[From: Bruce Nevin (Fri 93108 11:18:52 EDT)]
( Rick Marken (931007.1600) ) --
I just thought 1) the pair test was a good
example of "the test" and 2) it can be used to test your idea about
"contrasts". But things seem to get what seems to me to be too
complex too quickly
>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?
Well, you said that the answer I attempted was too complicated. Probably
my way of putting it was unnecessarily complicated, because I'm still
trying to figure all this out.
Martin's reworking of it is very helpful. The largish phonetic events
that a baby begins to control as "words" contrast with one another in the
same way as one person in the environment contrasts with another, such as
mommy and daddy: they differ in many respects, and it is their difference
at lower levels that amounts to "contrast". There may be some evidence
that their contrast is not a controlled perception, as for example the
fact that the same monosyllable can have quite different referents (ba
for bottle and bath and blanket), even while e.g. ba and ma are different.
Then ambiguity results in communicative error as the child learns the
game of requesting, etc. (Brunner, _Child's Talk_). Requesting ba
results in mommy bringing bottle instead of security blanket, or she says
"yes, you do need a bath, you're right" and pops her in the tub. "No,
not ba, BA!" appears to be a frustrating thing not to be able to get across.
The monosyllabic abbreviations, if that is what the child's "words" are,
have to be differentiated from one another in ways that adults identify
with adult words. But this is an outsider's view of the goal of
Reorganization creates new kinds of perceptual functions by factoring out
elements of contrast in what is heard and in what the child is struggling
to differentiate in speech. The process is supported by the fact that
the adults employ perceptual functions of precisely this sort to control
their adult pronunciations of what they say. Minds, great or not, run in
similar channels, to paraphrase the old catch phrase.
But the new phonemic perceptions (and their relations to lower phonetic
perceptions on the one hand and to words etc. on the other) do not
utterly replace the more primitive word/syllablel-as-phonetic-event
perceptions. It appears that both persist and continue to develop in
parallel. Imitation is a lively capability right alongside repetition,
and the one often precedes the other as we learn a new word.
I'm not sure what to do with the notion of words being categorically
different from other words. The category perceptions that are relevant
here are not word classes, but rather the phonemic categories.
It may be that the problem of keeping one word distinct from others is
(a) what drives reorganization and the development of phonemic perceptual
functions in the origin of language, but (b) after phonemic perception is
in place is relevant only to variations in pronunciation of the "same"
word. In careful speech we approach the hyperarticulated targets more
closely and as a consequence the sounds or sound-features perceived as
the phoneme categories are more greatly contrasted with each other; and
precisely because the above reorganizational development of the phoneme
perceptions has been successful, the words being pronounced are
phonetically more different from other words with which they might be
confused by a hearer. In less careful speech, we relax from those limits
of the physiological and acoustic envelope that help to define the
hyperarticulated targets of the phoneme categories; the specific phonetic
productions classified as one phoneme or another are closer together, but
they are still maximally different from one another within a reduced
physiological and acoustic envelope. And, correspondingly (as above),
the words being pronounced are phonetically less different from other
words and consequently more likely to be confused with them by a hearer.
In adult speech, perhaps we can say that the appearance of contrast
between words is a byproduct of control of contrastiveness among
phonemes. But to model language learning we have to start with
differences, and then contrasts, between phonetic productions that are
bigger than what we later identify as phonemes. And there is more than
one way to segment utterances into contrastive factors or phonemic
elements (this is called "the non-uniqueness of phonemic
representations"), and no good evidence for expecting that all adults
have arrived at the same representation of contrasts in the course of
reorganization as children. (Although received dogma in the field holds
that these elements are universal because they are hard-wired as part of
the human genetic endowment.) The appearance (to an outside observer)
that two speakers of the same dialect are controlling perceptions of the
same phonemes is due to the fact that in each case speaker A is
controlling for A's phonetic outputs (byproducts of A's control of A's
phonemic category/contrast perceptions) being recognized by speaker B as
instances of B's perhaps differently organized phonemic category/contrast
perceptions. The outside observer may see these phonetic outputs as "the
same" because she categorizes both A's productions and B's productions as
instances of her own phonemic category/contrast perceptions, which may
differ yet again from those of A and B. A of course perceives both A's
and B's productions as instances of A's phonemes, and conversely for B.
Is this still too complicated? Can you still offer your proposal as to
how to test these ideas by using the pair test as an example of the test?
Or do you still feel that this incapable of test? I guess (but I could
be wrong) that you have a counter-proposal in mind, one that is
presumably much simpler, and that you would like to test both. Could you
spell out how that counter-proposal handles the issues raised here?