On 05-03-01 07:01, Andrew Jarrette wrote:
> My question is, what was the probable pronunciation of
> Proto-Indo-European "r"? Specifically, was it pronounced trilled, with
> vibration, as in the majority of today's languages, or was it more like
> the commonest English pronunciation of "r", an alveolar or
> retroflex approximant, with no vibration? I know that English is rather
> special in preserving the original pronunciation of PIE "w" (although I
> hear that Flemish Dutch and a variety of Danish also preserve this
> pronunciation), and I wondered whether English was also alone in
> preserving the original pronunciation of "r", or whether English changed
> the original pronunciation of "r".
There's a long tradition of regarding the alveolar apical trill as THE
archetypal rhotic (also in the historical sense) and all other rhotics
as modern corruptions. This is nothing more than a traditional
prejudice. I'm right now finishing an article on the pronunciation of
/r/ in Old English, where I argue that it was mostly something else than
an apical trill, and that it varied dialectally just as much as it does
nowadays. There's growing consensus among those who have tackled this
problem (cf. Denton 2003 "Reconstructing the articulation of Early
Germanic *r", _Diachronica_ 20/1, pp. 11–43, available online, I
believe) that Germanic /r/ was pretty variable.
There is a recent article by J.C. Catford (2001) "On Rs, rhotacism and
paleophony" in the _Journal of the International Phonetic Association_
(31/2, pp. 171–185), where a number of changes involving rhotics in
several different IE languages are discussed, and the myth of the
original trill is somewhat debunked on grounds of phonetic
implausibility. If you ask me, phonetic variability is so characteristic
of rhotics in general that it would be stupid to insist on any
particular realisation as THE primitive one. Trills and taps are common
enough cross-linguistically, but early IE /r/'s may have had any number
of approximant variants, including retroflex and dorso-midpalatal ones
(certainly RUKI would be easier to understand and formulate if the
Indo-Iranico-Balto-Slavic /r/ was something of the latter kind, at least
syllable-finally).
Piotr