Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

When is a "Neem" not a "Neem"?: The influence of sentence structure on word-object association in infancy Lloyd, Valerie Louise 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1994-0049.pdf [ 1011.65kB ]
Metadata
JSON: 831-1.0087272.json
JSON-LD: 831-1.0087272-ld.json
RDF/XML (Pretty): 831-1.0087272-rdf.xml
RDF/JSON: 831-1.0087272-rdf.json
Turtle: 831-1.0087272-turtle.txt
N-Triples: 831-1.0087272-rdf-ntriples.txt
Original Record: 831-1.0087272-source.json
Full Text
831-1.0087272-fulltext.txt
Citation
831-1.0087272.ris

Full Text

WHEN IS A‘tNEEM NOT A “NEEM”?:THE INFLUENCE OF SENTENCE STRUCTURE ON WORD-OBJECT ASSOCIATIONS ININFANCYbyVALERIE LOUISE LLOYDB.A., The University of British Columbia, 1991A THESIS SUBMI’FiED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF ARTSinTHE FACULTY OF GRADUATE STUDIES(Department of Psychology)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIADecember 1993© Valerie Louise Lloyd, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of____________The University of British ColumbiaVancouver, CanadaDate tc- 1q93DE-6 (2/88)ABSTRACTThis research was designed to address the question of whether 14-month old infants’ability to associate word-object pairings is influenced by sentence structure. Lloyd, Werker, andCohen (1993) have shown that 14-month old girls notice changes in word-object pairings in anhabituation/dishabituation paradigm, but 8-, 10- and 12-month old infants and 14-month old boysdo not. Using video images, infants were habituated to two instances of word-object pairings. Aset of test trials was then shown: One in which the word-object pairing changed, and one inwhich it remained the same. The present research used the same habituation/dishabituationprocedure in two experiments with 14-month old infants. In Experiment 1 the nonsense wordswere presented alone, and in Experiment 2 the nonsense words were embedded in a set of carrierphrases. The first critical question was whether the infants would notice that the word-objectpairing had been changed. The second critical question was whether this ability varied acrossconditions: word in isolation vs word in a set of carrier phrases. The results from Experiment 1provide clear evidence that 14-month old girls, but not boys, are able to make word-objectassociations when the object labels are presented in isolation. Findings from Experiment 2, inwhich the target words were presented in a set of carrier phrases, indicate that the particularcarrier phrases presented in this study do not facilitate word-learning at this age.IITABLE OF CONTENTSAbstract iiTable of Contents iiiList of Tables ivList of Figures VAcknowledgement Vi1.0 INTRODUCTION p. 12.0 OVERVIEW OF RELATED RESEARCH p.73.0 A RELIABLE PROCEDURE FOR TESTING LEXICAL ACQUISITION ...p. 124.0 THE EXPERIMENTS p. 174.1 Experiment 1 p. 184.2 Experiments 2 p. 265.0 GENERAL DISCUSSION p.30References p. 34Appendix 1: Visual Stimuli used in Experiments 1 and 2 p. 38Appendix 2: Apparatus configuration for Experiments 1 and 2 p. 39iiiLIST OF TABLESChart 1: Sequence of Stimuli Presentation p. 23Table 1: Experiment 1: Design p. 24Table 2: Experiment 2: Design p. 28ivLIST OF FIGURESFigure 1: Results from Lloyd, Werker and Cohen (1993) study p. 16Figure 2: Results from Experiment 1: Target Words in Isolation p. 25Figure 3: Results from Experiment 2: Target Words in Carrier Phrases p.29VACKNOWLEDGMENTI would like to thank a number of people without whom the properties of this proposalwould not have emerged. Janet Werker and Les Cohen provided the intellectual foundation forthe studies proposed here. I am indebted to my thesis committee members: Janet Werker(advisor), Tannis MacBeth Williams and Richard Tees. In addition, Dare Baldwin, ReneeDesjardins, Caroline Johnson and Judith Pegg contributed valuable criticism with respect totheoretical issues and methodology. Diane Hobday and Carmen Swanson helped in booking andtesting the subjects for the Lloyd, Werker and Cohen (1993) study upon which this proposalstands, and Lynai Olsen assisted in preparation of the visual stimuli. I am grateful to MarcRomanycia for his generous technical support and philosophical inspiration. Most of all, I thankmy son, Jaman Lloyd-Zbitnew for his assistance with making the novel objects, his teachings inchild development, and the firm conviction that his mother will be “a professor in three moreyears”. This Master’s Thesis is dedicated to Victoria Watters, and the memory of Dr. MichaelChapman.viWhen Is A “Neem” Not A “Neem”?:The Influence Of Sentence Structure On Word-Object Associations In Infancy“In the beginning was the word. But by the time thesecond word was added to it, there was trouble. Forwith it came syntax...”(John Simon, 1980, p. 111)1.0 IntroductionThe set of experiments in this thesis proposal was designed to investigate theinfluence of sentence structure, that is, syntax, on lexical acquisition in infancy. We know thatby about nineteen months of age children are producing an average of fifty words, and that theyare beginning to combine them into 2-word utterances (Nelson, 1973). A longitudinal diarystudy by Benedict (1979) has shown that infants comprehend words in their ambient languagemuch sooner than they start to produce them, Although Benedict acknowledges enormousindividual differences in rate of acquisition, the infants in her study were able to comprehend anaverage of 50 words (both nouns and verbs) at 14 months, but they were only able to produce 10words. Similar estimates derived from diary studies have been provided by Snyder, Bates, andBretherton (1981) and Goldfield and Reznick (1990).The question the research described here begins to address is: Does sentence structureplay a role in infants’ acquisition of nouns? That is, are infants aged fourteen months able tolearn a word-object association just as easily when the word is presented in isolation as when it ispresented in the context of a carrier phrase, or is learning facilitated (or inhibited) if the noun ispreceded by a carrier phrase? The results of the experiments described here will help shed somelight on the role of syntax in lexical acquisition.For centuries philosophers, and more recently linguists and psycholinguists, have beeninterested in language learning because it is a uniquely human ability, and because it occurs sorapidly and apparently incidentally. The pinnacle debate that led to the current controversy overan explanation for language acquisition transpired during the late 1950s and early 1960s when1the strong behaviorist view promoted by B.F. Skinner (1957) was forcefully rebuked by NoamChomsky’s (1959) strong nativist position.Briefly, Skinner’s behaviorist position holds that language is learned through classicalconditioning involving such operations as simple stimulus-response associations, reinforcements,and extinctions. Verbal learning, according to the behaviorist view, involves the same laws oflearning as those shown to occur in animal learning; there is no special status given to humans’ability to learn and use language.Finite State Grammar is an example of grammar theory originating from the behavioristtradition. Left-to-right grammar , a “thinly disguised ‘stimulus-response’ theory”, according tosome critics (Ashcraft, 1989, p. 393), is a specific case offinite state grammar. This theoryproposes that each word in a sentence elicits a response-word, which then becomes the next wordin the sentence. The response-word is chosen based on some fixed probability of it occurringafter the stimulus word. For example, in the sentence “The old woman liked movies”, womanhas a ten percent chance of being chosen to follow old, and liked has a four percent chance ofbeing elicited by the word woman (Ashcraft, 1989, p. 392).Chomsky’s nativist argument against the behaviorist view of language acquisition is thatclassically conditioned learning cannot account for complex human behavior. The fixedprobabilities of finite state grammar do not account for the human ability to generate an infinitevariety of sentences, most of which the speaker has never before heard. This ability enables us toconstruct multiple sentences which all communicate roughly the same idea (e.g., I’m going towork now vs I’m off to the lab). Chomsky claims that finite state grammar only examines thesurface structure of language, whereas the underlying meaning of thoughts about objects, events,and ideas lies in the deep structure of language. The thoughts present in deep structure aretransformed into surface structure through transformational rules, which gives Chomsky’sgrammar the name transformational grammar, or phrase structure grammar.Chomsky further argues against the behaviorist approach with his criticism thatdescribing a grammar in terms of the probability with which words follow one another is2inadequate, since grammar is composed of semantic components as well as syntacticcomponents. By using left-to-right grammar, it is possible to generate many nonsense sentencessuch as “Colorless green ideas sleep furiously”, which follow the probabilistic syntactic rules, butare semantically anomalous.In order to account for the rapidity and ease with which human infants acquire language,Chomsky argues that there must be some innate components of language processing present, andthat these are unique to humans. The origin of language question, therefore, should not be “Islanguage learned through experience or innate mechanisms?”, but rather, “What are [thedescriptions of] these innate structures in particular domains?” (Chomsky, 1975, p.13).In order to account for the innate aspect of language learning, Chomsky proposed anLAD, or “language acquisition device” (LAD) (Chomsky, 1965) which has received considerabledebate (for reviews, see Golinkoff & Hirsh-Pasek, 1990 and Pinker, 1988). This view maintainsthat infants acquire language via an innately-specified LAD which must contain the following:(i) a universal phonetic theory that defines the notion “possible sentence”;(ii) a definition of “structural description”;(iii) a definition of “generative grammar”;(iv) a method for determining the structural description of a sentence, givena grammar, and;(v) a way of evaluating alternative proposed grammars.”