The purpose of this study was to determine whether a short listen-and-repeat training paradigm can be used to train vowel duration discrimination and production, and whether any learning effects are transferred to an untrained vowel or a non-linguistic sound. Similar training has previously been used to train vowel quality contrasts to young adults and children, with results showing up both in behavioral and psychophysiological measurements. Unlike vowel quality, segment duration can be considered to be a suprasegmental feature that is not directly dependent on any other acoustic feature of the sounds being trained. It is therefore plausible that it can be learned a separate skill and generalize to untrained segments, and even non-linguistic sounds. Participants were 18-30-year-old healthy adults with normal hearing,
who were not native Finnish speakers and had spent little time in Finland. The stimuli were semisynthetic Finnish pseudoword pairs /tite/-/ti:te/ and /tote/-/to:te/. A sinusoidal tone pair served as the non-linguistic stimulus. The behavioral measurements employed in the study were an oddball discrimination
task for all three stimulus pairs, and a listen-and-repeat production task for both of the vowel pairs. No feedback was given. The experiment was conducted in three sessions over three days. The first two consecutive days consisted of baseline measurements for all the stimuli and four blocks of production training. The third day, taking place 1-2 weeks after the second, consisted of full progress measurements. The results show that the training did induce changes in discrimination sensitivity and production of the trained length contrasts, though not all effects remained at the end of the experiment. This suggests that while the processing mechanisms related to the processing of duration contrasts are somewhat separated from the processing of vowel quality, it seems that they can be accessed with this kind of training.