Re: [tied] Re: Glottalic thought-experiments

From: Miguel Carrasquer
Message: 20706
Date: 2003-04-02

On Wed, 02 Apr 2003 18:16:57 +0200, Piotr Gasiorowski
<piotr.gasiorowski@...> wrote:

>> <t`> is -voiced +aspirated, and <t> is -voiced -aspirated, or, equivalently, <t`> is +voiceless -glottalized, <t> is +voiceless +glottalized.
>[+/-voiceless] (applied to obstruents) is a suspect feature from the point of view of phonological theory.

I know, but I thought I might as well invert both terms.

>But if we use [-/+voiced] instead, then [+glottalised] becomes an alternative way of saying [-aspirated]. The latter may be preferable if phonetic aspiration is obligatory while phonetic glottalisation is optional.

That seems to be the case in Modern (Eastern) Armenian. I suspect it
was also the case in Classical Armenian, although that's of course
impossible to prove. However, it's imaginable that over time, the
markedness of the feature can go back and forth between +/-aspirated
and +/-glottalized.

Does anyone know about a language that has ejective stops but no
voiceless aspirates? I suspect there is no such language.

In relation to evidence from Caucasian and American Indian lgs., while
searching the web for information about the degree of glottalization
of the Georgian ejectives (e.g. as compared to the Armenian ones)
[nothing found], I came upon the following abstract:

Sonya Bird, University of Arizona

Acoustic Properties of Lheidli Ejectives and their Effect on
Documentation Work CANCELLED

My paper focuses on the phonetics of the Lheidli ejective series, and
on how the difficulty in distinguishing them from voiced stops affects
through errors in their transcription documentation and consequently
language preservation efforts.

Lheidli is a dialect of Dakelh (Carrier) Athabaskan, spoken fluently
by only 4 elders in the northern interior of British Columbia.
Although there currently exists hardly any written material on
Lheidli, a project is under way to train a group of semi-speakers to
record interviews, transcribe and translate them, and eventually use
the materials collected as the basis of a dictionary, grammar, and
other resources. One of the issues that has come up in training the
semi-speakers is the lack of systematicity in their transcriptions:
words are often misspelled, or spelled differently from one token to
the next. This is due at least in part to the nature of the sounds

Like other Athabaskan languages, Lheidli has an ejective series ([t],
[k], [kw], [tl],[ ts], [tsh]). However, the acoustic and articulatory
properties of these sounds differ from those in other languages. As a
result of their unique features, Lheidli ejectives are perceptually
very similar to their voiced, non-ejective counterparts. For example,
it is often difficult to distinguish [t] from [d]. In this paper I
describe the Lheidli ejectives, and compare them to their Navajo
counterparts, which are more typical. Using acoustic data, I show that
Navajo ejectives involve larynx raising as well as glottal closure,
resulting in a clear ejective sound as the pressure behind the closure
in the oral cavity is released. In contrast to this, Lheidli ejectives
do not involve larynx raising, such that when the closure is released,
the effect is limited to some creaky-voicing at the onset of the
vowel. I account for the differences in the ejectives in these two
languages within current phonetic and phonological theory.

Having discussed the acoustic properties of Lheidli ejectives, I
present transcription data showing that the difficulty in perceiving
ejectives often leads to their being transcribed as voiced stops
instead. This has serious consequences for documentation work (and
language preservation) because many of the words in question are being
written down for the first time, to be used as a basis for the
creation of written resources on the language. If all ejectives end up
being transcribed as voiced stops, it will appear to future language
learners and researchers that Lheidli never had an ejective series.
This raises important questions about the goal of documentation work.
Should it be descriptive - i.e. capture only those distinctions that
are perceived by the remaining Lheidli semi-speakers - or should it be
prescriptive - i.e. capture the distinctions that we know have always
existed in the language? If we choose a prescriptive approach, how can
the semi-speakers be trained to distinguish between ejective and
voiced stops in their transcription? In the second part of my
presentation, I consider answers to such questions.

Miguel Carrasquer Vidal