[From Bruce Nevin (2003.08.07 17:08 EDT)]
Bill Powers (2003.08.07.0650 MDT)–
much of what I have called speech
“sounds” consists of kinesthetic sensations, not auditory ones.
The buzzing
sound of Z, for example, is felt in the tip of the tongue! Also, say
“the”
a few times and examine the “th” sound. It sounds partly like
your tongue
between your teeth. That’s not a sound.
Well, yes, but there is also sound of course, or Mary would be asking why
your sudden interest in thistles when you say something like this’ll
do.
It’s an interesting exercise, and one now of particular interest to you
perhaps for learning Mandarin, to train yourself to be more aware of
what’s going on during speech. As Rick says
Rick Marken (2003.08.06.1120)–
my subjective impression is that I don’t
control for articulatory patterns at all when I talk. I can barely tell
what the hell my mouth (oral cavity, tongue, palates, etc) is doing when
I speak. I think I just do whatever I have learned to do with my vocal
apparatus
You can’t control what you can’t perceive, yes, but maybe it’s also true
that its harder to learn to control what you’re not aware
of perceiving. And for reasons that I’ve suggested at other times, for
cultural conventions to work it may be necessary that they be controlled
out of awareness. We have good reason to mistrust those who we know have
a facility at manipulating the variables from which we deduce their
persona. But that’s another topic, and too far afield for now. More to
the point here, you can train your awareness of what you are doing as you
speak, and learn finer control of your ability to make different speech
sounds independently of “whatever you have learned to do with your
vocal apparatus” in order to speak English.
If you produce the initial fricative of this continuously without
moving your tongue back to form a vowel you certainly do feel the
vibration of the pressure wave produced by your vocal folds in the air
stream. You can turn your voice off and on while continuously forcing air
through the aperture between your tongue and your teeth, alternating
between the voiced dh of this (written with a lowercase
delta) and the voiceless th of thing (written with a
lowercase theta). Don’t say either this or thing, but just
alternate dhthdhthdhthdhth by producing a voice or not. (Drop into
saying this or thing from time to time until you are sure
you have it right.) While you’re alternating the voice on and off,
everything else stays the same. Because everything else is constant, you
can more clearly notice the sensations of the edges of your tongue
touching your teeth, but dropping away from contact just at the tip,
leaving a narrow, flat tunnel for the air to pass through. You can feel
the increase of air pressure when voicing stops, with its hissing
sound.
I think what we need are some other examples
of “instrumental” controlled
variables; that is, variables that are controlled in order to control
other
variables at the same level.
Nice term. And yes, I think you’re on the right track here. But instead
of instrumental controlled variables I’d call them instrumental
dependencies between controlled variables.
Rick Marken (2003.08.06.1120)–
Bruce Nevin
(2003.08.06 13:08 PDT)
Bill Powers (2003.08.05.0557 MDT)–
Neither sound nor articulation is prior to the
other. They are both at the same level of perception.
Yes.
But references for articulations are calibrated according to their affect
on sounds, not vice versa.
I think this is right:
articulations are controlled as the means of controlling sound.
This is confusing the source of the dependency between auditory input and
tactile input. What is going on is that perceptual inputs in one modality
are linked through the environment to perceptual inputs in the
other.
We control variables by means of dependencies in the environment all the
time. I’m making letters appear on a cathode ray tube by pressing my
fingers down on an array of little cuboid pieces of plastic.
The dependencies between the tactile sensations of speaking and the
sounds of speech are not in the hierarchy, they are in the environment,
as in other examples of ‘instrumental’ dependencies between controlled
variables such as those Bill identified:
Bill Powers (2003.08.07.0650 MDT)–
Example: I curl my fingers around a glass
in
order to hold it in position for drinking. The behavior of the fingers
and
the glass seems related in the same way the behavior of the tongue
is
related to saying “th”
As you said earlier, the dependency is a matter of physics - physiology
and aerodynamics - not a function of how the nervous system is hooked
up.
Here, I think you’re talking about the correlation of the visual
perception of the configuration of your fingers curling around the glass
with the kinesthetic perception of it. To pick up the glass you control
both at once, and you set the references for the kinesthetic perceptions
(configuration of hand and configuration of arm) according to the visual
perceptions (configuration of hand and location of hand) and the tactile
perception (touch of fingers on glass).
This is how I’m thinking about speech: a
kinesthetic modality being
controlled because of its effects on the auditory modality _at the
same
level_. Isaac Kurtzer sent me an abstract of a paper about an experiment
in
which articulatory variables were disturbed in a way that had no
measurable
or sensible effects on sounds. The paper reported that the disturbance
was
resisted. Of course in a hierarchical model that doesn’t prove that
articulation is the primary controlled variable (as the authors
thought).
I’m very interested in this! Can you send me the reference?
But it does suggest doing the opposite
experiment, varying the sound
without changing the articulation (altering phonemes before they are
heard,
and so forth). I’ve tried to get somewhere with that but have had
programming difficulties.
Last I knew, this had to be done in hardware (Remember Houde et al? Never
replied). Are processors fast enough now?
I think there’s no doubt that in speech the
highest priority variable is
auditory. But this does not make it a higher level variable, unless we
can
show that configurations in the sounds of speech are functions of
articulatory sensations, which I’m pretty sure is not the
case.
Just as in picking up the glass the highest priority variable is the
tactile sensation of fingers gripping glass, and that is obviously at a
lower level than either the visual or the kinesthetic configuration
perception!
There are some articulations, as you say, that
change when disturbed so as
to preserve the sound being made. But with this new consideration of
instrumentality, this does not have to mean that the auditory
variables
are of higher level. If your chin is held down against a rest, your
head
goes up and down to say “mama”. You can also say
“mama” with your lips
remaining slightly opened and using a crosswise finger to block and
open
the gap. There are physical interactions among different effects of
controlled variables, and we use them to produce the experiences we
want.
If one aspect of the situation changes we can usually find an
alternate
means of creating the same effect. This doesn’t mean there is a
hierarchical relationship among the various means at the same
level.
This perhaps ought to be cast in terms of
several
levels of perception in one modality being related to control of
several
levels of perception in another modality, because there are
sensations,
configurations, and transitions (etc) in both modalities at once, but
the
first set is controlled to have controlled effects on the second
set.
Yes indeed! This is perhaps most elaborate in the correlations between
our control of language perceptions all the way up and down the hierarchy
and our control of nonverbal perceptions all the way up and down the
hierarchy.
Rick Marken (2003.08.06.1120)–
The most convincing demonstration (to me) of
articulation as a means of producing intended sound patterns comes from
pressing my tongue against the bottom of my mouth … and then trying to
say words like “Paraguay”. When I’m affecting a Spanish
accent I roll the r in “Paraguay”. But I can’t do that
with my tongue pressed against the bottom of my mouth. But I can get
something like the sound of the rolled r by making a guttural roll in my
throat.
This is simple physics. Pulling your tongue tip down forces the dorsum of
the tongue up toward the velum and (dangling from it) the uvula. The flow
of air between the back of the tongue and the uvula makes the uvula flap.
Voilá! You are now producing a (Parisian) French r, a uvular trill. Also
heard in many dialects of German. And you have a demonstration of how the
‘same’ phoneme can be pronounced as an apical trill (tongue tip) in some
dialects and a uvular trill in others, and how the ‘same’ phoneme can
change its phonetic implementation among s, z, trilled r, uvular r, etc.
over time.
I came up with this articulatory compensation
for the lack of tongue rolling “automatically” in an effort to
hear “Paraguay” in something like the way I usually hear
it. And, indeed, I am able to hear something remarkably close to
the “rolled r” version of “Paraguay” using the
guttural substitute.
What I see here is a completely different articulation (guttural roll)
being used to produce an intended sound: “Paraguay” with rolled
r. This makes me think that perceptions of articulation are not
parallel controlled components of the words we speak. Articulations
are, I believe, varied as necessary to produce the sounds we intend to
produce.
Surely. But not in real time, but rather over a series trials until the
reference value is reset, as in setting a sight. A ventriloquist
practices a lot before being able to produce normal-sounding
speech without moving the lips. Having done all that practicing, she has
a set of kinesthetic references for the ‘ventriloquism dialect’.
Since there articulations are typically
observed when no disturbances (like food in the mouth) are present, it
looks like particular articulations are used to produce particular
sounds. But I think this apparent one-to-one mapping of action
(articulation) to result (sound) is an artifact of observing speech in
conditions where the need for variation in action is reduced or
eliminated.
Without practice, impediments to normal articulation result in failure to
produce the desired sounds. The young woman with the hole in her cheek
(reported in the body-piercing paper that I mentioned in my talk and in
the Festschrift) was unable to produce consonants like p and f until she
learned to hold her tongue against the inside of her cheek to plug the
hole and maintain sufficient air pressure. This was not an instant
variation of means to control the auditory variable, such that there was
no error in the acoustic output when the disturbance was introduced.
/Bruce
Nevin
···
At 09:34 AM 8/7/2003, Bill Powers wrote:
At 11:20 AM 8/6/2003, Richard Marken wrote:
At 11:20 AM 8/6/2003, Richard Marken wrote:
At 11:20 AM 8/6/2003, Richard Marken wrote:
