27_tomasello

8/6/2019 27_tomasello

1/11

Joint Attention and Early LanguageAuthor(s): Michael Tomasello and Michael Jeffrey FarrarSource: Child Development, Vol. 57, No. 6 (Dec., 1986), pp. 1454-1463Published by: Blackwell Publishing on behalf of the Society for Research in ChildDevelopmentStable URL: http://www.jstor.org/stable/1130423

Accessed: 09/06/2009 12:28

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=black.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Blackwell Publishing and Society for Research in Child Developmentare collaborating with JSTOR to digitize,

preserve and extend access to Child Development.

http://www.jstor.org
http://www.jstor.org/stable/1130423?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=blackhttp://www.jstor.org/action/showPublisher?publisherCode=blackhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/1130423?origin=JSTOR-pdf


2/11

Joint Attention and Early LanguageMichael Tomasello and Michael Jeffrey FarrarEmory University

TOMASELLO,ICHAEL,ndFARRAR,MICHAELEFFREY.ointAttention ndEarlyLanguage. HILDDEVELOPMENT,1986,57, 1454-1463. Thispaperreports2 studies thatexplorethe roleofjointatten-tional processes in the child's acquisitionof language.In the firststudy,24 childrenwere video-taped at 15 and 21 months of age in naturalistic nteractionwith their mothers.Episodes of jointattentional focus between mother and child-for example, joint play with an object-wereidentified.Inside, as opposed to outside,these episodes both mothersand childrenproducedmoreutterances,mothers used shorter sentences and more comments,and dyads engaged in longerconversations.Inside joint episodes maternalreferencesto objects that were alreadythe child'sfocus of attentionwere positively correlatedwith the child'svocabularyat21 months,while objectreferences thatattemptedto redirect the child'sattentionwere negativelycorrelated.No measuresfromoutside these episodes relatedto child language.In an experimental tudy,an adultattemptedto teach novel words to 10 17-month-old hildren.Wordsreferring o objectson which the child'sattentionwas already ocusedwere learnedbetter than wordspresented n an attempt o redirect hechild's attentional ocus.By the time children begin productivelanguage use, they have already establishedwith their caregivers a variety of social-communicative routines. Ninio and Bruner(1978) and Ratner and Bruner (1978) analyzedthe structure of these routines and demon-strated how such nonlinguistic interactions"scaffold" the child's early language. In ef-fect, these interactions provide the youngchild with a predictable referential contextthat makes both her and her mother's lan-guage immediately meaningful. In his theo-retical work, Bruner (e.g., 1981, 1983, 1985)has stressed that the underlying mechanismat work in these mother-child "formats" isjoint attention. Because young children do notpossess adult devices-either linguistic ornonlinguistic-for establishing the joint at-tention necessary for communication, recur-

rent interactive episodes help the infant to de-termine adults' attentional focus and thus theintended referent of their language. In thisway, formats support early communicative in-teractions and so facilitate the child's earlylanguage development (see also Bakeman &Adamson, 1984).Tomasello and Todd (1983) provided thefirst direct evidence that individual differ-ences in the ability of mother-child dyads to

establish and maintain a joint attentionalfocus are related to the child's subsequentlanguage growth. They videotaped mother-child dyads in their homes with a set of noveltoys at monthly intervals for a period of 6months, beginning with the child's first birth-day. The amount of time dyads spent in jointattentional episodes during the 6 months waspositively related to the child's vocabularysize at the end of this period. Several lines ofevidence, including cross-lagged correlations,supported the argument that these episodesfacilitated the child's early language develop-ment. This finding was replicated in a studycomparing singleton and twin children (To-masello, Mannle, & Kruger, 1986), in whichpositive correlations were found betweentime in joint attention at 15 months of age andvocabulary size at 21 months of age for eachgroup of children separately as well as for thesample as a whole.

