Computational models have explanatory power to understand cognition in individuals, but there is usually not enough training data to build such models with traditional machine learning. However, alternative architectures that model neural and cognitive processes in more detail can be more easily fit to individuals and can lead to mechanistic insights into human behavior. BiLex is such a computational model of the bilingual lexicon, containing a shared semantic system and phonetic language systems represented through self-organizing maps. In this study, BiLex was extended further into biological realism with excitatory and inhibitory lateral connections. This extension makes it possible to account for response times as well as common errors such as those based on semantic concept familiarity, typicality, and specificity. The Boston Naming Test was simulated in Spanish and English, and model performance approximated naming accuracy and error types in Spanish-English bilinguals with varying primary and secondary language histories. The results indicated that the approach captures semantic-level processes in bilingual lexical access, thus modeling the semantic cognition of individual subjects. In the future, such models may serve as a foundation for modeling impairments and treatment options.
Figures & tables
Figure 1: Laterally-connected BiLex model. The model simulated naming by receiving input to the semantic map, propagating activation through lateral connections and associative connections across all maps, and producing a response from the target language map. This architecture is motivated by the lexical-semantic structures in the brain, and can therefore serve as a foundation for modeling individualized performance, its breakdown in e.g. stroke and dementia, and identifying effective treatments ( Kiran et al., 2026 ) . Figure adapted from ( Peñaloza et al., 2019 )
Figure 2: Distributions of responses in the BNT naming task in humans vs. the model. One-sample Wilcoxon tests were conducted on the L2-English early-AoA group ( 2(e) , 2(f) ), with ”*” indicating statistically significant and ”ns” non-significant differences. The model approximated human accuracy and error distributions well qualitatively, and also quantitatively when sufficient data were available.
Figure 3: Regression analysis comparing BNT word frequency to naming accuracy rates in humans vs. the model. Frequency slopes indicate L1 vs. L2 naming accuracy gaps. Positive word frequency vs. accuracy trends were present in humans and the model (significant in only L2-English-early) across bilingual groups, suggesting the model simulated human behavior of greater accuracy when naming higher-frequency words.
Figure 4: Adjusting settling to fit individual performance. As tmax increased in the L2-Spanish-early simulation, naming accuracy increased in both English and Spanish from 75 to 400 steps. At tmax=600 to tmax=1000 , accuracy plateaued. Thus, tmax can be used to fit model performance to individual subjects.
Figure 5: Distribution of naming error types in the L2-Spanish-early model. Decreasing the model’s settling time ( tmax ) increased the percentage of no response errors. Increasing tmax enabled the model to produce a wider variety of human-like errors. For all groups, tmax=400 was chosen to balance error matching and overall naming accuracy.
Aphasia following stroke commonly produces systematic naming errors with characteristic profiles, but whether general-purpose language models not designed for clinical simulation can reproduce these patterns remains untested. We investigated (1) whether lesions or controlled perturbations to a multimodal language model can reproduce different types of errors in picture naming, and (2) whether the framework can reproduce the complete error profile of individual persons with aphasia (PWAs). Using LLaVA 1.6, we evaluated perturbation configurations that varied the layer, proportion, and amount of noise applied to model units. We examined 278 PWAs on the Philadelphia Naming Test, classifying responses into seven categories using a validated neural classifier. Six of seven response categories (correct, semantic, mixed, unrelated, neologism, no response errors) emerged at clinically-comparable proportions across distinct parameter space regions, with formal paraphasia being the exception. Searching the perturbation space revealed configurations that reproduced the individual error profile in at least six of seven categories for 97.8% of PWAs and in all seven categories for 79.5% of PWAs. Monte Carlo baselines confirmed that this matching reflects joint inter-category structure rather than marginal overlap. These results establish a quantitative framework for reproducing individual aphasic error patterns in picture naming. They suggest the potential for language models to serve as digital twins of individuals with post-stroke aphasia.
Yong Yang, Xiang Guan, Sophie Arheix-Parras +7
Department of Computer Science and Engineering, University of South Carolina · Department of Psychology, University of South Carolina · Department of Communication Sciences and Disorders, University of South Carolina +2
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.
Abderrahmane Issam, Irene E. Winther, Stefan L. Frank
Multilingualism is incredibly common around the world, leading to many important theoretical and practical questions, including whether multilingual acquisition leads to delays in learning and whether there are better and worse ways to structure multilingual input. Many correlational studies address these questions, but it is surprisingly difficult to get definitive answers because children cannot be randomly assigned to be multilingual and data are typically not matched between languages. We create matched mono- and bilingual datasets using synthetic data and machine translation to simulate a variety of controlled exposure conditions, pairing English with Spanish, Chinese, and Dutch input. We train GPT-2 models on these datasets and evaluate their performance on perplexity, grammaticality, and semantic knowledge. Across model scales, measures, and languages, bilingual models perform similarly to monolingual models in the first language, but show strong performance in the second as well. Control experiments show that our findings hold on real-world language data and even on shuffled second-language input. These results suggest that there are no strong differences between different bilingual exposure regimes, and bilingual input poses no in-principle challenges for agnostic statistical learners.