing] between the speaker's intention andmanifest phonation" (Butterworth. 1980. p. 156).and planning is the operation required to construct such a representation. Muchresearchin thisarea hasbeen concernedwithdetermining whatoutputunits-word.clause. phrase. and so on-correspond to the initial conceptual elementsof speech.Early arguments were based on pausal data, on the assumption that the systemwould translate a "conceptual unit" into speech and would then pause as anotherconceptual unitwasformulated (e.g., Boomer. 1965; Goldman-Eisler, 1958; Rochester& Gill. 1973). However. morerecentmodels donot require a complete representationof an utterance for speech to begin. and depict the detailed construction of theutteranceas more or less simultaneous with its phonological realization (e.g Ford.1982; Wiese. 1984).Planning can besubdivided. Micro-planning "isconcerned withpurelylocal functions, likemarking clause boundaries and selecting words . . . and. as it turns out.speakers only start to search for a word when it is needed for the next phrase"(Butterworth. 1980. p. 159).Macro-planning operatesat a higherlevel, and"concernsthe longrangesemantic andsyntactic organization ofa sizeable chunkofspeechandtherefore cannot be carried out locally" (Butterworth. 1980. p. 159). Reference hasalsobeenmade to pre-planning. which takesplacebefore speech, and co-planning.which occursat the same timeas speech (MacWhinney & Osser, 1977; ct. Rehbein,1987). Plans may be primarily oriented either to the meaning to be expressed insucceeding discourse or to the structureofthe utteranceto beproduced. The formermayextend at leastas far as 12clauses in advanceof the momentof speaking; thelatter occursmore on a clause by clause basis (Beattie. 1980; Butterworth. 1980;Holmes. 1984; ct. Gould. 1978).

Moniforing. InLl research. there are at least twoconceptions ofmonitoring incirculation (Levell, 1983, 1989). If the speakerhasdirectaccess to the components of


4/19

116 Graham Crookes

~ m @ w u n ~ m i l ~ ~ ~ ( ~ ~ mn r r n ~ @ ~ lU mmrnnl"nlnrm mrnmr-mmmmeproduction theory of monitoring" (Levelt, 1983, p. 46;and see, e.g.,Garnsey & Dell,1984). Alternatively, it may be that the speaker only has accessto the final resultofthe production process, but isableto "detectanystructuraldevianceswhichhemightas well havedetected in somebody else'sspeech, and he can moreover comparethederivedmessage withhisoriginal intention" (Levelt, 1983, p. 46). Thisis the "perceptual theory of monitoring" (see also Bock, 1982). Levelt conceived the monitor(following Laver, 1973, 1980) to be a component of the speech-processing systemwhich"compares parsedaspectsofinner andouterspeechwith (i) the intentions, andthe message sent to the formulator, and (il) criteriaor standards ofproduction. . . . [It]hasto dowith the detectionofspeech errors, syntactic flaws, etc.,but alsostandardsof rate, loudness, and other prosodic aspectsof speech" (1983, p. 50). It also has thefunction of making the speaker aware of production problems. Recently, however,terms haveproliferated: Berg(1986b, see also 1986a; Sternberger, 1985) notedShattuck-Hufnagel's (1979) "monitors;'whichboth disallow elementsand replacethembyothers in speech production-the function of "editors" in other models (ct. Baars,Motley, & MacKay, 1975; Motley, Baars, & Camden, 1983). Bergsuggested that forclarity'ssake, a distinction shouldbe made between the processes of (a) observing"utterance planning;' (b) vetoing material prepared for speech, and (c) replacingvetoeditemsbymorepreferable material. Theseprocesses he would defineasmonitoring, filtering, and editing, respectively.THE L2 MODELThe secondlanguagespeechproduction model is assumed (usually implicitly) tobebasically the same as that for L1 production (ct. Bialystok & Sharwood Smith, 1985;Wiese, 1984), thoughSL investigations with a production model orientation are rare(Wiese, 1984; but see de Bot, 1990; Faerch& Kasper, 1983; Jordens, 1986). Supportingevidenceforthe assumption of equivalence is to be found at the generaldescriptivelevelof temporalvariables, speecherrors, and pausalphenomena, notablyin thework of the Kassel Research Group (Dechert, Mohle, & Raupach, 1984; Dechert &Raupach, 1980, 1987) whichrepresents an important segmentof the psycholinguisticliterature dealing withL2 use.FaerchandKasper (1987) notedthat theKassel group'suse of this type of data "has not as yet had the impact it deserves on the wayperformance analysis is carried out in the international community of SL researchers" (p. 9). TheworkoftheKassel groupseemslittleutilized in L2 research; however,it establishes an adequate precedent for taking the structure of the SL productionsystemas basically the sameas that of the L1 production system, while recognizingthat there are both quantitative and qualitative differences-the competence it utilizesis lessextensive, and alsodifferent, consisting ofboth L1, fL, and L2 "rules:'Macro aspectsof theSL productionmodel are alsobasically the same.However, asa learner'sSL production systemisa very incomplete apparatus, planning andmonitoringare evenmore important. Theproduction ofsecondlanguage speechmaybedifficult, unfamiliar, accessible to consciousness rather than automatic, and involvingrisk(atleastto "face"), soplanningandmonitoring maybe moreextensively utilized


5/19


to cope with the greater demands and lesser resources 01 the L2learner.' They willalso be conceptually even more important, since they have a role in the developmentof the system. Planning is understood to be a part of any L2 model of speechproduction. For example, in a prominent study utilizing SLproduction, Hulstijn andHulstijn (1984)incorporated a standard LI production model (Clark& Clark, 1977)intheir discussion, in which planning has a major role, possibly along with monitoring,in the conversion 01 thought to speech. Monitoring, as a standard part of the humancognitive system, is equally likely to be involved in SLspeech production as it is inthe carrying out of any complex skill. The standard psycholinguistic position is thatthere is no reason to assume that monitoring works any differently in qualitativeterms for SLspeakers than for LI speakers (Hulstijn, 1989; and ct. Raupach, 1980).

L2 PRODUCTION AND L2 LEARNINGMoving lrom a synchronic to a diachronic, learning-oriented analysis 01 languageproduction systems, it may be noted lirstthat in L2studies there has been considerable interest in the SL learner's linguistic environment, or input, but much lessconcern with how that islearned (i.e.,intake), and the role 01 output [i.e.,productionor use) in the development ofSLproficiency has largely been ignored or denied (e.g.,Krashen, 1989).3 Recently, however, interest in this topic has strengthened in SLstudies, and has been supported by the development in psychology 01 a generallearning theory for cognitive skills, including language (Anderson, 1981; e1. Annett,1989),which emphasizes the role 01 use (or output, in SLterms).

Output in SL LearningOne of the first conceptualizations of SLlearning to contain an explicit discussion 01the role of output was Bialystok's early (1978)model 01 L2 learning, which allowedlor the development of "explicit" knowledge partly through a leedback loop fromproduction to knowledge.' Morerecently, Swain's "output bypothesis" (1985;Swain&Lapkin, 1989)-an attribution of weaknesses in Canadian immersion students' ESL toa lack of chances to use the language-has attracted attention to the topic (e1.Krashen, 1989). The development of SL communication strategy research (Faerch,1984;Faerch & Kasper, 1983;Poulisse, 1990)has constituted a third area 01 investigation 01 output, with less of a concern lor its effects on learning, however.Concurrentwork in variability has drawn attention to the role 01 learners' utilization 01 output indifferent forms of discourse (planned/unplanned; formal/inlormal) in SL development (Ellis, 1982; Tarone, 1982).

All these conceptions in varying ways recognize a role lor output in SLlearning,though earlier work in these areas was limited by its willingness to consider SLlearning as independent 01 all other learning processes. Although part 01 cognitivescience, second language acquisition studies have been relatively insensitive to therecent development of cognitive psychological learning theories (Mclaughlin, Rossman, & McLeod, 1983),including those that pertain to the development of proliciency in cognitive skills such as language.


