Piotr wrote
> There is an interesting argument about linguistic areas in Bob
Dixon's "Rise and Fall of Languages" (1997 [2nd edition 1999],
Cambridge University Press). Supposing, for the sake of the argument,
that Australia was peopled just once, ca. 50,000 BP, by a small
linguistically uniform group of people (say, a few boatloads) and that
any subsequent immigrations (before the arrival of the Europeans) were
not substantial enough to upset its linguistic structure -- how would
Proto-Australian have developed into the known Australian languages?
>
> The first phase would have been the steady growth of the initial
population and its geographic expansion, groups moving off into new
territory. After a couple of millennia the continent would have been
fully peopled and the relationships between its languages could have
been best captured through a family-tree model. In a word, we would
have had a single Australian family. Then a state of equilibrium would
have begun and continued for the remainder of the fifty millennia.
Languages would have died occasionally, and there would have been
small-scale splits from time to time, accounting for the observable
low-level subgroupings, due mainly to climatic changes and the
concomitant contractions and expansions of local populations. But the
total population would have been relatively stable and linguistic
diffusion (as well as other types of cultural diffusion) would have
led to the emergence of an areal prototype (not to be confused with
Proto-Australian):
>
> -- no fricatives or affricates;
> -- a rich set of place-of-articulation contrasts;
> -- a single row of oral stops;
> -- a nasal corresponding to every stop;
> -- a rich set of liquids (up to four laterals, two rhotics);
> -- a three-term number system in pronouns;
> -- nominative-absolutive inflection for pronouns,
absolutive-ergative for nouns;
> -- special avoidance styles in the presence of taboo relatives;
> -- etc. (plus a number of features shared regionally within smaller
overlapping diffusion zones).
On a related point, my lecturer in Aboriginal linguistics at the
University of Western Australia in 1970 claimed that the loss of
frictives and sibilants from Aboriginal Languages was due to the
widespread condition of Yaws, which leads to a constant sub-auditory
hiss, making these phonemes inaudible. I tend to be suspeicious of
his explanation but would be interested in what people on this list
think. Can phonemes change for this kind of reason? Are there any
PIE examples? (*TH, *DH for example).
> Lexemes are freely borrowed throughout Australia. Two adjacent
languages (no matter what their genetic relationship) typically share
about 50% vocabulary. This figure represents an equilibrium value,
which means that if two contiguous tribal languages that have split
recently share, say, 70% lexemes, processes such as the replacement of
tabooed words will gradualy bring the figure down to ca. 50%. Dixon
analyses those stabilising forces in his famous grammar of Dyirbal
(1972).
>
> The pattern produced by millennia of convergence in a situation of
equilibrium assumes the form of a common typological profile. There
are a number of small genetic groupings for which family trees are
reconstructable, but there is no evidence to justify higher-level
filiation (even Pama-Nyungan is a typological, not a genetic
grouping). If a common Australian family tree ever existed, diffusion
led to the complete erosion of its structure long ago. Therefore, the
derivation of the Australian languages from a single proto-language,
while not impossible, cannot be proved using linguistic data. If
Australia had been peopled by speakers of several languages belonging
to different families, they would have merged their typological
profiles during the period of equilibrium, producing exactly the same
kind of pattern.
>
> The reason why I'm describing the case of Australia in so much
detail is that it provides a prototypical scenario of processes that
have also occurred in other parts of the world. Your mesolithic
Balkan-Pontic "phylum" might be something like Pama-Nyungan, and the
more encompassing convergence area including virtually all the
languages (and cultures) of the ancient Middle East would be analogous
to the Pan-Australian grouping. In equilibrium areas traits
demonstrating common descent "diffuse away" over millennia and are
replaced by typological affinities resulting from areal contact and
convergence. If the term "Nostratic" corresponds to anything real, it
is probably the latter type of reality.
Thanks Piotr. This is my understanding of "Asianic", leading to
linguistic convergence between Tyrrhenian, Luwian, Neshite, Khattic,
Hurrian, Kartvellian and various Caucasian languages. Whether one can
add Elamite, Sumerian and Semitic into this melting pot, I don't know.
The questions lie in how much the two models "fit". I would suspect
that in the period from 12,000 BCE to 8,000 BCE a "tree" may have been
discernable, but thereafter the kind of areal typology would apply in
the Middle East, with a "tree typology" from 10,000 - 6,000 BCE still
discernable in the North Balkans/Pontic region, disappearing
thereafterwards.
Meanwhile we have the rapid expansion of agriculture in the Middle
East with a demic expansion from a core, and the process of
neolithicisation occurring for unrelated groups along the periphery of
expansion, leading in some cases, to full enculturation (so enabling
a strong degree of linguistic independence) and to rapid secondary
expansion. With these expansions, localised trees could be
constructed.
Eventually, with expansion of other languages (through trade and war),
this would tend to prune our trees so that eventually one twig may
survive (our linguistic isolates).
Thansk for the post, it makes far better sense to me than trying to
place everything into a single "tree-like" catch all (a la chez Glen).
Regards
John