Temporal Metrics and Perceived Fluency in L2 Speech: A Production-Based Analysis

Perceived fluency in second-language (L2) speech is often assumed to reflect stable speaking ability. However, fluency judgments are strongly influenced by temporal features such as speech rate, pausing patterns, and articulation timing. This study examines how measurable temporal metrics relate to perceived fluency outcomes. We analyze recorded L2 speech samples using automated transcription and acoustic processing tools. Extracted features include words per minute, articulation rate, pause frequency, and pause duration. These metrics are compared against fluency evaluation scores to assess which temporal features most strongly predict perceived fluency. Results show that speech rate measures are strong predictors of fluency judgments, while other temporal features contribute more modestly. The findings suggest that perceived fluency is closely tied to observable timing patterns rather than broader linguistic complexity. This work contributes a computational perspective to L2 fluency research by combining automated feature extraction with quantitative modeling. All analyses are based on reproducible scripts and aggregated metrics.