6/19

118

Cognitive Skills and How They Are LearnedGraham Crookes

Thoughdefinitions ofa cognitive skill are problematic.s Hemet's (1970) listofcharacteristics (cited inLevell, 1975) isindicative-a cognitive skill isa categoryofbehaviorthat has hierarchical organization, becomes automaticwith use, and requires feedback and anticipation for its operation. In the past, it has been referredto as anintellectual skill (Welford, 1976), separate from perceptual skills, whichsolely"codeand interpret incoming sensory .information . . . . Motor skills execute skilled movementefficiently butare relianton appropriate links betweensensoryinputand actionroutines. [An intellectual or cognitive skill] link[s] perception and action and [is]concernedwith translating perceptualinputintoa skilled response byusingappropriate decisions" (Colley & Beech, 1989, pp. 1-2). Descriptions of the nativespeaker'slanguage production systemsuggest that producing or comprehending speech is acomplextask involving many substages (Levell, 1978), whose hierarchical structurenecessitates the existence of plansor programs for the execution of an utterance. Itthus maybe described asa cognitive skill.Distinctions betweenknowing "what" andknowing "how" havebeen recognizedfor some time (c!. Ryle, 1949)-Qnemaindifference being in the accessibility of thisknowledge to consciousness. Motor skill learning theorists (e.g., Fitts, 1964) hadtreated this explicitly, recognizing that many skills pass through an early stage inwhichknowledge relatingto what isto be performedisavailable tothe learner inanexplicit, "declarative" mode,and onlylater becomesfully internalized, as "procedural" knowledge. However, after the heavy emphasis on learning in psychologythroughoutthe behaviorist period,cognitive psychology focused on the descriptionof existing cognitive structuresand skills (Andre & Phye, 1986), and itis only inthelastdecade that psychologists havebegun to develop a cognitive theory of learning,and specifically one that applies to cognitive skills.The generalcognitive model(ACT) developed byAnderson and associates (e.g.,Anderson, 1976, 1981, 1983) describes and makespredictions concerningthe learningof cognitive skills, regardlessof domain," A strengthof ACT is its use of productionsystems(Newell & Simon, 1972) to describe rule-governed cognitive behavior-asimple formalism that provides the meansand a requirement whereby allstepsin thedescription ofa complex cognitive process mustbe madeexplicit. The formalism aidsthe computersimulation ofcognitive skill operation,whichshould facilitate checkingthe accuracyofthe model (c!. Crookes, in press; Haugeland, 1981). Use ofACT* mayprovide one means for probingSL fluency development, somethinglargely avoidedthus far (c!. Clahsen, 1987; Rehbein, 1987; Sajavaara, 1987). ACT* describes thecollapsing or "compiling" of production systems by way of rules relating to thenumber of times a subsystem has been successfully utilized, so as to simulate thedevelopment of automaticity through repeated running-off of production systems.Theacquisition ofskill inthe performance ofspeech,aswithany other skill, "consistsessentially of automation of lowlevel plansor unitsof activity" (Levelt, 1978, p. 57),whichisclosely linkedto compilation (c!. McLeod & Mclaughlin, 1986). The utility ofthis particularaspect ofACThas been notedby Mohle & Raupach (1987), though itisnot tobe useduncritically, as the modelas a whole"isbasedon a somewhatnaiveunderstanding of processes underlying first and second language acquisition" (p.


7/19


1167; andd. Raupach, 1987). Empirical supportforAnderson's model isclearest withrespect to the effects of practice on various cognitive skilis: simulations involvingthe repeated running-off of production systems withspecified compilation rules produce success curves closely approximating the log-linear function widely found tocharacterize human skill learning(Neves & Andersen, 1981; Newell & Rosenbloom,1981).'Executive Control Processes in SL OutputSelection and ImprovementExecutive controlprocesses, notably monitoring and planning, have closely connected roles in the carryingout of complex behavior (De Lisi, 1987; Schofnick & Friedman, 1987). Prima facie, the morecomplex unfamiliar behavior is, the more important monitoring isfor it tobe carriedout successfully, and the more likely that someform ofplanningwill beneededin the initial phasesor occasions of use.Mclaughlin,Rossman, and McLeod (1983) referred to "controlled processes" being used in theinitial stagesof SL learning, at which time,attentionand "cognitive effort" (p. 145)must be expendedin carryingout language production. That is to say, at thisstagethe learner may both pre-plan an utterance and monitor its execution (see alsoFaerch & Kasper, 1983).Monitoring and planning may alsobe prominent in the learningof a cognitiveskill. Monitoring (Morrison & Low, 1983) provides a sourceor prerequisite forentry ofexplicit knowledge into the system. It also is involved in overseeing success ornonsuccess of utterances (MacWhinney & Anderson; 1986). Learnerdecisions, in theform of planning, partially determine which sectionsofan IL will be practiced. Sincewhatis not usedwill not get automatized, decisions here are important. Pre-planningofan utteranceprovides a wayforlesswell automatized sections of the system to beused,and thus forthe IL to be extended(subject, presumably, to a varietyofprocessingconstraints in the areaofsyntax; Pienemann, 1987). Initswidest sense, planningis necessary for the long-term success of goal-directed behavior, and monitoring isessential toseethat plansare effectively carriedout.

A role formonitoring in improvingSLoutput. Although the conceptofmonitoringhasbeen confused byKrashen's having initially emphasized it andsubsequently repudiated it, asa general cognitive process it cannotbe ignored. ThatSL learnersutilize monitoring inproducing outputisclear, both inthe earlywork that stimulatedinitial attention to the concept in SLA (Larsen-Freeman, 1975) and in more recentstudies suchas Hulstijn and Hulstijn (1984). In their investigation, forexample, adultSL learners ofDutch retold narratives ofabout four sentencesin length, whichwerepresentedto them inwritten form. It was first established that subjects could effectively respond to directions and feedback to placethe focus of their attentionon theinformational accuracy of their retellings. Then, when "requested to focus on thegrammatical correctness of . . . responses" (Hulstijn & Hulstijn, 1984, p. 31), subjectswereabletosignificantly increase the percentageofcorrectrealizations oftwoDutchword-order rules.


8/19

120 Graham CrookesSeveral typesof monitoring are recognized bySL investigators (though the more

m ~ m l l l [ l [ I ~ m l l l ~ [ [ m mUillfmlfl ~ m ~ ~ ~ ~ m l l l ] ~ ~ ][ m . l ~ m ,1986h). Morrison and Low (1983) suggested that when speakersare monitoring theirownspeech, theymaybe doingpre-articulatory and/or post-articulatory monitoring.The former may simply result in hesitant speech, whereas the latter may lead to"overt editing" (Hockett, 1967, p. 936) in the form of, for example, false starts andself-corrections (see also Levelt, 1983). Morrison and Low (1983) also argued thatmonitoring one's ownspeech involves verysimilar mechanisms tomonitoring that ofothers: "in both situations, an abstractimage held in the working memorystore isanalysedon the basisof storedinformation" (po 241). Theyadditionally bypothesizedthat "the act of detecting and subsequently repairing certain mistakes may havelongitudinal repercussions [i.e., learningJ" (1983, p. 244). Schmidt and Frota (1986)considered the latter hypothesis, but found no evidence in their data, collected on anadult learner ofPortuguese as an SL (Schmidt), that features that were self-eorrected(over a 6-month period) were those that improved. (However, this applies to postarticulatory monitoring only.) If the learner monitors his/her own SL speech, anutterance produced successfully on one occasion maybe notedand reusedthereafterwith increasing automaticity (e.g., Schmidt & Frota, 1986, p. 310; and Bahns, Burmeister, & Vogel, 1986, in thecaseof children's learningof ESL).

