sound and fury signifying control

[From Rick Marken (931005.0900)]

Bruce Nevin (Mon 93104 10:51:23 EDT) --

Were you unable to control a perception of an acoustical difference
between [sp`in] (with the aspirated voiceless [p`] of pin) and sbin (with
the unaspirated voiced [b] of bin)?

All I can control is the perceptual relationship between these sounds;
I might be able to control their acoustical relationship if I had a way
to perceive it (for example, using a sound spectrogram). I was able to
produce perceptions of spin and sbin that sounded different to me; they
both still sounded like "spin" (the event level perception was the same)
but I could hear the difference in sensations (I could also feel the
difference, which is probably important in the "pair" test). I imagine
that any person could tell that spin and sbin sound different (if the
person attends to the component sensations of the words) but the words
would still sound like the same word (at the event level). This just
shows that, at the event level (the word level) the sensations that
differentiate spin from sbin don't matter.

Bill Powers caught the fact that the reason I acknowledged the "pair
test" as a version of "the test" is because I saw that it is like
the coin game. The intended "word" (the first in the pair) is equivalent
to the intended pattern of coins. The variants of the words produced
by the "tester" (the second word in the pair) are like the changed
pattern of coins. In both cases, the testee only complains if the
variant results in a perception different from the intended one.
In both cases, also, the tester's hypoethesis about the controlled
variable is based on his or her own perception; in the coin
game, the tester might see the coins make a triangle and guess that
this is the controlled variable (triangular shape); in the pair test,
the tester might hear the prototype word as "spin" and guess that this
is the controlled variable. But the testee might be controlling a set of
sensations that sound like "spin" to the tester. So the tester tests this
by chaging the sensations to "sbin"; there is no resistance to this
disturbance of sensations so the tester retains the hypothesis that the
subject is trying to hear "spin" (becuase "sbin" sounds like "spin"
to the tester too). In the coin test, this is like moving one coin
upward to extend the apex of the tringle. If the subject cases about
the precise locations of the coins this will be a disturbance (as
"sbin" would be if the subject were controlling for precisely the
sensations of "spin"); if the subject is controlling for triangle,
however, moving the apex coin will not be a disturbance.

Is it the case that [sbin] and [sp`in] are both "unnatural" sounding and
"unnatural" feeling, relative to your ordinary pronunciation of spin as
[spin]?

Not really.just like accented variants of words.

For comparison, examine the perceptions that arise when substitute the
same [b] and [p`] segments in place of the m of "mud"?

OK.

I predict that
substituting the b results unequivocally in perception of a word, "Bud",
and substituting the p results unequivocally in perception of a non-word,
"pud" (I assume that's not a word for you).

It is a word (a sound event); it just has no meaning for me when I
just pay attention to its as a "word" event.

Is it the case that any of these three pronunciations fails to evoke the
word/meaning "spin"?

Yes. All three. I didn't think of "spinning" once. I just heard the
same word (event) each time.

How do you account for this difference in the perceptual effect of
substituting [p`] vs. [b] in a constant frame?

If I could account for it I would definitely be in contention for a
Nobel prize. But my GUESS is that it has to do with differential
neural computations that turn the different sound patterns into
quite different perceptual signals; the Nobel comes when I figure
out what those neural computations are.

Bruce Nevin (Mon 93104 15:44:57 EDT)--

"Contrast" is phonemic. It can be determined *ONLY* by applying
the Test with a speaker of a particular language.

So "contrast" just means " perceived difference". No problem.

Bill said:

The truth of the matter is that people perceive a contrast or
difference [sic] between pin and bin, but not between spin and sbin,
and nobody knows why, not even a linguist.

You say:

You give up too easily.

Perhaps there is an explanation due to the physiology and
acoustics (physics) of speech production?

But then you don't say what the explanation is. Yet you say that
Bill gives up too easily. Don't be so coy, Bruce. Tell us, why
do people perceive a difference between pin and bin but not
between spin and sbin?

But at root you are perfectly correct.

Ah, so it was just a bluff after all. Or do people perceive the
difference between between pin and bin but not between spin and
sbin because there is a _contrast_ in one case and not in the
other?

