Acoustic analysis and perception ratings of first and second language speakers’ Italian lexical stress

This study examines the acquisition of Italian lexical stress by adult L2 learners. L1 Italian speakers and beginner L2 Italian speakers were recorded reading aloud trisyllabic Italian words, e.g. COdice with antepenultimate syllable stress (“code”), moMENto with penultimate syllable stress (“moment”). We analyzed four acoustic-phonetic cues: duration, fundamental frequency (pitch is the perceptual correlate), amplitude, and spectral tilt (a measure of energy change over frequencies). We corroborated previous findings: L1 speakers used all four cues to differentiate between antepenultimate (strong-weak-weak) and penultimate (weak-strong-weak) stressed words. We found evidence of L2 speakers producing inconsistent patterns for all four cues. We then played these L1 and L2 recordings for L1 Italian speakers (N = 50) and asked them to rate the utterances using a visual analog scale (VAS). As expected, the L1 speech was rated higher (more fluent stress) than the L2 speech (less fluent stress). We modeled how the acoustic cues predicted VAS responses. Our findings highlight the roles of duration and pitch for L2 learners. We conclude with implications for learners and teachers of Italian.