Re: The Ambiguity of "Noise" [WAS: Parallelism]

If we are tripping over the ambiguity of the word 'noise,' it is my
fault for speaking imprecisely.

I *should* have said that the recording of a word like 'bad' had
three segments which had clear sound patterns characteristic of [b],
[@], and [d], and two interstital transitions.  And that the usual
assumption would be that those pure patterns, characteristic of the
phone in question, are important, because -- suprasegmental phonology
aside for the moment -- segments are supposed to be discrete units in
combination, processed as discrete units in combination..

Therefore, the expected result would be that given three sound
patterns, characteristic of [b], [@], and [d] in isolation, one could
perceive the word [b@d].  But in fact, it was difficult to do so.
Whereas, it was easy to recognize a word if the transitional pieces
(which have no status at all in traditional phonology) were there but
one of the phones was completely lost.

This suggests that words are stored and recognized as gestalt-like
sound contours, not as series of discrete units.  This is fairly
unremarkable from the point of view of a general psychology of
perception, but clashes with the 'digital' bias of much of 20th
century phonology.

If anything, this does not disprove the existence of phonemes;
phonemes don't even enter into it.  It casts doubt on the fundamental
status of phones themselves.

Oh, the imprecision I made before is that rather than speaking of
"intervals of sound recognizable as characteristic of particular
segments," I spoke of "intervals of clear tones" -- totally inaccurate
but easier to say, and I hoped people would divine my meaning.  But I
can see that that would give the impression that I was opposing
turbulent sound to non-turbulent sound, rather than 'identifiable as a
single phone' to 'not identifiable as a single phone' sound.

Ed Heil ------ edheil@postmark.net
--- http://purl.org/net/edheil ---

Jim Grossmann wrote:

> Apologies if you already took this point into consideration, but here goes:
>
> "noise" has more than one meaning in language studies.
>
> In information theory, it refers to a frequent concomitant to a signal that
> conveys no information, or something like that.
>
> In phonetics and acoustics, however, it refers to aperiodic sound produced
> by turbulance at some constriction of the vocal tract.
>
> Most consonants are noisy in the phonetic/acoustic sense.    Since noise
can
> be spread over different ranges of frequencies, different noises can be
> discriminated and can carry information.
>
> Frankly, it's simply inaccurate to allege that the proponants of the
> existence of phonemes equate the tonal or periodic componants of sounds
with
> the part of the sound that carries the information.   I have had a number
of
> academics tell me that the consonants carry most of the information, and
all
> of these academics believed in phonemes.
>
> Are we tripping over the multiple meanings of the word "noise"?
>
> Jim
>
>
> >> That doesn't mean that phonemes don't exist!  As you pointed out,
> >> there are three fairly pure tones.  Those are the defining
> >> characteristics of the word.  However, they're quite brief,
> >> especially stops, far too brief to perceive, so humans use that
> >> noise to figure out what the sounds are.  [b] [a] [d] do exist,
> >> but we infer the consonants that from the "noise" surrounding the
> >> vowel.
> >
> >If I can identify a word by listening to X, but not by listening to
> >Y, it seems perverse to insist that Y is signal and X is noise.
> >
> >But you seem to have firm opinions on this, and all I have is a
> >half-remembered anecdote, so I am not prepared to argue it.
>