Contrast is the fundamental, elementary datum for any science of
language. It can be determined only by the Test.

All the test can do is hone in on the perceptions people control.
A "contrast" in the test is nothing more than a perception that
does not match a reference (so when I arrange the coins in a square
it "contrasts" with the intended triangle and will be noted as
"different" or "error" by the subject). Contrast is just another
word for "error". When we want to say "pin" and we hear "bin" we
hear a contrast (an error). It's not the error (contrast) that is
fundemental to language; it is the fact that we can intend to have a
particular perception (pin, bin or spin), know when we are having
it and what to do when we are not.

Best

Rick

[Martin Taylor 931005 13:30]
(Rick Marken 931005.0900 to Bruce Nevin 93104 10:51:23 and 93104 15:44:57)

Contrast is the fundamental, elementary datum for any science of
language. It can be determined only by the Test.

All the test can do is hone in on the perceptions people control.
A "contrast" in the test is nothing more than a perception that
does not match a reference (so when I arrange the coins in a square
it "contrasts" with the intended triangle and will be noted as
"different" or "error" by the subject). Contrast is just another
word for "error". When we want to say "pin" and we hear "bin" we
hear a contrast (an error). It's not the error (contrast) that is
fundemental to language; it is the fact that we can intend to have a
particular perception (pin, bin or spin), know when we are having
it and what to do when we are not.

"I hear you"--"What do you hear?"--"I hear you want an argument"--"I didn't
say that"--"That's what I hear."

The message sent is not the message received (necessarily). Rick argues
that he can hear (and in production execute) a perceptible difference
between "spin" and "sbin", and that "Bud" and "Pud" are both "words" = "word
events". Bruce is trying to develop the idea of phonemic contrasts, and
Rick resists the perception. I think they are talking about different
levels of the talker/listener's hierarchy, and will never agree if they
continue to do so.

A phoneme is a category level variable. A phonetic difference is not.
What is it? I'm not sure, but probably it is at the level of configuration
or transition.

In my view of the hierarchy, there is a very significant difference between
the levels at and above category and the levels below. In fact, I admit
to some scepticism about the strict hierarchy of levels below (witness the
indecision about whether configuration is above or below transition). But
I think there is a real difference of effects when you get to category. At
that level, a perception is or is not that of a category member. One does
not control for the degree of belongingness to a category so much as for
the fact of belonging. A group of three lines something like this compressed
laterally "/-\" might be a represetation in perspective of a railway track,
a capital letter "A" or "H", or just "three lines used as an example." In
a different context it might be any or none of these. But if in a particular
context it is one (e.g. "H/-\T" makes it an A, under normal circumstances),
then it is not any of the others. There is no continuum at this level or
at any of the higher levels. In fact, there is hysteresis as one changes
the configuration between a good picture of, say, "H" and a good picture
of "A". The same configuration may be "A" if approached from one end of
the configuration continuum (at that level there IS a continuum), and as
an "H" if approached from the other. But it is never "0.3A and 0.7H".

Bruce used the continuum level (finessing "configuration-transition")
phonetic difference, to try to present the category level phonemic contrast.
Rick used the phrase "word event," which seems to me to be an oxymoron
(is there such a thing as a Co2moron?). "Event" lies within the set of
levels that have continuum representation. "Word" does not. A "word"
is specific. A configuration of letters or pattern of acoustic events
may change continuously, but a word is perceived to be one or other
of a set of possibilities (or as none of them). Phonemes contrast,
phonetic features differ.

···

=============
Quite apart from this specific case, I think that a discussion of the
differences between category-based control levels and continuum-based
levels might be quite fruitful. I believe that the structure of the
hierarchy changes at and above the category level. Below the category
level, I see at present no reason to question the strict upward flow
of perceptual signals and the counterflow of output signals. At, and
probably above, the category level, I think it will be found necessary
to postulate within-level connections among the perceptual signals, giving
rise to feedback loops totally within the perceptual structure. The same
might be true of the output signals, though I have no real reason to
suppose that it is.

I hope that the first part of this posting helps clarify the discussion
between Bruce and Rick. I'm not so hopeful about the last paragraph.

Martin