A second finding of these studies wasthat directiveness on the part of mothers-either verbal or nonverbal attempts to directthe child's attention or behavior-was nega-tively related to the proportion of objectlabels in the child's vocabulary. Others havefound a similar relationship (e.g., Della Corte,Benedict, & Klein, 1983; Nelson, 1973). Nel-

We would like to thank the mothersand the childrenwho made this study possible. Thanksalso to LaurenRio,NancyHodge,and SaraMannlefortheirhelp in datacollectionandcoding.Thisresearch was presented to the Boston Child LanguageConference on October25, 1985,and sup-ported in part by NIMH Grant no. 1 R03 MH38997-01A1. Address for correspondence: MichaelTomasello,Departmentof Psychology, EmoryUniversity,Atlanta,GA 30322.JeffreyFarrar s nowat the Department of Psychology, University of Denver, Denver, CO 80210.[ChildDevelopment,1986,57, 1454-1463.? 1986by the SocietyforResearch n Child Development,Inc.All rightsreserved.0009-3920/86/5706-0011$01.00]


3/11

Tomasello and Farrar 1455son (1981) hypothesized that this relationshipis due to the child's inferences about the func-tional significance of language based on theway people around him use it. If adults uselanguage primarily to refer to and categorizethe world (e.g., naming novel objects), thechild will infer that this is its primary functionand the acquisition of object labels will bevery important. Conversely, if adults are con-stantly using language for social-regulativepurposes (e.g., to greet, thank, exhort, pro-hibit), the child will infer that this is its pri-mary function and the acquisition of objectlabels will be less important.

Tomasello and Todd (1983) offered a dif-ferent interpretation. They argued that adultdirectiveness makes it more difficult for childand adult to establish a joint attentional focus.When the adult attempts to redirect thechild's attention in referring to an object, ifthe child is to determine the intended ref-erent she must shift her attention so as tocoordinate with the adult's. On the otherhand, when the adult's reference follows intothe child's already established attentionalfocus, the child need not actively make such adetermination; coordination of attention de-pends only on the adult's skill at determiningthe child's focus. These authors thus arguethat adult directiveness has an effect not somuch on the child's overall assessment of thefunction of language but rather on the learn-ing conditions surrounding the acquisition ofindividual words.

In the current view, then, joint attentionis important for early language both on the"macro" level of extended periods of adult-child interaction and on the "micro" level ofadult-child attempts to coordinate a specificpiece of language with a joint attentionalfocus on its intended referent. The currentstudy was designed to investigate these twotypes of attentional process in more detail.Two studies were conducted. The first wasbased on naturalistic observation of childrenjust beginning to learn language. In contrastto Tomasello and Todd (1983), who lookedonly at time spent in joint attentional epi-sodes, the current study focused on the lan-guage that occurred in these episodes. On themacro level, it was hypothesized that mother-child linguistic interaction would be facili-tated when the interactants were jointlyfocused on some aspect of the nonlinguisticcontext. It was thus expected that inside, asopposed to outside, episodes of joint attentionmother-child dyads would talk more and carryon longer conversations. Further, it was ex-pected that while children would be en-

couraged to use longer sentences, motherswould use shorter sentences in these epi-sodes because the intensity of these interac-tions encourages mothers' best Child Di-rected Speech speech register. In addition,because joint focus on the nonlinguistic con-text provides a predetermined conversationaltopic, mothers were expected to use less di-rective language inside joint episodes. On themicro level, it was hypothesized that objectlabels presented in an attempt to follow intothe child's attentional focus would facilitatethe establishment of joint attention and thusbe positively related to the child's use of ob-ject-names. Conversely, object labels pre-sented in an attempt to redirect the child'sattentional focus were expected to discouragejoint attention and thus be negatively relatedto the child's lexical acquisition. These rela-tionships were expected to be stronger insidethan outside macro-level episodes of joint at-tention.

