--- In phoNet@yahoogroups.com, "Richard Wordingham"
<richard.wordingham@n...> wrote:
> . . . Voiceless approximants are silent :)

By J. C. Catford's definition of approximant and resonant,
that's true of resonants, but not of approximants. Does
your definition differ from his? In his 'Fundamental
Problems in Phonetics" he writes

"In approximant articulation the articulatory channel
is somewhat larger, and flow through it is turbulent
only when voiceless, other-wise it is non-turbulent.
The cross-sectional area of the articulatory channel
ranges from about 20 mm² to around 80 to 100 mm². In
a typical approximant the articulatory channel is smaller
than the glottal area for voicelessness, but larger than
the mean glottal area for voice. It is possible that
voiced phonation for approximants is always 'tighter',
keeping the mean glottal area smaller, than for frica-
tives. Typical approximants are 'close' vowels, like
[i] or [u], 'semi-vowels' like [j], such as the y in
English yes, the voiced labio-dental approximant [v],
common in Indian languages, a very common type of
English /r/, and the 'liquid' [l]-sound of English and
most other languages. It will be found that all of these
sounds have the characteristic feature of approximants,
that is, they have non-turbulent flow when voiced; but
the flow becomes turbulent when they are made voiceless,
at about the same initiator power."
"As we pointed out above, the term approximant
was first used by Ladefoged in A Phonetic Study of
West African Languages (1964), where it is defined
(p. 25) as 'a sound which belongs to the phonetic
class vocoid or central resonant oral (Pike 1943)
and simultaneously to the phonological class consonant
...'. In his later work, Preliminaries to Linguistic
Phonetics (1971), Ladefoged revised his definition
to read (p. 46) 'Approximation of two articulators
without producing a turbulent air-stream.' Here he
tacitly eliminates the restriction of the term to
central resonant oral sounds, and indeed he includes
a lateral approximant [l] among his examples, and
later (p. 53) explicitly mentions such sounds. Our
definition of approximant differs from Ladefoged's
later definition only in that it implies an upper limit
to the area of the articulatory channel as well as a
lower limit: the approximant has non-turbulent air-flow
only when voiceless. This enables us to delimit the
category so as to exclude from it wide open front
vowels of the type [E, a], which are included in the
next class of stricture.
In resonant articulation the articulatory channel
is still larger, ranging from about 100 mm² upwards.
In a typical resonant the oral articulatory channel is
always larger than the glottal channel, whether for
voiced or voiceless phonation. In voiced resonants,
for example, more or less 'open' vowels, such as [E],
the trans-glottal jets of voice set the oral resonators
ringing, and the function of the open articulatory
stricture is merely to shape the oral resonating cavity
appropriately. In voiceless resonants, such as [E],
air-flow through the glottis is turbulent, and in this
case it is the turbulent hiss of audible breath that
sets the resonators ringing. In voiceless resonants
there must always be turbulent flow through the glottis;
if there were not, there would simply be silence. Flow
through the oral articulatory stricture may or may not
be turbulent. However,in a voiceless resonant the oral
channel is always larger than the glottal channel, and
consequently, even if ther is some oral channel
turbulence, the glottal turbulence is greater, and
its resultant hiss tends to mask the hiss of oral

(Note that the E above in "[E]" was an epsilon in
the original, and the v in "[v]" an upsilon.)