Integrated, Not Isolated: Defining Typological Proximity in an Integrated Multilingual Architecture

Area of Science:

• Cognitive Science
• Computational Linguistics
• Psycholinguistics

Background:

• Bilingualism research often uses a monolingual bias, comparing bilingual outputs to monolingual grammars.
• Previous studies focused on language switching rather than the integrated nature of bilingual systems.

Purpose of the Study:

• To explore typological proximity in bilingualism by examining integrated language systems.
• To propose a novel cognitive architecture for bilingual grammars.

Main Methods:

• Conceptualizing bilingual grammars as integrated source grammars mitigated by a common, combined grammar.
• Implementing this combined grammar in a parallel, distributed, and gradient architecture using a shared vector-space model.
• Employing compression through routinization and dimensionality reduction within the model.

Main Results:

• Bilingual grammars are proposed to be integrated, not isolated or simply conjoined.
• A shared vector-space model with compression mechanisms is suggested for representing bilingual cognition.
• The architecture aims for consistency with empirical findings in bilingual cognition and memory.

Conclusions:

• Bilingual minds may operate with a single, integrated grammar system rather than separate ones.
• This integrated architecture offers a new framework for understanding bilingual language processing and representation.
• The model provides a computational cognitive architecture consistent with observed bilingual behaviors.