Distinct production and neural activity of inflections in beginning L2 learners, advanced L2 learners and native speakers

The extent to which adult second language (L2) learners can achieve native-like neural processing of complex inflectional morphology remains unresolved. This study used event-related fMRI to examine neural mechanisms underlying the potential gradual achievement of L1-like proficiency in adult learners of Finnish (L2) with Russian as their first language (L1). Three groups participated: late beginner learners (N = 15, mean exposure to Finnish 24.9 months), late advanced learners (N = 15, mean exposure to Finnish 131.8 months), and native Finnish speakers (N = 15). Participants completed overt and covert word generation tasks. In the overt task, they spoke an inflected word aloud; in the covert task, they silently generated a form and judged a visually presented form as correct or incorrect. Stimuli included high-, medium-frequency, and pseudoword stems, with or without consonant gradation (stem allomorphy). Beginners had significantly higher error rates in both tasks than advanced learners and native speakers, indicating challenges in morphological encoding and recognition. Advanced learners performed comparably to native speakers. Furthermore, only advanced learners and native speakers exhibited increased BOLD responses in the LIFG and MFG for stems requiring consonant gradation, suggesting a mechanism for the lexical lookup of stems. These findings indicate that proficient adult L2 learners can develop native-like neural patterns of morphological processing, particularly in handling complex inflectional forms. This research contributes to understanding the neural mechanisms underlying morphological development in adult L2 acquisition and underscores the critical role of proficiency in achieving native-like processing of morphological complexity. • Advanced L2 learners show native-like brain activation during word inflection. • Beginners rely on word frequency, showing higher BOLD for high freq. and pseudo words. • CG stems cause more errors and stronger BOLD than NoCG stems across all groups. • Only natives and advanced learners show ROI activation differences for CG stems.