Presentation: Sarah
Summary: Thomas
Talker variability can
easily be adapted by human listeners. Studies have been done to investigate how
listeners accomplish the adaptation. Two approaches have been proposed
concerning this issue. Advocates for the normalization approach believe that
when listeners encounter variations in speech, they change the normalization
algorithm, adjust the incoming signal, and map the adjusted signal onto the
mental representation. On the other hand, supporters of the representation
approach hold the view that variations leave memory traces, and the
representations in the lexicon are adjusted accordingly. The goal of the
present study is to tease apart the predictions that can be made with these two
approaches. The target of this study is the raising of the vowel /æ/
before /g/ in words such as bag. In
the first experiment, two groups of participants listened to normal bag-type words and raised bag-type words respectively, and then
their identification of back-type
words was tested in an eye-tracking experiment. The rationale is that if the
adaptation to the raised bag-type
words facilitates the identification of the back-type
words, the representation approach is supported rather than the normalization
approach, according to which, adaptation on the bag-type variations should not have affected the adjustment of the
signal of the back-type words. However,
for the representation approach, it is possible that the competition between bag and back interpretations can be mitigated because of an adaptation to
the raised-bag stimuli that widens
the differences between these two words. This is exactly what the authors
found: the group of listeners adapted to the raised bag-type words identifies back-types
faster than the group that listened to the normal bag-type words. A second experiment of a within-subject design
replicated the finding. The authors conclude that adaptation to speech signals
involves more than adjustment on the incoming signal. Dynamic adjustment of the
representation in the lexicon also occurs as listeners cope with speech
variations.
