This study examines whether Korean second language (L2) learners of English can accurately reproduce native-like lexical stress patterns through imitation. Lexical stress in English is primarily marked by duration, intensity, and fundamental frequency (F0). Twenty-seven Korean L2 speakers participated in a three-stage experiment consisting of Baseline Speech (BS), Shadowed Speech (SS), and Memory-based Speech (MS). First, participants read scripted sentences (BS), immediately repeated native speaker speech models (SS), and reproduced the same sentences 24 h later without re-exposure (MS). Stimuli included 40 disyllabic words with either stressed-unstressed (SU) or unstressed-stressed (US) patterns. Target words were embedded in a carrier phrase to minimize coarticulatory effects and maintain consistent prosodic environments across tokens. Acoustic analysis revealed that L2 speakers’ productions became more native-like across stages. Duration emerged as the most robust and reliably retained cue over time while intensity showed the strongest convergence during immediate imitation. F0 patterns also exhibited improvements though the magnitude of change was comparatively smaller. Additionally, a small gender difference was observed, with male participants demonstrating slightly more accurate imitation. These findings suggest that Korean L2 learners are capable of perceptually guided phonetic imitation of English lexical stress with notable retention effects under memory-based conditions.
Bae et al. (Wed,) studied this question.