INTERSPEECH 2015
16th Annual Conference of the International Speech Communication Association

Dresden, Germany
September 6-10, 2015

Phonetic-Phonological Feature Emerges by Associating Phonetic with Semantic Information — A GSOM-Based Modeling Study

Mengxue Cao (1), Aijun Li (1), Qiang Fang (1), Bernd J. Kröger (2)

(1) Chinese Academy of Social Sciences, China
(2) RWTH Aachen University, Germany

Born with an innate neural architecture built specially for language learning, young children have the ability to distinguish sounds in a variety of languages. As they are exposed to native language environment, perceptual reorganization occurs, and native language system gradually establishes. Phonology knowledge, which is language-specific, emerges during this process. In this study, based on the further developed Interconnected Growing Self-Organizing Maps (I-GSOM) model, we present a series of computational modeling experiments which simulate the early language acquisition process for young children. A universal model that has the ability to distinguish phonemes in different languages (two in our case) is built based on English and Standard Chinese data. The native language learning is simulated based on Standard Chinese data with both phonetic and semantic inputs for children between 12 and 18 months old. Experiment results show that the conceptual based top-down process contributes to the reorganization of phonetic knowledge, and phonetic-semantic association helps the emergence of phonetic-phonological knowledge. It can be hypothesized from these findings that the phonetic-phonological interface does not appear as a clean cut within the speech processing system but as a broader zone located between sensorimotor and semantic processing.

Full Paper

Bibliographic reference.  Cao, Mengxue / Li, Aijun / Fang, Qiang / Kröger, Bernd J. (2015): "Phonetic-phonological feature emerges by associating phonetic with semantic information — a GSOM-based modeling study", In INTERSPEECH-2015, 1571-1575.