2018 Friday Poster 6603

Friday, November 2, 2018 | Poster Session I, Metcalf Small | 3pm

Matching number vs. linking roles: Using 3-participant scene percepts to understand infants’ bootstrapping
L. Perkins, T. Knowlton, A. Williams, J. Lidz

Understanding verb learning in infants requires understanding both how they represent scenes as events, and what principles they use for mapping sentences onto events. In Study 1, we identify a stimulus scene that 10-month-olds view as a 3-participant event. In Study 2, we test whether 20-month-olds think this can be described by a transitive sentence. This allows us to distinguish two bootstrapping hypotheses: an expectation that arguments match participants in number (Participant-to-Argument Matching, PAM) [1- 3] vs. an expectation of linking between grammatical and thematic relations (Thematic Linking) [4-5], e.g. of transitive Subject to Agent and Object to Patient.

Previous studies found that infants who heard novel transitive verbs preferred to look at events with 2 participants [1,3], but infants who heard novel intransitive verbs did not reliably prefer 1-participant events [3,6]. These results are consistent with Thematic Linking, for which transitive but not intransitive surface syntax is informative. But they are also compatible with PAM, if infants viewed the intended 2- participant stimulus as a 1-participant event [3,7]. To differentiate these alternatives, we normed infants’ representations of a TAKING scene, confirming that they view it as a 3-participant event, and then tested whether they allow it to be described with a 2-argument sentence.

Study 1 habituated 32 infants aged 9;21-12;15 (mean=10;23) to silent videos of a girl picking up a truck next to a boy (Fig.1). By hypothesis, this event is viewed as a 2-participant PICKING-UP. At test, one group saw the girl slide the truck instead of picking it up. Despite the different motion, this should be seen with the same participant structure. Another group saw the girl pick up the truck from the boy’s grip. This should be seen as a 3-participant TAKING. We found a significant interaction between phase and condition (F(4,30)=5.42, p<.03): the change in participant structure led to greater dishabituation than the perceptually salient manner change (Fig.2). Infants thus view this TAKING event under a 3-participant concept privileging the victim, along with the taker and taken.

Study 2 familiarized 19-to-22-month-olds with this TAKING scene paired with The girl pimmed the truck. Under Thematic Linking, pimming can describe the 3-participant TAKING, provided infants represent the girl as agent and the truck as patient. Under PAM, pimming cannot be TAKING: the sentence must express a nearby 2-participant concept like MOVING. At test, infants are prompted to find pimming by choosing between a TAKING scene and another scene where the girl moves the truck without the boy. Thematic Linking predicts a preference for TAKING. PAM predicts no preference: in both videos the girl moves the truck. Control groups are familiarized with ditransitive The girl pimmed the truck from the boy, which should describe TAKING, and intransitive The truck pimmed, which should describe the truck’s moving.

Fig.4 shows results from 31 participants, target n=72 to be tested by Fall 2018. While not yet reliable, these results provide suggestive support for Thematic Linking. Our full results will distinguish whether early verb learning relies primarily on argument number or thematic content.

References

[1] Naigles 1990. Children use syntax to learn verb meanings. J Child Lang 17. [2] Fisher et al. 2010. Syntactic bootstrapping. WIRES Cog Sci 1. [3] Yuan, Fisher, & Snedeker 2012. Counting the nouns. Child Dev 83. [4] Pinker 1984. Language Learnability and Language Development. [5] Williams 2015. Arguments in Syntax and Semantics. [6] Noble, Rowland, & Pine 2011. Comprehension of argument structure and semantic roles. Cog Sci 35. [7] Pozzan, Gleitman, & Trueswell 2015. Semantic Ambiguity and Syntactic Bootstrapping. Lang Learn Dev 12. [8] Werker et al. 1998. Acquisition of word-object associations by 14-month old infants. Dev Psych 34. [9] Gordon 2003. The origin of argument structure in infant event representations. Proc BUCLD 28. [10] Wellwood et al. 2015. Participant structure in event perception. Proc PLC 38. [11] Waxman et al. 2009. 24-month-old children’s interpretations of novel nouns and verbs in dynamic scenes. Cog Psych 59.