Typology and the phonetics of laryngeals

Jens Elmegaard Rasmussen jer at cphling.dk
Thu Apr 6 17:27:56 UTC 2000


On Tue, 4 Apr 2000, Ante Aikio wrote:

> [Quoting the 25 Mar 2000 posting by Herb Stahlke:]

>>   I'm puzzled about the near
>> absence of application of typology to the question of what the phonetic
>> values of the laryngeals might have been. [...]

> [Ante Aiko:] [...]
> I believe the IE loan words that show laryngeal reflexes in Uralic may
> tell something about the phonetic values of laryngeals. Since there are
> etymologies that show such substitutions as 1) *h[1-3] > Uralic *k, 2)
> *h[1-3] > Uralic *x (read *x as [Y] =gamma), 3) *h[1-2] > Uralic
> (retroflex) *S, it seems probable that some [x]-type sounds must be
> reconstructed (/x´ x xw/, perhaps?) Such phonetic values as e.g. [?] for
> *h1 proposed by e.g. Beekes 1995 seem problematic to me; a substitution
> [?] > [k] seems perfectly possible, but [?] > [Y] does not, let alone [?]
> > [S].

I believe the facts of IE are plain in themselves, and make very good
sense typologically as well. A voiced value of /H3/ is demanded by
*pi'be/o- 'drink'; the assimilation in Germanic *kwikw-a-z 'quick' demands
its being velar, and its o-coloration makes it round, ergo this was a
"rounded gamma", a voiced labiovelar fricative. The cases of /H2/
surfacing as /k/ (costa, koza), and its Anatolian reflexes makes it a
dorsal fricative, evidence like Skt. gen. patha's makes it voiceless and
h-like, and its a-coloration demands some degree of aperture; in sum, this
was a voiceless velar or postvelar fricative without lip-rounding. Both
/H2/ and /H3/ cause a preceding high vowel to undergo lowering in its
final part in Greek, Armenian and Tocharian, while /H1/ does no such
thing: the products of /iH2, iH3/ are Gk. /ja:, jo:/, Arm. /ya/, Toch.
/ya:/, while /iH1/ simply yields /i:/. That means no particularly marked
degree of aperture. Note that also a preceding /o/ is left unaffected,
/oH1/ giving simple /o:/ all over the place; this means no particularly
marked degree of closenes either. The riddle is solve by the observation
that, after /H1/ and /H2/, but not efter /H3/, the suffix *-tlo-/*-tro-
turns up aspirated: Lat. cri:brum, fa:bula, but po:culum. It is customary
to derive Lat. -br-/-bl- and Gk. -thr-/-thl- from IE *-dhr-/*-dhl-, but
Birgit Olsen suggests IE *-thr-/*-thl- with voiceless aspirates which are
not excluded by any material we know. Thus, both H1 and H2 aspirate, at
least a _following_ /t/, and /H1/ causes no "breaking" in the development
of i/u + H, while H2 and H3 do. The only phonetic value that accomodates
all of these observations is a plain [h], i.e. a simple voiceless
continuation of the preceding vowel.

   The set [h], [x], [{ghw}] is very close to Dutch which has [h], [x] and
[{gh}], while in many languages gamma is rounded, cf., e.g., the
development underlying the orthography of English law, corresponding to
Danish lov (Swedish lag). I have therefore bee preaching this set for
quite many years (on record since 1982, I see), often against heavy
criticism. I now see the very same set advocated, without explicit
reasons, in Meier-Bru"gger's Indogermanische Sprachwissenschaft. Has the
gospel been heard? At any rate, the agreement is nice to see.

Jens



More information about the Indo-european mailing list