Nicholas,

What you describe is called "acoustic phonetics." It goes back at least to Helmholtz, and the Bells (Alexander Melville, Alexander Graham) were pioneers. The "sound spectrograph" was developed during (or before?) WWII. and as soon as it was declassified, Martin Joos published a detailed exposition of its discoveries (*Visible Speech*, 1947). Major (near-)contemporary names in acoustic phonetics include Ilse Lehiste, Peter Ladefoged, J. C. Catford, and Ian Maddieson.

Crystal's *Cambridge Encyclopedia of Language* includes a full-page display of the largest consonant inventory known (a click language) alongside the smallest (Rotomas)..
--
Peter T. Daniels grammatim@...


----- Original Message ----
From: Nicholas Bodley <nbodley@...>
To: qalam@yahoogroups.com
Sent: Saturday, September 15, 2007 2:27:19 PM
Subject: Re: "FaYe" ... --> Human hearing, vowels, and consonants

On Sat, 15 Sep 2007 09:37:45 -0400, Don Osborn <...> wrote:

> Similarly, the info about 34 sounds may relate to something but not
> have any implications for the number of phonemes we can distinguish,
> nor for that matter the optimal number of characters in a script.
>
> But I would be interested to know more if anyone can clarify.

There's something important about human hearing that was discovered only
relatively recently, within the past few decades. Even in the late 19th
C., there was knowledge of the harmonic structure of steady-state sounds,
and even some ability to see their waveforms (by rotating-mirror polygons,
and fast-responding gas flames fed from sources that included acoustical
horns and diaphragms). It has long been realized that the tone color
(timbre) of a sound has a close relation to its harmonic structure --
their frequencies and relative amplitudes.

(Btw, *please* don't use the term "overtones" -- its use was discouraged
by Ellis, if not earlier; he translated Helmholtz' landmark 19th C.
treatise on acoustics. Use "partials", a term from acoustics. Harmonic
partials give steady sounds their character, and inharmonic partials, such
as those comprising the sound of a drumhead, for instance, also are
responsible for the character of those sounds.)

In addition to steady-state (and slowly-changing -- think of a
piston-engine airplane overhead) sounds, there are sounds that change too
fast for human ears to distinguish the structure of the changes as they
occur. These are called transients.

What was realized only compartively recently is that human hearing is
excellent at recognizing and distinguishing transients, even if we cannot
hear the details of them. Nevertheless, we can identify and distinguish
them very well. (I'm thinking "gestalt". :) )

I think it was Bell Labs. that did a remarkable experiment. They recorded
steady notes (most likely of the same pitch) played iirc on a violin and
on a trumpet. They took the recordings, and edited out the beginnings of
the notes, iirc fading up the volume of each. Most listeners were unable
to distinguish which instrument was which!

Perhaps mostly from that experiment, perhaps not only that, the scientists
learned that the onset -- the beginning -- of a sound is extremely
important for us to identify it. The onset of a sound usually contains
lots of transient information, and our hearing is very good at analyzing
that.

What's quite likely to be worthy reading:
<http://www.gi.alaska.edu/~Roederer/> (Bcc of this message to him, hoping
I'm not being too pretentious!)

See "To come soon!!". I bought and studied his first edition, and it was
quite memorable, as well as part of my self-education. It's possible that
Roederer invented the term "psychophysics" -- it's both.


+=+

It's no great leap to realize that vowels are comparatively steady-state
sounds; there were scientific devices made in the late 19th C. to create
vowel sounds, and that was before electronics.

Correspondingly, consonants tend (imho) to include transients, at least
some consonants (clicks, in particular!). Considering that human ears can
"analyze" transients so well, it's believable that they can recognize
quite a variety of consonants.

As to how many vowels we can distinguish, I suspect quite a few; consider
languages that have many vowel sounds (Swedish?), not to forget tonal
languages.

I'd conclude that that 34, or whatever it was, is unrealistically small,
although the process by which the author arrived at that number was not in
any sense dishonest; he (she?) might have been thinking differently from
what most of us assume.

I think there could be a very-worthwhile collaboration between linguists
and acousticians, although I suspect those titles are too generic. Don't
be scared off by mentions of the work of J. B. J. Fourier -- analysis,
synthesis, and diect and inverse transforms. If you don't understand, it's
not your fault! Fourier's concepts can be understood, and without any
mathematics to speak of. However, it does take an uncommonly-adept person
to do the explaining, someone to whom it is second nature to understand
the backgrounds of the audience.

My apologies for remaining typos.

HTH, a bit!

--
Nicholas B o d l e y
Waltham, Mass.
dilettante linguist
dilettante acoustician


groups.yahoo.com/group/qalam - world's writing systems.
To unsubscribe: qalam-unsubscribe@...
Yahoo! Groups Links