Disturbing Speech

CSGnet Archive CSG1993
1

[from Gary Cziko 931014.1850 GMT]

Responding to recent posts, mostly by Rick Marken and Martin Taylor:

All this discussion recently about closed-loop vs. "ballistic" models of
speech production got me thinking of some demos that should be quite easy
to do that might shed some light on this.

Some time ago I mentioned the demo of speaking while keeping the tip of
one's tongue pressed against the lower or upper teeth. For best effect, do
this while talking to yourself as you watch your face in a mirror. I have
no difficulty producing intelligible English in this manner, although it
does sound a bit thick and distorted (top teeth is worse). The facial
expressions that accompany the compensations necessary to produce speech
sounds are very funny (at least on my face). I take this as a
demonstration that speech is not produced via pre-planned behavior of the
articulators, but rather that the auditory perception is controlled in real
time.

But I also think that with an UNDISTURBED mouth we can speak by controlling
the propioceptive perceptions of our oral cavity and articulators. If so,
one should be able to speak intelligibly if all auditory feedback is
eliminated. I suppose this could be approached by providing a high
intensity white noise via headphones while the subject speaks softly (to
reduce bone transmission of sound). (Rock singers essentially have to do
this when their monitor speaker dies. But due to the nature of rock music,
it probably doesn't make their singing much worse). I'd like to try this,
but would be happy to find out that it has already been done.

But with zero auditory feedback, it should not be possible to produced
accurate speech if the mouth is disturbed (e.g., doing the tongue trick
mentioned above, or introducing a novel object into the mouth while being
subjected to high intensity noise). Now the usual proprioceptive
perceptions for producing speech sounds are no longer achievable, and with
no auditory feedback new alternative ones cannot be found.

My guess is that normal speech is some combination of real-time auditory
and propioceptive perceptual control, with auditory having precedence (in
the way that visual control has precedence over vestibular control in
maintaining balance, as demonstrated by my eye-ball-pushing demo). If
auditory feedback is knocked out, propioceptive control can take over, as
long as there are no disturbances to the vocal tract, in which case
(neither auditory nor propioceptive feedback) accurate speech is no longer
possible.

--Gary

P.S. Perhaps someone can fill me in on what novacaine does to one's mouth
(I've never let my dentist stick me with this). Does it disrupt both motor
and sensory functions? Is it hard to speak after having one's mouth shot
full of novacaine or other analgesic?

···

------------------------------------------------------------------
Gary A.Cziko Telephone: 217-333-8527
Educational Psychology FAX: 217-244-7620
University of Illinois E-mail: g-cziko@uiuc.edu
1310 S. Sixth Street Radio: N9MJZ
210 Education Building
Champaign, Illinois 61820-6990
USA
------------------------------------------------------------------

2

From Tom Bourbon [931014.1730]

[from Gary Cziko 931014.1850 GMT]

Responding to recent posts, mostly by Rick Marken and Martin Taylor:

All this discussion recently about closed-loop vs. "ballistic" models of
speech production got me thinking of some demos that should be quite easy
to do that might shed some light on this.

Some time ago I mentioned the demo of speaking while keeping the tip of
one's tongue pressed against the lower or upper teeth. For best effect, do
this while talking to yourself as you watch your face in a mirror. I have
no difficulty producing intelligible English in this manner, although it
does sound a bit thick and distorted (top teeth is worse). The facial
expressions that accompany the compensations necessary to produce speech
sounds are very funny (at least on my face). I take this as a
demonstration that speech is not produced via pre-planned behavior of the
articulators, but rather that the auditory perception is controlled in real
time.

Agreed. And it *is* a funny exercise. (Funnier on some faces than on
others. . ..)
. . .

But with zero auditory feedback, it should not be possible to produced
accurate speech if the mouth is disturbed (e.g., doing the tongue trick
mentioned above, or introducing a novel object into the mouth while being
subjected to high intensity noise). Now the usual proprioceptive
perceptions for producing speech sounds are no longer achievable, and with
no auditory feedback new alternative ones cannot be found.

My guess is that normal speech is some combination of real-time auditory
and propioceptive perceptual control, with auditory having precedence (in
the way that visual control has precedence over vestibular control in
maintaining balance, as demonstrated by my eye-ball-pushing demo). If
auditory feedback is knocked out, propioceptive control can take over, as
long as there are no disturbances to the vocal tract, in which case
(neither auditory nor propioceptive feedback) accurate speech is no longer
possible.

This experiment has been done, often. Impairment and loss of hearing
do have the effects you mentioned. Over time, the memories of
proprioceptive features of speech seem to become less clear (I do not like
that metaphor, but it will serve here) and the person makes speech sounds
that are more and more discrepant from their original forms. My experiences
when I tried to remember and reproduce the proprioceptive sensations of
tracking a triangular target (described in an earlier post today) were
frustrating; they pale by comparison with the task of maintaining
comprehensible speech in the absence of hearing. Repeating remembered or
imagined patterns, with precision, in the absence of significant
perceptions, is not easy, it it is possible at all. "Sort of close" isn't
the same as "doing the same thing."

P.S. Perhaps someone can fill me in on what novacaine does to one's mouth
(I've never let my dentist stick me with this). Does it disrupt both motor
and sensory functions? Is it hard to speak after having one's mouth shot
full of novacaine or other analgesic?

The pharmacologic effect is sensory; the total effect is -- you know that
one. Is it hard to speak? Yeth thur! (Drool)

Until later,

Tom

3

But with zero auditory feedback, it should not be possible to produced
accurate speech if the mouth is disturbed (e.g., doing the tongue trick
mentioned above, or introducing a novel object into the mouth while being
subjected to high intensity noise). Now the usual proprioceptive
perceptions for producing speech sounds are no longer achievable, and with
no auditory feedback new alternative ones cannot be found.

Stammering can be cured in some extent by delaying the auditory feedback
by an external microphone- audio amplifier - delay circuit - earphone device.
For normally speaking subjects the same device will cause stammering.
I have tried it and the effect is quite amazing: by increasing the
adjustable delay you at certain point hardly can produce intelligible
speech. The device is commercially available and in clinical use.

There has been a long lasting debate between auditory and motor theory
of speech perception (and some mixes). One of the newest additions is by
Johnson et al. "Individual differences in vowel production" (JASA 1993, 94 2
pp. 701-714): "It seems clear that to some extent speech motor movements are
not only goal directed, but also that the goal is an auditory one, leading
us to believe in the validity of an auditory theory of speech production"

I believe that the proper combination of production and perception aspects -
maybe via PCT - leads us to the right track in speech science.

Osmo.Eerola@hcc.utu.fi

···

from: Osmo Eerola 931015.0730 GMT
to: Gary Cziko 931014.1850 GMT