[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Request for help on individual-differences review.


One of my colleagues (Aimee Surprenant, a postdoc in our lab) and I are
working on a manuscript in which we review recent-and-not-so-recent studies
of individual differences in auditory capabilities among so-called "normal"
hearing listeners.  We would very much like to hear from any of you who might
have items that we might otherwise miss, that could usefully be included in
this review paper.  The central thrust of the paper is the failure of most
(but not all) measures of individual differences in psychoacoustic capabilities
to predict differences in speech processing.  We entertain the hypothesis
(1) that we simply have failed to measure the auditory abilities that are most
relevant to speech processing, and also give some consideration to (2) the
possibility that speech processing depends much more strongly on
the gross analysis of the waveform's spectral-temporal details than on the
ability to process its fine structure (the measures of psychoacoustic
capabilities might be more relevant to the latter).

I talked about these matters at a recent meeting of CHABA on speech processing,
and also at the AFOSR grantee's meeting in Dayton a couple of weeks ago. The
only new information in those talks was a couple of studies of the correlation
between individual differences in lip reading abilities and the ability to
recognize speech in a noisy background.  In each case we found the correlations
to be about 0.5....not impressivley large, but larger than the relation
between most psychoacoustic abilities and speech recognition scores.  We're
been interpreting the look-vs.-listen correlations as demonstating that
individual differences in speech recognition are to some degree modality
independent.  Modality-independent factors have been proposed in the past in
studies of lipreading, and we have found some support for those ideas
(speed of processing, capacity of working memory, etc.) by including a test
of sentence recognition on the basis of orthographic fragments (also correlates
about 0.5 with auditory speech processing, and somewhat less w. visual).

All this began because we found rather larger individual differences in our
psychoacoustic studies with tonal patterns than we had in earlier work with
simpler stimuli (single tones, noise bursts, clicks, etc.).  Others reported
similarly large individual differences in studies with profiles and a range of
other complex sounds.  We finally felt driven to some frank individual-
difference studies (e.g. Johnson, Watson, and Jensen, JASA, 1987) and then
developed a tape-recorded test battery including 8 subtests, 6 with nonspeech
and 2 with speech.  After using the Test of Basic Auditory Capabilities (TBAC)
with almost 1000 listeners (various groups of normal and impaired listeners)
it seems clear that there is a fairly strong "speech factor" to which at
least the nonspeech subtests on the TBAC are relatively insensitive.

So much for that.  Here is what we are looking for from any of you who might
be willing to contribute to this effort.

1) Do you know of any psychoacoustic measures that do correlate reliably with
speech processing?  (Jesteadt et al. have shown some evidence of such a
correlation with the Q of psychophysical tuning curves, for example)

2) Have you found any large individual differences in auditory discrimination
abilities for complex nonspeech sounds that ought to be added to our growing
list of such findings?

3) Do you know of prior reports of the look-vs.listen correlations for speech
processing? (we found only a couple of such papers with the appropriate data,
neither of which had noted the correlations...although they were about the same
size as the ones we found, when we calculated them)

For what its worth, we have generally been able to rule out IQ or other
general measures of intellectual ability, and have some evidence that
musicianship and language-learning abilities are not relevant (or very weakly)
...including the ability to learn foreign languages. Overall motivation also
does not seem to be a strong factor explaining the individual differences for
either speech or nonspeech discrimination performance, at least for unselected
college-student subjects.

The reason for this public request for help is that we imagine that many of
you may have run into the individual-differences problem in psychophysical
studies with complex sounds, but elected to de-emphasize them because of your
primary interest in stimulus determinants of performance.  We certainly did
that for a long time, but some of our groups of "normal" listeners seem to
include a distressing number of people who can't perform the tasks well
enough to produce orderly data.  Any help, from anecdotal reports to references
that we ought to know about but may not, would be most appreciated.  We will
make up a distribution list including all contributors, and pass around
whatever this "net" catches.

Best regards,

Chuck Watson
Indiana University