(Chomsky, 1965, p. 31)The strong version of Chomsky’s “innate hypothesis” states that very little experiencewith ambient language is required to trigger the LAD and to activate various levels of languageprocessing. This strong version is appealing because it maintains that the ability to use language(specifically, generative grammar) is quintessentially human, setting us apart from other animalson an evolutionary basis. The strong version of the “innate hypothesis”, however, does not standup as a scientific theory that generates testable and falsifiable predictions, since it is impossibleto disprove that individuals are born equipped with all the necessary language-processingmechanisms which merely “unfold” with chronological age and maturation. As a result circulardebate ensues, and any argument put forth in support of experiential language learning is subjectto the nativist criticism that it is not the linguistic experience at all that is responsible for3language learning, but rather the stimulation of certain linguistic conditions that “trigger” thetime-controlled LAD, which then results in abrupt changes in language use. For example, itmight be proposed that phonetic reorganization by infants around 9 months of age (Werker &Lalonde, 1988) may be a result of infants’ experience with their native language. A nativistcould argue against this explanation for phonetic reorganization, however, by claiming thatmaturation of the LAD, together with simple triggering from the input, accounts for theseperceptual changes.A weaker version of the “innate hypothesis” is more readily agreed upon by mostpsycholinguists. This version states that some innate structure is in place at birth and evenbecomes activated later on in development, but without substantial linguistic stimulation fromthe environment these mechanisms will not be able to function. Even Chomsky himselfconcedes that environmental stimulation is essential to language learning. He suggests that“. ..what we do know, then, or what we come to believe, depends on the specific experiences thatevoke in us some part of the cognitive system that is latent in the mind” (Chomsky, 1975, p. 6).There is considerable evidence, which will be discussed below, that experience with linguisticinput does have a significant impact on infant language development.The weaker version of the “innate hypothesis” further states that variations in amount andquality of ambient language in turn produce wide variation in the speed and accuracy with whichlanguage is acquired. Support for the weaker version is provided by Dunham and Dunham(1992), who suggest that maternal input accounts for some of the variance in the vocabulary ofinfants at thirteen and twenty-four months of age. The Dunhams have shown that, even thoughthe actual number of words produced by children who receive out-of-home care is similar to thatof children who do not receive out-of-home care, the correlation between maternal and infantlexicon is significantly higher for those infants receiving full-time care by their mothers.Psycholinguists leaning heavily on the “nurture” side of the “nature/nurture” debate,needless to say, have been hard-pressed to offer convincing evidence that experiential input aloneis sufficient for language acquisition because the rules of grammar are so abstract and seemingly4unlearnable in an exclusively experiential manner. The position taken in this research is not thatexperience alone is sufficient for language acquisition, but rather that certain kinds of experience-- certain forms of input -- might play a facilitative (or inhibitory) role in English languageacquisition.There has been abundant evidence that experience with some characteristics of speechsuch as intonation and word-stress may facilitate the language-learning process. For example,infant-directed speech is more rhythmic, has longer vowel duration, and has a higherfundamental frequency and more pitch variability than adult-directed speech (Fernald, 1989).Moreover, infants have been shown to prefer this infant-directed speech over adult-directedspeech, even in a non-native language (Pegg & Werker, 1993, in preparation; Werker, Pegg, &McLeod, in press).Aslin (1993) has reported that, in infant-directed speech, words are typically presentedwithin phrases or sentences rather than in isolation, and target words are typically placed inutterance-fmal position, even if it requires violating strict grammatical rules. One study byFernald and Mazzie (1991) has shown that English-speaking mothers in the United States stresswords in infant-directed speech if they are trying to teach their infants these words as objectlabels. Thus, target-word placement and word-stress within a syntactic framework may alsofacilitate lexical acquisition for the infant.Exposure to a particular language and infants’ ability to perceive segmental features ofspeech also may have a potentially facilitative influence on language development. Mehler,Jusczyk, Lambertz, Halsted, Bertoncini, and Amiel-tison (1988) demonstrated that neonates candiscriminate their own, familiar native language from a non-native language, and Moon, Cooper,and Fifer (in press) have further shown that newborns prefer their native language over a nonnative language. Jusczyk and colleagues found that 4 month old English-learning infants detectpauses that are incongruent with clause boundaries in both English and Polish, but by 6 monthsthey detect only pauses incongruent with English clause boundaries. Furthermore, by 9 monthsof age, English-learning infants detect violations to major phrase boundaries in English (Jusczyk,5Kemler Nelson, Hirsh-Pasek, Kennedy, Woodward, & Piwoz, 1992). By nine months of age,infants also show a preference for words that correspond to native language stress patterns(Jusczyk, Cutler, & Redanz, 1993). By the time they are 10 months of age, infants can alsodiscriminate minimal-pair contrasts in their native language with greater ease than they candiscriminate non-native minimal-pair contrasts (Werker & Tees, 1984; Polka & Werker, inpress). Thus it appears that very young infants are developing a sensitivity to some aspects ofthe rhythmic, segmental, and possibly syntactic structure of their native language.These are all examples of how infants are becoming sensitive to highly complex,prosodic, segmental, and phonetic properties of the linguistic structure within the first year oflife. It has been argued that this increasing sensitivity to the structure of the native languagemight facilitate later language development (Werker, Lloyd, Pegg, & Polka, in press). There isas yet, however, little empirical work assessing whether infants use linguistic structure tofacilitate word learning. There has been substantial research with children who are alreadyproducing speech, but not with infants who are just on the threshold of language acquisition.Studies with children older than two years of age tend to focus more on comprehension ofrelational aspects of linguistic structure. For instance, infants are starting to learn hierarchicalrelations such as object categorization, (Markman, 1989) at this age. Also apparent by this age isacquisition of semantic relations between objects and actions (Pinker, 1988) and syntacticrelations among different word classes (Gleitman, 1990).This leaves students of language acquisition wondering when the transition (if therereally is one) from speech perception to language comprehension and eventually production,takes place in the language learning process. At what age do grammatical (syntactic andsemantic) aspects of language start to affect lexical acquisition? What is the influence ofprosody on early lexical acquisition? Is there a concomitant effect of grammatical structure andprosodic features in language learning, or could it be that grammatical structure does not haveany influence until the infant is well into the second year of life and able to use language in amore meaningful, representational manner? The studies in this paper were designed to examine6this bridge from speech perception, when structural but not meaningful properties of languageare discriminated, to language comprehension, when speech is understood and used in ameaningful, intentional and communicative way. By examining the role of sentence structure inthe development of infants’ ability to form word-object associations, the studies discussed herebegin to clarify whether syntactic properties of infant-directed speech are related to lexicalacquisition for infants at fourteen months of age.First, an overview describing lexical access research that has been conducted with adultsand children will be presented. Next, current “bootstrapping” theories of lexical acquisition willbe reviewed, including specific research by Naigles (1990) with 24 month old children on verbcomprehension. Third, an explanation on how the present studies differ from previous researchwill be offered, and the procedure for a pilot study upon which these studies are based will bedescribed. Finally, procedural details and results from the two studies presented here will bedocumented.2.0 Overview of Related ResearchLexical access research with adults andyoung children. To date, no research withinfants has been undertaken that compares comprehension of a word when the word is presentedin isolation versus when it is presented in a sentence context, so I turned to the literature onlexical access in adults and school-age children. The effect of sentence structure on lexicalaccess in adults has been studied widely by cognitive scientists who are interested inunderstanding how knowledge is represented in the mental lexicon (see McClelland & Elman,1986; Klatt, 1979; and Marslen-Wilson, 1989 for examples of lexical access models). Forexample, a classic study by Warren and Warren (1970) demonstrated that adults are able torestore the part of a word that had been replaced with white noise based on the overall meaningof