This project aimed to get a better understanding of different aspects of speaker-controlled variability in production, and of their impact on speech intelligibility.

This research was timely because of an increasing focus on exemplar-based models of speech perception and on the impact of fine phonetic detail on speech perception. Also, it is increasingly recognised that speech perception research needs to be based on recordings of connected discourse produced with communicative intent, rather than highly-controlled scripted speech.

Within a specific communicative situation, it was hypothesised that speakers showing less variability in production (i.e. speakers who have ‘compact’ rather than dispersed’ constellations of traces within a phonetic category) would be more easily intelligible than speakers showing greater variability. However, across communicative situations, we expected that speakers who showed a greater ability to adapt their productions at a phonetic level to the communicative needs of the listener would be more intelligible than those who are less variable across speaking styles.