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

Abderrahmane Issam; Irene E. Winther; Stefan L. Frank

cs.CLOct 22, 2026

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

Authors: Abderrahmane Issam, Irene E. Winther, Stefan L. Frank

Organizations: Maastricht University · Radboud University

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.

Figures & tables

Figure 1: Difference in explained variance

\Delta R_{\text{pseudo}}^{2}

(scaled by

10^{4}

) in four reading measures as a function of SoE. The blue line represents the mean across all L2 reader subsets, while the red line denotes the native English (En-UK) control group. Faint lines indicate individual model evaluations. Positive values indicate a predictive advantage for the bilingual surprisal. With the exception of first fixation duration, bilingual surprisal consistently provides greater predictive power for L2 than L1 readers.

Figure 2: Per-word

\Delta

LL (scaled by

10^{4}

) between matched and mismatched models in four reading measures as a function of SoE. Solid lines and shaded bands represent subset means with 95% confidence intervals for the bilingual models (blue) and monolingual baseline (red), with faint blue lines indicating individual subset evaluations. Higher values indicate a stronger predictive advantage for the matched model relative the mismatched one. With the exception of first fixation duration, bilingual models match or outperform the monolingual baseline.

Metric	Mean $\Delta\text{LL}$		Sig. Ratio ( $p<0.05$ )
	Match	Mism.	Match	Mism.
First Fixation	1.07	1.33	17.0%	23.2%
Gaze Duration	3.23	1.78	55.0%	30.4%
Go-Past	3.32	1.69	53.7%	30.8%
Total Fixation	6.53	4.00	77.3%	64.3%

Table 1: Nested model comparisons (

m_{\text{both}}

) evaluating the unique contribution of matched versus mismatched bilingual surprisal averaged across all SoE models and language subsets. With the exception of First Fixation, matched models’ mean

\Delta LL

and significance ratio are higher than the mismatched models.

Figure 3: Per-word

\Delta

LL (scaled by

10^{4}

) between matched and mismatched models in four reading measures as a function of SoE, grouped by Latin (top) and Non-Latin (bottom) L1 scripts. Solid lines and shaded bands denote subset means with 95% confidence intervals for the bilingual models (blue) and monolingual baseline (red). Faint blue lines indicate individual subset traces. Early measures (First Fixation and Gaze Duration) reveal a clear divergence in predictive behavior between the two script groups.

Figure 4: Bilingual surprisal advantage

\Delta R^{2}

(scaled by

10^{4}

) across readers’ English proficiency (LexTALE %) in four reading measures. Curves show quadratic fits per SoE model; circular markers denote binned means (

\pm 95\%

CI) at SoE=48k. Faint dots indicate individual observations, and the horizontal dotted line denotes zero gain. Inset labels indicate the significance of the quadratic proficiency term in a WLS with L1-subset fixed effects and standard errors clustered by L1 subset (

{}^{*}p<0.05

{}^{**}p<0.01

{}^{***}p<0.001

{}^{\text{ns}}\text{not significant}

). While early measures show negligible non-linearity, late measures (Go-Past, Total Fixation) show an inverted-U trajectory peaking among intermediate-proficiency readers (

65\text{--}80\%

Appendix figures & tables6 assets

Supplementary material from the paper’s appendix.

Appendix

Subset	Language & Site	Samples	Readers
hi_iiith	Hindi (Hyderabad)*	77,522	53
en_uk	English (UK)	66,041	50
bp	Brazilian Portuguese	62,608	54
ge_zu	German (Zurich)	58,949	47
hi_iitk	Hindi (Kanpur)*	58,642	45
ru_mo	Russian (Moscow)*	57,817	49

Appendix

Table 2: MECO L2 subsets, corresponding L1 languages and recording site (* denotes Non-Latin), total number of samples (before filtering) and number of readers.

Hyperparameter	Value
Layers	12
Embedding size	768
Hidden size	768
Intermediate hidden size	3072
Attention heads	12
Attention head size	64

Appendix

Table 3: Architectural and optimization hyperparameters for the GPT-2 bilingual models.

Figure 5: Pairwise Pearson correlation of English surprisal values between all L1 bilingual models across the SoE stages. The dark blue line represents the mean correlation, the shaded band shows 95% confidence interval, and the light blue lines show individual pairwise comparisons.

Figure 6: Explanatory power (

\Delta R^{2}

over the baseline, scaled by

10^{4}

) of monolingual and of bilingual surprisal across readers’ English proficiency (LexTALE %) in four reading measures. Faint dots indicate individual observations, and the horizontal dotted line denotes zero gain. Inset labels indicate the significance of the quadratic proficiency term in a WLS with L1-subset fixed effects and standard errors clustered by L1 subset (

{}^{*}p<0.05

{}^{**}p<0.01

{}^{***}p<0.001

{}^{\text{ns}}\text{not significant}

). Both bilingual and monolingual models exhibit a pronounced inverted-U relationship with proficiency.

