Graduation Year


Document Type




Degree Granting Department

Communication Sciences and Disorders

Major Professor

Ruth H. Bahr


Bilingual word learning, MouseTracker, Nonword processing, Phonotactics, Wordlikeness


Phonotactic processing is foundational to the word processing task in both monolingual and bilingual children (Li & Farkas, 2002; Pierrehumbert, 2001; Shook & Marian, 2013; Storkel & Morrisette, 2002). While the use of phonotactic information in word processing in monolingual children is relatively well documented, it is less well understood in bilingual children. The purpose of this study was to investigate how bilingual kindergartners process the phonotactic probabilities of their two languages. A set of nonwords was developed that manipulated the strength of phonotactic probability across both Spanish and English while also controlling the language environment of the experimental task (i.e., whether children were tested in Spanish or English). Hence, this study allowed for a unique investigation into how bilingual children process two languages and their associated phonotactic probabilities. Specifically, this study provided answers to: whether or not bilingual children benefitted from a high probability processing advantage, if the phoneme systems of two languages were stored as one unit or separate units, and if there was an effect of language environment (i.e., an assimilation effect, Burki-Cohen et al., 1989).

By varying the phonotactic probabilities of nonwords and the language environment), the answers to several research questions were sought. First, language exclusive nonwords (nonwords that had phonotactic probabilities unique to English or Spanish) were used to investigate the presence of a high phonotactic probability processing advantage in bilingual children. Second, high/low nonwords (nonwords with a high phonotactic probability in one language and a low phonotactic probability in the other language) were compared with the language exclusive nonwords to determine if the phonotactic systems of a bilingual child's two languages are stored together such that they interact during word processing. Finally, ambiguous nonwords (those with equal phonotactic probability in both languages) were used to investigate the influence of language environment on phonotactic processing. The nonwords were created by manipulating phonotactic probabilities in each language, recorded by two bilingual speakers, standardized for fundamental frequency and synthesized to become phonetically and acoustically ambiguous. Wordlikeness judgments in each language were obtained from monolingual English and bilingual Spanish-English adults. These results determined that adults were processing the varying phonotactic probabilities of the nonwords as designed and the words were appropriate stimuli for use in a word sorting task with bilingual children.

In an attempt to replicate aspects of a natural language environment, the current study first divided children into two bilingual testing groups: one where mostly English was spoken and another where mostly Spanish was spoken. Children watched cartoons illustrating the need for sorting nonwords into two languages before completing the word sorting tasks. The experiment was presented using MouseTracker (Freeman, 2011), which recorded the participant's response and mouse cursor movement (as a measure of decision complexity) as the child selected either Spanish or English.

Mixed level modeling results indicated significant differences in language choice but not decision complexity across the nonword types. First, bilingual children sorted language exclusive nonwords by focusing on whether the word was more probable in English or Spanish than whether the nonword had high or low probability within a language. Hence, these participants did not appear to benefit from a high phonotactic processing advantage. When children were sorting the high/low nonwords, they tended to ignore the fact that the nonwords had phonotactic probability in both languages, and treated them as belonging to the language in which they had the highest phonotactic probability. This finding would suggest that bilingual children do not appear to store the phonotactic systems of two languages together. Finally, results showed no effect of language environment when children were sorting the ambiguous nonwords. Overall, it appears that bilingual children focus on the overall phonotactic probability of a nonword (i.e., whether it is more probable in Spanish or English) during processing, while ignoring any dual phonotactic probabilities from two languages. These results are incorporated within a proposed model of bilingual word processing and a brief discussion of how these findings can be expanded to explain bilingual word learning is provided.