I've read it. To sum up: - PIE originally had just two series of stops: voiceless and voiced - The voiced series had the allophones [B, D, G] and [b, d, g]. - The allophonic voiced stops became phonemic voiced aspirated stops. - /B/ merged with /w/ (He claims that evidence of this is shown by the unusually high number of roots that start with *w, which were formerly *b before merging.). /D/ and /G/ had nothing to merge with, and simply hardened to become phonemic /d/ and /g/. - He argues that the occurrence of Grassman's Law in both Greek and Sanskrit shows that it was originally a PIE thing, not a separate and identical innovation.
My thoughts/concerns: - I haven't seen any frequency data for PIE phonemes. I would very much like to. - With only 2 series of stops, that gives 2*2=4 types of CVC root forms. After the phonemicization, that gives 3*3=9 types, 3 of which are illegal. He doesn't say why roots of the form *deg became *dhegh, and why *teg and *dek didn't become *tegh and *dhek. As for why/how *degh and *dheg appear at all, he says they ""occur under highly constrained conditions, and are shown by Grassman (1863/1967) to be secondary developments of *dhegh."" What exactly these conditions are, he doesn't say. (Again, this is the important part, and really shouldn't be left out...) - He doesn't mention *kw and *gw at all; I assume he thinks that they were just allophones of *k and *g. - I'm not sure why he thinks Celtic fits into "group C" when PCelt had the *gw->*b and *gwh->*gw shift. - Ditto for Balto-Slavic with Winter's Law.
Afterthoughts: - I really want to see some frequency data (that also notes phoneme position in the root). Are the root forms *degh and *dheg really so rare? If they are, I haven't noticed it. -- I could write a program to do this, but I don't think I want to make this in an excel spreadsheet, which will mean it's a little less portable when I'd share it. -- Writing the dictionary of roots would be long and tedious, but necessary to get an fair view of the frequency. Also, I'd have to go through multiple dictionaries to get all the roots. (I'm eager for Lubotsky's Pokorny 2.0 project to be finished in a couple years so they'd all be compiled in a single dictionary.)
|