Patrick Ryan wrote:
> What, precisely, bars a development /m/+/d/ > /Nd/?
Nothing. In fact, that change, and related ones, has taken place in many
Austronesian languages. In one case, in the Moluccan area, it affected
conjugated verbal forms:
*ku+táNis 'I weep'
*mu+táNis 'you weep'
*na+táNis 'he weeps'
by pretonic V-deletion, these became **ktaNis, mtaNis, ntaNis, then by
assimilation (and other rules) > taNi, daNi, daNi (even later taNi, raNi
raNi)--most of the modern languages have generalized one form or the other.
Similarly, the stative prefix *ma- lost its vowel in pretonic position,
producing a variety of m+C clusters; in the case of m+stop, then **m
assimilated to the stop position, leading to m+b,p, n+t,d, N+k,g etc.
This last change produced, originally, initial prenazalized phonemes in
Proto-Oceanic, usually on verbal forms, though with lots of variation as to
whether a given language shows a reflex of plain stop or prenas.stop.
Once prenas.stops were permitted in initial position, an interesting
development arose by analogy (at least according to one hypothesis): given
a form like "tapa" perhaps a noun or adj. meaning 'XXXX', a doublet "ntapa"
was created to mean 'something similar to, or, a different aspect of, XXXX'.
Sometimes the prenasalization seems entirely random-- e.g. while many, many
Oceanic languages reflect the word for 'ear' (*taliNa) with a reflex of *t-,
Fijian exceptionally has "daliga" [ndaliNa], with /nd/ < **nt.
***
(PR)
> What, precisely, in your opinion, is the phonetic difference between
> /n/+/d/ and /Nd/?
...
> [W]hat criteria should we use to judge whether the [nd] is phonemic (/Nd/)
> or simply compositional (/n/+/d/)?
It depends, I suppose, on how the language is analyzed phonologically,
whether nasal+stop is considered a cluster or a unit. That in turn might
depend on privilege of occurrence. If a language like Fijian can be analyzed
as CVCV, and N+stop can occur both medially and initially, then it follows
that N+stop should be considered a unit--as indeed the Fij. writing system
does: written "b, d, q" are phonetically [mb, nd, Ng] ("g" is plain [N]).
(Whereas in related Malay/Indonesian, where homorganic N+stop are permitted
only in medial position, and word structure is CV(N).CV(C), they are best
viewed as clusters. That apparently was also the case in Proto-Austronesian,
where, besides medial *Nasal+C, a few other sounds could occur before medial
C. [also in Ml/Indo.])
***
(Brian):
> Pre-nasalized stops, in those languages in which they
> _actually_ occur are phonemic.
(PR)
Which says nothing about the source of the phoneme.
(ME)
True, it doesn't. In my AN exs., they do arise from original clusters that
became phonemic through reanalysis of the basic word structure, from
clustered **N+CV(N).CV(C) to unit **[NC]V.[NC]V = CVCV, and the process was
then extended by analogy so that any initial stop could be prenasalized,
regardless whether it could originally be prefixed or not.
Bantu languages apparently permit initial clusters of non-homo. nasal + C
(like mtu 'person, man' IIRC)-- in this case, of course, the m- is a
separate morpheme (classifier) and (I think) syllabic. I guess our Engl.
"mgonna ~Ngonna" would belong in this category. As for the Sino-Tibetan
exs., I don't know.
PR:
> But I am not aware that there is a formal phonetic difference between /nd/
> (actually /n/ + /d/) and /Nd/.
There is, but it's subtle and difficult to pin down :-( It seems to depend
on what used to be called "juncture" phenomena -- the old "nitrate" vs
"night rate" or "catch it" vs "cat sh*t" exs.; perhaps different cons./vowel
allophones, perhaps speed of articulation, and, essentially, where the
native speaker claims there's a syllable boundary. A Brazilian will surely
claim "samba" is [sam]+[ba], whereas a Fijian would call it [sa]+[mba] (Fij.
saba 'to slap'). I suspect a spectrogram would show a difference.
In initial position, the difference seems to lie in how prominent the nasal
articulation is-- whether it has any syllabicity (which I think would
correlate how quickly the velum is raised in the transition from nasality to
the stop release).
My apologies for not being more on-IE-topic.
Roger Mills