The second study was a lexical trainingstudy designed to provide experimental cor-roboration for the findings on the micro level,that is, for the relationship between direc-tiveness and lexical acquisition. Children intheir second year of life were presented withnovel object words either in an attempt to re-direct their attentional focus or, alternatively,in an attempt to follow into their current at-tentional focus. It was predicted that childrenwould learn new words more easily whenthey were presented in the latter condition. Ifthis were indeed the case, it would provideexperimental corroboration that adult direc-tiveness is associated with slower early lex-ical acquisition, and it would provide evi-dence for the operation of attentional factorsin this process.Study 1Method

Subjects.-Twenty-four white, middle-class children-equal numbers of firstbornsand later-borns, males and females-were re-cruited by personal contact from local day-care facilities. Children were all between 12and 18 months of age at recruitment (meanage = 14.6 months) and, according to mater-nal report, had begun productive languageuse.Observational procedure.-Each moth-er-child dyad was videotaped at home for aperiod of 15 min on two occasions, once whenchildren were 15 months and once when theywere 21 months of age. Dyads were pwovidedwith a set of novel toys and given no specialinstructions except to "Do what you normally


4/11

1456 Child Developmentwould do." A research assistant and camera-person were present at each session. Motherswere told at recruitment that we were inter-ested in their child's language development,and they were instructed at that time to beginnoting the child's normal language practices.At each of the taping sessions, mothers wereinterviewed about their child's use of lan-guage.

Coding procedure.-Each videotapewasfirst coded for episodes of joint attentionalfocus. As defined by Tomasello and Todd(1983), these were episodes that met the fol-lowing conditions: (1) they began with onemember of the dyad initiating interactionwith the other, (2) both members then visu-ally focused on a single object or activity for aminimum of 3 sec.(either member could lookaway briefly during an extended interaction),and (3) at some point during the joint focus(possibly at initiation) the child directed someovert behavior toward the mother (especiallya look to the face) as evidence that he wasaware of their interaction, thus excludingmere onlooking. An example might be: thechild hands the mother a spoon, looking toher face; she places it in a cup; he takes it out,mouths it, and puts it back in the cup, lookingto the mother; they continue this until some-one (usually the child) shifts attention. Hadthe child played with these objects alone, thiswould not have been a joint attentional epi-sode even if the mother was visually focusedon the objects throughout.

The language inside and outside theseepisodes was of interest. It was coded in twoways. First, each videotape was transcribedby a team of two research assistants. An inde-pendent assistant coded the transcripts forspecific language measures and then, usingthe joint attention coding, tabulated data onthe language measures separately for insideand outside the episodes of joint attentionalfocus. Language measures for both motherand child were: number of utterances andMean Length of Utterance (MLU). Formothers only, the proportional distribution ofutterances into comments, questions, and di-rectives was also determined. For childrenonly, the total number of words and object-labels per minute was also determined. In ad-dition, two measures of the dyad's conversa-tional behavior were of interest: number ofconversations (a conversation was defined asadjacent utterances on a common topic) andmean number of child turns per conversation(as an indication of conversation length). Thecomparison of these measures inside and out-side episodes of joint attentional focus con-stituted the macro level of analysis.

The second coding, on a more microlevel, concerned attentional factors associatedwith maternal reference to objects. In a sepa-rate coding of the videotapes, an independentteam of two coders established for each ma-ternal reference to an object (in which the ob-ject word received some prosodic stress): (a)whether or not it was made in an attempt tofollow into the child's ongoing attentionalfocus (i.e., visual), as opposed to an attempt toredirect her attention or behavior, (b) whetheror not the mother gestured or provided someother nonverbal indication of her attentionalfocus while making the reference, and (c)whether or not the child actually visuallyfocused on the object at the time of the objectreference. In this way, each object referencewas assigned one of eight unique patternsgenerated by a factorial combination of thethree dichotomies. Each of these patterns wasdesignated by a sequence of three "+" or" " symbols, one for each of the three crite-ria used in their determination. For example,an object reference in which the adult fol-lowed into the child's attentional focus, ges-tured, and the child focused on the object suc-cessfully was designated by "+ + +". Thesedata were then tabulated separately for insideand outside periods of joint attentional focus.

