bialystok (1987), influences of bilingualism

15
  http://slr.sagepub.com/ Second Language Research  http://slr.sagepub.com/content/3/2/154 The online version of this article can be found at:  DOI: 10.1177/02676 5838700300205  1987 3: 154 Second Language Research Ellen Bialystok Influences of bilingualism on metalinguistic developme nt  Published by:  http://www.sagepublications.com  can be found at: Second Language Research Additional services and information for http://slr.sagepub.com/cgi/alerts Email Alerts: http://slr.sagepub.com/subscriptions Subscriptions: http://www.sagepub.com/journalsReprints.nav Reprints: http://www.sagepub.com/journalsPermissions.nav Permissions:  http://slr.sagepub.com/content/3/2/154.refs.html Citations: What is This?  - Dec 1, 1987 Version of Record >> 

Upload: cucu-rauss

Post on 02-Nov-2015

221 views

Category:

Documents


0 download

DESCRIPTION

Bialystok (1987), Influences of Bilingualism

TRANSCRIPT

  • http://slr.sagepub.com/Second Language Research

    http://slr.sagepub.com/content/3/2/154The online version of this article can be found at:

    DOI: 10.1177/026765838700300205 1987 3: 154Second Language Research

    Ellen BialystokInfluences of bilingualism on metalinguistic development

    Published by:

    http://www.sagepublications.com

    can be found at:Second Language ResearchAdditional services and information for

    http://slr.sagepub.com/cgi/alertsEmail Alerts:

    http://slr.sagepub.com/subscriptionsSubscriptions:

    http://www.sagepub.com/journalsReprints.navReprints:

    http://www.sagepub.com/journalsPermissions.navPermissions:

    http://slr.sagepub.com/content/3/2/154.refs.htmlCitations:

    What is This?

    - Dec 1, 1987Version of Record >>

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • Influences of bilingualism onmetalinguistic developmentEllen Bialystok York University, Ontario

    Address for correspondence: Ellen Bialystok, Department of Psychology, York University,Downsview, Ontario, Canada.

    The relationship between metalinguistic awareness and bilingualism is inter-preted in terms of a framework which defines metalinguistic awareness asconsisting of two processing components: analysis of linguistic knowledge, andcontrol of linguistic processes. It is argued that bilingualism enhances only thelatter of these processing components, so global assessments of metalinguisticability by bilingual subjects are bound to lead to inconsistent results. Somestudies are reported in which these two processing components are separated.Bilingual children are shown to be superior to monolingual children onmeasures of control of linguistic processes.

    The search for a relationship between metalinguistic development andbilingualism has produced a variety of findings. Evidence has beenpresented in support of the argument that bilingualism facilitatesmetalinguistic development (as well as other language and academicdevelopments) (Ben-Zeev, 1977; Cummins, 1978; lanco-Worrall,1972); that it is unrelated to metalinguistic development (Ben-Zeev,1977; de Avila and Duncan, 1981; Rosenblum and Pinker, 1983) andthat it impedes progress in metalinguistic achievement (Macnamara,1967; Palmer, 1972). Two possible interpretations of this lack of con-sistency are: (1) that the facilitating effects of bilingualism areconfined to certain types of metalinguistic problems; and (2) that thefacilitating effects of metalinguistic performance are confined tocertain types of bilingualism. An evaluation of these possible inter-pretations of the relationship between bilingualism and metalinguisticdevelopment requires a more precise definition for metalinguistic skillthat can be related to bilingualism in specific ways.

    Consistent with current notions of intellectual functioning in whichminimally two components relating to executive processing anddeclarative knowledge are identified (Brown, Bransford, Ferrara, andCampione, 1983; Sternberg and Powell, 1983), metalinguistic ability

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 155

    has been defined in terms of two underlying skill components(Bialystok and Ryan, 1985). These skill components refer to separableaspects of language processing which, we claim, are involved in allaspects of language use. Different uses, however, depend on each skillcomponent to various extents. Metalinguistic problems are languagetasks expressly designed so that their solution depends upon relativelyhigh levels of one or the other of these skill components.The two skill components are analysis of knowledge, which is the

    ability to construct explicit representations of linguistic knowledge,and control of processes, which is the ability to control linguisticprocesses by intentionally selecting and applying knowledge to arriveat a solution. The development of each is considered to be gradual andcan be characterized by values on a continuum. Moreover, the deve-lopment of each is assumed to advance in response to different experi-ences. Bilingual children, it is argued, have an advantage only in thesecond of these and should therefore demonstrate superior perform-ance only on those metalinguistic tasks specifically designed to requirehigh levels of control.

    Analysis of knowledge is the skill component responsible formaking explicit those representations that had previously beenimplicit or intuitive. In cognitive development, Karmiloff-Smith(1986) describes the childs development of a skill as being based onthree ordered phases of representation: implicit; explicit 1, in whichthe implicit knowledge is structured but not accessible to con-sciousness ; and explicit 2, in which conscious access becomes possible.During the process of language acquisition, children are constantlyanalysing and structuring knowledge, moving it, in Karmiloff-Smithsterms, through the three phases of representation. At any point intime, different aspects of the childs linguistic knowledge will be repre-sented in each of the three phases. The result is the appearance of acontinuum, with representations of linguistic knowledge occupyingall points. The reason for that continuum, however, is that discreteaspects of linguistic knowledge are subjected to continual structuringand analysis. It is the process responsible for this structuring that isindicated by the skill component of analysis of knowledge.The language uses encountered by children during various stages of

    language acquisition require different levels of analysis of theirlinguistic knowledge. To participate in conversations, children easilyproduce and understand sentences for which they have no explicitknowledge of structure (Bowerman, 1982). Explication of thatstructure has been shown to be a necessary, but not sufficient, condi-tion for writing coherent texts (Scardamalia and Paris, 1985) and forthe elaboration of new categories of knowledge required for learningto read (Menyuk, 1981). Metalinguistic tasks are usually contrived so

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 156

    that the solution is achieved only if the relevant structure is knownexplicitly. Some aspects of structure which emerge are knowledge ofthe units of speech (words, syllables, phonemes; see Ehri, 1979),understanding of the relationship between forms and meanings (e.g.the sun-moon problem studied by Piaget, 1929; Scribner and Cole,1981; Vygotsky, 1962), and awareness of syntax (e.g. grammaticalityjudgement and correction problems, de Villiers & de Villiers, 1974).Accordingly, although children appear to possess a relatively sophis-ticated command of linguistic concepts, their explicit knowledge ofthese concepts is inadequate for solving most metalinguistic tasks.High levels of analysis of knowledge are associated with constructssuch as conscious or explicit knowledge.The development of control of linguistic processes reflects the

    childs ability to apply operations intentionally in the solution to aproblem. Solving a conservation of liquid problem, for example,requires that the child directs attention to both the height and width ofthe beaker, in spite of the compelling perceptual impression that onlythe height is relevant. In language use, attention is usually focussed onthe meanings, a strategy that is effective for ordinary conversation(Hakes, 1980). But certain uses of language require control over thatattention so that it can be directed to specific aspects of the language.The ability to switch back and forth between forms and meanings,between graphemes and phonemes, between words and intentions, forexample, is a crucial part of fluent reading (Lesgold and Perfetti,1981). Metalinguistic tasks typically require children to focus on formand sometimes ignore or suppress meaning. Deliberately controllingattention in this way requires control of processing. Just as childrengradually acquire this skill in other cognitive domains, so do theyacquire it in using language. The knowledge of procedures for solvinga variety of language problems and the ability to execute thosesolutions through appropriate attentional focus is the function ofcontrol of linguistic processing. High levels of control of processingare associated with constructs such as intentionality.

