Age of various language families

[Claire Bowern]

----------------------------Original message----------------------------

No doubt other Australianists will reply too so I'll keep this brief.

> Anyway, there are two points that I found especially intriguing. The first
> is that isoglosses in Australia display no bundling whatsoever (or so Dixon
> claims -- much of what I know about Australian languages comes from him,
> and I know he's controversial to say the least among Australianists). And
> in a well-behaved, nicely branching family, we would of course expect
> subgroups, and hence isogloss bundling. If his claim is true, this is most
> interesting. I recently tried to discuss it with a well-known Australianist
> who made me disappointed by simply saying that "we just _know_ that
> Pama-Nyungan is a family, and that Dixon is wrong", without being able to
> deliver a single argument. (If there are Australianists on the list, I'd
> love to hear the relevant arguments).

Many arguments for the status of Pama-Nyungan as a family stem from trying
to show that it is a subgroup of a higher-level family (proto-Australian
or something else). Dixon argues against this, quite rightly, in my
opinion, since the evidence for this is I believe limited to 'initial
laminisation' (some words with n- (apico-dental) in non-Pama-Nyungan
languages appear with nh- (a lamino-dental nasal) in Pama-Nyungan) - this
work is due to Nick Evans. Another piece of evidence of this type is the
forms that are reconstructible only to Pama-Nyungan languages and which
are distinct from a set of 'Northern' (ie, in many non-Pama-Nyungan
languages) pronouns. THis work was done by Barry Blake.

Recent soon-to-be-published work by Barry Alpher looks at
defining regular sound correspondences within Pama-Nyungan. He shows that
there are regular sound correspondences between Pama-Nyungan languages
that are not shared by at least some of the non-Pama-Nyungan families (eg
Gunwinyguan). There is also work of Alpher's on verb conjugations, and
further work in progress on this topic by Harold Koch.

Other common similarities in Pama-Nyungan languages, besides basic
vocabulary, include an alternatin between -ngu and -lu ergative suffixes
and -nga and -la locative suffixes, a dative -ku and a first person dual
(usually inclusive) of the form ngali.

Harold Koch and are in the process of editing a book of papers dealing
with (sub)grouping methodology and evaluating the evidence for someof the
lexicostatistical classifications which have stood (largely untested)
since the mid1960s. The papers were presented at ICHL in a workshop last
year. The book includes Barry Alpher's paper, mentioned above.

So Pama-Nyungan is based on a good deal more than Sprachgefuhl.

>
> The other point I found interesting is the one which provoked my recent
> question on the list. It goes something like this: Indo-European is
> generally believed to be X years old, and has split up into Y different
> languages. Mankind has been speaking for 20 (or whatever) times as long as
> Indo-European has existed. Therefore, if the splitting rate IE is
> representative, there ought to be (even if we take language death into
> account) umpteen gazillion languages spoken on earth today. Clearly, this
> isn't the case. So there.
>

I did some calculations to work out the amount of language death required
for us to reach a modern figure of approx 6000 languages if we assume
different rates of splitting. For the Indo-European rate and Dixon gives
it, one needs to assume an extinction rate of a bit under 42% per
generation. On his more modest splitting rate, the figure is about 12%.
That is, you have to assume that a linguistic tree bifurcates at the same
rate across its branches and at each generation 12% or so of the languages
die, allowing 'fractional' languages.  Now, I have no idea whether these
numbers are plausible estimations, how we would get enough reliable data
to test it, or whether the quesiton is even meaningful, but if you take
Dixon's model at face value, those are the rates required.

Claire Bowern


-----------------------------
Department of Linguistics
Harvard University
305 Boylston Hall
Cambridge, MA, 02138