the sentence (i.e., The *eel was on the orange). Further, Salasoo and Pisoni (1985) haveshown that when all but the initial 150 milliseconds of a word is replaced with a “noise7envelope,” adults fail to identify the word when it is presented in isolation, but identify itcorrectly when it is presented in a normal sentence context.More pertinent to my work are two studies conducted by Marsien-Wilson and Tyler(1980) and Tyler and Marslen-Wilson (1981). In the first study, adults’ speed of response to atarget word presented in three different sentential contexts was measured. The sentence contextswere: normal prose, syntactic prose, and random word-order prose. An example of a normalprose sentence is “Some thieves stole most of the lead off the roof,” where lead is the targetword. A syntactic prose sentence would be syntactically correct, but semantically anomalous, ormeaningless: “No buns puzzle some in the lead off the text.” The random word-order conditionwould be syntactically and semantically anomalous, such as “Some the no puzzle buns in leadtext the off.” Adults’ reaction time to the target word was faster in the normal prose condition,slower in the syntactic prose condition, and slowest in the random word-order condition.The same study was conducted with 5 to 10 year old children, with modification of thesentences to match the children’s language ability (Tyler & Marslen-Wilson, 1981). The patternsof response for the children were identical -- response time for the normal prose condition wassignificantly faster than either the syntactic or random word-order conditions. The two latterconditions did not differ significantly. Thus, like adults, children were fastest at recognizingwords presented in normal prose sentences. Unlike adults, children were no faster in thesyntactic prose than the random word order condition.Although their work with children has not continued, the results obtained by Tyler andMarsien-Wilson are intriguing and point to a significant role played by semantics and syntax inword recognition. Since there is such a strong effect of sentence context on lexical access, thissuggests that there could possibly be a similar effect operating in lexical acquisition. Such aneffect would be consistent with the notions of syntactic and semantic bootstrapping advanced byGleitman and Pinker, discussed below, because both bootstrapping hypotheses suggest thatsemantic and syntactic relations do influence word acquisition, albeit to differing degrees and atdifferent times in linguistic development. Thus the methodology Marsien-Wilson and Tyler8(1980) and Tyler and Marsien-Wilson (1981) used to assess the influence of syntactic context onword recognition in children and adults can be profitably extended to assess the role of semanticand syntactic context on word acquisition in infancy and early childhood. To do so, however, areliable procedure for assessing word-learning in very young children is required. A detaileddescription of such a procedure will be discussed in Section 3.0 below.“Bootstrapping” theories of lexical acquisition. There are two current theories relevantto this research on how children acquire language: the “syntactic bootstrapping hypothesis”proposed by Lila Gleitman, and the “semantic bootstrapping hypothesis” advocated by StevenPinker. The debate between Gleitman and Pinker concerns how syntactic structure andunderlying semantic relations catapult young children from a pre-linguistic state into a state ofefficient language use within the first few years of life. Gleitman states that, “...although the[two hypotheses] are distinct, to hold that one of them is implicated in learning is not to deny thatthe other one is, too. Quite the contrary. It is very likely that they operate in a complementaryfashion” (Gleitman, 1990, p. 30). Pinker, similarly, denounces the pointless debate between“semantophiles” and “syntactophiles” “over whether syntax or semantics [is] in some sense moreimportant.... [for] even a cursory examination of adult linguistic abilities shows that adultcompetence involves rules that dictate purely formal, syntactic properties of sentence elements”(Pinker, 1988, pp. 107-108). The interesting theoretical question is not whether semantic orsyntactic bootstrapping occurs, but rather how and when each of these processes contribute to theacquisition of language.The evidence that Pinker offers in support of his “semantic bootstrapping hypothesis” isdata collected by Roger Brown in his book A First Language (1973). In this book, Browncharacterizes Stage 1, or telegraphic speech, as two-word utterances. An example of telegraphicspeech is mommy cookie. He further describes these utterances as expressions of “semanticrelations, such as ‘agent-object,’ ‘entity-location,’ and ‘possessor-possessed’ “ (Pinker, 1988, p.107). Pinker suggests that the words representing each of these semantic relations correspond toonly one type of grammatical device. Agents of action are expressed as subjects, and patients are9expressed as objects of action. Words that cannot be learned by this “word-to-world”correspondence are henceforth classified through grammatical linking rules based on the positionthey occupy in phrase structures or inflectional paradigms. An example would be a child’sdeduction that in Ideas interest Mary; ideas is a noun because it occupies the same syntacticposition that dog does in The dog chewed the bone (Pinker, 1989). Thus, further semantic-syntactic relations that are too abstract to learn through direct observation are acquired throughChomsky’s proposed “rich deductive structure’: given one fact about an element in alanguage, many others are either invariably true or have a high probability of being true” (Pinker,1988, p. 108).The specific prediction in Pinker’s “semantic bootstrapping hypothesis” are that thechild’s first words should correspond to concrete one-on-one syntactic-semantic correlationsassumed to launch the learning process, i.e., first nouns should be names for things; first verbsshould refer to actions or changes of state; first subjects should refer to agents of actions; firstobjects should refer to patients, etc. This process of language acquisition, according to Pinker,explains why “Stage 1 speech would not be a blind alley in the road toward adult mastery oflanguage. Rather, it would be a necessary first step in using innate knowledge to acquire anessentially formal system with circumscribed areas of semantic predictability” (Pinker, 1988, p.110). Since the semantic bootstrapping hypothesis predicts that one of the earliest linguisticabilities infants develop is to learn nouns as labels of concrete objects, the words presented to theinfants in the present two studies are nouns (albeit nonsense) that label accompanying objects. Inthe first study they are presented alone, and in the second study the same nouns are presented incanonical sentences.Gleitman, recall, does not dispute that semantic bootstrapping might explain some formsof word-learning. She does, however, argue that many of the scenes children encounter areambiguous and consistent with multiple interpretations, especially in verb-learning. In thesecases syntactic structure is the only way the child can deduce the meaning of verbs. Rather than10word-to-world mappings, Gleitman holds that sentence-to-world mappings are required toaccount for the full range of meanings acquired in early childhood.A hypothetical example of a multiply interpretable scene offered by Gleitman is one inwhich little George pushes his toy truck. Each time he pushes it, he hears his caregiver say“George pushes the truck.” However, whenever George pushes the truck, it moves. So how is itthat George figures out that “push” means “push” and not “go” (as in “George is making the truckgo”)? According to Pinker’s semantic relations hypothesis, either interpretation would be correctsince the agent-object relation is the same irrespective of interpretation. Gleitman claims theevidence is in the syntactic structure of the sentence. (Compare the syntactic structure of thesentences: “George pushes the truck” vs. “George is making the truck go”).These bootstrapping hypotheses are the most well-developed arguments put forth forlexical acquisition in early childhood. The empirical data held up in support of each hypothesis,however, are from research with children two years and older -- on verb-learning in Gleitman’scase, and verb- and noun-learning in Pinker’s case; and there are more data available fromresearch on production than comprehension of words, because it is relatively easier to collectspeech samples from toddlers than to assess comprehension in younger infants. In order to gaina more complete understanding of the initial word-learning process in infancy, then, it isnecessary to test the comprehension of infants much younger than two years. It is particularlynecessary to test whether there is an effect of sentence context on simple noun acquisition ininfancy, since it has been argued that nouns are produced earlier than verbs (Nelson, 1973).Perhaps information provided by sentential structure assists the infant in word-learning muchearlier than the evidence supporting the semantic and syntactic bootstrapping theories wouldhave us believe.One piece of evidence from a comprehension study that provides support for Gleitman’ssyntactic bootstrapping hypothesis, that syntax is essential to verb-learning, comes from a studyconducted by Naigles (1990) in which 24 month old children are shown a video of two people:One person is dressed up as a rabbit and one as a duck. In the video, the rabbit pushes the duck11back down every time the duck pops up from a squatting position. Simultaneously, the rabbitand duck both make circles in the air with their right arms. One group of subjects hears a voicesaying “The rabbit is gorping the duck” and the other group hears “The rabbit and the duck aregorping.”After this training period each child is shown two visual stimuli in a preferential-lookingprocedure: One where the rabbit is pushing the duck down, but they are not wheeling their arms;and