"syllabicity"

[Ralf-Stefan Georg]

>But, unlike PIE, the Sanskrit
>1-vowel system is very nearly true, since Sanskrit simply has no other
>elements than /a/ that cannot be called a consonant, while IE has /e/ and
>/o/, and in my opinion also /a/.

I take it that I have distinguished myself earlier by ignoring these simple
facts, which of course I shouldn't have (I wasn't following the discussion
attentively). At the same time I still feel uneasy with the label
monovocalic system. While I understand the theoretical background, which
leads to this label here by taking into account that all other syllabic
nuclei found in Sanskrit are variants/allophones of phonemes which also
have non-syllabic variants, i.e. consonants by definition, the danger of
this label is that it may lead people to believe that there might be
languages with only one possible syllabic nucleus, the phonetic realization
of which would be entirely predictable from the context. Since this is
still not the case for Sanskrit, I keep to my unease with this notion.
Again: it is theoretically OK, no doubt about that, but still, the
*phonemes* which can, under certain circumstances, surface as "vowels" are
more than one, not just one, with /a/ being the only one which always
surfaces as a vowel.

In short, we should differentiate between two kinds of "monovocalic"
systems: one, where, as in Sanskrit, only one phoneme has only vocalic
allophones, but certain others have consonantic and vocalic ones, and a
system, where only one phoneme *can* have one or several vocalic allophones.
I still view the latter as typologically impossible, resp. unheard of,
while I admit (of course) that the former description fits the Sanskrit
data (and they won't go away by ignoring them, as Jens puts it rightly).

St.

Stefan Georg
Heerstrasse 7
D-53111 Bonn
FRG
+49-228-69-13-32