PIE Accent (was Re: Stress placement systems)

On Thu, 21 Sep 2006 10:18:01 +0100, R A Brown <ray@...>
wrote:

>H. S. Teoh wrote:
>> That's very interesting. Perhaps they are relics of a pitch accent
>> system in PIE?
>
>That would seem to suggest it. But, alas, I have not kept up with the
>latest thinking on PIE. I note that 'Stress System Database' gives:
>{quote}
>Indo-European (protolanguage) 	12..89/1L 	Halle & Vergnaud 1987:72
>Syllable "weight" determined by lexical accent
>{unquote}
>
>I note "weight" is put between quotes which suggests to me that it has
>nothing to do with syllabic quantity. But there is no reference to a
>pitch accent.
>
>Maybe others on the list are more informed regarding PIE word accent.
To answer H. S. Teoh's question, they are indeed relics of a pitch-accent
system in latest PIE.

At the latest stage of PIE, the accent system was lexically and
morphologically determined.  To the best of my knowledge, there were no
productive alternations at that point.  Quantitative ablaut, however,
points to an earlier state of affairs, where the accent was both 1) based
on stress rather than pitch, and 2) was "fixed" in the sense of following
a phonological pattern.  Evidence in favor of this hypothesis includes the
following:

1. Forms that exhibited quantitative ablaut were a closed class.
2. Productive forms had consistent (i.e. non-alternating) full-grade
vocalism.  This is especially true for the thematic paradigms and the
causative verbs.
3. Syllabic resonants, *i, and *u generally did not receive accent.
Instances in descendant languages where they do are the result of
analogical levelling and/or "re-tonicity" (accenting a previously
unaccented word).

Late PIE accent was also not true lexical accent, but phrasal accent.
Cues for this state of affairs include the Greek grave accent, Vedic
accentless substantives, and the treatment of verbal accent in both Greek
and Vedic.  Given that PIE was primarily an SOV language, the unmarked
verb position was sentence/clause-final.  This position also tends to be
the lowest on the "prosodic slope", at least in indicative
sentences/clauses.  PIE indicative verbs, then, were treated prosodically
as proclitics.

- Rob