Re: Motor Theory of Speech Perception

From: Harnad, Stevan (harnad@cogsci.soton.ac.uk)
Date: Wed Apr 30 1997 - 20:08:15 BST


> From: Taylor, Karl <krt196@soton.ac.uk>
>
> > (27) The findings about perception by chinchillas:
> >
> > A. are irrelevant to theories of speech perception
> > B. support the motor theory of speech perception
> > C. *contradict the motor theory of speech perception
> > D. support the locus theory of speech perception
> > E. contradict the locus theory of speech perception
>
> I can remember this from the lecture:
>
> Some people proposed a new theory, Motor Theory, to explain
> speech perception (which says that it is based on speech
> production), involving categorical perception.
>
> Someone found out that preverbal infants possess categorical
> perception.
> The Motor Theorists said "Oh. Well, I guess categorical
> perception is so important that it is hard wired in our
> species."
>
> Someone else then found out that even chinchillas have this
> thing.
> The Motor Theorists said "Really?! Well, perhaps speech
> simply took advantage of a pre-existing sensitivity in our
> early ancestors."
>
> But how does this contradict the Motor Theory?
> Speech perception is still based on speech production.

Good exact reconstruction of the whole story. But remember that the
motor theory was supposed to EXPLAIN something: It was supposed to
explain why certain sound-differences sound so different even though
they are physically similar. For example, Ba vs Pa, or Ba vs. Da. vs Ga.
Acoustically, in both cases these are just values along a continuum.
But we don't hear them as being on a continuum (as we do, say,
different loudnesses); we perceive them as being distinct in the way
that colours are distinct (rather than as just gradual changes in one
colour, or in shades of gray).

The motor theory said: Physically they may be on a continuum, but in
order to PRODUCE those sounds, the mouth has to make very different
movements, and these movements are NOT on a continuum (the way, for
example, vowels are). So it is the discontinuous mouth movements you
need to make to produce them that make you hear them discontinuously
too.

Now when the preverbal infants showed that they perceived
discontinuously too, the Motor theorists saved the theory by saying that
since it was so important for us to perceive speech clearly, the
boundaries have become inborn, but their origin was still the motor
production that all members of our species share.

Then came the data about the chinchillas, who turn out to have the
ba/pa boundary too, even though they never said, never would say, never
COULD say the sounds "ba" and "pa" in their lives (or in the lives of
its their ancestors).

Did motor theory crawl away and die? No: Theories only disappear when
they are replaced by better theories, and this was only negative data -- so
it DID contradict motor theory -- but, since you have "proof" only in
mathematics, the Motor theory could be "saved" again by assuming that it
was still a motor process that made ba and pa so different, but it was a
motor process that was somehow taking advantage of an already existing
previous discontinuity IN PERCEPTION. That's the critical point:

The motor theory was supposed to be explaining perceptual differences
with production differences. But now the motor theory itself had to prop
itself up on a prior perceptual difference!

So although I said in lecture that people still hold onto the motor
theory, there is no question that the chinchilla results contradicted
it, and forced the theorists to twist themselves into a very
uncomfortable posture in order to save the motor theory.

In mathematics, if something is contradicted, it must be rejected; in
science, it can sometimes be saved, but always by weakening it with
further assumptions. That's what happened with the motor theory of
speech perception.



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:52 GMT