RESOLVING COMPETITION IN BILINGUAL STATISTICAL LEARNING PARADIGMS
Open Access
- Author:
- Mitchel, Aaron Daniell
- Graduate Program:
- Psychology
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- August 21, 2007
- Committee Members:
- Daniel J Weiss, Thesis Advisor/Co-Advisor
Daniel J Weiss, Thesis Advisor/Co-Advisor - Keywords:
- language acquisition
speech segmentation
statistical learning
face processing - Abstract:
- One of the fundamental issues in bilingual language research has been determining how multiple language representations develop over the course of acquisition. In an effort to explore this, previous research adapted a traditionally monolingual statistical learning word segmentation task to provide input from two artificial languages, thereby simulating the early stages of bilingual acquisition. The goal of that study was to determine whether learners could form multiple representations for each input language and encapsulate information in each of the input speech streams, an ability that would facilitate learning in cases where statistical interactions between the languages might interfere with proper learning. The results demonstrated that adult learners can track two sets of statistics at once (i.e., encapsulating the information within each language), suggesting that they can form multiple representations when confronted with multiple language input. This process was facilitated by an indexical cue of speaker voice. In the absence of such an indexical cue, learners combined input across both languages, resulting in reduced learning when presented with statistically incompatible language pairs. The present study examines whether the process of forming multiple representations in bilingual segmentation tasks can be facilitated by visual cues such as videos of faces speaking the artificial languages. Previous research has demonstrated that synchronous visual displays (such as faces) can facilitate infants' performance on a segmentation task. Here, two artificial languages with incompatible statistics were paired with videos of two synchronous, dynamic faces. With the faces serving as an indexical cue to language, participants learned each language significantly above chance, replicating previous results. Further experimental conditions demonstrate that this effect hinges on indexical synchronous information being available to the learner. Neither a static visual cue of color background nor a dynamic display of faces that produce both languages (and therefore not indexical) facilitated segmentation. These results suggest that faces are particularly effective indexical cues for facilitating the encapsulation of statistical information in a bilingual word segmentation task, underscoring the potentially important role for audio-visual synchrony in the course of normal language acquisition.