Newest papers

Bilingual Language Models Capture Cross-Linguistic Influence in Second-Language Reading

Abderrahmane Issam, Irene Elisabeth Winther, Stefan L. Frank · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.31620 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

Mientras que el modelo de lenguaje monolingüe (LM) es ampliamente establecido como un predictor de los tiempos de lectura, no puede capturar la dinámica de lectura no nativa (L2) que es modelada por el primer lenguaje del lector (L1). En este trabajo, investigamos si la transferencia cross-lingual dentro de los modelos bilingües se alinea con los prejuicios L1 impulsados por humanos durante la lectura L2. No evaluamos los modelos bilingües en los datos de rastreo de ojos de los lectores de 12 diferentes antecedentes L1 que leen textos ingleses. Para aislar la transferencia específica L1, comparamos los modelos bilingües entrenados en un L1 que coincide con el fondo L1 de los lectores tanto en una línea de base inglesa monolingüe como en modelos bilingües entrenados en otros L

Machine translation (Meta M2M100); the English original follows.

Abstract

While monolingual language model (LM) surprisal is widely established as a predictor of reading times, it cannot capture the non-native (L2) reading dynamics that are shaped by the reader's first language (L1). In this work, we investigate whether the cross-lingual transfer within bilingual models aligns with human L1-driven biases during L2 reading. We evaluate bilingual models on eye-tracking data of readers from 12 diverse L1 backgrounds reading English texts. To isolate L1-specific transfer, we compare bilingual models trained on an L1 that matches the readers' L1 background to both a monolingual English baseline and bilingual models trained on other L1s. We demonstrate that matching the readers' L1 improves the prediction of L2 reading times. This predictive advantage is concentrated in late-stage eye-tracking measures, modulated by L1-L2 script overlap, and peaks among intermediate-proficiency readers. These findings provide evidence that bilingual LMs capture aspects of cross-linguistic influence, and that such influence is present in L2 reading behavior.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Reading (process), Computer science, Artificial intelligence, Natural language processing, Linguistics, Baseline (sea)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange