[From: Bruce Nevin (Thu 93107 14:35:14 EDT)]
Rick Marken (931007.0915) --
>I'd appreciate a response to that, rather than flip dismissal of a single
>sentence interpreted contrary to the context in which I said it.I didn't dismiss your sentence; I tried to correct your understanding
of how the test works
One more time.
(1) I should have said "dismissal of my post based upon a single sentence
from it, interpreted contrary to the context in which I said it." I
started to say something like that, but it was too complicated, and
in simplifying I introduced a fatal opportunity for misunderstanding.
My fault.
(2) The post that you dismissed was not about the test. As the
subject header and the opening paragraph stated, it was an analogy
between our conversation and the coin game. It was an
attempt to go up a level and look at our conversational process and the
current state of our evolving discussion from a perspective outside it,
by drawing an analogy to the communicative process used in the coin game.
Therefore, your attempt to correct my understanding of how the test works
was based on a misunderstanding of what I was saying. If you got an
impression of how well or how poorly I understand how the test works from
the post from which you extracted that one sentence, you did so by
effective use of the imagination loop. I didn't put it in that post.
You did.
Now if you're willing to go back and read what I said with this new
understanding, fine. Otherwise, I give up. Maybe going up a level by
means of referring to our ongoing interactive conversational process is
just not the kind of perception you like to pay attention to, who knows.
You can be convinced that I really don't understand the test if you like.
If you're inclined to put energy into getting me to have the correct
reference perceptions regarding how the test works, fine, I'll probably
learn something, and the cost to me is at worst finding creative ways to
say yes I already knew that. Even that could be fun, to a degree.
I am currently more interested in your answer to the question I posed
yesterday -- the last question in this paragraph:> I can think of ways to test whether
> a person is controlling for "pin" or the contrast between "pin"
> and some other word, like "bin". Before proposing such tests, I would
> like to know if my understanding of "controlling for contrast" is
> the same as yours. Is my description of controlling for contrast
> correct? That is, does a person control for contrast by controlling
> for a perception of a word as being categorically different than
> all other words with which it might be mistaken?
Sorry, I missed this. The volume of mail around these issues has kind of
swamped me, and I've had trouble keeping up. I don't have a clear and
simple answer to this. My best attempt to clarify the relation between
phonemic contrast perception, phonemic category perception, and phonetic
detail perception is in a post I sent this morning.
Thinking of it as a contrast between a particular word "pin" and some
other particular word, like "bin" might account for some variation of the
sort that Martin adduced for donkey vs. flunkey. I don't have any good
evidence about that, Martin says he does. What I have in mind is what I
touched on in the post yesterday about the article in _Language_,
"carefulness" or distinctness of pronunciation, ranging from lax and
relatively indistinct to hyperarticulated. The evidence in the article
suggests that the reference perceptions for vowels are for
hyperarticulated speech, in which the vowels are maximally different from
each other phonetically. But I suspect these are not fixed "targets",
but rather, for a given vowel, whatever happens to be as different as
possible from the other possible vowels, when they are also each being
controlled for maximum difference from the others. This is the converse
of the gather program. You could write a program called disperse in
which the agents maximized their distance from one another within a field
defined (a) by physiological constraints (with acoustic
consequences)--can't move the tongue any higher, farther forward,
whatever, without the sound of turbulent air flow and (b) by social
convention prohibiting some of the physiological possibilities--e.g. we
don't use the combination of tongue in forward position and lips rounded
in English, as for the [y] sound of French tu or German Ueber. This
program would plot the hyperarticulated formant positions of the vowels.
If the formants are plotted with the higher frequencies toward the origin
of a Cartesian graph and the lowest frequencies out on the horizontal and
vertical axes, those positions for first and second formants map onto a
picture of the location of the highest elevation of the tongue within the
oral cavity. See the _Language_ article for examples. Similar things
could be done for maximizing the perceptual distance among the consonant
segments. In speech that is less distinct, the distances between
phonemes is reduced.
I explained how I think this starts out with error in keeping words
(coincidentally syllables) distinct from one another in such a way that
adults recognize them, and how reorganization that follows on that
persistent error results in development of the sound categories with the
above exaggerated exemplars as reference perceptions.
We can see that we are dealing with categories rather than degrees of
perceiving a given hyperarticulated phoneme target because of the ease
with which we "map" from one dialect to another with hyperarticulated
targets that may be markedly different. F'rinstance, the specific sound
that is the a of can in one dialect "maps" to a specific sound that (for
a speaker of the first dialect) is an instance of the a of late followed
by a yun syllable: CAyun. Yet speakers of the two dialects very quickly
and easily normalize these differences and recognize their very different
pronunciations as repetitions of the word "can".
Bruce
bn@bbn.com