It is important to note that these twolevels of analysis are independent, not only inthe methodological sense that they werecoded independently, but also conceptually.Though on the surface it would seem that ajoint attentional focus on the macro levelwould automatically imply that the mother'sobject references would follow into thechild's focus, this is not necessarily so. First,joint attentional episodes sometimes involveseveral objects (e.g., placing blocks in a bowl).If a mother directed the child's attention toone of these objects and the child was focusedon another, then this was considered a direc-tive inside a joint episode. Second, a mothercould make an attempt within a joint episodeto redirect her child's attention to outside ob-jects. If the child did not attend, or attendedonly briefly and then returned to the object ofjoint focus, this also was counted as a direc-tive within a joint episode. Conversely, itcould also happen that a mother could followinto her child's focus when not in a joint at-tentional episode. If this did not result in anextended (3-sec) period ofjoint focus, this wascounted as an attempt to follow into thechild's attention outside a joint attentional ep-isode.

The language interview used to assessthe child's language development at 15 and21 months of age was an adaptation of the


5/11

Tomasello and Farrar 1457Bates (1979) interview which, in addition toutilizing spontaneously generated informa-tion, prompts the mother to provide examplesof the child's language use by asking herabout specific contexts in which children talk.(For example: What does she do when shewants food? Any special foods? What aboutwhen she wants her bottle? When she's in thehigh chair? At the refrigerator? In the store?)From this interview, a vocabulary list (includ-ing pat phrases) was compiled. Vocabularysize was computed, as well as the proportionof the child's lexical items that were objectlabels (i.e., general nominals as defined byNelson, 1973). This latter measure was usedin an attempt to capture language-acquisitionstyle independently of sheer size of vocabu-lary.

Reliability was computed for each of themeasures by having a second team of assis-tants code 20% of the subjects and computethe percentage of their agreement with theoriginal coders. Reliabilities were as follows:judgments of joint attentional episodes (dura-tions had to be within 3 sec) agreed at 84%;child language measures (including conversa-tion) agreed at 88%-100%; maternal languagemeasures (including types of object refer-ence) agreed at 82%-100%.ResultsAt both time periods, mother-child dyadsspent about two-thirds of their interactiontime inside joint attentional episodes andabout one-third of their time outside these ep-

isodes. Because of this difference, all mea-sures of frequency were divided by the appro-priate measure of time to yield a "per-minute" frequency. All other measures wereproportions of one language measure relativeto another. Each mother and child languagemeasure was analyzed with a 2 x 2 repeated-measures ANOVA, using joint attentionalstate (inside and outside) and child age (15and 21 months) as independent variables.

Child and dyad language.-Table 1 pre-sents means and standard deviations for allchild and dyad language measures. All ofthese measures were higher inside than out-side the joint attentional episodes; for four ofthe six measures the difference was statisti-cally significant. Inside, as opposed to out-side, joint attentional episodes children pro-duced more: utterances per minute, F(1,23) =11.72, p < .05; words per minute, F(1,23) =10.02, p < .05; and words referring to objectsper minute, F(1,23) = 17.16, p < .01. Thechild's average number of turns per conversa-tion was higher inside as opposed to outsidejoint attentional episodes, F(1,18) = 16.01, p< .01 (only 19 dyads had conversations).Child age produced several main effects andinteracted with joint attentional state for sev-eral of these measures, as shown in Table 1.In each case of interaction, differences be-tween the values inside and outside joint at-tentional episodes were greater when thechild was 21 months of age.

TABLE 1LANGUAGEMEASURESNSIDEANDOUTSIDE OINTATTENTIONALPISODESATBOTHCHILDAGES

15 MONTHS 21 MONTHSInside Outside Inside OutsideLANGUAGEMEASURES Joint Episodes Joint Episodes Joint Episodes Joint Episodes

Child:Utterances (per min) .... 1.0 (1.4) .6 (.82) 3.7 (3.1) 1.6 (1.3)*abMLU ................. 1.2 (.24) .9 (.83) 1.3 (.29) 1.1 (.42)Words (per min) ........ 1.2 (.17) .8 (1.0) 4.9 (4.7) 2.0 (1.1)*aObject labels (per min) .. .6 (.10) .5 (.75) 1.8 (2.0) .8 (1.6)**aDyad:Conversations (per min) . .5 (.51) .4 (.41) .9 (.87) .7 (.69)aAverage child turns ..... 1.7 (.73) 1.0 (.86) 4.5 (2.2) 2.4 (1.7)**aMother:Utterances (per min) .... 16.9 (8.2) 8.6 (5.6) 12.1 (5.1) 9.1 (4.7)**MLU .................. 3.9 (.67) 4.2 (.65) 4.4 (.51) 4.9 (.10)*a% Comment ........... .56 (.11) .46 (.18) .48 (.09) .41 (.20)*a% Question .......... .29 (.12) .32 (.15) .36 (.11) .46 (.20)*a% Directive .......... .15 (.07) .22 (.15) .16 (.08) .13 (.10)b* Inside and outside joint episodes different,p < .05.**Inside and outside joint episodes different,p < .01."a5 monthsdifferent from 21 months,p < .05.bInteractionbetween joint episode and child age, p < .05.


6/11

1458 Child DevelopmentBecause of the relatively high variabilityon some of the child and dyad measures, sub-jects were also assessed on an individualbasis. The pattern of results remained the

same. Out of 24 children, 20 had highervalues inside joint attentional episodes on theutterances per minute and words per minutemeasures (p < .01, sign test), and 16 childrenhad more object labels (p < .08). Seventeen ofthe 19 children who had conversations hadhigher values inside joint episodes on boththe conversations per minute and the averagelength of conversation measures (p < .01, signtest).Maternal language.-Table 1 also pre-

sents means and standard deviations for ma-ternal language measures. Like their chil-dren, mothers produced more utterances perminute inside as opposed to outside joint at-tentional episodes, F(1,23) = 15.87, p < .001.Their MLUs, however, were shorter insidethan outside the episodes, F(1,23) = 3.90, p


7/11

Tomasello and Farrar 1459language measures, both inside and outsidejoint attentional episodes. If the "opposite"cross-lagged correlations--that is, maternalmeasures at 21 months with child languagemeasures at 15 months-are similar to thosereported above, then it is likely that thechild's language is influencing the object ref-erence types as much as the reverse. How-ever, of the 32 "opposite" cross-lagged corre-lations, only one was statistically significant:frequency of the + + + model inside joint ep-isodes correlated with child vocabulary size at.55, p < .05.DiscussionThere were three main findings in thisstudy. The first was that during periods ofjoint attentional focus both mothers and chil-dren talked more, the dyad engaged in longerconversations, and mothers used shorter sen-tences and more comments. It is tempting toconclude from this that, as hypothesized, pe-riods of joint attentional focus in some wayscaffold early mother-child linguistic interac-tion. However, another plausible hypothesisis that the causality is in the opposite direc-tion: the dyad's ability to interact linguis-tically is a major factor in the establishmentand maintenance of joint attentional episodes.There is undoubtedly some truth to this.However, it is not the case that language is anecessary condition for a joint attentionalfocus-virtually every dyad had some jointinteractions with no language. Nor is it thecase that language is sufficient for joint visualattention-all dyads had linguistic interac-tions outside of joint attentional episodes.Also, it is important to note that while thechild's linguistic competence increasedacross the two observation sessions, the timein joint interaction did not. Thus, for all ofthese reasons the causality could not flow ex-clusively from language to joint attention. Themost plausible interpretation, then, is that thedirection of influence is "transactive": jointattentional episodes scaffold the prelinguisticchild into language, which helps the child es-tablish and maintain these episodes, whichfacilitates further linguistic interactions.

The second finding was that the types ofobject references mothers made inside theepisodes of joint attentional focus were re-lated to the child's subsequent language de-velopment, whereas these same measuresoutside the joint episodes did not correlate.This is despite the fact that there was no sys-tematic difference between the types of mod-els given inside and outside joint attentionalepisodes. However, because periods ofheightened linguistic activity for the child

corresponded to joint attentional episodes, itmay be that children are more tuned in tomaternal language when they themselves arespeaking or when they are engaged in conver-sations of a certain length. Thus, again, lan-guage may be part of the cause as well as theeffect. Again, it is probably best to think intransactive terms. In this case, the causal fac-tor may best be conceived as periods of jointattention, which involve linguistic as well asnonlinguistic elements.

