Measuring L2 Speech Fluency Based on Syntactic Distribution of Pauses
Résumé
Pauses play an important role in structuring speech and making it comprehensible. However, when they occur at unexpected positions-such as within syntactic phrases-they can hinder understanding and become what we often refer to as disfluencies. This paper presents a novel metric for measuring second language (L2) speech fluency, based on the extent to which speakers place pauses at unexpected syntactic locations. We compare different approaches: one categorizing pauses by constituent types (clauses and phrases), and another using a continuous measure of syntactic distance between adjacent words. These scores are applied to conversation-based spontaneous speech produced by Japanese learners of English (CEFR levels B1 to C1) and by native English speakers. Results indicate that pauses tend to shift toward higher syntactic boundaries as proficiency increases. All tested measures positively correlate with proficiency, but syntactic distance yields a substantially stronger correlation.
| Origine | Fichiers éditeurs autorisés sur une archive ouverte |
|---|---|
| Licence |