[An-lang] Proto Dialect chains

[Andy Pawley]

Here are brief responses to some points made by commentators in the
last few days.

1. The discussion has begun to spread in more directions than I can
handle. As was mentioned earlier, my main concerns were with
(a) different notions of what counts as a 'subgroup', where the key
variable is, specifically, the pattern of distribution of known
innovations over a set of languages and geography, and
(b) the implications that different distributional patterns have for
drawing inferences about regional variation in earlier stages in the
history of these languages.

I grant the existence of variation between idiolects, styles and
social dialects but I am not concerned with these in this context
insofar as they do not have geographic correlates. I grant that it is
often hard to establish innovations but that is a whole other issue
and I wanted to keep it separate.

2. John Terrell refers to a recent paper:

Terrell, J. E. , Hunt, T. L., J. Bradshaw (2002). "On the Location of
the Proto-Oceanic Homeland."
Pacific Studies 25(3):57-93

noting that this "explores the ins & outs" of Malcolm Ross's
notions of innovation-defined subgroups' vs 'innovation-linked subgroups'"

Coincidentally, when the Terrell et al. paper appeared a couple of
months ago I'd just written an essay with a very similar title
('Locating Proto Oceanic).  I'll attach the text as a PDF file. (The
maps can't be included at the moment, for technical reasons, but I
can mail them to anyone who is desperate.)  This essay is an
introductory chapter to the The Lexicon of Proto Oceanic vol. 2. The
Physical Environment, by M. Ross, A. Pawley and M. Osmond, which is
in press with Pacific Linguistics. It should be out later this year.


3. Isidore Dyen writes:
>>I see no difficulty with regarding your P-S example as a
subgroup, provided, of course, that P-S represent different languages.
Languages are dialect geographically discrete and thus each one has a
boundary with every other language. If you do not know whether the P-S
members are discrete then you are involved with a dialect geographical
problem which needs to be resolved because if the so-called languages
are not discrete, those pairs or triplets ....that are not separated
from each other constitute a single language and therefore belong
together by virtue of that fact; their exclusively shared traits (EST)
to the extent that they are innovations can be taken only as far back
as an earlier stage of their language, not to a proto-language that
that language might share with other languages, if any, in your P-S
set. If each member of your P-S set is a (discrete)language their
respective boundaries more likely occurred sequentially. Since they
became different languages upon the formation of boundaries among them,
the times of the formation of those boundaries determine the sequence.
If the formation of a boundary between two languages depends on the
total disappearance of  cross-pairs of mutually intelligible speech-
types (or idiolects), then it is clear that the simultaneous appearance
of boundaries is highly unlikely. Thus it is a convention (underline)
in comparative linguistics to treat boundaries as occurring
simultaneously if the evidence does not permit assigning a sequence.
This convention need not be confusing, but there can be controversy in
adjudging the evidence.

-- These points seem unexceptional to me, except for the conclusion
drawn in the first sentence. A subgrouping hypothesis is a hypothesis
about a period of unified development in the past. My view is that if
there is no innovation common to all the members of P-S, but only a
pattern of overlapping innovations linking the four languages then
P-S is, plainly, a subgroup only in a weak sense. The evidence in
this case indicates that, far from being unified, there was
well-marked regional variation at the time(s) that P-S became
discrete languages, with extreme ends of the chain showing no unified
development. For my purposes, it is useful to distinguish between
different degrees of unified development.


4. Waruno Mahdi comments:

>>So, strictly speaking, in instances with overlapping alignments
as described by Andy Pawley, at best only such with one certain
distribution pattern could theoretically be innovations of a
common precursor protolanguage. In all the differently distributed
patterns, these cannot be innovations in the actual sense which
we think we are implying when we use the term. The question, then,
is, which, if any, of the various observed distributions are we to
assign to an innovation?

>>I think, it would perhaps be more consistent if we stopped using
the term 'innovation' but talk of 'common features' instead.
It's less sexy, I agree, but at least we wouldn't be potentially
deceiving our own selves.

-- If I've read Waruno right, this is a remarkably pessimistic view,
which I can't agree with. We all know that establishing how languages
have changed over time and space can be a very tricky business and
that one should always proceed with caution. We start by noting the
pattern of 'shared/common features' without presuming always to know
which ones are innovations. (Though in some cases we do know
immediately, because some basic facts about the history of the
language family are firmly established.)  But we don't leave it
there. There are well-tried methods which can (sometimes) sort out
innovations from retentions and borrowings. Waruno is saying that
there are certain circumstances where these distinctions are
impossible to make with any confidence. Fair enough, but let's not
throw the baby out with the bathwater.

Andy Pawley
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/an-lang/attachments/20030722/e64e1ecb/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Locating_Oceanic_PDF.pdf
Type: application/pdf
Size: 70101 bytes
Desc: not available
URL: <http://listserv.linguistlist.org/pipermail/an-lang/attachments/20030722/e64e1ecb/attachment.pdf>
-------------- next part --------------
_______________________________________________
An-lang mailing list
An-lang at anu.edu.au
http://mailman.anu.edu.au/mailman/listinfo/an-lang