----- Original Message -----
Sent: Tuesday, December 28, 1999 12:46 PM
Subject: Glottal theory and other points
Manuel wrote:
> What are the fundamentals of the glottal
theory? Is it typologically
> acceptable? - I remember your remarks about
nostratic's huge bunch of
> consonants.
> Is there compatibility
between glottal hypothesis and laringeals theory?
> How is it explained
the fade of *b? Is it similar than Celtic and Armenian
> case?
>
Personally I always find PIE a little bit strange regarding the consonants
> inventory: there was a complete development of stops while a very poor
> show-up of fricatives.
> By the way, how can an analytic language
become a synthetic one? I always
> have thought of this phaenomenon as
resembling an endless circle: synthetic
> (flective) become analytic
("atomic" like Chinese; then, in an agglutinative
> process finally
returns to synthetic. Am I right?
> Kindest regards and wishes of the
Happiest New Year
1. The glottalic hypothesis posits a different
reconstruction of the three PIE "rows" of stops -- different, that is, from the
classical triad of plain voiceless (*t), plain voiced
(*d), and aspirated voiced (*dh) series.
According to the best-known variant of the hypothesis, proposed by Gamkrelidze
& Ivanov, PIE had voiceless, voiced and glottalised stops; those belonging
to the first two rows had aspirated and plain phonetic variants (allophones) in
various contexts (*[t]/*[th], corresponding to
traditional *t, and
*[d]/*[dh] for traditional
*dh); the glottalised row (*t') is the
counterpart of traditional *d. "Glottalised" is usually
inerpreted as "ejective", i.e. accompanied by a closure of the glottal folds and
an upward movement of the larynx, so that air pressure builds up in the mouth
and the stop is released with an sharp "pop".
The glottalic hypothesis and the laryngeal
theory are in principle independent; the validity of the one does not depend on
the validity of the other.
There are a number of known languages with such
rows of stops as t--th--t' or
t--d--t'. or From a
typological point of view the glottalic theory is a very elegant solution. As
opposed to the traditional reconstruction, it offers a convincing explanation of
the absence of roots like *deg (glottalic
*t'ek') by proposing a constraint that prohibits the occurrence
of more than one ejective per morpheme. It also accounts for the absence or
rarity of PIE *b (*p'): languages with an
ejective series often lack a labial member of that row. The main problem with
PIE ejectives is that the comparative evidence is more consistent with the
standard reconstruction. The existence of "ample typological precedent" for the
glottalic system is not generally regarded as a guarantee of a more accurate
reconstruction. After all, many familiar sound systems display statistically odd
features. If you want my personal opinion, I consider the glottalic hypothesis
an interesting alternative proposal; but as long as it is justified primarily by
theoretical considerations, not by comparative data, it will remain an exercise
in paper linguistics.
2. Although PIE is sometimes described as having
only one fricative (*s), the "laryngeals" (or at least some of
them) were fricatives too. In my opinion two or three "laryngeals" are
required to explain the known correspondences and ablaut patterns. All phonetic
reconstructions of those PIE phonemes are somewhat speculative, but my personal
preference is for a system containing a glottal fricative (or "aspirate")
*h (for what is often transcribed *H1) and a
velar fricative *x (for *H2 and
*H3; perhaps a labiovelar fricative *xw for
the latter). Even a system with three fricatives (*s,
*x, *h) would be richer than that of Ancient
Greek (s, h) and as rich as that of Classical
Latin (s, f, h).
There are languages with large fricative
inventories; e.g. English and Polish have nine such phonemes each, with
four voiceless/voiced pairs plus one voiceless "odd man out". Ubykh (an
Abkhaz-Adyghean [= North-West Caucasian] language) is analysed by some experts
as having 80-85 consonant phonemes; more than 30 of them are fricatives,
contrasting as voiceless vs. voiced or plain vs. labialised, with pharyngealised
and palatalised counterparts for several of them -- a really formidable
inventory. I suspect some the Ubykh "phonemes" are artefacts of
phonological analysis. Unfortunately the last speaker of Ubykh (employed as
informant by three generations of linguists) died in 1992, so nowadays we're
restricted to secondary sources. Whatever the truth about Ubykh, its system was
complex enough to qualify as extremely odd -- that's for sure. At the
opposite extreme, there are many sound systems without any fricatives
whatsoever (as in nearly all Australian languages). All known languages have
stops, but fricatives (and affricates) are not indispensable.
3. You are right; function verbs in analytic
constructions often develop into clitic particles agglutinated to the main word
of the phrase and may eventually be grammaticalised as inflections. Then
inflections may be reduced phonetically and lost, and their loss is compensated
for by syntactic means, so that e.g. prepositional phrases replace case
forms.
Piotr