Re: PIE suffix =t in food?

> > > > Another good loan-giver here would be Germanic since their *s was consistently render'd with MPF *s^.
> > >
> > > Interesting. Predorsal *s?
> >
> > If by "predorsal" you mean "slightly backed alveolar" then yes, that seems to be a common realization for /s/ in languages with a single sibilant. It's what we have in modern Finnish, for example.
>
> Michelena argued for a similar situation in Latin on the basis of Basque borrowings ("Lat. /s/: el testimonio vasco", RFE 86:474-89, 1967). Most of these continue Lat. /s/ with the Bq. predorsal /z/, not the apical /s/. Thus Bq. zucu 'sap' < Lat. succum, gauza 'matter, affair' < causam, gaztelu 'castle' < castellum, etc.
>
> One apparent exception is Bq. soka 'rope', but since this is of Celtic origin, it probably entered Basque from Gaulish, not Latin. The evidence is meager, but it suggests that Gaul. /s/ was apical, which in turn makes it likely that there was another sibilant. That was probably not the reflex of *-st-. I think Bq. bost 'five' is best explained as borrowed from Celt. *bosta: 'palm, fist' (not the other way round!), so *-st- was maintained as a cluster in the Gaulish dialect which supplied loanwords to ancient Basque. More likely the reflex of IE *-tt- was not apical *-ss- in that dialect; perhaps it was interdental *-TT- or some kind of laminal, and sibilant crowding forced ordinary /s/ to be apical.

The situation in Old Finnish may have been similar as well, with *tts : ts > †TT : T motivating backing of /s/. This could then have spread eastwars, motivating the Karelian shift *s > s^.

Then again I'm not sure which variety of /s/ occurs in Estonian, Votic, Veps, etc. (even if all of them also having /s´/ and most also /s^/ predicts a front /s/) - the backing could also date back to Late Proto-Finnic, in connection to *s^ > h.

> > For completeness: around the same time *st was also being substituted by *s, and another notable exception is that masculine stems yield *-as : *-aha- regardless of origin (Gmc, Baltic, II).
>
> Word-final *-s must have sounded stronger, at least when words were uttered in isolation. Thus Basque has gorputz 'body' < Lat. corpus

against simple /z/ in other positions.

Possible. The standard explanation links this to consonant gradation, but it seems not very strongly supported (stem-medial *s does not alternate with *h). The fact that there is also a near-complementary distibution of *-s (mostly in *-as) versus *-h (mostly in *-eh) leaves me particularly suspicious.

> > > > Many (most?) papers from Koivulehto, who has been doing the most work in this area, are in German. This compilation is probably the best starting place:
> > > > http://www.sgr.fi/sust/sust237.html
> > >
> > > Thanks very much for this list. The most intriguing title is the one about words with Fi. -aav- corresponding to Gmc. *-aww-. If these came indeed from Gmc., they must have been borrowed before the Verschaerfung, when Gmc. laryngeal residues still constituted distinct phonemes, with the vowel-lengthening occurring later within Proto-Finnic.
> >
> > K. actually suggests just what stlatos noted: that there was no sound change as much as a length substitution due to *ww not being a valid cluster in pre-Finnic.
>
> Are there other examples of such a length-metathesis?

No, Finnic deals with geminates just fine in general, this would be more connected to avoidance of *w + continuant clusters. These were mostly resolved by metathesis (*ws *wr *wj > *sw *rw *jw are known) but that's obviously not possible here.

> > > But the Gmc. shift *-o- > *-a- had already occurred (otherwise Fi. *-oov- from Early PGmc *-oHw-, no?)
> >
> > Loaning before laryngeal loss would probably predict *-okk- or possibly *-ohv-. It sounds like you're thinking of the reconstruction of Proto-Uralic *-VxC- for Finnic *-VVC-, but this has actually been explained otherwise recently; and at any rate vocalization there would have been a much too ancient development to be fed by Germanic loans.
>
> I was thinking (tacitly) of MPF *-aaw- from EPGmc *-aHw-, regardless of the origin of NATIVE Finnic *-VVC-.

Anyway, no, I don't think there would be any reason to expect that as a substitution.

> > It would seem so, yes, but I'm not following how that matters for the aav~aww layer?
>
> I made the tacit assumption that PGmc laryngeals lasted about equally long in different positions, which is unwarranted. Odd things can happen with laryngeals.

Certainly. The original _olut_ "beer" even has a very curious doublet _kalja_ "ale" which seems older by certain features but younger by others.

_j.