The role of planning in improving SL output. The importance of practiceeffects in a skill-learning model of SL development also allows a role for planning."Formal practice" (Bialystok, 1978) hasbeen found to transform "explicit knowledge"ofan SL into "implicit knowledge:' In that study, Bialystok equated formal practicewithexplicit teachingand acceptedthatwhatwaspracticed wasteacher-determined.It is possible, however, forthe intlividuallearnerto makea conscious decision to use.a particularword, phrase,or sel of utterances (ct. Bartelt, in press; Schmidt & Frota1986, pp. 269, 319). Discussing Bialystok's position, Sharwood Smith (1981) arguedthat

someaspects of second language performance can inprinciple be planned fromthe start entirely onthebasis of explicit knowledge. . .. Let us also suppose thatthis type of activity is repeated again and again. In such situations, it is surelyreasonable tosuppose that a certain number ofstructures planned and performedslowly and consciously can eventually develop into automatized behaviour. (p.166)In these cases, he stated, "utterances initiated by explicit knowledge can providefeedback into implicit knowledge" (p, 166), andwemay assume that itwill be learnerdecisions (and additional factors such as task demands) that determine what getspracticed and automatized. Several recent studies provide preliminary indicationsthat ensuring that SL learners produce planned speech results in their producingdiscourse that ismorecomplex (Crookes, 1989), moreaccurate(Ellis, 1987), andmoreexplicitly structured ("with discourse moves . . . markedmore overtly and elaborately"; Williams, 1990, p. 2).


9/19

Second Language Speech Production Research

Communication Strategies and SL Production121

The prominent area of interlanguage research dealing with communication strate-gies, thoughobviously initially the study 01 what learnersdo in certain problematiccircumstances (e.g., Faerch& Kasper, 1983; Kellerman, inpress; Kellerman, Ammerlaan, Bongaerts, & Poulisse, 1990) is directly concernedwith production, and thuspotentially withitseffects on learning. Bialystok and Sharwood Smith (1985) pointedout the difficulty of distinguishing between learning strategies and communicationstrategies: "the effect 01 employing a particular strategyin a given context may beeither one of learningmore about the language, or one 01 solving an immediatecommunication problem, or both" (p. 114). Theyrecognized in passing that thewayalearner engages in SL production, for example, when utilizing a strategysuch as"rehearsal;'can havethe effect of facilitating learning. Linking the area of learning/communication strategyresearchto thatofSL production, they remarked:

Even ifit is necessary to describe separately the processes responsible for lan-guage useunder conditions ofdifficulty .. . differently from theprocesses respon-sible when nosuch problems are perceived . .. theprocesses of language produc-tion are probably notdifferent. . . .Our description of strategies, then, must becompatible with someoverview ofhow thesystem operates inordinary languageproduction. (1985, pp. 114-115)Theyalsodrewattention to the importance to the SL speaker of "executive procedures for integrating and coordinating aspects of knowledge in the production oflinguistic responses" (p. 114). Forthem, strategies can relate to "the expansion andanalysis of linguistic information" (p. lIS) andto speechproduction underthenormalconditions 01 time and situational constraints. It should be emphasized (followingBialystok & Sharwood Smith, 1985) that since there is reason to expect research inthisarea to becongruentwithSL production models, it shouldbeuselullor the study01 general SL perlormance. However, as Ellis and Roberts (1987) observed: "theresearchhasbegunto document what strategies learnersuse in communication butthese have not been systematically related to language development or variablelanguage use. Inother words, the relationship betweenuse and development is stillpoorly understood" (p. 27). (See Poulisse, in collaboration with Bongaerts & Kellerman [19901, forrecentworkdeveloping our understanding of thisrelationship.)RESEARCH TECHNIQUES USED TO STUDY L2 PRODUCTIONData Collection ProceduresData collection procedures may range lrom those placing little restriction on theindividuals producing thespeechtobe described, using relatively unconstrained, freespeechsamples, to those limiting production to imitation 01 given models (elicitedimitation, EI), or completion of partial phrases (utterance completion, UC). In EI,speakersare askedto repeat utterances that they hear, and their successes, failures,and errors constitute the data for the investigation (see Chaudron & Russell, 1990).


10/19

122 GrahamCrookesSpecification of the models for imitation enables the inVeStirtor, havinr collected

the data, to identify exactlywhatcannotbe repeated. Several studies haveusedwhatLennon (1984) termed "Oral Reproduction"-thepresentation of an oraldiscourse tosubjects intheir first or secondlanguage, withthe requirement that theyreproduce itin the second language. fn UC tasks, as the name suggests, subjects respond to thebeginning of an utterance (presented orally or in writing) bY completing it orally.Theymayor maynot also be exposedto a stimulus word, or frame, providing anindication ofwhat is tobe usedin the completion. Datagathered from thisprocedureusually are "errors" or interlanguage forms (Hulstijn & Hulstijn, 1984), and avoidances are possible; data couldalso be latencies-time from presentation of utterancefragment to initiation of production.A related technique that has occasionally been used is elicited translation (e.g.,HOlscher & Mohle, 1987; Swain, Dumas, & Naiman, 1974). Subjects are presentedwitha language sample (written or spoken) and asked to translate it (either writingdownthe result, or producing anoral redaction). Thereisno call forthe generation ofnew semantic content, and in this respect the production is controlled. The dataproduced in the taskthat are mostrelevant to understanding production are primarily those obtained from the think-aloud protocols that the subjects utter as theyareworking on the translation task. Theyrelatemoreto the process than to the productofthe SL production system.Studies of the processes attendant on speech production can try to tap directlyintothe cognitive system throughintrospective or retrospective reports. The application of this methodology to SL studies in general has been extensively discussedelsewhere (notably, Ericsson & Simon, 1987; Faerch& Kasper, 1987). Inone ofthe asyet few such investigations whose primary focus isthe productive system, Bartelt (inpress) comparednonnative speakers' (NNSs') retrospective accounts of their production processes with Levell's (1989) model of speech production. Retrospection hasalsobeen applied to the investigation of compensatory strategies in SL oral production (specifically, forthe purposeof localizing such strategies; Poulisse, Bongaerts, &Kellerman, 1987).Research DesignThe simplest question that can be asked of SL speech production is, "Whatdoes itlook like?" In answering thisquestion, investigations can range from simply lookingat elementsofspeechproduction that fall intoa single category or setofcategories ofinterest to attempting a complete analysis ofan extendedspeechsamplewithas full adescription of itscontextually situatednatureas possible. Descriptive unitsthat havebeen particularly utilized in the construction of SL production models have relatedprimarily to utterance-level matters such as rate ofspeech, pause lengths, and falsestarts, onwhich arguments about structural characteristics of the production system,suchas planning, havebeen constructed (e.g., Fathman, 1980; Seliger, 1980).Ata slightly broader level, description intended to develop ideasaboutSL production has useddiscourse analysis, usually focusing on complete stretchesof speech,oftenbetween dyadsof NNSs, followed by qualitative analysis intended to discover


11/19

SecondLanguage Speech Production Research 123

underlying processes, such as transfer (e.g., Faerch& Kasper, 1989), for example.Many studies of communication strategies havethese characteristics (for review, seePoulisse, Bongaerts, & Kellerman, 1984), as have a numberof investigations of SLnarrative(e.g., Dechert, 1980, 1983, 1984; Dechert& Raupach, 1987; Lennon, 1984).I would like to be able to move on from descriptive investigations' to state,taxonomically at least, the characteristics of experimental SL production work, butthere isso littleof it that I feel thiswould be premature. I therefore turn to a generalcritical discussion of researchtechniques usedin thisarea.Methodological Problems of SL Production ResearchThere are so manyproblems in doing SL production research (andSL research ingeneral) that nostudyis immune from criticism. Indeed, anysocial sciencemethodologyhasinherentweaknesses, andit isnecessary toconduct studies ofthe sametopicusing different techniques (Brinberg & McGrath, 1985; Chaudron, Crookes, & Long,1988) in order to he in a position to state findings strongly. This having been said,there are alsoweaknesses in existing SL production workthat can be pointedout tostrengthenfuture SL investigations.The studies listed here' constitute muchof the bulkof descriptive SL productionstudies. Theyutilize relatively controlled data, particularly pause lengths, false starts,repeals, and intonationally defined "runs of speech;' but they are not explicitlyhypothesis-testing in nature.An initial problem is that manyof these investigationsreportthe data from a small numberofsuhjects (1, Raupach, 1984; 3,Mohle, 1984; 8,Dechert, 1980; etc., which is to some extent justified by their exploratory nature)though others use a larger sample (12, Lennon, 1984; 18, Brenzel, 1984). All SLresearcherswill be sympathetic to the problem ofsmall samplesize,thoughthisdoesnotmeanwecan absolve ourselves ofthe lackof representativeness it mayimply (orthe needto dealwiththe problems ofthe lawof small numhersand the absenceofpower in statistical tests it creates; d. Cohen, 1977; Kraemer & Thiemann, 1987).However, more important in an SL context is the variability of subjects. Hulstijn(1989), in a mostuseful exposition of experimental methodsusedinHulstijn (1982),discussed the importance of and procedures for selecting subjects whose performance isat a level appropriate to the mailer to be investigated (syntactic level, taskperformance, etc.), which both earlier descriptive and more recent experimentalstudies (e.g., Crookes, 1989) in thisareahavenotdone.Also in thecontextof samplesused, it is notable that many studies havedrawnon the Kassel corpus-SL productionscollected between1979 and 1982 "from a varietyofAmerican, English, French,and German students" (Dechert, Mohle, & Raupach, 1984, p. 7). It can be veryfacilitating for researchto havea corpus of data or protocols alreadycollected-butit alsomayprevent a researchprogram from moving from description to the investigation of specific hypotheses. It is not usually the case that an existing corpuscan be variedenough to provide the data for new investigations, and it may leadresearchers to tailorthe investigation to fit thedata,rather than the (preferable) otherway round.Because ofthe demands of experimental work, it isoftendesirable to beginwork


