Item

Abstract

To deal with various sources of variability in the speech signal listeners compensate for acoustic context in the recognition of sounds (e.g., vowel normalization, compensation for speaking rate). These mechanisms have mostly been studied in isolation but show striking similarities in their workings and presumed earliness of operation during word processing. This raises the question whether compensation for speaking rate and spectral context are reflections of a common compensation mechanism. An indication of a general mechanism could be the simultaneous use of the two types of contexts in online word recognition. Alternatively, a different timing would support the presence of underlyingly different compensation mechanisms. To test these hypotheses two eye-tracking experiments investigated the time course of the influence of rate and spectral context on the perception of the Dutch //-/a:/ contrast (e.g., tak-taak, “branch”–“task”) which is cued by spectral and durational information. As expected, both types of context affected vowel perception. Moreover, both effects occurred at similar points in time early during word processing. A comparison of context effects to the uptake of spectral and durational cues on the vowels also showed a similar time course of processing. This suggests that compensation processes take place at or before the level of processing at which acoustic cues to phonemes are processed. It further supports the view that compensation for speaking rate, compensation for spectral context, and possibly compensation for other types of context such as pitch, take place at the same processing level as part of a general compensation mechanism.