The third main finding concerned thespecific relationships between object refer-ence types and the child's language. Insidethe joint attentional episodes, three of the fourobject reference types that followed into thechild's attentional focus correlated positivelywith the child's subsequent lexical develop-ment, whereas one of the directive types cor-related negatively. It is interesting to notethat the only directive type that correlatedpositively was the one in which the mothermade her attentional focus clear by gesturingand in which the child focused successfullyon the referent object (- + +). It is puzzlingat first glance that the + + - and the + - -reference types correlated positively with thechild's subsequent vocabulary since, in these,the child was not focused on the object at theprecise moment the name was provided.However, by definition of follow-in, in both ofthese reference types the child was focusedon the object when the mother began her ut-terance. These two types thus indicate situa-tions in which the child looked away from theobject before its name was uttered. Manytimes this simply meant that the child lookedto the mother's face as she spoke and returnedto a focus on the object soon thereafter.Though in some cases the child shifted hisattention permanently, most often these refer-ence types do represent an instance of jointattentional focus and thus should facilitate thechild's word learning.

Once again in this third finding, how-ever, either direction of influence is possible.In contrast to the current hypothesis that theobject reference type affects the child's lexicalacquisition, it is possible that the correlationsare due to the child's influence on themother: linguistically competent children in-duced mothers to provide certain types of ob-ject references. However, the cross-laggedcorrelations argue against this interpretation.Frequency of the - - - object reference typeat 15 months correlated negatively with childlanguage at 21 months, but child language at15 months did not correlate with object refer-ence types at 21 months. This pattern indi-


8/11


9/11

Tomasello and Farrar 1461TABLE 3

MEANS AND STANDARDDEVIATIONS OF CHILD PERFORMANCEMEASURES ASAFUNCTIONFTRAININGONDITIONSummedacross All 4 Sessions)Child PerformanceMeasure Follow-in Direct

Frequency of spontaneous production .... 40 (.50) .40 (.50)Frequency of elicited production ......... 10 (.30) .50 (.50)Percent comprehension .................. 50 (09) 32 (10)*Percent comprehension (follow-up) ....... 64 (13) 36 (13)**Conditionsdifferent,p < .05.

a function of attentional strategy training con-dition. Because there was no systematic effectof session, the values used for analysis andpresented in Table 3 are the values obtainedby summing across the four training sessions.There was no effect of the placement of theword in the sentence (middle or end) or of theorder of conditions. These were therefore ex-cluded from further analysis.Children comprehended the modeledwords better in the follow-in condition, t(9) =2.41, p < .05. In this condition, the childrenaveraged correct responses on 50% of the

trials overall: the mean score was 4.0 out ofeight trials per child per condition (two wordsfor four sessions). This proportion wassignificantly above the chance performance of25% (assuming children always picked an ob-ject, which they did not), p < .05, whereas theproportion of correct responses in the directcondition was not. There were no statisticallyreliable effects found for either of the produc-tion measures, which were both quite low inboth conditions; of the 40 trained words (fourper child, 10 children) there were only 14 pro-ductions, and these came from only five chil-dren.

Due to illnesses and absences, onlyseven of the 10 children could be given fol-low-up testing within a few days of the 2-week interval. Again, children spoke very lit-tle, and so neither production measureproduced differences. (However, it should benoted that all of the three productions re-corded in the follow-up session were from thefollow-in condition.) Analysis of the compre-hension task produced a significant differencein favor of the follow-in condition: 64% to36%, t(6) = 4.58 p < .05. Analysis of individ-ual subjects confirmed this trend: six of theseven children had better comprehensionscores in the follow-in condition, and theother child had equal scores in both condi-tions. The probability of this occurring bychance alone is less than .05, sign test.