Metric	SoE	$N$	$\beta_{\text{quad}}$	$p_{\text{quad}}$	$\Delta\text{AIC}$
First Fixation	0	480	+0.002	0.464	$-$ 1.5
	16k	480	+0.002	0.494	$-$ 1.7
	32k	480	$-$ 0.001	0.793	$-$ 2.0
	48k	480	+0.003	0.436	$-$ 1.5
Gaze Duration	0	480	$-$ 0.007	0.129	$-$ 0.2
	16k	480	$-$ 0.007	0.093	+0.0

Appendix

Table 4: Model comparison evaluating non-linear proficiency moderation across eye-tracking measures.

\beta_{\text{quad}}

denotes the estimated coefficient of the quadratic proficiency term, and

p_{\text{quad}}

denotes its robust HC3 significance.

\Delta\text{AIC}=\text{AIC}_{\text{linear}}-\text{AIC}_{\text{quadratic}}

, where values

>+2.0

indicate that the quadratic model improves on the linear baseline by more than the penalty for its extra parameter ( Burnham and Anderson, 2002 ) . Boldface marks exposure stages where both criteria favor the quadratic model (

p_{\text{quad}}<0.05

\Delta\text{AIC}>2.0

Figure 7: Bilingual surprisal advantage (

\Delta R^{2}=R^{2}_{\text{bilingual}}-R^{2}_{\text{monolingual}}

, scaled by

10^{4}

) modeled against composite L2 proficiency (PC1 extracted across LexTALE, Vocabulary, Spelling, TOWRE word, and TOWRE nonword; with 52.5% variance explained) in four reading measures. Curves depict second-order polynomial regressions across SoEs; markers denote empirical binned means (

\pm 95\%

CI) at SoE=48k. Inset labels indicate the significance of the quadratic proficiency term in a WLS with L1-subset fixed effects and standard errors clustered by L1 subset (

{}^{*}p<0.05

{}^{**}p<0.01

{}^{***}p<0.001

{}^{\text{ns}}\text{not significant}

). Faint points represent individual participants, and the dotted horizontal line marks zero gain.

Cross-Lingual Transfer

Cross-lingual transfer enables models to apply knowledge or capabilities learned in one language to tasks in another.

See all 256 papers

Is Cross-Lingual Transfer in Bilingual Models Human-Like? A Study with Overlapping Word Forms in Dutch and English
Apr 8, 2026Iza Škrjanec, Irene Elisabeth Winther, Vera Demberg +1Cross-Lingual Transfer Multilingual Language Models
Saarland University, Germany · Zuse School ELIZA, Germany · Radboud University, the Netherlands
Beetle: A Bilingual Model Suite for Modelling Second-Language Processing
Sep 18, 2026Suchir Salhan, Catherine Arnett, James Michaelov +1Multilingual Language Models Language Model Pretraining
University of Cambridge · EleutherAI · MIT
Probing for Reading Times
Apr 20, 2026Eleftheria Tsipidi, Samuel Kiegeland, Francesco Ignazio Re +4Computational Psycholinguistics Language Model Probing
ETH Zürich · Toyota Technological Institute at Chicago · University College London

Apr 8, 2026J/K move · Enter open · S save

Is Cross-Lingual Transfer in Bilingual Models Human-Like? A Study with Overlapping Word Forms in Dutch and English

Iza Škrjanec, Irene Elisabeth Winther, Vera Demberg, Stefan L. Frank

Saarland University, Germany · Zuse School ELIZA, Germany · Radboud University, the Netherlands

Cross-Lingual Transfer Multilingual Language Models

Open paper AskPDF arXiv

Bilingual speakers show cross-lingual activation during reading, especially for words with shared surface form. Cognates (friends) typically lead to facilitation, whereas interlingual homographs (false friends) cause interference or no effect. We examine whether cross-lingual activation in bilingual language models mirrors these patterns. We train Dutch-English causal Transformers under four vocabulary-sharing conditions that manipulate whether (false) friends receive shared or language-specific embeddings. Using psycholinguistic stimuli from bilingual reading studies, we evaluate the models through surprisal and embedding similarity analyses. The models largely maintain language separation, and cross-lingual effects arise primarily when embeddings are shared. In these cases, both friends and false friends show facilitation relative to controls. Regression analyses reveal that these effects are mainly driven by frequency rather than consistency in form-meaning mapping. Only when just friends share embeddings are the qualitative patterns of bilinguals reproduced. Overall, bilingual language models capture some cross-linguistic activation effects. However, their alignment with human processing seems to critically depend on how lexical overlap is encoded, possibly limiting their explanatory adequacy as models of bilingual reading.

Abstract

Cross-Lingual Transfer

Explore similar work

Is Cross-Lingual Transfer in Bilingual Models Human-Like? A Study with Overlapping Word Forms in Dutch and English

Beetle: A Bilingual Model Suite for Modelling Second-Language Processing

Probing for Reading Times