2ed Sta

Syntactic bootstrapping

an ethemt evedentitingrele

ate s

intransitive sentences did not. These results demonstrate that 2-year-olds infer important

arn twor

xampe presLockreferer exareferefor n

information alone.In word learning tasks, sighted children can coordinate

their sensitivity to syntactic information with their obser-

sentence (e.g., the boy mooped) or a transitive sentence(e.g., the boy mooped the girl). Toddlers then viewed twoscenes, one depicting an event with one actor, the otheran event with two actors. When asked to nd mooping,toddlers looked longer at the two-actor event in the transi-tive, than in the intransitive, condition. Performance in a

0010-0277/$ - see front matter 2009 Elsevier B.V. All rights reserved.

* Corresponding author. Tel.: +1 847 467 0737.E-mail address: s-arunachalam@northwestern.edu (S. Arunachalam).

Cognition 114 (2010) 442446

Contents lists available at ScienceDirect

Cognit

.e lsedoi:10.1016/j.cognition.2009.10.015sighted children. Landau and Gleitman (1985) have ob-served that they do so in part by using the syntactic con-texts in which words appear. Of course, sighted childrentoo recruit syntactic information to glean broad aspectsof word meaning. For example, they expect that verbs tak-ing sentential complements (e.g., the boys [verb] that thetiger will pounce) refer to mental states rather than activ-ities (e.g., Asplin, 2002; Gleitman, 1990; Papafragou, Cassi-dy, & Gleitman, 2007). But it is an open question howmuchlearners can glean about verb meaning from syntactic

the observed scene that is most compatible with the mean-ing conveyed by the sentence.

Recent evidence has gone further, demonstrating thatsimply hearing a novel verb in a sentence, without anaccompanying scene, helps toddlers to infer some informa-tion about its meaning, and to use that information later tond an appropriate referent when a visual scene becomesavailable (Yuan & Fisher, 2009). Yuan and Fisher rst intro-duced 2-year-olds to two actors, engaged in a dialogue inwhich a novel verb was mentioned either in an intransitiveIntuitively, we assume that to lenovel word, children must hear theneously observing its referent, for ein the presence of a cat, or /kik/ in thaction (e.g., Augustine, 398/1992;1984). But visual access to potentialcontexts and language learners. Fodren lack visual access to potentialrive at virtually the same meaningscomponents of meaning from syntactic structure alone, using it to direct their subsequentsearch for a referent in a visual scene.

2009 Elsevier B.V. All rights reserved.

he meaning of ad while simulta-le, hearing /kt/ence of a kickinge, 1690; Pinker,nts varies acrossmple, blind chil-nts, yet they ar-ovel words as do

vation of the visual world. This coordination is especiallyimportant in verb learning (e.g., Fisher, 2002; Gleitman,1990; Naigles, 1990), where syntactic information inherentin the sentence structure provides a linguistic zoom lensto help the learner detect what is currently being ex-pressed about an ongoing event or a state or relation(Gleitman & Fisher, 2005: p. 132). In this zoom lens met-aphor, syntactic information, available concurrently withthe visual scene, focuses learners attention on the part ofLanguage acquisitionWord learning two participants performing a causative action. When asked to nd mooping, toddlers

who had heard transitive sentences chose the causative scene; those who had heardBrief article

Meaning from syntax: Evidence from

Sudha Arunachalam *, Sandra R. WaxmanDepartment of Psychology, Northwestern University, Evanston, IL 60208, Unit

a r t i c l e i n f o

Article history:Received 29 June 2009Revised 16 October 2009Accepted 20 October 2009

Keywords:

a b s t r a c t

When toddlers viewbeen shown to helpfurther to reveal thasupports toddlers idialogues incorporathe absence of anyviewed two candid

journal homepage: www-year-olds

tes

vent while hearing a novel verb, the verbs syntactic context hasidentify its meaning. The current work takes this nding one stepn in the absence of an accompanying event, syntactic informationcation of verb meaning. Two-year-olds were rst introduced tonovel verbs either in transitive or intransitive sentences, but invant referent scenes (see Yuan & Fisher, 2009). Next, toddlerscenes: (a) two participants performing synchronous actions, (b)

ion

vier .com/ locate/COGNIT

S. Arunachalam, S.R. Waxman / Cognition 114 (2010) 442446 443control condition revealed that this preference for the two-actor scene reected their interpretation of the transitiveverb, per se, and not the presence of the two nouns alone.Thus, even before seeing an event, when toddlers heard anovel verb in a sentence, they associated the number ofnouns in the sentence with the novel verb, and broughtthis into correspondence with the number of actors inthe subsequently presented visual scene.

Toddlers spontaneous ability to match the number ofnouns with the number of participants in the event de-scribed by the verb is impressive; to succeed, they had tohave used linguistic information alone to determine thenumber of participants to associate with the verb. But re-cent work also reveals that toddlers can glean more fromsyntactic context than the number of likely participants.For example, when a novel verb is presented in conjunc-tion with a visual scene, 26-month-olds can determinewhether a verb taking two noun phrases refers to a causa-tive scene (e.g., a duck pushing down on a rabbits head,forcing him to squat) or synchronous scene (e.g., a duckand bunny each twirling one arm in circles) by notingwhether the verb appears in a transitive (e.g., the duck isgorping the bunny) or intransitive sentence (e.g., the duckand the bunny are gorping) (Naigles, 1990; see also Bunger& Lidz, 2004; Fisher, 2002; Naigles & Kako, 1993). Thus,when a novel verb is introduced in conjunction with visualscenes, toddlers use the syntactic structures in which twonoun phrases occur to infer the relation between twoparticipants.

But can toddlers use syntactic structure to zoom in onthe relation between participants even when the verb isintroduced without accompanying visual information?The current experiment addresses this issue directly. Fol-lowing Yuan and Fisher, we presented the linguistic streambefore providing any visual information, but following Nai-gles, we introduced sentences with two nouns and subse-quently tested learners interpretations by showing themvisual scenes with two participants. At issue is whethertoddlers can use syntactic information in absence of anevent, to form a representation of the novel verb that al-lows them to determine not just the number of partici-pants involved, but also the relation in which theparticipants will stand to each other.

First, toddlers heard a novel verb, presented in eithertransitive (e.g., the boy is going to moop the lady) orintransitive sentences (e.g., the boy and the lady are goingto moop). Only later were they given relevant visual infor-mation. This consisted of two test scenes, presented side-by-side, each involving two actors. In one test scene, theactors were engaged in a causative event (e.g., a boy spinsa girl in a chair); in the other, the same actors were en-gaged in a synchronous event (e.g., a boy and a girl eachwave one hand in circles). Crucially, while the test sceneswere presented, no syntactic or semantic informationwas available to help the toddler infer which scene de-picted mooping. Therefore, their choice of test eventshad to be guided by the syntactic information they hadheard before exposure to the events.

We reasoned as follows: if toddlers can use the syntac-tic structure in which a novel verb is presented to infermeaning, then those who had heard the novel verbs intransitive sentences should choose the causative eventwhen asked to nd mooping more often than thosewho had heard intransitive sentences. Those who hadheard intransitive sentences should choose either scene.See Naigles and Kako (1993) for evidence that althoughtoddlers have a clear preference to interpret transitive sen-tences as referring to causative events, they accept bothsynchronous and causative events as referents of intransi-tive sentences, a pattern that likely reects the fact thatintransitive sentences are compatible with both speci-cally synchronous meanings (e.g., waving hand in circles)and more general meanings (e.g., playing).

