Herr Dunkel wrote:
Adam Hyllested did some fun work on Indo-Uralic
. It looks quite good, but I've only skimmed it instead of reading in detail.
It's a rather thought-provoking article with some clever new comparisions, but as usual, it has some problems with using roots found only in one subfamily, or irregularly behaving roots found in only one end of a family — which will be frequently loans rather than inherited even from the PIE/PU level.
In several cases superior loan etymologies have already been proposed. E.g. "PU *kurk- 'throat, neck'" (#32) is only attested in Finnic *kurkku ~ Mordvinic *kərga, which neither correspond to each other nor are possible root shapes for inherited vocabulary, and can be explained satisfactorily as Germanic (*kʷerkō) and Slavic (*kъrka, IIRC?) loans respectively.
WeepingElf wrote:What regards the laryngeals and why *h2 doesn't colour *o, this is easy to understand in terms of phonological features.
*h1 does not add or remove any vowel quality features; it only lengthens the vowel when it follows it.
*h2 deletes the feature [+front], thus changing *e into *a; as *o is already [-front], its quality doesn't change.
*h3 deletes the feature [+front] and adds the feature [+round], thus changing *e into *o; as *o is already [-front] and [+round], its quality doesn't change.
*h4, if it existed at all, behaves like *h2.
I'm seriously skeptical on if *h₃ can be considered [+round]. Mainly because I'm also skeptical on if this can be considered to have been distinctive for *o (versus *e and *a) at the time laryngeal coloring occurred. Coloring was clearly completed much earlier than the loss or possible mergers of laryngeals, and might mesh together with early vowel system developments in a complex fashion.
There's amazingly little evidence that *o was more round than *e; if anything, the widespread mergers with *a, no apparent examples of *kʷ → *k before *o specifically, and Brugmann's Law lead me to suspect that it's actually *o that was the low vowel /a/. *a itself, if this existed separately, would then have been something else like *æ or *ə. The development of these into /o a/ in Greek + Italic + Celtic would be a common type of chainshift.
The difference between the coloring effects of *h₂ and *h₃ I'd guess could then have been chronological rather than phonological. E.g.
1) *h₃e → *h₃a
2) *a → *o
3) *h₂e → *h₂a
Cf. English. OE /ēaɣ/ goes to ModE /aɪ/ in eye
, while OE /ēax/ goes to ModE /eɪ/ in eight
; this isn't because /ɣ/ was more strongly a-coloring, it's because this started on the whole vocalization → monophthongization → (re)diphthongization path earlier.