| Incidental word learning: Two-year olds can infer the referent of a novel word “on the fly” using linguistic and contextual cues
Renate Zangl* and Deena Skolnick**
[*Postdoctoral Research Associate at Stanford, **former Honors students at Stanford, Deena is now a graduate student at Yale]
When encountering a novel word, adults use multiple linguistic and contextual cues to infer its referent. An adult hearing See the APPLE on the toma when seeing an apple resting on an unfamiliar object would infer that toma was the name of the novel object without further clarification. For children just beginning to understand speech such indirect word learning is difficult; however, from two years onwards children increasingly learn new vocabulary indirectly through inferential processes, drawing on diverse sources of information including prior word knowledge and cues from the visual scene. In three experiments we investigated whether two-year-olds could learn the name of a novel object mentioned only incidentally in a situation in which attention was strongly directed, both visually and verbally, to another more salient object already familiar to the child.
Two-year-olds were tested in a “looking-while-listening” procedure as they looked at pictures while listening to speech naming one of the pictures. Through high-resolution analyses of eye-movements, we monitored the time course of comprehension as the sentence unfolded. On both Teaching/Target and Teaching/Distracter trials, the child saw salient familiar objects positioned on or by two unfamiliar structures (monochromatic pedestal or arch). For half the children, the pedestal was the novel target object and the arch was the distracter, reversed for the other half. In the speech accompanying Teaching/Target trials, the name of the familiar object was emphatically stressed, while the novel target word was deaccented (e.g. Look at the KITTY on the deebo). On Teaching/Distracter trials, the familiar object was also stressed but the novel object was not named (e.g. Look at the DOGGY overthere) On Testing trials, children saw the arch and pedestal paired without familiar objects and were asked Where’s the DEEBO?
In the first two experiments, one picture showing a familiar object positioned on or by a novel object was presented on each teaching trial, with the preposition (on vs. by the deebo) used correctly in Expt.1 (n=32) and incorrectly in Expt.2 (n=34). In Expt.3 (n=32), the processing load on teaching trials was increased: two pictures were presented simultaneously, each showing a familiar and a novel object. Here the child had to first identify the appropriate familiar object to determine which of the two novel objects was most likely the one also mentioned in the sentence.
In Expts.1 and 2 children correctly identified the novel target object on testing trials, although it was never the focus of the sentence during teaching and was never singled out or labeled directly. In Expt.3 two-year-olds showed they could make the correct mapping under even more complex processing conditions. In this challenging learning context, they were able to infer that the unstressed novel word mentioned incidentally in the sentence referred to the unknown object that was closest in proximity to the familiar object also named in the sentence. Even more impressive, this rapid inferential learning happened “on the fly” as the child scanned the scene and processed the spoken sentence, integrating visual and auditory information from moment to moment.
|