    It is the development of this second skill component that seems to bemost affected by bilingualism. Vygotsky (1962) suggested thatbilingual children should be more advanced than monolingualchildren in solving Piagets (1929) sun-moon problem. Becausebilingual children have the experience of two linguistic systemslabelling the same conceptual system, the arbitrary connectionbetween forms and meanings is more readily apparent. Moreover,these children have more experience attending to formal linguisticfeatures that may change even though meanings are constant, as indeciding between languages, attending to different phonologicalsystems and choosing the correct label for an object. In addition, their

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 157

    clearer representation of linguistic and conceptual information asseparate structures makes problems involving selective attention tolinguistic features less difficult for bilingual children. It is the clearconcepts of word meaning and word form and their relation thatenable the child to solve metalinguistic problems based on the separa-tion of forms and meanings. And it is these problems for whichbilingual children are claimed to be advantaged.The general hypothesis, then, is that bilingual children should be

    more advanced in solving metalinguistic problems requiring highlevels of control of processes than are monolingual children, butequivalent to monolingual children in solving metalinguistic problemsrequiring high levels of analysis of linguistic knowledge. Comparisonsare difficult, however, because many factors conspire to make subjectgroups unequal in any global comparison. Social class and languagebackground are known to be relevant in determining performance onmetalinguistic tasks, irrespective of differences in bilingualism (e.g.Cummins, 1976). The more important factor, however, may be levelsand types of literacy. Our claim is that literacy is a significant factor inthe growth of analysis of linguistic knowledge (Bialystok and Ryan,1985; Ehri, 1979). Accordingly, different experiences with literacymay influence childrens levels of analysis of knowledge, thus inter-vening in a simple comparison of the way in which children withdifferent bilingual experiences solve metalinguistic problems (since inspite of efforts to control for differences, metalinguistic problems arecertain to involve analysis of knowledge as well as control ofprocessing). Nonetheless, the assumption underlying this research hasbeen that attention to experimental controls and cautious interpreta-tion of findings reveals important patterns that contribute to anunderstanding of the relation between bilingualism and metalinguisticawareness.

    A variety of evidence has shown a bilingual superiority on metalin-guistic problems demanding high levels of control of processes.Vygotskys prediction, for example, that a bilingual advantage wouldappear for the sun-moon problem has been demonstrated by Scribnerand Cole (1981) and lanco-Worrall (1972). Further, lanco-Worrall(1972) found that bilingual children were more sensitive than mono-lingual children to semantic relations between words when the mean-ings competed with phonetic similarity. Finally, middle class bilingualchildren were more successful in solving a metalinguistic problem inwhich words needed to be substituted into a sentence creating non-sense, requiring that the child ignore the meaning (Ben-Zeev, 1977).Working class children showed no advantage relative to their mono-lingual peers on the same problem.The tasks used in these studies all depend on levels of control to a

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 158

    greater extent than they do on levels of analysis. Comparisons aredifficult, however, because the tasks are still different from eachother, so it is difficult to interpret results in terms of these skill com-ponents. Our approach has been to construct versions of a meta-linguistic problem so that each version (task) differentially requiresgreater levels of analysis of linguistic knowledge or control of linguis-tic processes for its solution. It is no doubt the case that each tasksomehow requires both skill components; the argument is simply thatthe primary component can be determined. Comparisons can then bemade among the tasks both in terms of their relative difficulty for aspecific group of children and for the relative difficulty of a singleversion by different groups of children.Judgement of sentence acceptability is one problem that we have

    adapted in this manner (Bialystok, 1986a). Judging grammaticalityhas often been reported to be easier than correcting deviant sentences(de Villiers and de Villiers, 1972; Gleitman, Gleitman and Shipley,1972; Hakes, 1980; Scribner and Cole, 1981). This difference weattribute to the greater demands for analysis of knowledge given bythe correction task. Similarly, manipulating the formal structure ofsentences that contain semantic anomaly has been shown to be moredifficult than applying similar manipulations to meaningful sentences(e.g. Ben-Zeev, 1977). The anomalous sentences, in our terms,distract the child from attending to form.

