Phoneme input reveals more bilingual learning differences than text
Cross-Linguistic Effects in Bilingual Phoneme BabyLMs
Computation and Language
Summary
People who learn two languages from birth often mix influences from both languages. The authors trained small bilingual language models using speech sounds (phonemes) instead of written words, to make the learning more like how babies hear language. They kept English as one language and changed the other language among four different ones with varying similarity to English. They found that using speech sounds showed bigger differences in how the models learned grammar related to the first language, while how they learned words early on matched how similar the sounds were between the two languages.
What this means in practice
- •For speech technology developers: Improve bilingual speech recognition by training models on phonemic input reflecting different language pairs.
- •For natural language processing engineers: Design bilingual language models that better capture grammatical learning trajectories influenced by the first language using phoneme-based training.
Authors
Nikitas Theodoropoulos, Maria Lymperaiou, Giorgos Filandrianos
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.