Infants Encode Phonetic Detail during Cross-Situational Word Learning.

Paola Escudero,Karen E Mulak,Haley A Vlach

doi:10.3389/fpsyg.2016.01419

Abstract

Infants often hear new words in the context of more than one candidate referent. In cross-situational word learning (XSWL), word-object mappings are determined by tracking co-occurrences of words and candidate referents across multiple learning events. Research demonstrates that infants can learn words in XSWL paradigms, suggesting that it is a viable model of real-world word learning. However, these studies have all presented infants with words that have no or minimal phonological overlap (e.g., BLICKET and GAX). Words often contain some degree of phonological overlap, and it is unknown whether infants can simultaneously encode fine phonological detail while learning words via XSWL. We tested 12-, 15-, 17-, and 20-month-olds’ XSWL of eight words that, when paired, formed non-minimal pairs (MPs; e.g., BON–DEET) or MPs (e.g., BON–TON, DEET–DIT). The results demonstrated that infants are able to learn word-object mappings and encode them with sufficient phonetic detail as to identify words in both non-minimal and MP contexts. Thus, this work suggests that infants are able to simultaneously discriminate phonetic differences between words and map words to referents in an implicit learning paradigm such as XSWL.

Highlights

Learning new words is a difficult task
Because looking remained above 50% only for the first word and interstimulus interval segment, we focused our analysis window there. This resulted in a 1.5 s analysis window, which encompasses the shaded region on Figure 4
We examined whether infants can simultaneously discriminate phonological differences among words and learn word mappings during XSWL

Summary

Introduction

Learning new words is a difficult task. In any one moment in time, the world presents learners with a seemingly infinite number of potential referents for just one word (Quine, 1960). Despite the ambiguity and difficulty of the task, infants are able to acquire new words, with Australian English-learning infants going from saying zero or a few words at 12 months, to over 100 words 8 months later, to over 400 words 8 months after that (Kalashnikova et al, 2016). To understand how infants become such remarkable word learners, much research has examined the cognitive processes that contribute to word mapping. Research on word mapping has historically focused on how learners resolve ambiguity in word-referent mapping in one moment in time. In naturalistic language learning environments, infants must resolve ambiguity across many moments in time. Recent research has shifted toward examining how learners resolve ambiguity across learning events. This work has revealed that learners can track the co-occurrence of words and referents across time

Objectives

Results

Conclusion