    Three studies, each involving about 120 children, were conducted.In Studies 1 and 2, children were in Junior Kindergarten (5 years old),Grade 1 (7 years old), and Grade 3 (9 years old). In addition,approximately half of the children in each grade were bilingual. InStudy 1, the bilingual children came from homes in which English wasnot spoken at all; in Study 2, the bilingual children were English-speaking students enrolled in French immersion programmes. InStudy 3, children were in Grade 2 and Grade 3, and half the children ineach grade were bilingual in that they spoke a language other thanEnglish at home (c.f. Study 1). In all cases, testing was conducted inEnglish.

    Children were asked to judge or correct sentences for their syntacticacceptability, irrespective of meaningfulness. This instruction wasexplained to the children through a series of example sentences whichvaried in their grammaticality and meaningfulness. A puppetcharacter was used to present each sentence and children were told totell the puppet if he said the sentence the right way and that it wasfun to be silly, so only grammaticality was to be judged. A highdemand for analysis of knowledge was operationalized in terms ofcorrecting, as opposed to judging, grammatical structure and ofdetecting ungrammatical, as opposed to grammatical, sentences

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 159

    (Bialystok, 1979). A high demand for control of processing was opera-tionalized in terms of solving the problem under conflicting condi-tions, namely when the value of the grammaticality was incongruentwith the value of the meaningfulness. There were four types ofjudgment tasks. Meaningful sentences could be grammatical (Why isthe dog barking so loudly?) (GM) or not (Why the dog is barking soloudly?) (gM); and anomalous sentences could be grammatical(Why is the cat barking so loudly?) (Gm) or not (Why the cat isbarking so loudly?) (gm). The two conflicting sentences (gM,Gm) arenot equivalent in their demands. For gM, the intact meaning does notcreate much distraction, and so the problem is to have sufficientlyanalysed knowledge to recognize the grammatical error. Thus, thesesentences test the level of analysis of linguistic knowledge. For Gm,the anomalous meaning is a compelling distraction, and the instruc-tion to focus only on form requires intentional effort to separate theform and meaning, ignoring the latter. Once this separation isachieved, there is little burden on analysed knowledge, since thesentence is grammatical. Thus, these sentences test the level of controlof processing.

    Across the three studies, the manipulation of the control andanalysis requirements in the different versions of the grammaticalityjudgement task was reflected in problem difficulty. The correctiontask was more difficult than the judgement task and the incongruentitems in the judgement task were more difficult than their congruentcounterparts. Similarly, ignoring the anomalous meanings in the cor-rection task according to the instructions, in order to receive the scorefor meaning corrections, was more difficult than repairing thegrammar. The interpretation is that these two factors, analysis ofknowledge and control of processes, constitute an important aspect ofthe structure of this metalinguistic task, and ability in these two skillcomponents is prerequisite to solving the problems.The effects of bilingualism in the solution to these problems are

    evidenced as interactions between language and the ability to solvecertain of the items. Changes in control demands were studied indifferences among the four types of sentences in the judgement task asa function of the different language groups. The data obtained fromStudies 1 and 2 are shown in Figure 1; Study 3 replicated the patternsin Study 1.The sentence judgements requiring the greatest levels of control are

    the Gm items in which the meaning is made salient by means ofanomaly, but the task requires attending to the intact grammaticality.The bilingual children in all three studies consistently judged these Gmitems more accurately than did monolingual children, at all ages testedand for all types of bilingualism.

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 160

    Figure 1 Mean number correct judgements out of six for judgement task by group

    The difference between monolingual and bilingual children for thegM items is a measure of analysis of linguistic knowledge. Here thedifference depended on the childs type of bilingual experience. InStudies 1 and 3, the monolingual children judged these items betterthan did bilingual children. These were the studies in which childrenwere bilingual by speaking one language at home while being educated(and socialized) in English. These children were weaker thanmonolingual children for problems requiring richer levels of analysisof linguistic knowledge. The children who were bilingual throughFrench immersion programmes performed equivalently to theirmonolingual peers on these gM items, indicating no relative weaknessin their level of analysis of linguistic knowledge.

