How Language Experience Shapes Short-Term Memory: An Embedded Computational Account of the Lexicality Effect in Monolingual and Bilingual Speakers

The lexicality effect, better memory for words than for non-words, is a hallmark of verbal short-term memory and a clear demonstration that prior linguistic experience supports immediate recall. However, how this effect depends on participants’ proficiency with the language itself remains poorly understood. Across two experiments, we examined how language proficiency shapes immediate memory for French high- and low-frequency materials in three groups: Bilingual French L1 speakers, Bilingual English L1 speakers, and English monolinguals. In Experiment 1 (serial recall), the lexicality advantage was largest for Bilingual French L1 participants, reduced or reversed for Bilingual English L1 participants, and inverted for English monolinguals. In Experiment 2 (order reconstruction), which substantially minimizes item-based retrieval demands, a similar albeit muted pattern emerged, indicating that linguistic experience influences memory even when redintegration demands are low. To explain these findings, we extended the Embedded Computational Framework of Memory (eCFM) by varying the representational richness of participants’ lexical spaces. The model reproduced all key empirical patterns, showing that differences in the availability of semantic and orthographic dimensions are sufficient to generate the observed gradient across proficiency groups. These results provide a mechanistic account of how language experience and bilingualism shape the structure of the lexicality effect, linking long-term representational richness to performance in short-term memory tasks.