one in which the rabbit and duck are wheeling their arms, but the rabbit is not pushing theduck down. The child is then asked: “Where’s gorping now? Find gorping!” The results wereconclusively in favor of the children understanding the syntactic meaning. Every child testedlooked longer at the screen that matched the syntactic condition in which they were trained thanthe screen that did not match. This study is supportive of Gleitman’s claim that “...if the syntacticstructures are truly correlated with the meanings, the range of structures will be informative fordeducing which word goes with which concept” (Gleitman, 1990, p. 30).In order to extend this line of research from verb-learning, the studies reported on hereexamine the influence of syntactic structure on noun-learning. Even in a scene which is notmultiply interpretable, such as a single object moving across a video screen, sentence contextmay still assist (or interfere) with noun-learning. Since it is known that infants begin tocomprehend words much earlier than twenty-four months of age, and since we have developed aprocedure which is reliable for fourteen month olds, this is the age group that we chose to test inour research.3.0 A Reliable Procedure for Testing Lexical AcquisitionIn addition to the observational data on word comprehension collected from diary studiesmentioned at the beginning of this paper, there have been attempts to assess word comprehensionin infants from 10 to 18 months of age using a preferential-looking technique. In this procedure,the infant is typically given a choice of two objects presented on two separate video monitors,and the amount of visual fixation to the object that matches the word presented is compared to12the amount of visual fixation to the object that does not match the word. Oviatt (1980) claims tohave developed a reliable preferential-looking procedure to measure comprehension in 10 montholds. In this procedure, two video images are presented side-by-side in front of the infant, who isseated beside his/her parent and one of the experimenters. Then the name of the object ispresented, and the amount of time the infant looks at the matching or mis-matching picture isrecorded. One criticism of Oviatt’s procedure is that the infant’s parent presents the speechstimulus, which may introduce external contextual cues and other uncontrolled variables. Theseexternal cues may ultimately influence the infants’ direction of visual gaze more than the speechstimuli would have if they were presented in a controlled manner (i.e., audio-taped format). Asecond line of research on infant word comprehension is provided by Reznick (1990), who hasdeveloped a preferential-looking procedure assessing word comprehension in 8 to 20 month oldinfants. Unfortunately, the words presented to the infants in this study were inappropriate for theyoungest age-group (i.e., infant, woman, butterfly, purse, etc.), making the results -- that theolder infants understood more of the words than the younger infants -- unsurprising. Finally,Golinkoff (1987) has successfully assessed word comprehension using a more controlledpreferential-looking procedure, but in our laboratory this procedure has only proven reliable forinfants older than 18 months of age (Werker & Pegg, 1992).A series of preferential-looking and habituation/dishabituation studies conducted byHumphrey and Tees (1979, 1980) revealed that the habituation/dishabituation procedure may bemore sensitive than the preferential-looking procedure for assessing infants’ ability to detectauditory/visual correspondences. Indeed, there is a substantial history ofhabituafion/dishabituation studies, testing such diverse perceptual abilities in infancy as colorperception (Bornstein & Kessen, 1977), sound localization (Zelazo & Weiss, 1989) and objectpermanence (Baillargeon, 1991).Younger and Cohen (1989) have adapted the habituation/dishabituation procedure from atechnique used to assess infants’ ability to discriminate different exemplars of a single perceptualdomain to one that investigates infants’ ability to learn correlations among several features of an13exemplar within a domain. They used this modified procedure to conduct a series of visualcategorization studies with which they tested infants’ ability to correlate several visual features ofa single object. A further adaptation by Lloyd, Werker and Cohen (1993) of thehabituation/dishabituation procedure tests infants’ ability to detect correlations across modalities-- in this case, between words and objects. This procedure uses controlled auditory-visual stimuliand is reliable in assessing word-object associations in infants as young as 14 months of age.In the experimental procedure used in the word-object association study by Lloyd,Werker, and Cohen (1993), infants were repeatedly shown a video tape of two familiar, attractivetoys-- a truck and a dog -- that moved horizontally or vertically across the video monitor,respectively. One object was paired with the nonsense word “111” (i.e., truck/”lif”) and the otherwith the nonsense word “neem” (i.e., dog/”neem”). Following this familiarization phase, infantswere shown two test trials presented in counterbalanced order: One maintained the familiarword-object combination (i.e., truck/”lif”), and the second presented a new word-objectassociation (i.e., dog/”lif”). The critical question was whether the infants would look longer atthe new, “switched” word-object combination than at the familiar, “same” combination. If theydid, it would indicate they had formed an association between a specific nonsense word and afamiliar object.Ninety-six infants were tested; 32 at each of three ages, 8, 10 to 12, and 14 months. Boysand girls were represented approximately equally in each age group. Half of the subjects in eachage group were presented with the “switch” test trial first, and the other half were shown the“same” test trial first. The infants were presented the familiarization events until their fixationtime had dropped by a set criterion, or until they had received a total of 20 14-sec trials. Ananalysis of the familiarization data indicated approximately half the infants at each age and ofeach gender reached the habituation criterion. Thus habituation status also was entered as afactor in the analysis.The results revealed an interesting developmental pattern. Infants at 8- and 10 to 12-months of age showed no evidence of dishabituating to the “switched” word-object association.14Even those who reached habituation criterion did not look significantly longer at the “switch”than at the “same” test trial. Thus the habituators at these earlier ages had apparently learned theindividual word and object features during the familiarization phase, rather than the word-objectpairings.As might be expected, infants at any age who failed to habituate did not look significantlylonger at the new word-object combination. To control for the possibility that the infants whodid not dishabituate to the “switch” trial had not merely become fatigued by the end of the study,a completely novel word-object pairing was presented at the beginning and end of the testingsession. Visual fixation to these pre- and post-test trials was compared, and found to not besignificantly different1.At 14 months of age, however, infants who had habituated to criterion lookedsignificantly longer at the “switched” word-object combination than at the “same” pairings,showing clear evidence of having learned the word-object associations t(14) = 2.267, p<.05 (onetailed). Upon closer scrutiny of the data, it appeared that almost all of the female 14-month oldinfants had dishabituateci to the “switch” test trial, and almost all of the male 14-month oldinfants had not. With gender entered as a between-group variable, it became clear that thefemales were responsible for the significant difference between the last block of trials in thehabituation phase and the “switch” trial in the test phase, t(7)=4.613, p<.OO2 (two-tailed),whereas the same comparison for the males did not yield a significant difference, t(6)=.173,p>.05 (two-tailed). Furthermore, the girls also looked significantly longer at the “switch” than“same” test trials, t(7)=5.922, p<.OO1 (two-tailed) (see Figure 1).1. The results from this habituation/dishabituation study do not preclude the possibility that word-objectassociations may be made at earlier ages in other testing procedures (for example, in situations where the infant isgiven the opportunity to play with and act on the object).1514-Month Old Habituators by GenderL10.5o 10k1n 9g 8.5T 8I 7.5me6.5Last Block SameFigure 1: Results from Lloyd, Werker and Cohen (1993) showing dishabituation to theswitch in word-object pairing by 14-month old girls.Similar gender differences have been found in the literature on language comprehensionin infancy (Huttenlocher, Haight, Bryk, Seltzer, & Lyons, 1991; Oviatt, 1985; Reznick, 1990;Reznick & Goldfield, 1992) and in production (Nelson, 1973). Often, however, the sameresearchers only find gender differences in some, but not all of the experiments conducted(Oviatt, 1980; Reznick, 1990; Reznick & Goldfield, 1992). Still other reports on early languagecomprehension indicate no differences at all in language comprehension and production(Benedict, 1979; Goldfield & Reznick, 1990; Oviatt, 1982; Snyder, Bates & Bretherton, 1981).At least one comparison study on gender differences in linguistic performance indicatesgirls are more precocious than boys in language production before 2 years of age, after whichtime the difference disappears (Maccoby & Jacklin, 1974), but much data have been collectedsince the time of the analysis, and it is clear that more meta-analytical studies are required beforeany substantive conclusion can be made about gender differences in early language acquisition.A meta-analysis conducted by Hyde and Linn (1988) claims that any gender differences reported16SwitchTrialby Maccoby and Jacklin (1974) have now disappeared. Even though there is considerablecontroversy over how prevalent and large gender differences are in early language acquisition,when