    In general, the bilingual children in Study 2 performed moreimpressively than did the bilingual children in Studies 1 and 3. Thereare two differences between the bilingual groups in these studies. Thefirst, and most salient, is that the bilingual children in Studies 1 and 3were being tested in their second language, while the monolingualchildren in those studies and the bilingual children in Study 2, were

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 161

    tested in their native language. While this is no doubt a relevant differ-ence, it must be noted that it did not serve uniformly to suppress thescores of these children relative to their monolingual peers; thebilingual children scored higher than the others on the tasks con-sidered to involve control of processing.The second difference between the groups is that the children in

    Study 2 were also biliterate, reading in both English and French. Thisbiliteracy, coupled with the fact of being native speakers of English,would have the effect of increasing their level of analysis of knowledgeof English relative to the bilingual children in Studies 1 and 3. Thustheir advantage is specific: these bilingual children perform betterthan the other two groups of bilingual children (and possibly betterthan some monolingual children) on tasks demanding high levels ofanalysis of knowledge. But children who were bilingual in any sense,uniformly performed better than monolingual children on tasksdemanding high levels of control of processing. Thus a facilitatingeffect of bilingualism may be mediated by specific levels of mastery ofanalysis of knowledge achieved through other means. For this reasonit appears that on some tasks, e.g. high control tasks, bilingualismexerted a facilitating effect while on other tasks, e.g. high analysis ofknowledge tasks, bilingualism created some disadvantage.Another problem examined in this paradigm is counting or

    segmenting words in sentences (Bialystok, 1986b). Childrens failureon these problems has been interpreted as evidence of their lack ofunderstanding the concept of word (Ehri, 1979; Fox and Routh, 1980;Papandropoulou and Sinclair, 1974). Such a concept, however, iscritical to metalinguistic notions of language structure. In terms of thecurrent conception of metalinguistic ability, the difficulty with theseword concept tasks can be traced to the two skill components ofanalysis and control. Thus, removing the barriers imposed on theproblem by excessive levels of analysis and control should allow accessto the solution by younger children who, in more holistic definitionsof metalinguistic ability, would be excluded. Moreover, those aspectsof the solution most governed by control of processing should besolved more readily by bilingual children than by monolingual ones.Four versions of a word concept task were constructed which varied

    in their demands on the two skill components. A more analysedconcept of word is one in which the boundaries that isolate words areclear. In a less analysed concept of word, the child may realize that aword is some unit of language but be uncertain as to the definingproperties. Hence one difficulty in counting the number of words in asentence is in deciding on the appropriate boundaries: is a syllable aword; is a morpheme a word? Mehler (1982) has shown that a strategyof syllabic segmentation is primary, even among infants, so an early

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 162

    and relatively unanalysed concept of word may tie the boundaries tosyllables. The simplest problem, then, would be to count the numberof words in a sentence in which all the words were monosyllabic. Herea correct solution could be achieved through the application of avague and possibly slightly incorrect hypothesis of word boundaries.Counting tasks can be made progressively more difficulty for thechilds analysis of knowledge by gradually including sentences thatcontain polysyllabic words (alligator) for which a syllable hypothesismust be abandoned, and double morpheme words (rainbow,sunshine) for which a meaning unit hypothesis must be abandoned.The difficulty for control of processing in the word count problems

    is again in keeping attention fixed on forms, in spite of the temptationto examine meanings. It is difficult to see the formal constituent struc-ture of a sentence when there is an integrating and overridingmeaning. Problems could be made simpler for their control demands,then, by removing the meaning, removing therefore the linguisticfeature competing for the childs attention. This was achieved byscrambling the words in the sentences produced in the various analysisof knowledge conditions. If children could successfully count thevarious kinds of words in the different analysis of knowledge condi-tions when the sentence had been scrambled, then their failure on themore traditional sentence conditions could not be attributed to theirlack of (analysed) knowledge of words. If children could count thewords equally in both sentence and scrambled conditions, then themeanings were not presenting an inordinate challenge to their capaci-ties for selective attention.A study was conducted with 62 children in Junior Kindergarten and

