This content is archived, it could be outdated.
Matthias Sjerps defends PhD on November 28
How do listeners manage to deal with variation in speech? They rely on cues in the speech context, Matthias Sjerps, PhD at MPI's Language Comprehension Department, concludes. On November 28 at 15:30, in the Radboud University aula he will defend his thesis 'Adjusting to different speakers: Extrinsic normalization in vowel perception'.
November 25, 2011
Speakers vary widely in the shapes of their vocal tract and in their speaking styles. Two different speakers producing the same vowel can produce very different sounds, but listeners seem to understand with little effort. How do they manage?
Early perceptual processes
Listeners tune in to speakers’ speech through normalization, Matthias Sjerps describes in his thesis. While most research in speech perception focuses on cues that are present in the speech signal itself, properties of the surrounding context play an important role on the perception of those cues. Sjerps describes how relatively basic auditory processes cause strong contextual influences on the perception of speech cues. These perceptual adjustments have a positive influence on speech perception because they allow listeners to adjust for part of the variability in the speech signal, especially the variability that is caused by differences in speakers' vocal-tract lengths.
Influences of language exposure
In his experiments, Sjerps investigated these processes in speech stimuli but also for stimuli that do not sound like speech. Similar normalization effects were found with non-speech signals, but effects were generally stronger with speech signals. Moreover, a comparison across native speakers from different languages indicated that a listener's background language has an influence on the strength of normalization processes. These findings suggest that, even though normalization has an early perceptual origin, this process is in part shaped by linguistic experience.

