Repetitive synchronous imitation as a tool for investigating speech timing

Fred Cummins & Juraj Simko

University College Dublin, Ireland

Repetitive Synchronous Imitation (RSI) was developed by Gabor Harrar (U. Stockholm) as a teaching tool for imparting Swedish prosody to L2 learners. In RSI, a user listens to a short looped phrase and attempts to speak in synchrony with that phrase. They hear a mixture of their own voice and the model voice. This exerts a very strong pressure to adopt the same timing and intonation as the model. We have just begun using RSI as an investigative tool to map the space of volitional modulation of speech prosody. By varying the rhythmic structure of model phrases, we can probe subjects' abilities to synchronize with a given pattern. Patterns may be from familiar of unfamiliar languages, and may incorporate signal processing to modify the temporal and melodic content. Our initial experiments aim at finding subjects' preferences for loop timing, and attempting to relate that preference to the rhythmic content of the model phrase. We will present initial results from these investigations, and we hope to discuss possible uses of RSI for the empirical investigation of rhythm in language. One such use we intend to explore is the use of RSI to force users to intentionally modify their prosody as much as is possible, so that we can look for residual features of their speech which might serve as indices of speaker identity in a biometric application. We look forward to jointly developing other uses of this experimental technique.