Cross-Linguistic Effects in Bilingual Phoneme BabyLMs

2026-09-29

AI Summary

This study addresses the limitation that existing language models predominantly rely on textual input, failing to simulate authentic child speech acquisition, while cross-linguistic effects in bilingual acquisition remain unclear. To bridge this gap, this work introduces phonemic representations into a bilingual BabyLM framework for the first time, constructing strictly data-constrained English-German and English-Swedish models based on phonemic input. Through corpus phonemization preprocessing and multilingual comparative analysis, it investigates how different first languages (L1) shape grammatical and lexical learning trajectories. The findings reveal stronger L1 grammatical transfer effects under phonemic input conditions and demonstrate that early lexical acquisition disparities align closely with phonemic inventory similarity. These contributions offer novel insights into children’s bilingual cognitive mechanisms.

Abstract

Cross-linguistic effects are a central topic in bilingual first-language acquisition. Artificial learners can help investigate L1-L2 interactions by enabling controlled comparisons across language combinations and learning conditions. Recent work explores this direction by training bilingual language models under developmentally plausible constraints. However, human and model learners still diverge in fundamental ways, with one major difference being input modality: children learn primarily from spoken input, whereas language models are typically trained on orthographic text. To reduce this gap, researchers have trained models on phonemic representations of speech. In this work, we combine these research directions to train bilingual BabyLMs with phonemic input. We keep English fixed as the L2 and vary the L1 across German, Swedish, Persian, and Basque, selected to represent contrasting combinations of syntactic and phoneme-inventory distance from English. Our results show stronger L1-related variation in grammatical learning trajectories under phonemic than orthographic input, while early lexical differences align with phoneme-inventory similarity.

Problem

• Cross-linguistic effects
• Bilingual first-language acquisition
• Phonemic input
• BabyLMs
• Input modality

Innovation

• Bilingual BabyLMs
• Phonemic Input
• Cross-Linguistic Effects
• First Language Acquisition
• Language Models