This study investigated the longitudinal improvement in fluency of three Japanese participants who studied at a graduate school in the UK. In addition to quantitative analysis of videotaped conversations focusing on core components of fluency, such as speech rates, dysfluency markers and pauses, qualitative analysis, including protocol analysis and journal analysis, were also employed to discuss why they did not improve in fluency as originally expected. The analyses revealed that these participants did not have as many chances to communicate with native speakers as originally expected, and that a sociocultural factor of being excessively afraid of making mistakes might be a reason that prevented their natural improvement.
Index Terms. fluency, speech rates, dysfluency markers, pauses, sociocultural elements