This study examines sentence comprehension in two in-situ heritage Slovene communities in Italy’s border-shift context, comparing the Gorizia/Trieste (N = 66) and Natisone Valley (N = 43) adult populations with monolingual Slovene speakers. While both heritage groups operate in Italian-dominant bilingual environments near the Slovenian border, they differ in Slovene exposure due to varying institutional support. Using a sentence–picture verification task, this study examines comprehension of non-canonical word order, relative clauses and cataphoric dependencies. The Gorizia/Trieste group performed comparably to monolinguals, demonstrating strong sensitivity to case morphology and syntactic processing. Conversely, the Natisone group showed reduced accuracy and longer response times across all domains, reflecting case system vulnerability and processing limitations linked to reduced exposure. These findings highlight the important role of active maintenance and institutional support, independent of geographical proximity to the mainland, in shaping heritage language grammars and their use.

Studies on bilingual individuals indicate that both first-language (L1) and second-language (L2) processing recruit linguistic and simulation systems, with L1 processing typically showing greater activation of simulation systems (often referred to as preference systems) and L2 processing relying more heavily on linguistic systems. When bilingual speakers switch between languages, the language used initially may influence the activation of preference systems in the subsequent language. The present study examines whether L2 proficiency moderates this influence. Employing a 2 × 2 × 2 × 2 mixed design, we investigated how L2 proficiency shapes embodied effects during L1 and L2 sentence comprehension under two language-switching conditions (L1 → L2 and L2 → L1) in bilingual participants. Results revealed that low-proficiency bilinguals demonstrated embodied effects in L1 comprehension across both switching directions. In contrast, highly proficient bilinguals showed such effects only when switching from L1 to L2 but not from L2 to L1. These findings suggest that, in the L2 → L1 condition, the linguistic system activated during L2 processing in highly proficient bilinguals exerts a stronger influence on subsequent L1 processing than in low-proficiency bilinguals, highlighting the role of proficiency in modulating cross-linguistic embodied effects.