Re: [tied] PIE *ts ?

On Sun, 23 Feb 2003 13:26:03 +0000, "Glen Gordon"
<glengordon01@...> wrote:

>Miguel:
>>Yes. I thought about tt -> tst, but there is no way of telling
>>how old this phenomenon is in relation to the possible borrowing
>>of "6" from Semitic.
>
>Actually, there is a way. The Semitic borrowings must necessarily
>have been introduced into IE during the neolithic. So, we know
>the general range to be between 6000 and 4000 BCE for *sweks and
>other such loanwords. In fact, 6000 BCE is probably more precise,
>if I do say, but you can reach your own conclusions. It must
>surely have been well enough before 4000 BCE, if this is to be the
>approximate date of the fracture of IE.

Assuming for a moment that *swek^s was also in Anatolian, yes: the
word must have been borrowed before the breakup of IE. That's the
terminus ante quem, which I would also place in the 4/5,000 BCE area.

The terminus post quem is more difficult to assess. Do only Neolithic
peples have use for a word for "six"? I have no hard evidence for
that.

>However, the affricatization of *tC to [tsC] and *dC- to [dzC]
>would seem to date to a time after which the voiceless "fortis"
>*t: became voiced *d. I doubt that *dC- would develop as readily
>if *d were still a voiceless inaspirate. (By the way, the
>topological considerations such as a lack of **b that are used
>to prove that IE *d was in fact *t? by the Glottalic IE
>camp show that, at the very least, *d must have once been
>unvoiced in Pre-IE.)
>
>So, we can pinpoint this to the early Late IE period (approx. 5000
>to 4500 BCE). This would be a reasonable date as well because of
>the fact that this phenomenon appears to have existed in an
>as-yet-unfractured IE and yet it cannot be terribly early since
>it doesn't appear there was enough time for the affricate to
>become a distinct phoneme yet.

The [ts] in the combination /tt/, /dt/ is common PIE (it's in
Anatolian), so the terminus ante quem is the same as for *swek^s.
Again I'm not confident about any kind of terminus post quem, although
the fact that it doesn't have phonemical status makes it probable that
it was not _very_ long before the breakup of IE. On the other hand,
subphonemic allophones can remain stable for thousands of years.

>In a nutshell, I really doubt that there was a dental affricate
>in IE at the time when *sweks was borrowed. The dental affricate
>*c that would explain the t/s alternation would also have already
>eroded to *-s and *-t- by then. Afterall, by lacking a dental
>affricate, we can see why *ks would be used to pronounce the
>correlating Semitic phonemes of that word.

It is likely that the t ~ s alternation passed trough a stage /t/ >
/ts/ > /s/. In my own opinion, it was specifically /tW/ > /tsW/ >
/sW/, and the fact that Semitic /s^/ was borrowed as PIE *sW shows
that the stage /tsW/ had already passed.

>Miguel:
>>But Toch. tkam. and possibly Hitt. tekan show that this *ts was
>>definitely post-PIE.
>
>Doesn't Hittite /tekan/ come from an inanimate *dHegH(o)m?

The plene spelling te-e-kan suggests that the source is indeed
*dhég^hom, but *dhg^hóm (like elsewhere in IE) cannot be fully
excluded (which is why I said "possibly").

>Why can't Tocharian /tk-/ come from an affricatized *dHgH-?

Tocharian A has tkam. < *dhg^hó:m and cka:car < *dhugh&2té:r, while
Tocarian B has kem. and tka:cer. If we accept Winter's theory that *d
gives /c/ in Tocharian, and that *dh..Ch.. > *d..Ch.. (Grassmann's
law), then Tocharian A has *dhugh&te:r > *dugh&2te:r > cäka:ce:r >
cka:car as expected, and Toch. B. has simplified the cluster ck- to
tk-. But what happened to the cluster *dhg^h-? If something like
Grassman or Bartholomae had applied we would expect A *ckam. and B
*tkem.. We also don't have metathesis (dhgh > ghdh) as in Greek
khtho:n. Ungrassmanized unmetathesized *dhgh- would have given Toch
A. tk- and B. with simplification k-, which is exactly what we find.
What would affricatized *dzgh have given? I don't know. I'd expect
ck- or s'k-, but it doesn't matter. Occam's razor (!) tells us that
unmodified *dhgh- already gives the correct result, so why complicate
things with affrication?

=======================
Miguel Carrasquer Vidal
mcv@...