a gender difference is reported, it is almost always the case that girls are shown to be moreprecocious than boys. It is for this reason that the gender difference found in Lloyd, Werker, andCohen (1993) was taken seriously, and that an attempt to replicate these findings was undertakenin Experiment 1.The results from the Lloyd, Werker, and Cohen (1993) study provided clear evidence thatgirls at 14 months of age can discriminate novel from familiar word-object pairings. They alsoshowed that this ability changes across age. Hence, their procedure seemed particularly usefulfor testing whether carrier phrases facilitate, inhibit, or have no effect whatsoever on word-objectmapping in infancy.4.0 The ExperimentsThe research described here extends that of the Lloyd, Werker, and Cohen (1993) studyin several ways. First, controls are introduced for object novelty. The truck and dog shown alongwith the audio stimuli in the Lloyd, Werker, and Cohen (1993) study were objects with which theinfants may have had some familiarity. Even if they did not have experience with the particularobjects used, the infants may have had experience with other exemplars of the objects, and word-categories to which these objects belong (i.e., doggy; car-car). Thus some infants may alreadyhave had category labels for the objects and, according to Markman’s (1989) mutual exclusivityprincipal, after one label is learned for an object, further labels will be assumed by the infant torefer to a feature of the object, such as color or shape. Thus, mutual exclusivity may haveinterfered with the infants’ learning a second label for the object in the Lloyd, Werker, and Cohen(1993) study. Second, the direction in which the objects move across the video screen iscontrolled in the present experiments. The objects used in the Lloyd, Werker, and Cohen (1993)study moved in different directions across the screen: The truck moved horizontally and the dogmoved back-and-forth toward the viewer. Thus the infants may have dishabituated to the17“switch” trial in the word-object pairing based on the association between the word and directionof movement rather than the association of the word with the object itself. In the present studies,the objects were novel, and they both moved in the same direction-- horizontally across thevideo screen.By using a strategy similar to that developed by Marslen-Wilson and Tyler (1980),infants’ ability to associate objects with words presented in isolation was compared to theirability to form such associations when the words were presented in a normal prose sententialcondition. Since infants comprehend an average of fifty words at 14 months (Benedict, 1979),and 14-month old girls have the ability to make word-object pairings when the words arepresented in isolation, whereas 8-, 10- and 12-month olds do not (Lloyd, Werker & Cohen,1993), 14-month old infants were tested in this set of experiments. It was reasoned that it isimportant to first establish whether 14-month old infants, particularly girls, are influenced bysentence structure while making word-object associations before going on to examine whethersentence structure influences infants at younger ages.In summary, two separate experiments were conducted. The first tested lexicalacquisition of nonsense words presented in isolation (with novelty of object and direction ofmovement controlled), thus providing a replication and extension of Lloyd, Werker, and Cohen(1993); and the second tested acquisition with a new group of infants when the same nonsensewords were presented in phrase-final position within a set of normal prose carrier phrases.4.1 Experiment 1The purpose of Experiment 1 was to replicate our previous results showing that 14-monthold girls can make associations between isolated nonsense words and objects, but with novelrather than familiar objects used as the visual stimuli, and with the direction of motioncontrolled. The infants learned a pair of word-object combinations by viewing them on a videotape monitor. They were subsequently tested on whether they detected a violation of one of theoriginal word-object combinations. The infants’ visual fixation to the habituation and test trials18were recorded as the dependent measure. Completely novel visual stimuli were used, and eachobject moved in a horizontally across the video screen. The speech stimuli remained the same asin the Lloyd, Werker, and Cohen (1993) study.MethodSubjects. Sixteen 14-month old infants (± 3 weeks) were required for the first study, withfemales and males represented equally, Their gestational age at birth ranged from 37 to 43weeks, and their ambient language environment was composed of a minimum of 80% English.Eight girls and 8 boys successfully completed the study and habituated to criterion in Experiment1; one boy and one girl were tested in each order. The data collected from 14 infants were notincluded in the analysis: 3 girls and 4 boys due to fussiness; 1 girl whose mother pointed duringthe test trials; and, 3 girls and 3 boys due to non-habituation. Overall attrition rate was fortyseven percent.The infants were recruited mainly by visiting new mothers at Grace Hospital inVancouver, but also through community service announcements in local newspapers and onradio and television programs. Permission to contact the parents by phone when their infantswere the appropriate age was obtained, and the parents were given details of the experimentalprocedure at the time they were selected for participation. Those who eventually took part in theexperiment were given an infant scientist t-shirt and an infant scientist degree in appreciation fortheir participation.Stimuli. The audio stimuli were seven exemplars of each of two nonsense consonantvowel-consonant (CVC) words: “neem and ttlif” for the habituation and test phase thals. Anadditional word, “pok”, was also recorded for the pre-test and post-test trials (see Chart 1 andTable 1). These exemplars were chosen because they differ in vowel category and theconsonantal dimensions of nasality, voicing and place dimensions (Ladefoged, 1975).Maximally different nonsense words were used to ensure optimal discrimination.19The infant-directed CVC speech exemplars were identical to those used in the Lloyd,Werker, and Cohen (1993) study. These tokens were recorded in a sound-proof booth by afemale speaker who is fluent in English. While recording this speech, the speaker was instructedto imagine that she was speaking to a 14-month old infant to ensure appropriate infant-directedspeech and prosodic contours. Each of the seven isolated word exemplars was approximately.7-seconds in duration, with a 1.5-second interval between words. This produces audio files thatare each 14 seconds in duration.2The visual stimuli were three novel objects video-taped against a black background andtransferred to laser-disc format. Two of the objects, made from modeling clay (FIMOReg.TM),were used for the habituation and test phase trials and the other object, a commerciallymanufactured plastic water wheel, was used for the pre- and post-test trials (see Appendix 1). Tocontrol for brightness, size, and color, and to ensure that one figure was not inherently moreattractive than the other, the clay objects were composed of equal amounts of each of three colors-- red, blue, and yellow. These bright, primary-colored novel objects occupied approximately13.5° vertical and 13° horizontal visual angle. The duration of each video file was s so that theysynchronized with the digitized audio files.Apparatus. The testing session took place inside a small, quiet, and dimly lit lab room(2.3 m x 2.7 m) in the Department of Psychology at the University of British Columbia. Theinfant sat on the parent’s lap facing a Mitsubishi HC 3905, 45cm video monitor with 640 dothorizontal by 480 line vertical resolution (see Appendix 2), A Bose 101 speaker was locateddirectly below the monitor, through which the audio stimulus was played at approximately 70dB.During testing, the video image was approximately 70cm from the infant’s eyes and thevideo monitor was surrounded by black draping from floor to ceiling, with a 6cm hole centered2 Initially, six exemplars placed 3 seconds apart, to match the number of tokens in Experiment 2, were piloted-tested onthree infants. All of the infants in the pilot tests failed to complete the study due to fussiness, presumably because this audiostimulus sounded unnatural. The audio stimuli were then placed 1.5 seconds apart, as in the Lloyd, Werker & Cohen (1993)study, and the collection of data proceeded at a normal pace.20above the monitor behind which was placed a Panasonic PV-S770K closed circuit camera, Theclosed circuit system was used to monitor the infant’s visual responses on-line and to record thesession on video tape for the purpose of reliability coding. The parent listened to female vocalmusic over a pair of Koss Pro/4AA headsets connected to a Panasonic RX-C5700 portable stereoin an adjacent room. The purpose of the music was to mask the audio stimuli (a female’s voice)presented to the infant.The experiment was controlled by a Macintosh IIfx computer interfaced with a SonyLDP- 1550 laser disc player. The video segment for each visual image from the laser disc playerwas synchronized with a digitized audio file by the computer and the signals were transmitted tothe monitor and speaker, respectively.The experimenter monitored the infant’s visual gaze on a NEC PM-1271A monitor in theobservation room and recorded the duration of gaze by depressing a pre-programmed key on thecomputer keyboard when the infant was looking at the visual stimulus and releasing it when theinfant looked away. The experimenter was blind to the audio stimuli presented and whether thethai was an habituation or test thal, but monitored the visual stimuli presented to the infant, sincethis was not recorded on the video tape. After each 14-second thal, the program switched thestimuli to a flashing red light without sound. Once the infant was looking at the flashing light,the experimenter depressed a second key to initiate the next thai. Visual