12/19

124 Graham Crookes

1989),and perhaps there isa tendency to be less concerned with methodology, andparticularly sample size, in such studies. InSL research, we now have a plenitude ofsuch studies. Sometimes this approach is given a label (e.g., "idiographic;' Dechert,1984; or "interpretative;' Rehbein, 1987),but this point is not always faced squarely,and these two terms are neither defined nor discussed in the two references mentioned. The investigation may be referred to as a case study; Mahle (1984), forexample, is direct about this point, and about the limitations of the study in question,recognizing that the significance of anyone of the temporal variables investigated inthe study is "dependent on the distribution and salience of the others as well as thelinguisticproperties of the text . . . syntactic complexity or degree of idiomaticity" (p.28).However, since data from a case study can be extracted from the full array of

protocols in both defensible and indefensible ways, this is not sufficient (Atkinson &Delamont, 1985; Merriam, 19881. One of the few explicit discussions of the procedures to be used in marshalling an argument from small-scale analyses of discourse isJackson (1986), who provided a critical exposition of the method of analytic induction (Denzin, 1970). The whole question of the procedures to be used in the analysisof discourse raises complex philosophical questions (pertaining to the challengepresented to "standard" scientific methodology bY hermeneutic approaches; el.Smith, 1989). However, investigators need to explain, at the least, whether theiranalysis deploys, either explicitly or implicitly, an analytic system with defined categories that more than one individual can use. They need to consider documentingthe reliability of their analysis-a preliminary approach to which is to report interobserver agreement as to the assignment of elements of discourse to categories (as inLennon's [1984J use ofChafe's "spurts"; el. Crookes, 1990). If these standard discourseanalysis procedures are not used, are we faced with a common one-person investigation of a phenomenon using middle-of-the-roadanalysis of a traditional descriptive(applied) linguistics type (e.g., Towell, 1987), or is the investigator utilizing sometechniques of conversational analysis, without, say, committing completely to thephilosophical positions these imply (Hopper, Koch, & Mandelbaum, 1986; Sigman,Sullivan, & Wendell, 1988)? Is the validity of the analysis to be established throughgenerally accepted social science practices, or through those of a more Verstehennature? (See Jackson & Jacobs, 1983; Jackson, 1986; Jacobs, 1986;Smith, 1988, whorefers to these as "humanistic"; and Grotjahn, 1991.)Issuesof reliability in temporal variable studies of production, involving, for example, pauses per utterance, mostly come down to the use of the appropriate experi

mental apparatus (servochart plotters, oscillominks; el. Dechert, 1980; Rowe, 1986)and are relatively uncontroversial. One issue of validity seems in particular moreproblematic. ManySL production studies utilize pauses in discussions of planning. Inone ofthe more substantial of these, Lennon (1984)was initially quite clear about theproblems of utilizing pauses to investigate planning, noting their "multifunctionality"(p. 50).However, he subsequently interpreted pause lengths as directly indicative ofplanning, as is not uncommon in such studies (el. Fathman, 1980). This was anassumption of early Ll production studies (e.g., Goldman-Eisler, 1958) but ignores


13/19


any possibility that speakers can plan what they are saying as they are saying it.However, Mohle (1984) wasclear about thewayinwhich interpretation ofpatternsofpausing as indicative of planning differences across different aspectsof discourse isprimarily "an assumption for which no direct evidence is offered here" (p. 37).Investigations of executive control functions (otherthan initial groundbreaking pilotstudies) will in general need to provide independent evidence that the indicators ofthe cognitive processing to be probedactually do directly signal that process (guidedretrospection beingoneway, for example).Thegeneral issue oftaskisonetowhichSL investigators are becoming increasingly sensitive (Chaudron, 1985; Crookes, 1986; Ellis, 1987; Hulstijn, 1989), soit needbemerelytouchedon here.Onone hand, investigators shouldcontrolfor task,so as tobe able todeal witha singlediscourse type.On the other, theyshouldusemorethanone taskso as to preventtheir results beingattributable to the task itself. Ideally, weneed a variety of experimental taskswhose characteristics are fully understood, asituation unlikely to be arrived at in the immediate future (el. Chaudron, 1985).However, in the absence ofsuch guidelines, we havethe responsibility to perceivetasksor speech elicitation procedures in general not merelyas ways of gettingoursubjects to speak,but asbundles of largely unexamined auxiliary hypotheses, whosepotential effects we must either investigate or attempt to control for. Ellis (1987)emphasized controllingdiscourse type while manipulatingplanning." Hulstijn (1989)notedthat "itis necessary todistinguish between taskand task requirement" (p.29}in other words, evenwithin a single task, different demandconditions can resultinwidely differing language production. Theimportance ofLI baseline data in settlingthis point hasbeen emphasized byKellerman, Ammerlaan, Bongaerts, and Poulisse(1990). In thisarea, then,we should lookto see morestudies focusing directly on themeasurement aspects ofspeechproduction.CONCWSIONIt maybe that SL production researchis nowat the pointwhere the field isready tomove from primarily descriptive research to greater use of experimental investigations ofa moreobviously hypothesis-testing, theory-developing nature. The handfulof such SL production studies (e.g., Crookes, 1989; Ellis, 1987) are hy no meansimmune from criticism, but such criticism is of the sort that can be made of anyregular pieceof educational research.11 Despite some reservations expressed hereabout earlier SL production studies, criticism cannot legitimately be leveled at theinvestigators, who usually havebeen explicit about the limitations of this research.However, a standardphilosophy of social science would assume that the targetof theenterprise is the development of explanations basedon generalizations that evolveout ofsupported hypotheses. IIthis position isaccepted, it isnecessary for investigations to be directed to the many preliminary positions that have been sketched indescriptivework, and to probethemmoredirectly ina hypothesis-testing manner. IIthis isdone,wemayexpect to see SL production researchmoving from its currentposition oflowvisibility toonemoreconsonant withitsimportance inSL learningasa whole.