    Grade 1, the latter group being divided between English programmeand French immersion and, therefore, somewhat bilingual. Childrenwere individually tested and orally presented with a series of eightsentences and eight scrambled strings. Following each sentence orstring, they were asked how many words it contained. (Children had aset of markers to help them count while they listened to the sentence orstring.) The results in Figure 2 are collapsed across the differentanalysis of knowledge conditions and show the effects of the controlmanipulation of sentence versus string on the three groups. Thebilingual children could count the words with equivalent accuracywhether they were in real sentences or in scrambled strings. Themonolingual children, even in Grade 1, counted better when the wordswere arranged in strings. Their performance in the high controlsentence condition was inferior to that of their bilingual peers in theFrench immersion programme.

    These two problems, grammaticality judgment and word concept,are examples of the way in which metalinguistic tasks can be made

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 163

    Figure 2 Mean number correct judgements out of eight for sentences and strings by group

    more precise to examine separable components of metalinguisticability. The results obtained from these and other tasks permit anumber of generalizations about the relationship between linguisticawareness and bilingualism. The tasks themselves were madesystematically more difficult by the experimental manipulations.These manipulations have been interpreted as increasing the demandson control and analysis. Metalinguistic problems are difficult, then,because they demand particularly high levels of mastery of these skillcomponents.The effect of bilingualism was seen primarily on the development of

    control of processes. This effect, however, was mediated by a numberof factors, possibly including biliteracy. In this sense, it may be thatthe additional boost to analysis of knowledge offered by biliteracy iseither a catalyst or even a precondition for the potentially facilitatingeffects of bilingualism to be demonstrated.The relation between metalinguistic ability and bilingualism, then,

    is construed as a set of conditions that hold between the childsmastery of the skill components of analysis of knowledge and controlof processes and the demands posed by specific metalinguistic tasks.

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 164

    These skill components are considered to be the basis of language pro-ficiency, and it is through mastery of these components that advancedforms of language use become possible. Through their commonreliance upon this underlying set of skill components, differentaspects of language proficiency have important relationships to eachother. Thus literacy, with its special connection to analysis ofknowledge, is an important constraint on the relationship betweenbilingualism and metalinguistic development. Learning to read in twolanguages has often been shown to be easy for children who haveadvanced (in our terms, analysed), conceptions of language. This rela-tion, too, may be interpreted in terms of the way in which individualchildren recruit their ability to analyse linguistic knowledge andcontrol attentional processes to master new problems that imposespecific demands on these abilities. These relationships are examinedin other studies in this research programme using different meta-linguistic tasks, different operationalizations of the constructs anddifferent subject samples. It is in the synthesis of all these studies that,it is hoped, a clearer understanding of the cognitive basis of languageproficiency and its reflection in special language skills will bediscovered.

    Acknowledgement t .This article is based on a paper presented at the Language AcquisitionResearch Seminar, Utrecht, The Netherlands, in September 1985. Theresearch reported in this paper was funded by Grant A2559 from theNatural Sciences and Engineering Research Council of Canada.

    References

    Ben-Zeev, S. 1977: Mechanisms by which childhood bilingualism affectsunderstanding of language and cognitive structures. In P.A. Hornby,P.A., editor, Bilingualism: psychological, social and educationalimplications, New York: Academic Press.

    Bialystok, E. 1979: Explicit and implicit judgments of L2 grammaticality.Language Learning 29: 81-103.

    1986a: Factors in the growth of linguistic awareness. Child Development57: 498-510.

    1986b: Childrens concept of word. Journal of PsycholinguisticResearch 15: 13-32.

    Bialystok, E., and Ryan, E. 1985: Towards a definition of metalinguisticskill. Merrill-Palmer Quarterly 31: 229-51.

