|From:||Vasiliy Chernov <bc_@...>|
|Date:||Monday, November 27, 2000, 21:01|
On Wed, 22 Nov 2000 20:30:28 -0600, Eric Christopherson
>On Wed, Nov 22, 2000 at 11:18:24PM +0000, Keith Alasdair Mylchreest wrote:
>Once again, Keith wrote:
>> I'm not sure I entirely believe this, [t']
>> I can manage, a stop with double articulation, alveolar and glottal, but
>> [(t)s'] ? and [T'] ?? let alone [hl'], I can't see how you can be makinga
>> stop in one part of the vocal tract and a sibilant/fricative in another
>I have a real hard time with those too, but I've been assured that they do
>exist. I try to approximate them by pronouncing e.g. [s] with [?]
>immediately afterwards (sometimes I put the glottal stop before _and_after
>it, which I intuit might be closer to the real thing, but I'm not sure).
Maybe, the following will help a little bit. The way I see it - a bit
simplicist and very subjective in many points, but seems to work for me.
1) Consider the difference between voiced and voiceless stops. It's easy
to notice that it's chiefly based on the timing of voice onset (which
you can comfortably control if you put your finger onto your Adam apple
and feel the vibration). In voiced the voice turns on early (before
opening the occlusion), and in voiceless - with a delay.
2) The same voice onset timing is involved in more subtle distinctions.
Thus, if it nearly coincides with the opening of the occlusion, you get
plain voiceless stop (e. g. [t] in _stop_); if it is delayed a bit more,
you get an aspirate (like [t_h] in _top_). If you have your voice on from
the very start of articulation, you get 'fully voiced' stops (typical
e. g. of most Romance and Slavic langs). If you try to turn voice on
noticeably earlier than you open the occlusion, you get a preglottalized
voiced stop - kinda inevitably.
See where I'm pointing? It's like a continuous gradient in which different
langs use different areas as phonemic features.
3) Now consider a system which uses more than two zones in this continuum
as phonemic features. More specifically, a system distinguishing two
rows in the 'voiceless' span. It's natural that it'll try to emphasize
the difference in voice onset timing in some way.
One way is to stress the delay in one series, producing a very strong,
distinct aspiration (maybe emphasized with some pharyngeal friction or
adding some affrication to the stop). Thus you get a lang opposing
'plain voiceless' to 'aspirated voiceless'.
An alternative strategy is to stress the sharp voice onset coinciding
with the opening of the occlusion. And here it's time to tell you about
4) the way I was tought to pronounce ejectives by a Georgian speaker.
His instructions (for glottalized [t']) were roughly as follows:
a) make your tongue to form an occlusion, as if preparing to pronounce
a [t], so that no air could pass through it;
b) prepare to push air out, gradually increasing air pressure in your
c) now make the air to break through very abruptly, jumping to the
fully voiced vowel articulation. If you manage it properly, you'll
feel something *like* a glottal stop coinciding with vowel onset;
d) you are nearly there. What you've got is an unnatural, hypercorrect
Georgian [t']. Now you only need to relax and repeat it with less tension,
without screaming like mad.
5) Some comments I'd like to add.
It is very common that glottalized stops are opposed to aspirates. For
example, Georgian 'plain' stops are in fact lax aspirates.
It is also common that langs having ejectives use them to render foreign
plain (unaspirated) stops (e. g. Ancient Greek loans in Georgian,
Ethiosemitic, Aramaic, etc.).
'Glottalized' is a feature more common for occlusives than for fricatives.
Glottalized affricates are more common than the respective fricatives. In
particular, it seems that all Semitic langs that preserve the glottalized
quality of their 'emphatics' realize [s'] as [t_s'] (which is often
obscured by the traditional transcriptions modeled on Arabic).
Glottalized stops are indeed different from clusters with glottal stop.