14/19

126

NOTES

Graham Crookes

1.I do not,of course,mean here that misconception ofmonitoring indicatedin thewritings ofKrashen(e.g.,1981, 1982. 1985).Amoreuseful senseof the term,congruentwithhowit isconceivedinotherareasofthebehavioralsciences, isoutlinedforSLstudiesbyMorrison and Low (1983; d. Wiese, 1984). Fordiscussionsof how the term "monitoring" has been misused in some writings on SLdevelopment, see, e.g., Crookes(1988), Gregg(1984), andMcLanghlin (1987).2.The comparisonhere isbetweenthe adult L21earneroperatingunder normalconditions and the fullycompetent (i.e., adult) Ll user operating under normal conditions. However, a child using his/her firstlanguagemayhavejustas muchneedto exertmetacognitive controlskills suchas planningandmonitoring asthe secondlanguagelearnerdoesin his/her normallydifficult circumstances. Similarly, the adultL1 usermayencounter demandingcommunication conditions, when it will bemore important than usual to plan andmonitorutterances.3.However, it is true that in interlanguage studies,output,in the sense of the collected utterances oflearners, hasbeen the predominantdata sourcefor investigatingSLcompetence.4. See alsolaterversions(e.g., Bialystok, 1982; Bialystok & Ryan, 1985), whichusedthe term"analyzed"instead of "explicit;'and add a seconddimension, "control;'to the characterization of learners' knowledge.Mohle and Raupach (1987) provided a very useful comparison of these dichotomies, as well as that ofknowledge/control (Bialystok & Sharwood Smith, 1985), and howthey relate to Anderson's (1981) declarative/proceduraldistinction.5.A detailedanalysisofwhatcan be classified asa skillisprovided by Downing and Leong(1982) andsummarizedinDowning (1984), whichgivesa 20-heading listof skillcharacteristics. Anderson (1976, 1981,1983) seemedto avoida definition, as didO'Malley, Chamot, and Walker (1987). See alsoMclaughlin (1987),and Mclaughlin, Rossman, and McLeod (1983) for furtherdiscussion. Initial attempts to introducethis topicintoSL-related discussion canbe markedin the publication ofLevelt's (1978) article inthe firstissueofStudiesinSecond Language Acquisition (apparentlywidelyignored, alongwithMcDonough's [19811 discussion oftheapplicabilityof the conceptto SLlearning).6.O'Malley, Chamot, and Walker(1987) have been prominentin fostering discussion of ACT- in SLcontexts(andsee alsoFaerchand Kasper, 1985, 1986, 1989; Raupach, 1987). Concurrentconceptualdevelopments in applied linguistics, notably the work of Bialystok (1978, 1982) and SharwoodSmith (1981), haveprobably fostered the acceptability ofACT- inSLstudies,though these researchers havestated that they donot findthe theoryentirelycongruentwiththeirwork(Bialystok &Sharwood Smith,1985, n. 2),andgenerallydidnot discuss itat allor attemptto integrateit with their work.Fluencyisthe targetformostSLlearning,andSLresearchersand theoristsmustunderstandit assuch.Onthefaceofit, fluency would appear tobe primarilyperformance-based innature,and equivalenttoautomaticity. The connectionis strongenough, I believe, to make such formulations as ACT useful, but knowledgerepresentations (i.e., analyzed or not)and mattersof control(Bialystok & Sharwood Smith,1985; SharwoodSmith&Kellerman, 1989) must alsobe consideredin investigating this topic,which cannot for reasons ofspacebe consideredfurther here.7.A preliminaryattempt has been made to applythis kind of model to learninga first languagebyMacWhinney and Anderson (1986), though limitations of the linguistic analysis used preventedAnderson's(1983)modelfrombeinggivena goodtestin thisinstance.S.Fora helpful generaldiscussion ofdescriptive research,see BrinkandWood(1989).9. Including at leastAppel(1984), Appeland Goldberg (1984), Brenzel(1984), Dechert(1980, 1983, 1984,1987), Dechertand Raupach(1989), Fathman (1980), Grosjean (1980), Lennon(1984), Mohle (1984), O'Connell(1980), Raupach(1980, 1984), Rehbein ( 1 ~ 8 7 ) , Saiavaar nd Lehlnnen (1980), and Seliger (1980). I excludefromthis grouping communication strategy studies,Which, although dealingwith L2use, havein the pastbeen oriented lesstowardthe developmentofa modelofL2production or learning.10.Thoughhavingoperationalizedplanningintermsof timehe neglectstoprovidean independentcheckthat planningtookplace.II. Spacedoesnot permit treatment of these moregeneral issueshere, thoughit must be admitted thateducational research as a whole rarely meets research methodology canons in full.Outstanding areas ofdifficulty are statistical hypothesis testingprocedures, notablythose applied to contingency tables (suchasthoseusingchi-squared statistics to test for goodness-of-fit; see Sternberger, 1989); the overattention giventoalpha levelas comparedto beta levelin exploratoryresearch (Cohen, 1982; Cowles & Davis, 1982; Davis &Gaito, 1984; Freiman,Chalmers, Smith, &!Kuebler, 1978; Rosenthal & Rubin, 1985; Ryan, 1985); smalleffectsizestestedwithunderpoweredtests;and a reluctanceto replicatestudies(Santos, 1989; ct. Shaver&Norton,1980; see alsoCohen,1977; Daley &Hexamer; 1981; Kraemer& Thiemann, 1987).


15/19

Second Language Speech Production Research

REFERENCES

127

Anderson, J.R.(I976). language, memoryand thought. HiUsdale, NJ: Erlbaum.Anderson, 1.R. (Ed.). (1981). Cognitive skills and their acquisition. Hillsdale, NJ: Erlbaum.Anderson, J.R. (1983). The architecture ofcognition. Hillsdale, NJ: Erlbaum.Andre,T.,& Phye, G.D. (1986). Cognition, learningand education. InG. D.Phye& T.Andre[Eds.], Cognitiveclassroom learning: Understanding, thinking andproblem solving (pp. 1-19). Orlando, FL: Academic.Annett, J. (1989). Training skilled performance. InA.M. Colley & J. R. Beech (Eds.), Acquisition andperformance of cognitive skills(pp. 61-84). New York: Wiley.Appel, G. (1984). Improving secondlanguageproduction. InH.W. Dechert, D.Wible,& M. Raupach (Eds.),Second language productions (pp. 186-210). Tiibingen: Gunter Nan.Appel, G., & Goldberg, M. (1984). Referential choice in second languagenarrative production. In H. W.Dechert,D.Mohle, & M. Raupach (Eds.), Second language productions (pp. 132-155). Tubingen: Gunter

Narr.Atkinson, P., & Delamont, S. (1985). Bread and dreams or bread and circuses: A critiqueof 'case study'researchin education. InM. Shipman (Ed.), Educational research: Principles, policies andpractice (pp. 2645). London: FalmerPress.Baars, B.J.,Motley, M. T.,& MacKay, D.G. (1975). Outputeditingfor lexical statusin artificially elicited slipsofthe tongue.Journal of Verbal Learning andVerbal Behavior, 14,382-391.Bahns, J., Burmeister, H., & Vogel, T. (1986). The pragmatics of formulas in L2 learner speech: Use anddevelopment. Journal ofPragmatics, 10,693-723.Bartelt, G. (inpress). Theethnography ofsecondlanguageproduction./RAL.Beattie, G. W. (1980). Theroleof language production processes in theorganization ofbehaviorin face-to-laceinteraction. InB.Butterworth (Ed.), language production (Vol. 1,pp.69-107). London: Academic.Berg,T. (1986a). The aftermathof error occurrence: Psycholinguistic evidencefrom cut-offs. language andCommunication, 6(3). 195-213.Berg,T. (198Gb). The problems of language control: Editing, monitoring, and feedback. Psychologicalsearch, 48, 133-144.Bialystok, E.(1978). Atheoretical modelofsecond language learning. language Learning, 28,69-84.Bialystok, E.(1982). Onthe relationship betweenknowing and usinggrammatical forms. Applied linguistics,3.181-206.Bialystok, E.,& Ryan, E.R. (1985).Ametacognitive framework forthe development of language skills. InD.L.Forrest-Pressley, G.E.MacKinnon, & T. G.Walker (Eds.), Metacognition, cognition, andhumanperfor-mance(Vol. 1,pp.207-252). New York: Academic.Bialystok, E.,& Sharwood Smith, M. (1985).lnterlanguage is not a stateof mind.Applied Linguistics, 5, 101117.Bock, J.K. (1982). Toward a cognitive psychology ofsyntax: Information processing contributions to sentenceformulation. Psychological Review, 89,1-47.Bock, K.(1989). Closed-class immanence insentence production.Cognition, 31, 163-186.Boomer, D.S.(1965). Hesitation and grammatical encoding. languageandSpeech, 8, 14&-158.Brenzel, E. (1984). Summarizing in a second language. InH. W. Dechert, D.Moble, & M. Raupach (Eds.),

