From: Piotr Gasiorowski
Message: 19055
Date: 2003-02-23
----- Original Message -----
From: "Miguel Carrasquer" <mcv@...>
To: <cybalist@yahoogroups.com>
Sent: Saturday, February 22, 2003 6:23 PM
Subject: Re: [tied] Re: "Will the 'real' linguist please stand up?"
> There is, as far as I know, no /ts/ in PIE. This may be accidental, or it may reflect an earlier soundlaw with got rid of /ts/ in favour of, say, /k(^)s/. In either case, if /ts/ was not in the langauge when the word was borrowed, that may explain why it was adopted with /ks/ instead of /ts/.
Even if we leave aside "older" (pre-PIE) *ts, whose existence can be inferred through internal reconstruction (*t/*s alternations in comparatively reconstructed PIE), I think a case could be made for "younger" *ts as a marginal phoneme in late PIE (with restricted distribution, like that of /Z/ or /N/ in English).
Phonetic [ts] was the realisation of *t before another *t across a morphological boundary (and possibly before other voiceless stops): *-tt- [-ts.t-] (. = syllable boundary), and this [ts] developed exactly like the final cluster in words like *po:d-s [po:ts] (ending up as /s/ in most branches but as /t/ in Sanskrit and /ts/ in Hittite). I also suppose that the development of *-tt- > -ss- as in Italic, Celtic, Germanic and Albanian was via progressive assimilation: *-ts.t- > *-ts.ts- > -s.s- .
A monosegmental analysis of *ts could furthermore help to explain some "thorny" metathetic developments like *tk (> *tsk > *kts by metathesis) > ks, kt . In Indo-Iranian, this *ts may have participated in the RUKI change (below, *C and *J stand for palatal stops ~ affricates):
*kts > *kts^ > *ks^ (> Iranian xs^, Indic ks.)
*k^ts > *Cts^ > *Cs^ (> Iranian *s^, Indic ks.)
[dz] (as in [-dz.d(H)-]) could then be regarded as a voiced allophone of *ts.
In the "earth" word,
*dzg^H > *Jdz^H > *Jz^ (> Iranian *J > Av. z, Indic *gz^ > ks., but preconsonantally *Jz^m- > *Jm- > Indic jm-)
Piotr