Publication Type: Conference Paper/Unpublished Manuscript
Abstract: Backgrounds and aims: Two competing theories of lexical acquisition attempt to describe the process by which infants learn the appropriate range of a word’s meaning: The taxonomic constraint suggests that infants attribute newly-learned words to whole categories rather than individual items. Alternatively, the exemplar-based account suggests that words are initially used in association with the specific items they occur with and only after considerable experience with several members of a category are infants able to extend the use of the word to refer to a wider range of items. In an inter-modal preferential looking (IPL) task, the interaction between cognitive categories and early words was investigated.

Methods: In a test – train – test – test paradigm, 10-month-old infants were taught words for two pictures, both typical members of basic adult linguistic categories, which were selected to be familiar but unknown to infants of this age. Prior knowledge of the two words was tested in the initial test block.
Trials lasted 5000ms and were divided into two parts with the presentation of the target word beginning half-way through. Testing blocks presented the two pictures simultaneously naming one of them. Pre-training and post-training test blocks were exactly identical. The final testing block differed only in that the two training pictures were replaced by two pictures that infants had not seen before, but were also typical members of the same adult linguistic categories. The training block consisted of 8 presentations, with each picture- word pairing appearing four times. Infants’ looking behaviour during the task was recorded and analysed off-line.

Key Results: Contrary to expectations, significant naming effects were found in the pre-training test block, using three different IPL measures. Naming effects with infants of this age have not been demonstrated before using this method. Items analyses and comparisons between groups of infants will also be reported (in progress).

Conclusions: These results demonstrate that 10-month-old infants are capable of recognising picture-label associations without any explicit training and respond to the IPL task successfully, at least when familiar items are used. Even at this early stage of lexical acquisition they demonstrate preliminary linguistic categorisation abilities which enable them to generalise the use of known words to pictures they have never seen before, in unfamiliar contexts. Implications for the two theories presented above will be discussed.