DiscussionThe main finding of the training studywas that the follow-in strategy producedgreater word learning, as measured by com-prehension, than the direct condition. Thevery small amount of production does not per-mit firm conclusions. Though it is possiblethat more training would have produced moreproductions, other lexical training studieshave obtained results with this amount oftraining (cf. Schwartz & Terrell, 1983). Morelikely, the small amount of production wasprobably due to the children's general shy-ness alone with strangers.The results of this study help to explainthose of Study 1. By themselves, the correla-tions of that study could be explained if itwere the case that the child was (a) particu-larly attracted to some objects, (b) thus"primed" to learn their names, and (c) playedwith these objects most often. If this were thecase it would mean that when mothers fol-lowed in, it would most often be attractiveobjects the child was playing with, and thusconditions for word learning would be max-imal. Objects named in a directive mannerwould be those of little or no interest (thechild was not playing with them) and solearning conditions would be less favorable.The results of Study 2, however, cannot beexplained in this way since in this study ob-jects were randomly assigned to conditions.Together, then, the results of the two studiesare most economically explained by positinga facilitative effect of joint attentional pro-cesses.

Another possible explanation for the cor-relations of Study 1 is Nelson's (1981) func-tional hypothesis. It is possible that childrenwith more directive mothers were learningthat the primary function of language is social-regulative, and thus they were less interestedin learning object names. Children of motherswho more often followed-in to their attentionattributed to language more cognitive sig-


10/11

1462 Child Developmentnificance and thus learned more object labels.This interpretation is not plausible in Study 2,however. In this study the same child learnedwords differentially depending on how theywere presented. This could not be the prod-uct of one overall hypothesis about the func-tional significance of language. It is of coursepossible that Nelson's hypothesized mecha-nism is at work in the real world (and Study1), while the current finding is a laboratoryphenomenon; or it is possible that the twomechanisms are both operative, though ondifferent levels. Once again, however, themost economical explanation of the two stud-ies together is in terms ofjoint attentional pro-cesses.General Discussion

The current studies, in combination withfindings of previous research, suggest thatjoint attention is important to early languageacquisition in two ways. First, relatively ex-tended episodes of joint attentional focus be-tween adult and child provide important non-linguistic scaffolding for the young child'searly linguistic interactions. This effect seemsto extend well into the second half of thechild's second year of life, beyond the veryearliest stages of communicative develop-ment where most previous research has con-centrated (e.g., that of Bruner and his col-leagues). Further, what happens in theseepisodes seems to be of special importancefor acquiring new language. Keith Nelson(e.g., 1982) has argued that, in general, whenlearning conditions are favorable the child'sacquisition of novel linguistic structures mayoften be based on a single, or at most a veryfew, adult exemplars of that structure. The re-sults of the current study suggest that for theinitial phases of lexical development, rela-tively extended episodes of joint attentionalfocus between child and adult may constitutean important part of such conditions. This ispresumably because such episodes are pe-riods when the child is attentive, motivated,and best able to determine the meaning of hermother's language (cf. Ninio & Bruner, 1978;Ratner & Bruner, 1978).

Within joint attentional episodes, itwould seem to be important that the adult talkabout the object on which the child isfocused, rather than constantly trying to redi-rect the child's attention. Roth (1985) hasshown, in fact, that when mothers follow intotheir child's attentional focus, they are morelikely to elaborate semantically on previouschild utterances. In the current interpretation,the important factor in all cases is the relative

ease with which the child is able to establishthe attentional focus of the adult and thus thereferential context of her language. It is inter-esting to note in this regard that somethingvery similar to this also operates at the levelof conversational interaction. Olsen-Fulero(1982) has demonstrated that directivenesshas an adverse effect on early mother-childconversations. If conversational topic may bethought of as analogous to an object of jointvisual attention, then these results parallelthose of the current study.