    Bowerman, M. 1982: Starting to talk worse: clues to language acquisitionfrom childrens late speech errors. In Strauss, S., editor, U-Shapedbehavior growth, New York: Academic Press.

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 165

    Brown, A.L., Bransford, J.D., Ferrara, R.A., and Campione, J.C. 1983:Learning, remembering, and understanding. In Flavell, J.H. andMarkman, E.M., editors, Carmichaels manual of child psychology,Volume III, New York: Wiley.

    Cummins, J. 1976: The influence of bilingualism on cognitive growth: asynthesis of research findings and explanatory hypotheses. WorkingPapers on Bilingualism 9: 1-43.

    1978: Bilingualism and the development of metalinguistic awareness.Journal of Cross-Cultural Psychology 9: 131-49.

    de Avila, E.A. and Duncan, S.E. 1981: Bilingualism and the metaset. InDuran, R.P., editor, Latino language and the metaset, Norwood, N.J.:Ablex.

    de Villiers, P.A. and de Villiers, J.G. 1972: Early judgments of semanticacceptability by children. Journal of Psycholinguistic Research1: 290-310.

    1974: Competence and performance in child language: Are childrenreally competent to judge? Journal of Child Language 1: 11-22.

    Ehri, L.C. 1979: Linguistic insight: threshold of reading acquisition. InWaller, T.G. and MacKinnon, G.E., editors, Reading research:advances in theory and practice, New York: Academic Press.

    Fox, B. and Routh, D.K. 1980: Phonemic analysis and severe reading dis-ability in children. Journal of Psycholinguistic Research 9: 115-19.

    Gleitman, L.R., Gleitman, H., and Shipley, E.F. 1972: The emergence of thechild as grammarian. Cognition 1: 137-64.

    Hakes, D.T. 1980: The development of metalinguistic abilities in children.New York: Springer-Verlag.

    Ianco-Worrall, A. 1972: Bilingualism and cognitive development. ChildDevelopment 43: 1390-400.

    Karmiloff-Smith, A. 1986: From metaprocess to conscious access: evidencefrom childrens metalinguistic and repair data. Cognition 28: 95-147.

    Lesgold, A.M., and Perfetti, C.A., editors 1981: Interactive processes inreading. Hillsdale, NJ: Erlbaum.

    Macnamara, J. 1967: The effects of instruction in a weaker language.Journal of Social Issues 23: 121-35.

    Mehler, J. 1982: Studies in the development of cognitive processes. InStrauss, S., editor, U-Shaped behavior growth, New York: AcademicPress.

    Menyuk P. 1981: Language development and reading. In Flood, J., editor,Understanding reading comprehension, Newark, Del.: InternationalReading Association.

    Palmer, M. 1972: Effects of categorization, degree of bilingualism andlanguage upon recall of select monolinguals and bilinguals. Journal ofEducational Psychology 63: 160-64.

    Papandropoulou, I. and Sinclair, H. 1974: What is a word? Experimentalstudy of childrens ideas on grammar. Human Development 17: 241-58.

    Piaget, J. 1929: The childs conception of the world. London: Routledge &Kegan Paul.

    Rosenblum, T. and Pinker, S.A. 1983: Word magic revisited: Monolingual

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from

  • 166

    and bilingual childrens understanding of the word-object relationship.Child Development 54: 773-80.

    Scardamalia, M. and Paris, P. 1985: The function of explicit discourseknowledge in the development of text representations and composingstrategies. Cognition and Instruction 2: 1-39.

    Scribner, S. and Cole, M. 1981: The psychology of literacy. Cambridge, MA:Harvard University Press.

    Sternberg, R.J. and Powell, J.S. 1983: The development of intelligence. InFlavell, J.H. and Markman, E.M., editors, Hundbook of childpsychology, Volume III, New York: Wiley.

    Vygotsky, L.S. 1962: Thought and language. Cambridge, Mass.: MIT Press.

    by F Nico on October 30, 2012slr.sagepub.comDownloaded from