fixations were stored bythe computer, and both thal-by-thal scores and a block mean summary sheet were automaticallycomputed for each infant.Procedure. The infants were tested in a modified habituation paradigm identical to thatused by Lloyd, Werker, and Cohen (1993). Tn this procedure, during the habituation phase theinfant was shown two word-object combinations presented alternately (i.e., Image 1/”lif’, Image2/”neem”) and in semi-random order until looking time decreased to a set criterion (.65 of thefirst block of four trials), or a maximum number of 20 trials were completed. During the testingphase, two test trials were presented: One maintained the word-object pairing used in the21habituation phase (i.e., Image 1/”lif”) and the other was a new word-object combination (i.e.,Image 1i’neem”). The dependent variable was the amount of time the infant looked at thestimuli. As a control for whether the infant was disinterested in the test stimuli, or had becomefatigued, a novel word-object pairing was presented before and after the experimental stimuli.The infant was seated on the parent’s lap in front of the television monitor and audio speaker.During testing, the parent was made blind to the visual and audio stimuli by wearing a cap witha visor attached that covered her/his eyes and a pair of headsets through which female vocalmusic was played.Two infants were assigned to each of eight orders in the experimental condition (Table1), Recall that each session began and ended with a 14-second pre-test thal of a novel object (aplastic water wheel) paired with the nonsense word “pok”. Next, a minimum of 2 and amaximum of 5 blocks of four 14-second trials were presented. Each block consisted of twoword-object pairings, each presented twice, and randomly ordered by the computer program. Ared, flashing light without sound was presented between trials to recapture the infant’s attention.In reminder, habituation was established if visual fixation time on the third or fourthblock of trials was less than 65% of that for the first block of four trials. If the infant hadreached habituation criterion after the third or fourth block of trials, the test phase began. If theinfant had not reached habituation criterion, a fifth and final habituation block of four trials waspresented, then the test phase began. The test phase was composed of two trials: One word-object pairing was the same, and one was switched. The order of same and switched trials wascounterbalanced across infants, providing a within subject control group. The total testing timewas approximately five minutes.Coding Criteria. For on-line coding, the experimenter depressed a pre-programmedcomputer key when the infant was looking at the video image, and released it when the infantlooked away. The computer was programmed such that timing was accurate to within .25seconds. A maximum score of 14 seconds was possible for each trial.22Reliability coding. A second coder scored twenty percent of the infantS video-tapedresponses to the experimental stimuli, again with girls and boys equally represented. A Pearsonproduct-moment pairwise correlation of on-line and off-line trial scores had to equal or exceed95% for the data to be considered reliable. This level of agreement was achieved in all instances.Experimental Design. Each testing session began with a pre-test, followed immediatelyby a maximum of 20 habituation trials. Next, two test trials were presented, and the sessionended with a post-test.CHART 1Sequence of Stimuli PresentationHabituation Phase TestPhaseTrialsS1 2 3 4 5 6 7 8 9 10 11 12 L 14 15 1 17 18 19 20Block 1 Block 2 Block 3 Block 4 Block 5In reminder, the speech stimulus for the pre- and post-test was the word “pok. Audio 1file consisted of seven different exemplars of the word “hf” and Audio2 consisted of sevendifferent exemplars of the word “neem”. The 8 stimulus orders are presented in Table 1, below.Recall that two infants, one girl and one boy, were tested in each order, rather than four subjectsper order as in Lloyd, Werker, and Cohen (1993), since almost all of the infants in this studyreached habituation criterion.23TABLE IExperiment 1: DesignOrder Habituation Phase Test PhaseSame Switch1 Image 1/lif Image2lneem Imagel/lif Image2/lif2 Image 1/hf Image2/neem Image2/ neem Image 1/neemSwitch Same3 Image 1/hf Image2/neem Image 1/neem Image2/neem4 Image 1/hf Image2lneem Image2/hif Image 1/hfSame Switch5 Image2/lif Imagel/neem Image2lhif Imagel/lif6 Image2/lif Image 1/neem Image 1/neem Image2/ neemSwitch Same7 Image2llif Image 1/neem Image2/neem Image 1/neem8 Image2/lif Image 1/neem Imagel/lif Image2/lifResults and discussionIsolated word condition. A one-way repeated-measures analysis of variance (ANOVA)was conducted on the data collected from the 8 girls and 8 boys who completed the study, andreached habituation criterion. Gender was entered as a between-groups factor, and duration ofvisual fixation to the “final block” of habituation trials, “switch” test thai and “same” test thaiwas then compared. There was a significant interaction between gender and trials, F(2,14) =5.724, p< .01. On the basis of the previous study (Lloyd, Werker & Cohen, 1993), the girls wereexpected to look significantly longer at the “switch” test trial than the “same” test trial and “finalblock” of habituation thals, but were not expected to look significantly longer to the “same” testtrial compared to the “final block” of trials. Indeed, when the girls’ data were analyzedseparately, looking time to the “switch” test trial was significantly higher than the “same” testtrial, t(7) = 6.66, p<.00l (one-tailed), and looking time to the “switch” test trial was significantlyhigher than the “last block” of habituation trials, t(7) = 2.287, p <.03 (one-tailed). The differencein looking time between the “last block” of habituation trials and the “same” test trial was notsignificantly different (see Figure 2).24As expected, the boys did not look significantly longer to the “switch” test trial comparedto the “same” test trial or to the “final block” of habituation trials compared to the “switch” testtrial. In addition, to test whether those who did not dishabituate to the “switch” test trial had notmerely become tired or disinterested in the study, a correlated t-test comparison was donebetween the pre- and post-test and found to be non-significant. Furthermore, it had been decideda priori to eliminate the data from those infants who looked at the pre- and post-test thals for lessthan half the trial duration, but all the infants in this study exceeded the 50% criterion.Exueriment 1: Target WordsPresented in Isolation9.5L 98.50k 8i 7.51m 5.5e 54.5Last Block SameFigure 2: Results from Experiment 1 showing dishabituation to the switch in word-objectpairing by 14-month old females when object novelty and direction of movement iscontrolled.The above results confirm the results of the Lloyd, Werker, and Cohen (1993) study, andestablish that 14-month old girls are able to make word-object associations in a short trainingperiod, even with novel objects and even when direction of movement is controlled. Thus theprocedural basis for Experiment 2, which would test infants’ ability to learn word-objectassociations when the word is presented in a set of carrier phrases, was firmly established.*p<03—— Females—0-——— MalesSwitchTrial25As in the Lloyd, Werker, and Cohen (1993) study, only the female subjects in Experiment1 showed an indication of associating the word and object. This justifies conducting a time-lagcross-sectional study as a follow-up to Experiment 1 to investigate the validity of a genderdifference, which will be completed as a separate project to this thesis. Using this strategy, the14-month old females and males who participated in Experiment 1 will be re-tested at 16 monthsof age in the same condition as they were tested in at 14 months of age. The purpose of thisstudy will be to ensure that the girls maintain the ability to make the word-object associations,and determine whether the boys are yet able to make the associations. A new group of 16-montholds will also be tested to confirm the reliability of the procedure, and to control for any test-retest confounds.In summary, the results from Experiment 1 replicate the Lloyd, Werker, & Cohen (1993)study, and extend the finding, that 14-month old girls are able to learn word-object associationsin a controlled experimental setting, to include novel objects that move in the same direction.4.2 EXPERIMENT 2With the results of Experiment 1, we were in a position to ask the next question: “Whatrole does sentence structure play in the association of words with objects in infancy?”Experiment 2 was designed to test this question using the same methods as in Experiment 1,except the audio stimuli were a sequences of carrier phrases that each ended with the target wordrather than a series of target words presented in isolation.MethodSubjects. A new sample of eighteen 14-month old infants (± 3 weeks) were tested withfemales and males represented equally. Eight girls and 8 boys successfully completed andhabituated to criterion in Experiment 2; 1 boy and 1 girl were tested in each order, as well as 1additional girl and boy tested in randomly selected orders3. The data collected from 9 infantsTwo addilional infants were tested, and their data included in the analyses, because their appointments hadalready been scheduled at the time of study completion26was not included in the analysis: 4 boys due to infant fussiness; 1 boy due to a technical error,and 4 boys due to non-habituation. Overall attrition rate was thirty-three percent.Stimuli. The visual stimuli were the same as in Experiment 1, except the trial durationwas increased to 19 seconds to accommodate the longer audio files. The auditory stimuli variedin that the target word was embedded in a set of three carrier phrases each repeated twice pertrial. Each trial was now 19 seconds in duration to accommodate the addition of carrier phrases.Stress was consistently sentence final in all cases and prosody was controlled across carrierphrases.Apparatus, procedure, coding criteria, and reliability: Same as in Experiment 1 above.Experimental Design. In Experiment 2 the target words were embedded in syntacticallyand semantically correct carrier phrases. which were:Audio 1: Look at the ljf*. Do you see the hf? What a nice hf’Audio2: Look at the neem. Do you see the neem? What a nice neem!