Second language productions (pp.69-113). Tlibingen: GunterNarr.Brinberg, D.,& McGrath, J. E.(1985). Validity andtheresearch process. Beverly Hills, CA: Sage.Brink, P. J. (1989). Exploratory designs. In P. J. Brink& M. J.Wood (Eds.), Adoonced design innursing research(pp. 141-160). Newbury Park.CA: Sage.Brink, P. J.,& Wood, M. J. (1989). Descriptive designs. In P. J.Brink& M. J.Wood (Eds.), Advanced design innursing research (pp.123-140). Newbury Park,CA: Sage.Butterworth, B.(1980). Someconstraints onmodels oflanguageproduction. InB.Butterworth (Ed.), Languageproduction (Vol. 1,pp.423-459). London: Academic.Calfee, R.(1981). Cognitive scienceandeducational practice. Review ofResearch inEducation, 9, 3-73.Carroll, D.W. (1986). Psychologyof language. Pacific Grove, CA: Brooks/Cole.Chafe, W. (1980). Somereasonsforhesitating. InH.W. Dechert& M. Raupach (Eds.), Temporal variables inspeech (pp.169-180). Hague: Mouton.Chaudron. C. (1985). Intake: Onmodels and methodsfor discovering learners' processing of input.Studies inSecond language Acquisition, 7,1-14.Chaudron, C., Crookes, G., & Long, M. H. (1988). Reliability and validity in second language classroomresearch (fechnicalReport No.8). Honolulu: University ofHawaii, CenterforSecondLanguage ClassroomResearch, Social Science Research Institute.


16/19

128 Graham CrookesChaudron, C.,& Russell, G.(1990). The validity of elicited imitation asa measure of second language competence. Manuscript submitted forpublication.

b.hsen, It (Jk Natural language development: Acquisitional processes leading 10 flnency in speechproduction. In H. W. Dechert & M. Raupach (Eds.), Psycholinguistic models of production (pp. 67-75).Norwood, NJ: Ablex.Clark, H.H.,& Clark, E.V.(l977).Psychology and language. New York: HarcourtBraceJovanovitch.Cohen, J.(1977). fbwer analysis forthe behavioral sciences (2nded.).Hillsdale, NJ: Erlbaum.Cohen, P. (1982). To be or not to be: Control and balancing of Type I and Type II errors. Evaluation andProgram Planning,S, 247-253.Colley, A.M., & Beech,J.R. (1989). Acquiring and performing cognitive skills. InA.M. Colley & J.R.Beech(Eds.), Acquisition andperformance of cognitive skills (pp.1-16). New York:Wiley.Cowles, M., & Davis, C.(1982). Onthe origins ofthe .0Slevelofstatistical significance. American Psychologist,37,553-558.Crookes, G.(1986). Tasks: A cross-disciplinary review IlechnicalReportNo.4). Honolulu: University ofHawaii,Social Science Research Institute, CenterforSecond Language Classroom Research.Crookes, G.(1988). Planning, monitoring, andsecond language development (fechnicalReportNo.6).Honolulu: University of Hawaii, Social Science Research Institute, Center lor Second Language Classroom Research.Crookes, G. (1989). Planning and interlanguage variation. Studies in Second Language Acquisition, 11,367383.Crookes. G.(1990). Theutterance, and other basicunitslor discourse analysis. AppliedLinguistics, 11, 183199.Crooks, G.[in press). Theory formats end second language acquisition theory. University ofHawaii Working

Popers in ESI., 9(3).Cutler, A. (Ed.). (1982). Slips ofthe tongue and language production. Hague: Mouton.Daley, J.A.,& Hexamer; A. (1981). Statistical powerin researchin English education. Research in the teachingofEnglish, 17, 157-164.Davis, C.,& Gaito, J. (1984). Multiple comparison procedures within experimental research.Canadian Psychol

ogy, 25,1-13.DeBot,K. (1990). A bilingual production model: Levelt's 'Speaking' modeladapted. Manuscript submitted lorpublication.Dechert, H.W. (1980). Contextual hypothesis-testing procedures in speechproduction. InH.W Dechert& M.Raupach (Eds.), Towards a cross-linguistic assessment ofspeech production (pp. 101-121). FrankfurtLang.Dechert, H.W (1983). How a storyisdone ina secondlanguage. InC.Faerch& G.Kasper (Eds.), Strategies ininterlanguage communication (pp. 175-195). London: Longman.Dechert.H. W. (1984). Individual variation in language. InH.W Dechert,D. M6hle, & M. Raupach (Eds.),Second language productions (pp. 156--185). Tubingen: GunterNarr.Dechert,H.W (1987). Understanding producing. In H.W Dechert& M. Raupach (Eds.), Psycholinguisticmodels of production (pp. 229-238). Norwood, NJ: Ablex.Dechert,H.W,M6hle, D., & Raupach. M. (Eds.). (1984). Second language productions. Tubingen: GunterNarr.Dechert,H. W., & Raupach, M. (Eds.). (1980). Towards a cross-linguistic assessment of speech production.Franklurt: Lang.Dechert,H.W., & Raupach, M.(Eds.). (1987). Psycholinguistic models ofproduction. Norwood, NJ: Ablex.Dechert,H.W., & Raupach, M. (Eds.). (1989). Transfer in language production. Norwood, Nf:Ablex.DeLisi, R. (1987). A cognitive-developmental modelof planning. InS.L. Friedman, E.K. Scholnick, & R.R.Cocking (Eds.), Blueprints forthinking (pp.79-109). Cambridge: Cambridge University Press.Dell, G.S. (1985). Positive feedback in hierarchical connectionist models: Applications to language production.Cognitive Science, 9, 3-23.Dell, G.S. (1986). Aspreading-activation theory 01 retrieval in sentenceproduction. Psychological Review, 93,

283-321.Dell, G.S. (1989). The retrieval ofphonological forms in production: Tests ofpredictions from a connectionistmodel. InWMarslen-Wilson (Ed.), Lexical representation andprocess (pp.136--165). Cambridge, MA: MITPress.Denzin, N. K.(1970). Theresearch act.Chicago, IL: Aldine.Downing, J. (1984). Task awareness in the development of reading skill. In J. Downing & R. Valtin (Eds.),Language awareness and learning toread(pp. 27-55). New York: Springer-Verlag.Downing, J..& Leong, C.K.(1982). Psychology of reading. New York: Macmillan.Ellis, R. (1982). Discourse processes in classroom second language development. Unpublished doctoralthesis,University ofLondon.Ellis, R.(1985). Avariablecompetence modelofsecond languageacquisition.IRAL, 23,47-59.


17/19

SecondLanguage Speech Production Research 129Ellis, R. (1987). Interlanguage variability in narrative discourse: Styleshifting in the use ofthe past tense.Studies inSecond languageAcquisition, 9, 1-20.Ellis. R.(1988). Theroleofpracticeinclassroom learning.AlLA Review, 5,20-39.Ellis, R., & Roberts, C. (1987). Two approaches for investigating secondlanguageacquisition. InR. Ellis (Ed.),