1. Methods

1.1. Participants

Forty typically-developing toddlers (20 males; meanage of 27.3 months, ranging 24.929.9) were included inthe nal sample. We focused on 27-month-olds becauseat this age, toddlers actively add verbs in their own spon-taneous speech and have demonstrated success in similarexperimental tasks (e.g., Naigles, 1990; Yuan & Fisher,2009). Toddlers were recruited from Evanston, IL and sur-rounding areas, and were acquiring English as their nativelanguage, with less than 25% exposure to another lan-guage. Parents completed the MacArthur long form vocab-ulary checklist: words and sentences (Fenson et al., 1993).Mean production vocabulary was 442 words (range: 50681); there were no differences between conditions invocabulary. To be included in the nal sample, toddlershad to (a) point correctly on at least two (of four) pointinggames and training trials, and (b) point clearly on at leastone test trial. Fourteen toddlers who failed to meet thesecriteria were excluded; another 13 were excluded due tofussiness, and 2 to experimenter error.

2. Materials

2.1. Visual stimuli

In the Dialogue phase, toddlers viewed digitized videorecordings of two live actors speaking. In the test phase,they viewed videos of different actors performing actions;half of the actions involved two human actors, and half in-volved one human actor and one inanimate object. Videoswere presented on a 20-in. television screen.

2.2. Auditory stimuli

A female native speaker of American English producedthe speech stimuli (described in Table 1) using child-direc-ted speech. Speech was recorded in a sound-attenuatedbooth, synchronized with the visual stimuli, and presentedon a speaker centered below the visual display.

2.3. Apparatus and procedure

Toddlers played freely with toys while the caregiversigned a consent form and completed the MacArthur

imental trials). One training trial involved an unergative

Test

Baseline (24 s) Response (24 s)

ve connow w

dy an? The

e boy

s. They

444 S. Arunachalam, S.R. Waxman / Cognition 114 (2010) 442446checklist. The toddler and caregiver were then brought intoan adjoining room where the toddler was seated in an in-fant seat, 16 in. from the television screen. The caregiversat behind the toddler and was requested not to talk orotherwise interact with her during the session. One exper-imenter controlled the experimental procedure from be-hind a curtain; another sat next to the toddler to elicitresponses. We asked toddlers to indicate their choice ofscenes by pointing. We chose pointing as a measure be-cause it is a more active behavioral response than tradi-tional looking-time methods, has been successfully usedat this age in word learning tasks (Bernal, Lidz, Millotte,& Christophe, 2007; Maguire, Hirsh-Pasek, Golinkoff, &Brandone, 2008), and converges well with looking-time(Arunachalam & Waxman, 2009). Pointing was recordedwith a video camera centered above the screen.

Toddlers rst participated in a warm-up game designedto encourage them to point to the screen. Two video clipsof Sesame Street characters were presented on the screen,

Table 1Representative set of stimuli.

Dialogue phase

Linguisticstream

Transitive condition IntransitiA: You know what? A: You kB: What? B: What?A: The lady mooped my brother A: The laB: Really? The lady mooped yourbrother?

B: Reallymooped?

A: And the boy is going to moop the girl A: And thmoop

B: Oh yes. He is going tomoop her

B: Oh ye

Observationalstreamside-by-side, and the experimenter asked the toddler topoint, once to a particular character (e.g., Elmo), and onceto a particular action (e.g., dancing). If a toddler was reluc-tant to point or pointed incorrectly, the experimenter dem-onstrated the correct response and encouraged the toddlerto do the same.

Next, each toddler participated in six trials, each featur-ing a different verb. Two training trials (involving familiarverbs, sleep and hug) were followed by four experimentaltrials (involving novel verbs). Because the training trialswere designed to familiarize toddlers with the experimen-tal procedure, the structure of the training and experimen-tal trials was identical. Each trial included two phases:dialogue and test. For experimental trials, toddlers wererandomly assigned to either the transitive or intransitivecondition. Toddlers in both conditions saw exactly thesame video scenes, but heard different auditory stimulisee Table 1. The four experimental trials were presentedin one of two random orders, balanced across conditions.The leftright positions of the two types of test scene werecounterbalanced across trials.intransitive verb (sleep), and the other involved a transi-tive verb (hug). The experimental trials involved eithertransitive sentences, e.g., the boy is going to moop thelady (transitive condition) or conjoined-subject intransi-tive sentences, e.g., the boy and the lady are going tomoop (intransitive condition). Each dialogue consisted oftwo six-sentence video clips, averaging 34 s, includingeight mentions of the verb. Dialogue videos appeared inthe center of the screen.

