Semantic native-likeness in highly proficient second-language users

• Boris Kogan ^[1] ; Lucía Agulla ^[1] ; Martín Dottori ^[1] ; Lucía Amoruso ^[1] ; Leticia Vivas ^[2] ; Adolfo M. García ^[1]
  1. [1] Universidad de San Andrés

     Universidad de San Andrés

     Argentina

     
  2. [2] National Scientific and Technical Research Council (CONICET), Buenos Aires, Argentina
• Localización: Revista Argentina de Ciencias del Comportamiento ( RACC ), ISSN-e 1852-4206, Vol. 16, Nº. Extra 3 (Suplemento (Septiembre)), 2024 (Ejemplar dedicado a: Actas de resúmenes de la XIX Reunión Nacional y VIII Encuentro Internacional de la Asociación Argentina de Ciencias del Comportamiento), págs. 151-152
• Idioma: inglés
• Enlaces
  • Texto completo (pdf)
• Resumen

  • Introduction: Native-likeness (similarity between non-native and native users of the same language) has been shown to depend on second language proficiency (L2p).However, relevant evidence comes mainly from phonological or syntactic phenomena, prompting an underexplored question: can higher L2p also entail a more native-like organization of semantic memory? Aims: Analyze spontaneous conceptual associations in bilinguals varied across L2p levels and examine whether potential L2p effects on semantic native-likeness are robust at both the group and the individual level. Methods: We asked high and low L2p bilinguals to describe concepts in L2 and L1 via a classical feature-listing task. To capture semantic native-likeness, we calculated the number of shared features between the L2 norms’ noun-concepts (representing the prototypical semantic structure of the language) and the participants’ L2 responses (reflecting their individual semantic structure). This was done through the validated common element correlation (CEC) metric. In addition to L2 semantic native-likeness (L2-L2 CEC), we calculated the same metric for three control conditions, based on the overlap between L1responses and L2 norms (L1-L2 CEC), L2 responses and L1 norms (L2-L1 CEC), and L1 responses and L1 norms (L1-L1 CEC). Finally, we examined whether semantic native-likeness scores discriminated between high and low L2p levels at the group level (via inferential statistics) and at the individual level (via machine learning analyses). Results: L2-L2 CEC was significantly higher in L2p bilinguals, revealing more native-like semantic patterns. Moreover, such an index yielded robust subject-level classification in machine learning analysis. None of the control indices differed between groups. Discussion: Our study indicates that native-likeness may be observed at the semantic level and that this phenomenon is sensitive to L2p. Extensions of this approach could further illuminate the interplay between bilingual semantic memory and language-experience variables.