Second language acquisition in context (pp.3-29).Englewood Cliffs, NJ: Prentice-Hall.Ericsson, K. A., & Simon, H.A. (1987). Verbal reportson thinking. InC.Faerch& G. Kasper (Eds.), Introspectionin second language research (pp. 24-53).Philadelphia:Multilingual Matters.Faerch, C.(1984). Strategies in production and reception-Some empirical evidence. InA.Davies, C.Criper, &A.P. R.Howatt (Eds.), Interlanguage (pp.49-70).Edinburgh: Edinburgh University Press.Faerch, C" & Kasper, G. (1983). Plansand strategiesin foreign language communication. InC.Faerch& G.Kasper (Eds.), Strategies in interlanguage communication (pp. 20-60). London: Longman.Faerch, C.,& Kasper, G.(1985). Procedural knowledge as a componeatof foreign languagelearners'communicativecompetence. InH.Bolte & W. Herrlitz (Eds.), Kommunikation im Sprachunterricht (pp. 169-199).Utrecht: Rijksuniversiteit.Faerch, C.,& Kasper, G. (1986). Cognitive dimensions of languagetransfer. InE.Kellerman & M. SharwoodSmith(Eds.), Crosslinguistic influence in second language acquisition and performance (pp. 49-65).Oxford: Pergamon.Faerch, C.,& Kasper, G.(1987). Fromproductto process-Introspectivemethodsin secondlanguage research.In C. Faerch & C. Kasper (Eds.), Introspection in second language research (pp. 5-23). Philadelphia:Multilingual Matters.Faerch, C.,& Kasper G.(1989). Transfer in production: Some implications for the interlanguege hypothesis. InH.W. Dechert& M. Raupach (Eds.), Transfer in language production (pp. 173-193). Norwood. NJ: Ablex.Fathman, A K.(1980). Repetition andcorrectionasan indication ofspeechplanningandexecutionprocessesamong second language learners. In H. W. Dechert& M. Raupach (Eds.), Towards a cross-linguisticassessment of speech production (pp. 77-86).Frankfurt: Lang.Fitts, P. M. (1964). Perceptual-motor skill learning. InA.W. Melton (Ed.), Categories of human learning (pp.244-285). New York: Academic.Ford, M. (1982). Sentenceplanningunits: Implications forthe speaker'srepresentation ofmeaningful relationsunderlying sentences. In1.Bresnan (Ed.), Thementalrepresentation of grammatical relations (pp. 797827). Cambridge, MA: MIT Press.Foss, D.J.,& Hakes, D.T. (1978). Psycholinguistics. Englewood Cliffs, NJ: Prentice-Hall.Freiman, 1.A.,Chalmers, T. C.,Smith, H.,Jr.,& Kuebler, R. R.(1978). Theimportance ofbeta,the typeIIerrorandsamplesizein thedesignandinterpretation ofthe randomized controltrial.NewEngland Journal ofMedicine, 299,690-694.Fromkin, V. (1971).The non-anomalous natureofanomalous utterances. lnnguage, 47,27-52.Garnsey, S.M., & Dell, G.S. (1984). Someneurolinguistic implications ofpreartlculatory editingin production.Brain andLanguage, 23,64-73.Garrett,M. f. (1975). Theanalysis ofsentenceproduction. InG.H.Bower (Ed.), Thepsychology of learning andmotivation (Vol. 9,pp. 133-177).NewYork: Academic.Goldman-EisJer, F. (1958). Speechproduction and the predictability ofwordsin context.Quarterly Journal ofExperimentalPsychology, 10,96-106.Gould, 1.D.(1978). An experimental studyofwriting, dictating and speaking. InJ.Requin ( E d . ~ Attention andPerformance Vll(pp. 299-319). Hillsdale, NJ: Erlbaum.Gregg, K. (1984). Krashen's Monitor andOccam's Razor. Appliedlinguistics, 5, 79-100.Grosjean, F. (1980). Linguistic structuresand performance structures: Studies in pausedistribution. InH. W.Dechert& M. Raupach (Eds.), Temporal variables inspeech (pp. 91-105). Hague: Mouton.Grotjahn, Rudiger. (1991). TheResearch Programme Subjective Theories:Anewapproachinsecondlanguageresearch.Studies inSecond lnnguage Acquisition, 13,187-214.Haugeland, J. (1981). Thenatureand plausibility of cognitivlsm.fn J. Haugeland (Ed.), Mind design (pp. 243281). Montgomery, Vf: Bradford Books.Herriot, P. (I970). Thepsychology of language. London: Methuen.Hockett, C. F. (1967). Where the tongueslips, there sJip I. In V. A F1romkin (Ed.), Speech errors as linguisticevidence (pp. 93-119). Hague: Mouton.Holmes, V. M. (1984). Sentenceplanningin a storycontinuation task.lnnguage andSpeech, 27,115-134.HOlscher, A, & Mohle, D. (l987). Cognitive plansin translation. InC.Faerch&G.Kasper (Eds.), Introspectioninsecond language research (pp.113-134). Philadelphia:Multilingual Matters.Hopper, R., Koch, S.,&Mandelbaum, J.(1986). Conversation analysis methods. InD.G. Ellis & W. A.Donohue(Eds.), Contemporary issues in language anddiscourse processes (pp.169-186). Hillsdale, NJ: Erlbaum.Hulstijn. J.H. (I982). Monitor use by adult second language learners (Doctoral dissertation, University ofAmsterdam). University Mkrofilms International, RPD82-70028.


18/19

130 GrahamCrooke.Hulstijn, J.H. (1989). A cognitive viewof interlanguage variability. InM. R. Eisenstein (Ed.), The dynamic

I 1 J u a ~ l p IUlltijJtumHulstijn, J. H.,& Hulstijn, W. (1984). Grammatical errorsas a function of processing constraints and explicitknowledge.1JInguage Learning, 34,23-43.Jackson, S. (1986). Building a casefor cliaims about discourse structure.In D. Ellis & W. A.Donohue (Eds.),Contemporary issues in language anddiscourse processes (pp.129-147). Hillsdale, Nl Erlbaum.Jackson, S.,& Jacobs, S. (1983). Strategyand structurein conversational infiuence attempts.CommunicationMonographs, 50,285-304.Jacobs. S.(1986). How tomakean argumentfrom examples indiscourse analysis. InD.Ellis& W.A.Donohue(Eds.), Contemporary issues in language anddiscourse processes (pp. 149-167).Hillsdale, NJ: Erlbaum.Jarvella,R.J.,& Deutsch, W. (1987). An asymmetryin producing versusunderstanding description of visualarrays.InA.Allport, D.MacKay, W. Prinz.& E.Scheerer(Eels.), Language perception andproduction (pp.41-59).Orlando. FL: Academic.Jordens,P. (1986). Production rulesin interlanguage: Evidence from caseerrorsinL2German. InE.Kellerman& M. Sharwood Smith(Eds.], Crosslinguistic influence in second language acquisition (pp. 91-109). New

York: Pergamon.Kellerman, E. (inpress). Compensatory strategiesin second language research:A critique, a revision, andsome (non-)implications forthe classroom. In R. Phillipson, E.Kellerman, L.Selinker, M.Sharwood Smith,& M.Swain (Eds.)Foreign/second language pedagogy research. Clevedon, U.K.: Multilingual Matters.Kellerman, E.,Ammerlaan, A.,Bongaerts, T., & Poulisse, N. (1990). System and hierarchy in L2 compensatorystrategies. InR.Scarcella, E.Andersen.& S.Krashen (Eds.), Developing communicative competence (pp.163-178). New York: Newbury House.Kempen, G.,& Hoenkamp, E.(1982). Anincremental procedural grammarforsentenceformulation. CognitiveScience, u, 201-253.Kluwe, R.H. (1987). Executive decisions and regulation of problem solving. In f. E.Weinert & R.H. Kluwe(Eds.), Metacognition, motiootion, andunderstanding (pp. 31-64).Hillsdale, NJ: Erlbaum.Kraemer, H.C.,& Thiemann, S.(1987). How marry subjects?Newbury Park, CA: Sage.Kraemer, H.C.,& Thiemann, S.(1989). Astrategytousesoftdataeffectively in randomized controlled clinicaltrials.Journal ofConsulting andClinical Psychology, 57, 148-154.Krashen, S.D.(1981). Second language acquisition andsecond language learning. Oxford: Pergamon.Krashen, S.D.(1982). Principles andpractice insecond language acquisition. Oxford: Pergamon.Krashen, S.D.(1985). Theinputhypothesis. Oxford: Pergamon.Krashen, S. D. (1989). Weacquire vocabulary and spelling by reading: Additional evidencefor the inputhypothesis. Modern Language Journal, 73(4), 440-464.Lackey, N.R., & Wingate, A. L. (1989). The pilotstudy:Key to research success. In P. J. Brink& M. J.Wood(Eds.), Advanced design in nursing research (pp. 285-292). Newbury Park,Ck Sage.Lapointe, S.G., & Dell, G.S.(1989).Asynthesis of recentworkin sentenceproduction. InG.N. Carlson & M. K.Tanenhaus (Sds.), Linguistic structures in language processing (pp. 107-156). Boston: Kluwer.Larsen-Freeman, D. (1975). The acquisition of grammatical morphemesby adultESL students.TESOL Quarterly, 9,409-419.Laver, J.D.M. (1973). Thedetection and correctionofslipsofthe tongue. InV. A.Fromkin (Ed.), Speech errorsas linguistic evidence (pp. 132-143). Hague: Mouton.Laver, J.D.M. (1980). Monitoring systemsinthe neurolinguistic controlofspeech production. InV. A.Fromkin(Ed.), Errors in linguistic performance (pp. 287-305). New York: Academic.Lennon, P. (1984). Retelling a storyinEnglish asa secondlanguage. InH.W. Dechert,D.Mohle, & M. Raupach(Eds.), Second language productions {pp. 50-68).Tubingen: GunterNarr.Levelt, W. J.M. (1975). Systems, skills and language learning. InA. J.VanEssen & J. P. Menting (Eds.), Thecontext of foreign-language learning (pp.83-99).Assen, Netherlands: Van Gorcum.Levelt, W. J.M. (1978). Skill theoryand languageteaching. Studies inSecond Language Acquisition, 1,53-70.