One final point should be made. All ofthe measures in the current study were of vi-sual, not auditory, attention. This is quite sim-ply because visual attention is most easily ob-servable. It is possible, for example, that thechild was indeed attending aurally to an ob-ject when she was coded as not attending-for example, when the mother shook a rattlethat the child recognized. It is also possiblethat the child was not attending aurally to themother's language in some cases-a situationnot dealt with in the current study. Undoubt-edly, a systematic account of auditory atten-tion is necessary for a thorough understandingof the role of attentional factors in the lan-guage acquisition process.Individual differences in early languageacquisition present a challenge and an oppor-tunity for researchers. As Katherine Nelson(1981) has pointed out, explaining these dif-ferences may play a crucial role in discov-ering the basic cognitive and social processthat underlie language development. Thus farexplanations have centered on such factors ascognitive style differences among children(Bretherton, McNew, Snyder, & Bates, 1983),social-interactional differences among moth-

er-child dyads (Nelson, 1973), and differencesamong the social environments of children(i.e., the amount of interaction with fathers,siblings, peers, strangers, etc.; see Mannle &Tomasello, in press). It is safe to assume thateach of these has some role to play. In thisstudy we have attempted to identify and ex-plore another set of factors that, like theothers, may be fundamental for language ac-quisition and at the same time contribute toindividual differences. Joint attentional pro-cesses are clearly worthy of future researchattention.ReferencesBakeman,R., & Adamson,L. (1984).Coordinatingattention to people and objects in mother-infantandpeer-infant nteraction.Child Devel-opment,55, 1278-1289.Bates,E. (1979).Theemergenceof symbols:Cogni-


11/11

Tomasello and Farrar 1463tion and communication in infancy. NewYork:Academic Press.

Bretherton, I., McNew, S., Snyder, L., & Bates, E.(1983). Individual differences at 20 months:Analytic and holistic strategies in language ac-quisition.Journalof ChildLanguage,10, 293-320.

Bruner, J. (1981). The pragmatics of acquisition. InW. Deutsch (Ed.), The child's constructionoflanguage (pp. 35-56). New York: AcademicPress.Bruner, J. (1983). The acquisition of pragmatic com-mitments. In R. Golinkoff (Ed.), The transitionfrom prelinguisticto linguisticcommunication(pp. 27-42). Hillsdale, NJ: Erlbaum.Bruner,J. (1985).Child'stalk:Learningto use lan-

guage. New York: Norton.Della Corte, M., Benedict, H., & Klein, D. (1983).The relationship of pragmatic dimensions ofmothers' speech to the referential-expressivedistinction. Journal of Child Language, 10, 35-44.Mannle, S., & Tomasello, M. (in press). Fathers,siblings, and the bridge hypothesis. To appearin K. E. Nelson & A. van Kleek (Eds.), Chil-dren's language, Volume 6. Hillsdale, NJ: Erl-baum.Nelson, K. (1973). Structure and strategy in learningto talk. Monographs of the Society for Re-

search in Child Development, 38(1-2, SerialNo. 149).Nelson, K. (1981). Individual differences in lan-

guage development: Implications for develop-ment and language. Developmental Psychol-ogy, 17, 170-187.Nelson, K. E. (1982). Theories of the child's acqui-sition of syntax. Annals of the New YorkAcademyof Sciences,345, 45-69.Ninio, A., & Bruner, J. (1978). The achievementand antecedents of labelling. Journal of ChildLanguage, 5, 1-16.Olsen-Fulero, L. (1982). Style and stability inmother conversational behavior. Journal ofChild Language,9, 543-564.Ratner, N., & Bruner, J. (1978). Games, social ex-change, and the acquisition of language. Jour-nal of Child Language,5, 391-402.Roth, P. (1985). Timing and function of maternalspeech to 12-month-olds. Manuscript sub-mitted for publication.Schwartz, R., & Terrell, B. (1983). The role of inputfrequency in lexical acquisition. Journal ofChild Language, 10, 57-66.Tomasello, M., Mannle, S., & Kruger, A. (1986).The linguistic environment of one to two year oldtwins. DevelopmentPsychology, 22, 169-176.Tomasello, M., & Todd, J. (1983). Joint attentionand lexical acquisition style. First Language, 4,197-212.

Documents

27_tomasello