* word stress is shown in bold.As in Experiment 1, 8 combinations of habituation and test thals were presented, with “same”and “switch” thals counterbalanced in the test phase:27TABLE 2ExDeriment 2: DesianOrder Habituation Phase Test PhaseSame Switch1 Imagel/Audiol Image2/Audio2 Image 1/Audio 1 Image2/Audiol2 Image1/Audio 1 Image2/Audio2 Image2/ Audio2 Image 1/Audio2Switch Same3 Image 1/Audio 1 Image2/Audio2 Image 1/Audio2 Image2/Audio24 Image 1/Audio 1 Image2/Audio2 Image2/Audio 1 Image 1/Audio 1Same Switch5 Image2/Audio 1 Image l/Audio2 Image2/Audio 1 Image 1/Audio 16 Image2/Audiol Image 1/Audio2 Jmagel/Audio2 Image2/ Audio2Switch Same7 Image2/Audio 1 Image 1/Audio2 Image2/Audio2 Image l/Audio28 Image2/Audio 1 Image l/Audio2 Image1/Audio 1 Image2/Audio 1Results and discussionCarrierphrases condition. As in Experiment 1, a one-way repeated-measures analysisof variance (ANOvA) was conducted on the data collected from the 9 girls and 9 boys whocompleted the study and reached habituation criterion. Gender was entered as a between-groupsfactor, and duration of visual fixation to the “final block”, “switch” and “same” trials was thencompared. There was no significant interaction between gender and test trials, F(2,16) = .433,p> .05. On the basis of the previous study (Lloyd, Werker, & Cohen, 1993), the girls wereexpected to look significantly longer at the “switch” test trial than the “same” test trial and “finalblock” of habituation trials, but were not expected to look significantly longer at the “same” testtrial compared to the “final block” of habituation trials, Therefore, an analysis of the girls’ datawas conducted separately, which compared looking time to the “switch” test trial with lookingtime to the “same” test trial. There was not a significant difference between “switch” and “same”test trials, t(8) = 1.224 , p>.05, nor were any of the other comparisons of looking timesignificantly different for the girls’ or the boys’ data (see Figure 3).Although the comparison between “switch” and “same” test trials for the girls’ data was not significantlydifferent, seven of nine girls looked longer at the “same” than the “switch” trial, regardless of order in which the testtrials were presented. This pattern of results is intriguing, and may warrant further attention.28Again, to test whether those who did not dishabituate to the “switch” test thai had notmerely become tired or disinterested in the study, a correlated t-test comparison was conductedbetween the pre- and post-test thai looking times and found to be non-significant. All of theinfants in this study also exceeded the 50% criterion set a priori for the duration of the pre- andpost-test thai looking times.ExDeriment 2: Target Words Presentedin a Set of Carrier PhrasesL 11.5—— Femalesm 6.5 —U--——— Malese5.5Last Block Switch SameTrialFigure 3: Results of Experiment 2 showing that neither 14-month old girls nor boysdishabituated to the switch in word-object association when the target words werepresented in a set of carrier phrases.In consideration of the above results, it was concluded that the carrier phrases inExperiment 2 did not facilitate infants’ ability to make word-object associations at 14 months ofage. Further support for this conclusion comes from additional analyses of the data: a one-waybetween-groups ANOVA comparing proportion of looking time to trial duration for the “switch”test trial between Experiments 1 and 2 revealed no significant difference in dishabituation; a chisquare analysis showed that the number of infants that habituated to criterion was notsignificantly different for the two experiments; and, a chi-square analysis comparing the number29of infants that habituated by the third block of thals revealed no significant difference betweenExperiments 1 and 2. In fact, the overall pattern of habituation was remarkably similar amongthe infants who participated in these two experiments.5.0 General DiscussionResults from the studies presented here enrich our understanding of the role sentencestructure plays in first language acquisition. Two studies have been conducted. The first tested14-month old infants’ ability to associate isolated nonsense words with novel objects, and hasprovided additional evidence that at 14 months of age, girls, but not boys, are able to make theseword-object pairings. The second study was identical, except the target words were embedded inthe context of a set of carrier phrases. Results from the second study indicated that neither 14-month old boys nor girls were able to make word-object pairings in this more complex condition.There was no difference in overall patterns of habituation between the two studies; theonly difference found was that the girls dishabituated to the switched word-object pairing in thefirst but not the second experiment. There was a difference, however, between the pattern ofhabituation in Experiments 1 and 2, and the pattern of habituation in the Lloyd, Werker, andCohen (1993) study. Most of the infants in Experiments 1 and 2 habituated to the novel objects,(73 and 83 percent, respectively) whereas only half of the infants in each age-group in the Lloyd,Werker, and Cohen (1993) study habituated to the familiar objects. This indicates that perhapsmutual exclusivity was playing a role in the Lloyd, Werker, and Cohen (1993) study -- theyounger infants, and the 14-month old boys may have been in the process of learning labels forthe familiar objects, which may have interfered with their learning of new object labels.Further research is required to address several issues unearthed by Experiments 1 and 2.First, it is important to establish the validity of a gender difference in word-learning at this age.Even though the finding of a gender difference first reported in Lloyd, Werker, and Cohen(1993) has been replicated with Experiment 1, the extent of the gender difference and thepossibility that confounding variables were responsible for the results needs to be examined in30further research. Specifically, a follow-up time-lag sequential study will provide us with moreinformation about the reliability of a gender difference in word-learning.Second, it is important to investigate whether the 14-month old girl& ability to learnword-object pairings in Experiment 1 is, in fact, linguistic. Perhaps the girls were simplyresponding to a sound-object association, and the fact that the sound was linguistic was not animportant factor in their performance. An important subsequent study would be to present thesame objects to girls of this age, except this time the objects would be paired with arbitrary, non-linguistic sounds rather than speech sounds. If the results of such a study are significantlydifferent from those in Experiment 1 in either the speed of habituation, the number of girls whohabituate, or the degree of dishabituation, this would provide a stronger indication that the resultsof Experiment 1 do, indeed, reflect object-labeling rather than the construction of simple object-sound associations.Third, it is important to modify the speech stimuli in Experiment 2 to determine whetherthe infants failed to dishabituate to the switched word-object pairing because the target wordswere embedded in carrier phrases, or because the particular carrier phrases used in thisexperiment made the task demands too high for 14-month old infants. For example, there werethree different types of carrier phrases in the speech stimuli in Experiment 2: Imperative: “Lookat the neem”; Interrogative: “Do you see the neem ?“; and, Exclamatory: “What a nice neem!”Perhaps the memory requirements imposed by the variety of carrier phrases in the second studywas more responsible for the null results of Experiment 2 than the sentential context itself, Afurther study, in which only one type of carrier phrase is repeated several times per trial wouldlessen the memory load on the infants and have more construct validity in addressing thequestion of whether sentence structure influences word-learning in infancy.There are a few things to be said regarding the bootstrapping hypotheses discussed in theintroduction to this paper. According to Pinker’s semantic bootstrapping hypothesis, infantslearn their first nouns through word-to-world semantic operations, starting with simple subjectobject relations. The results from Experiment 1 and 2 appear, upon first glance, to support31Pinker’s position that the relationship between word class and the object is more important thansentence structure to word-learning. A more direct test of Pinker’s semantic bootstrappinghypothesis would be to teach infants object labels when the objects are in semantic relations,such as agent-patient relations. After they have learned the word-object pairings, the word-object pairings could be switched and the infants’ response to this change measured to detenninewhether semantic relations influence infants’ ability to make word-object associations.According to Gleitman’s syntactic bootstrapping hypothesis, infants learn their first wordsby means of sentence-to-world mappings. Although the experiments discussed in this paper didnot test this hypothesis directly, the results from Experiment 2 provide no evidence that infants at14 months of age are taking advantage of sentence structure to learn object labels. To be fair toGleitman’s position, however, these studies did not assess word acquisition when the target wordis presented in one type of sentential condition compared to when the target word is presented inanother type of sentential condition (e.g., classifying the target word as a noun vs. verb). To testGleitman’s syntactic bootstrapping hypothesis directly, a comparison between two types ofsentences applied to a multiply interpretable scene, such as in Naigles (1990) would be required.In