Infant Word Recognition and Word Learning is Predicted by Socio-economic Status

• In Event: Conditions of social vulnerability and language development

Abstract

There is mounting evidence that social-economic indicators influence the course of early language acquisition (e.g., Golinkoff, Hoff, Rowe, Tamis-leMonda & Hirsh-Pasek, 2019. A seminal study by Hart and Risley (1995) demonstrated that children who enter preschool from higher socio-economic strata are better positioned for language learning than those from lower strata. Less emphasis has been placed on how socio-economic influences language acquisition in infancy. Here, we report the results of two experiments measuring word learning and word recognition in a multilingual, multi-ethnic, and economically diverse cohort of Singaporean infants (Singapore Longitudinal Early Development Study).

In the first study, 45 infants from 18 to 20 months were tested on their capacity to bind newly learned labels to objects using the Switch paradigm (Werker, Cohen, Lloyd, Casasola, and Stager, 1998). Infants were taught two dissimilar sounding words /pʰiŋ/ or /lan/, associated with novel objects, during a habituation phase. During the test phase, infants were presented with a Same trial (the object-word pairing which they were taught) and a Switch trial (a novel object-word pairing). During the test phase, fixation times to Same trials were compared to fixation times to Switch trials, yielding a significant novelty effect (t(44) = 3.89, p<.0001. In a hierarchical linear regression, total conceptual vocabulary size, paternal and maternal education, and Hollingshead scores were entered as predictors and novelty effects as the dependent variable. Results indicated that over and above variation in vocabulary and parental education, Hollingshead scores predicted infants’ novelty responses, t = 2.80, p=.008. No other predictors were significant. Variation in SES accounted for 17.3% of the variance in novel word learning.

In a second study, 37 infants were tested on their ability to recognize familiar words at 18-20 months. They were presented with familiar word-object pairings in a preferential looking paradigm either in a neutral carrier phrase (e.g. “Can you see the dog?”) or in a leading carrier phrase “Can you drink the juice?”). Infants’ eye movements to the labelled object and to an unnamed distractor were coded. The dependent variable consisted of fixation to the labelled object prior to labelling and after labelling. A significant elevation signifies word recognition and is termed a ‘naming effect’. On average, infants demonstrating naming effects in both neutral and leading phrases (all p-values <.003). Naming effects were submitted to a hierarchical linear regression. English vocabulary size, paternal and maternal education, and Hollingshead scores were entered as predictors. For words in neutral sentences, the regression model was significant Maternal education positively predicted naming effects over and above effects of SES, English vocabulary, and paternal education, accounting for 12.1% of the variance. For the less challenging task of identifying words in leading phrases, the regression model with the same predictors was not significant. Results demonstrate that variation in SES variables can account for significant variation in emergent lexical processes – word recognition and novel word learning – that drive the later growth of vocabulary. Notably, SES effects were more pronounced in more challenging tasks (novel word learning and word recognition in neutral frames).

Authors

• Leher Singh, National University of Singapore
  Presenting Author

• Jean Yeung, National University of Singapore
  Non-Presenting Author