2.3.2. Test phaseToddlers then saw the two test scenes side-by-side. On

the training trials, the event type was held constant across2.3.1. Dialogue phaseEach trial began with a scene of two women having a

conversation in which they used either a known verb (onthe two training trials) or a novel verb (on the four exper-

dition Look! Wow! Wheres mooping?hat?

d my brother moopedlady and your brother

and the girl are going to

are going to mooptest scenes. On the sleep trial, both test scenes depictedone participant: (1) a woman sleeping, and (2) the samewoman crying. On the hug trial, both scenes depictedtwo participants: (1) a woman hugging a toy, and (2) thesame woman lifting a box. On each experimental trial, bothtest scenes depicted the same two participants (e.g., a manand a woman), with the event type differing across the twotest scenes: (1) a synchronous event (e.g., man and womaneach wave one of their own hands in circles), and (2) acausative event (e.g., the same man spins the same womanaround). On all four experimental trials, each test scene de-picted two moving participants.1

For both training and experimental trials, the test phasebegan with a 24 s inspection period, during which toddlersheard Look! Wow! and had an opportunity to inspect thetest scenes, both of which were novel to them. The screen

1 On half of the experimental trials, one participant was animate and theother inanimate; on the others, both were animate. On animateinanimatetrials, the animate participant was always the agent of the causative action.Because there were no effects of animacy, we collapse across this factor inthe analysis.

then went blank for 1.5 s, during which time the novel verbwas presented. Importantly, at this point, the verb was pre-sented with no syntactic information to indicate its mean-ing: In both conditions, toddlers heard, e.g., Wheresmooping? Next, the test scenes reappeared for 24 s, andinfants heard: Do you see mooping? Find mooping! Theexperimenter seated next to the toddler repeated the ques-tion and encouraged her to point (e.g., Can you showme?). Neutral feedback (e.g., Good pointing!) was givenon all trials, regardless of the childs response. Notice that iftoddlers choices differ as a function of condition, these dif-ferences must be attributable to the syntactic information(i.e., transitive or intransitive sentences) provided beforethe test scenes appeared.

2.3.3. CodingAll pointing responses were veried by a second condi-

tion-blind coder from the video recordings. Agreement be-tween coders was 100%. We analyzed toddlers rst point

intransitive sentences (Naigles & Kako, 1993; Yuan & Fish-er, 2009). Recall that intransitive sentences are compatiblewith both specic synchronous meanings like wavinghand in circles and more general meanings like playing.Clearly, then, by 27 months, toddlers successfully extractinformation about a new verbs meaning from the syntac-tic context in which it is introduced, even in the absence ofvisual information, and recruit this information later whenthey are shown candidate events. Toddlers ability to cullan initial representation of a novel verbs meaning in thedialogue phase, on the basis of syntactic context alone, isstriking. It reveals an early ability to do more than countnoun phrases and match them to event participants (Yuan& Fisher, 2009). Counting noun phrases has been argued tobe a central ability in early verb learning (Fisher, 2002;Lidz, Gleitman, & Gleitman, 2003; Yuan & Fisher, 2009),but the current results reveal that toddlers use syntacticinformation to encode more than just participant number;they also use syntactic information to home in on the kind

e scen

S. Arunachalam, S.R. Waxman / Cognition 114 (2010) 442446 445on a given trial, with one exception: one toddler spontane-ously corrected his rst response on one trial, explicitlysaying, No, thats mooping, pointing to the other scene.

2.4. Results and discussion

The results, depicted in Fig. 1, provide the rst docu-mentation that even in the absence of a visual scene, 27-month-olds can use syntactic information to zoom in ona particular relation between event participants. As pre-dicted, toddlers in the transitive condition were morelikely to point to the causative scene (M = .67) than thosein the intransitive condition (M = .43), F(1, 38) = 10.06, p