Levelt, W. J.M. (1983). Monitoring and self-repair in speech.Cognition, 14,41-104.Levelt. W. J.M.(1989). Speaking. Cambridge, MA MITPress.MacWhinney, B.,& Anderson, J.R. (1986). The acquisition of grammar. In I. Gopnick & M. Gopnick (Eds.),From models tomodules (pp. 3-25).Norwood, NJ: Ablex.MacWhinney, B.,& Osser, H. (1977). Verbal planningfunctions in children's speech.Child Development, 48,978-985.McClelland, J.L.,Rumelhart, D.E.,& Hinton, G.E. (1986). Theappeal ofparalleldistributed processing. InJ .L.McClelland, D.E.Rumelhart, & thePDPResearch Group, Parallel distributed processing (Vol. 1,pp. 3-44).Cambridge, MA: MIT Press.McDonough, S.H. (1981). Psychology in foreign language teaching. London: Allen & Unwin.McLaughlin, B.(1987). Theories ofsecond language learning. London: Edward Arnold.


19/19

Second Language Speech Production Research 131Mclaughlin, 8., Rossman, T.,& Mcleod, B. (1983). Second language learning: An information-processingperspective. Language Learning, 33,135-158.Mcleod, 8., &Mclaughlin, B.(1986). Restructuring or automaticity? Reading ina second language.lnnguageLearning,36,109-123.Merriam, S.B.(1988). Case studyresearch ineducation. SanFrancisco: Jessey-Bass.M6hle, D. (1984). Acomparison of the secondlanguagespeechproduction of different nativespeakers.InH.W. Dechert,D. Mahle, & M. Raupach (Eds.), Second language productions (pp.26--49). Tubingen: GunterNarr.M6hle, D., & Raupach, M. (1987). The representation problem in lnterlanguage theory. In W Lorscber & R.Schulze (Eds.), Perspectives on language in performance(pp. 1158-1173). Tlibingen: Gunter Narr.Morrison, D.M., & Low, G. (1983). Monitoring and the second language learner. In J.C. Richards & R.W.Schmidt (Eds.), funguage andcommunication (pp. 22S-250). London: Longman.Motley, M. T., Baars, B. 1., & Camden, C. T. (1983). Experimental verbalslip studies: A review and editingmodelof language encoding. Communication Monographs, 50,79-101.Neves, D. M., & Anderson, J. R. (1981). Knowledge compilation: Mechanisms for the automatization ofcognitive skills. InJ . R Anderson (Ed.), Cognitive skills and their acquisition (pp. 57-84). Hillsdale, NJ:Erlbeum.Newell, A, & Rosenbloom, P. S. (1981). Mechanisms of skill acquisition and the law ofpractice.In J. R.Anderson(Ed.). Cognitive skills andtheir acquisition (pp.1-56).Hillsdale, NJ: Erlbaum.Newell, A.,& Simon, H.A.(1972). Human problem solving. Englewood Oiffs, NJ: Prentice-Hall.O'Connell,D.C.(1980). Cross-linguistic investigations ofsome temporal variables in speech.In H.W. Dechert

& M. Raupach (Eds.), Towards a cross-linguistic assessmentofspeech (pp. 23-38).Frankfurt: Lang.O'Connell, D.C.,&Wiese, R (1987). Thestateof theart: Thefateofthe start.In H.W. Dechert& M. Raupach(Eds.), Psycholinguistic models ofproduction (pp.3-16).Norwood. NJ:Ablex.O'Malley, J.M., Chamot, A. 0., & Walker, C.(1987). Someapplications of'Cognitive theoryto secondlanguageacquisition. Studies inSecond funguage Acquisition, 9, 287-306.Pienemann, M. (1987). Psychological constraints on the teachability of languages. InC.Pfaff (Ed.), First andsecond language acquisition processes (pp. 143-168). Rowley, MA: Newbury House.Poulisse, N., in collaboration withBongaerts, T.,& Kellerman, E.(1990). Theuseof compensatory strategies byDutch learners ofEnglish. Dcrdrecht: Foris.Pouiisse. N.,Bongaerts, T., & Kellerman, E. (1984). On the use of compensatory strategies in lnterlanguegeperfonnance./nterlanguageStudies Bulletin, 8, 70-105.Poulisse, N.,Bongaerts, T., & Kellerman, E. (1987). Theuseof retrospective verbalreportsin the analysisofcompensatory strategies. InC.Feerch& G.Kasper (Eds.), Introspection in second language research (pp.213-229). Philadelphia: Multilingual Matters.Raupach. M. (1980). Cross-linguistic descriptions of speech performance as a contribution to contrastivepsycholinguistics. InH.W. Dechert& M. Raupach (Eds.), 1bwards a cross-linguistic assessment of speechproduction (pp. 9-22).Frankfurt: Lang.Raupach, M. (1984). Formulae in speechproduction. InH.W. Dechert,D.Mahle, & M. Raupach (Eds.), Secondlanguage productions (pp. 114-137). Tublngen: GunterNarc.Raupach, M. (1987). Procedural learningin advancedlearners of a foreign language. In J.A Coleman & R.Towell (Eds.), Theadvanced language learner (pp. 123-155). London: CILT.Rehbein.J. (1987). On fluency insecondlanguagespeech.In H.W. Dechert& M. Raupach (Eds.), Psycholinguistic models of production (pp. 97-105). Norwood, NJ: Ablex.Rochester, S.R.,& Gill, J. (1973). Production ofcomplexsentencesinmonologues and dialogues. Journal ofVerbal Learning andVerbal Behavior, 12,203-210.Rosenthal, R, & Rubin, D.B. (1985). Statistical analysis: Summarizing evidence versusestablishing facts.Psychological Bulletin, 97,527-529.Rowe, M. B.(1986). Wait time:Slowing downmaybe a wayofspeedingup! Journal of Teacher Education, 36,43-50.Ryan, T. A (1985). "Ensemble-adjusted p values": Howare theyto be weighted? Psychological Bulletin, 97,521-526.Ryle, G.(1949). Theconcept ofmind. Oxford: Methuen.Sajavaara, K. (1987). Second language speech production: Factors affecting fluency. InH.W. Dechert& M.Raupach (Eds.), Psycholinguistic models of production (pp. 45-65). Norwood, NJ: Ablex.Sajavaara, K., & Lehtonen, J. (1980). The analysis of cross-language communication: Prolegomena to thetheory andmethodology. InH.W. Dechert& M. Raupach (Eds.), Towards a cross-linguistic assessment ofspeech production (pp.55-76).Frankfurt: Lang.Santos,T.(1989). Replication inappliedlinguistics research.TESOL Quarterly, 23, 699-702.Schmidt, R.w., & Freta,S. N. (1986). Developing basicconversational abilityin a secondlanguage: A case

1991 crookes ssla

Documents