Learning words in two languages: Effects of distributed exposure and competition

• In Event: Bilingualism - early childhood

Abstract

Bilingual children often have smaller single-language vocabularies than their monolingual peers, but the mechanism underlying this difference is poorly understood (Thordardottir, 2011). According to the weaker-links hypothesis (Gollan & Acenas, 2004) , dual-language input reduces language-specific exposure and leads to weaker links between wordforms and semantic representations in bilinguals. An alternative hypothesis roots the effects of dual-language input on vocabulary in competition between languages (Benitez et al., 2016). Crucially, there is little empirical work that has tested either of these hypotheses. Across two experiments, we investigated the effect of dual-language input on children’s word learning. The crucial questions were whether reduced exposure or competition in the dual-language condition would lead to reduced learning, and whether an addition of a speaker cue to language membership would mitigate these effects.
Twenty-six Spanish-English bilingual children completed Experiment 1, and 17 different children completed Experiment 2 (in progress; see Table 1). In both experiments, children learned one set of novel noun-object pairs in a single-language condition and another set in a dual-language condition. In the single-language condition, children learned English-like words in 6 blocks. In the dual-language condition , children learned Spanish-like words in the first 3 blocks and English-like words for the same objects in the last 3 blocks . In Experiment 1, all novel words were recorded by the same English-Spanish bilingual speaker. In Experiment 2, English novel words were recorded by a native English speaker and Spanish novel words were recorded by a native Spanish speaker, testing the impact of the one-person, one-language method of bilingual exposure. Children’s retention was tested via a 2-alternative forced choice recognition task after each block. Condition order was counter-balanced for both experiments.
Two models were constructed to analyze recognition data from each experiment. The first model (A), testing reduced exposure, contrasted performance in 6 blocks of the English-only condition to 3 English blocks in the dual-language condition. If reduced exposure impacts word learning, we expected better performance in the English-only condition. The second model (B), testing competition, contrasted performance in the first 3 blocks of the English-only condition to the 3 English blocks in the dual-language condition, equating the number of English exposures across condition. If competition impacts word learning, we expected better performance in the English-only condition even after matching the number of exposures. For Experiment 1, both Model A (B=0.54, SE=0.12, p<.001) and Model B (B=0.50, SE=0.22, p=.02) indicated a significant effect of condition. Participants performed better in the single-language than the dual-language condition. For Experiment 2, only Model A indicated a significant effect of condition (B=0.67, SE=0.25, p=.006). Across Experiments 1 and 2, children learned above chance in both conditions.
These findings suggest that both reduced exposure and competition between language-specific wordforms can yield reductions in bilingual word-learning performance. The addition of speaker cue in Experiment 2 mitigated the effect of competition, providing preliminary support for the one-person, one-language strategy. Broadly, findings suggest while distribution of exposures reduces learning compared to single-language input, bilingual children can and do learn successfully from dual-language input.

Authors

• Emma J. Libersky, University of Wisconsin - Madison
  Presenting Author

• Caitlyn Slawny, University of Wisconsin - Madison

• Margarita Kaushanskaya, University of Wisconsin-Madison