"syllabicity"

[Rich Alderson]

On 29 Apr 1999, Jens Elmegaard Rasmussen <jer at cphling.dk> wrote:

>On Tue, 27 Apr 1999, Patrick C. Ryan wrote:

>> Rich continues:

>>>Thus, Lehmann violates a major principle when he asserts that any stage of
>>>Indo-European lacked a phonemic vowel:  If a phone is present in a language,
>>>it has a psychological status in the lexicon, and while it may alternate
>>>with other sounds in the language because of morphological rules or
>>>unconstrained processes, it cannot be denied phonemic status.

>I think he violates an even more fundamental rule: If a segment is opposed
>to zero, it exists! Thus, since even an extremist monovocalic IE phonology
>would oppose a 3sg in *-t to a 2pl in *-te, it must have a phoneme /e/.
>This of course does not detract from the stimulating effect of the book -
>just look at us!

I believe that we have stated the same argument from two slightly different
perspectives.

>[... PCR:]

>> But, why all the fuss about monosyllabicity when Sanskrit provides us with
>> the next logical outcome of a language that, at an earlier stage, was
>> monovocalic (at least, phonemically).

>I believe this is right even synchronically, barring words of marginal
>phonological integration: In Sanskrit,
>	[a:] is identical with /a/ +/a/
>	[i] is a realization of /y/
>	[u] is a realization of /v/
>	[i:] is identical with [i] + [i], thus a realization of /yy/
>	[u:] is identical with [u] + [u], thus a realization of /vv/
>	[r.] is a realization of /r/
>	[r.:] is identical with [r.] + [r.], thus a realization of /rr/
>	[l.] is a realization of /l/
>	[e:] is a realization of /ay/
>	[o:] is a realization of /av/
>	[a:u] is a realization of /aav/
>	[a:i] is a realization of /aay/

One can just as easily argue that [y] is a realization of /i/, [w] realizes
/u/, and so on, when the vocalic phoneme is collocated with another vocalic
phoneme, if one wants to accept this sort of deep-level phonology.  The better
analysis, in my opinion, is to accept that the vowels /i(:) a(:) u(:) e: o:/
are all phonemic, and deal with the other surface realizations as the result of
*morphophonemic* rules, not phonological rules or processes.

>Thus, in Sanskrit, short /a/ is the only true vowel demanded to allow an
>unambiguous notation of all (normal) words. This is a one-vowel system of the
>kind dismissed as a typological impossibility for PIE. - I rush to add that
>the acceptability of this analysis for Sanskrit does not make it correct for
>PIE which, for completely independent reasons, appears to need at least the
>vowels /a, e, o/ on the phonemic level - and even long /a:, e:, o:/ and
>underlying /i, u/ (opposed to /y, w/!) on an abstract morphophonemic level.

>In Sanskrit, as in PIE, the rules stipulating a given sonant/semivowel to
>appear syllabic or nonsyllabic are relatively clear. Such an element is
>nonsyllabic when contiguous with a vowel, otherwise it is syllabic. Only
>Sievers and a touch of analogy compromise predictability.

It is the compromised predictability that requires us to see both vowels and
resonants as phonemic, with some morphological rules creating interactions
between them, in a synchronic description of Sanskrit.

								Rich Alderson