summary, the results provided strong evidence that 14-month old girls are able to learnword-object associations in a controlled, experimental setting. The results of the Lloyd, Werker,and Cohen (1993) study, that 14-month old girls, but not boys, are able to learn word-objectassociations in a controlled experimental procedure were replicated in Experiment 1, Thesefindings are consistent with similar findings of gender differences in the literature on earlylanguage acquisition with older infants. Thus the development of a reliable procedure to testword-learning in infants as young as 14 months of age is a useful methodological contribution.The results from Experiment 2 indicated that the particular set of carrier phrases used didnot facilitate word-learning in 14-month old infants, but rather inhibited the girls’ ability to detectchanges in word-object pairings. It is too early, however, to abandon the notion that carrierphrases, under some circumstances and for infants of certain ages, might facilitate word-learning.The results from these studies provide a strong foundation for designing further studies in early32language comprehension. We are now in a better position to examine when, and under whatspecific conditions, young infants are able to decide when a “neem” is, or is not, a “neem”.33REFERENCESAslin, R. N. (1993). Segmentation of fluent speech into words: learning models and the role ofmaternal input. In Developmental Neurocognition: Speech and Face Processing in theFirst Year ofLife, B. de Boysson-Bardies et a!. (Eds.), (pp. 305-3 15). Netherlands:Kluwer Academic Publishers.Bailargeon, R. & DeVos, 3. (1991). Object permanence in young infants: Further evidence.Child Development, 62, 1227-1246.Benedict, H. (1979). Early lexical development: comprehension and production. Journal ofChild Language, 6, 183-200.Bomstein, M. H. & Kessen, W. (1977). The categories of hue in infancy. Annual Progress inChild Psychiatry and Child Development, 44-50.Brown, R. (1973). A first language. Cambridge, MA: Harvard University Press.Chomsky, N. (1959). A review of Skinner’s Verbal behavior. Language, 35, 26-58.Chomsky, N. (1965). Aspects of the theory ofsyntax. Cambridge, MA: MIT Press.Chomsky, N. (1975). Reflections on language. New York: Random House.Dunham, P. & Dunham, F. (1992). Lexical development during middle infancy: A mutuallydriven infant-caregiver process. Developmental Psychology, 28, 414-420.Echols, C. H. & Newport, E. L. (1992). The role of stress and position in determining firstwords. Language Acquisition, 2, 189-220.Fernald, A. (1989). Intonation and communicative intent in mothers’ speech to infants: Is themelody the message? Child Development, 60, 1497-15 10.Fernald, A. & Mazzie, C. (1991). Prosody and focus in speech to infants and adults.Developmental Psychology, 27, 209-221,Gleitman, L. (1990). The structural sources of verb meanings. Language Acquisition, 1(1), 3-55.Gleitman, L., Landau, B. & Wanner, E. (1988). Where learning begins: Initial representationsfor language learning. In F. Newmeyer (Ed.), The Cambridge linguistic survey (pp. 150-193). New York: Cambridge University Press,Goldfield, B. A. & Reznick, J. S. (1990). Early lexical acquisition: rate, content, and thevocabulary spurt. Journal of Child Language, 17, 17 1-183.Golinkoff, R. M. & Hirsh-Pasek, K. (1990). Let the mute speak: What infants can tell us aboutlanguage acquisition. Merrill-Palmer Quarterly, 36, 67-92.Golinkoff, R. M., Hirsh-Pasek, K., Cauley, K. M., & Gordon, L. (1987). The eyes have it:Lexical and syntactic comprehension in a new paradigm. Journal of Child Language, 14,23-45.34Humphrey, K., & Tees, R. C. (1980). Auditory-visual coordination in infancy. Some limitationsof the preference methodology. Bulletin of the Psychonomic Society, 16, 213-216.Humphrey, K., & Tees, R. C. (1979). Auditory-visual integration of temporal relations ininfants. Canadian Journal ofPsychology, 33, 347-352.Huttenlocher, J., Haight, W., Bryk, A., Seltzer, M., & Lyons, T. (1991). Early VocabularyGrowth: Relation to Language Input and Gender. Developmental Psychology, 27, 236-247.Hyde, J. S. & Linn, M. C. (1988). Gender differences in verbal ability: A meta-analysis.Psychological Bulletin, 104, 53-69.Jusczyk, P. W., Cutler, A. & Redanz, N. J. (1993). Infants’ preferences for the predominantstress patterns of English words. Child Development., 64, 675-687.Jusczyk, P. W., Kemler Nelson, D. G., Hirsch-Pasek, K., Kennedy, L., Woodward, A. & Piwoz,J. (1992). Perception of acoustic correlates of major phrasal units by young infants.Cognitive Psychology, 24, 252-293.Klatt, D. H. (1979). Speech perception: A model of acoustic-phonetic analysis and lexicalaccess. Journal ofPhonetics, 7, 279-312.Ladefoged, P. (1975). A course in phonetics. NY, NY. Harcourt Brace Javanovich, Inc.Lloyd, V. L., Werker, J. F. & Cohen, L. B. (1993). Age changes in infants’ ability to associatewords with objects. Poster presented at the 60th Anniversary Meeting of the SocietyforResearch in Child Development, New Orleans, Louisiana.Maccoby, E. & Jacklin, C. N. (1974). The psychology ofsex differences. Stanford, CA:Stanford University Press.Markman, E. M. (1989). Categorization and naming in children: Problems of induction.Cambridge, MA: MIT Press.Marslen-Wilson, W. (1989). Access and integration: Projecting sound onto meaning. InWilliam Marsien-Wilson (Ed.), Lexical representation and access (pp. 3-24).Cambridge, MA: MIT Press.Marsien-Wilson, W. & Tyler, L. K. (1980). The temporal structure of spoken languageunderstanding. Cognition, 8, 1-7 1.McClelland, J. L. & Elman, J. L. (1986). The TRACE model of speech perception. CognitivePsychology, 18, 1-86.Mehler, J., Jusczyk, P., Lambertz, G., Haisted, N., Bertoncini, J., & Amiel-tison, C. (1988). Aprecursor of language acquisition in young infants. Cognition, 29, 143-178.Moon, C., Cooper, R. P., & Fifer, W. P. (in press). Two-day-old infants prefer their nativelanguage. Infant Behavior and Development.Naigles, L. (1990). Children use syntax to learn verb meanings. Journal of Child Language, 17,357-374.35Nelson, K. (1973). Structure and strategy in learning to talk. Monographs of the Society ofResearch in Child Development, 38, No.149.Oviatt, S. L. (1980). The emerging ability to comprehend language: An experimental approach.Child Development, 51, 97-106.Oviatt, S. L. (1982). Inferring what words mean: early development in infants’ comprehensionof common object names. Child Development, 53, 274-277.Oviatt, S. L. (1985). Tracing developmental change in language comprehension ability beforetwelve months of age. PRCLD, 24, 87-93.Pegg, J. E., Werker, J. F., & McLeod, P. J. (1992). Preference for infant-directed over adult-directed speech: Evidence from 7-week-old infants. Infant Behavior and Development,15, 325-345.Pinker, S. (1988). Learnability theory and the acquisition of a first language. In Frank S. Kessel(Ed.), The development of language and language researchers: Essays in honor ofRoger Brown (pp. 97-119). Hillsdale, NJ: Erlbaum.Pinker, S. (1989). Language Acquisition. In Michael I. Posner (Ed.), Foundations of CognitiveScience (pp. 359-399). Cambridge, MA: MIT Press.Polka, L. & Werker, J. F. (in press). Developmental changes in perception of non-native vowelcontrasts. Journal ofExperimental Psychology: Human Perception and Performance.Reznick, J. S. (1990). Visual preference as a test of infant word comprehension. AppliedPsycholinguistics, 11, 145-166.Reznick, 3. S., & Goldfield, B. A. (1992). Rapid change in lexical development incomprehension and production. Developmental Psychology, 28, 406-413.Salasoo, A. & Pisoni, D. B. (1985). Interaction of knowledge sources in spoken wordidentification. Journal ofMemory and Language, 24, 210-23 1.Simon, J. (1980). Paradigms lost: reflections on literacy and its decline. NY, NY. CrownPublishers.Skinner, B. F. (1957). Verbal behavior. New York: Appleton-Century Crofts.Snyder, L. S., Bates, E., & Bretherton, I. (1981). Content and context in early lexicaldevelopment. Journal of Child Language, 8, 565-582.Tyler, L. K. & Marslen-Wilson, W. D. (1981). Children’s processing of spoken language.Journal of Verbal Learning and Verbal Behavior, 20, 400-416.Warren, R. M. & Warren, R. P. (1970). Auditory illusions and confusions. Scientific American,223, 30-36.Werker, J. F., Pegg. J. E. & McLeod, P. J. (in press). A cross-language comparison of infantpreference for infant-directed vocal interactions: English and Cantonese. InfantBehavior and Development.36Werker, J. F. & Lalonde, C. E. (1988). Cross-language speech perception: Initial capabilitiesand developmental change. Developmental Psychology, 24, 1-12.Werker, J. F., Lloyd, V. L., Pegg, J. E. & Polka, L. (in press). Toward a more completeunderstanding of the role of the input on infant speech processing. In J. Morgan & K.Demuth (Eds.), Signal to Syntax. Lawrence Erlbaum Associates.Werker, 3. F. & Pegg, J. E. (1992). Infant speech perception and phonological acquisition. In C.Ferguson, L. Menn, and C. Stoel-Gammon (Eds.), Phonological Development: Models,Research, and Implications (pp. 285-3 12). Parkton Maryland: York.Werker, J. F. & Tees, R. C. (1984). Cross-language speech perception: Evidence for perceptualreorganization during the first year of life. Infant Behavior & Development, 7, 49-63.Younger, B, & Cohen, L. B. (1986). Developmental change in infant’s perception ofcorrelations among attributes. Child Development, 57, 803-8 15.Zelazo, P. R., Weiss, M. 3. (1989). Recovery and dishabituation of sound localization amongnormal-, moderate-, and high-risk newborns: Discriminant validity. Infant Behavior &Development, 12, 321-340.37Appendix 1: Objects shown during Pre/Post4est (water wheel), andHabituation/Dishabituation phases (modelling clay figures). 38Testing BoothComposite MonitorsObserverComposite or Mac MonitorSpeakerParent with Child on LapAudio Cassette PlayerObservation Booth00Laser Diskand PlayerAudio AmplifierMac II with DVA 4000Appendix 2: Apparatus configuration for Experiments 1 and 2. 39

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0087272/manifest

Comment

Related Items