Conlang and natlang typological trends, standards and enclaves (was: synthesis index)

Hi taliesin & Jim,

On Tue, 4 Apr 2006 taliesin the storyteller wrote:
>
> * Yahya Abdal-Aziz said on 2006-04-04 08:57:29 +0200
[snip]
> > Well, I hadn't heard of the "synthesis index" before, so did a google
> > for it which returned thousands of (mostly chemical) results, but very
> > few to do with linguistics.  One of the few was Bickell & Nicholls,
> > Typological Enclaves, at:
> > http://www.uni-leipzig.de/~autotyp/download/enclaves@ALT5-2003BB-JN.pdf
>
> Now this was a very, very interesting presentation! Ignore the
> synthesis index stuff, it's about how language "trends" spread, and had
> some examples of weird stuff that I hadn't seen before.
>
> So, are our conlangs like the enclaves or like the area standard?
Yes, I agree that this is indeed fascinating stuff!
The map presents quite a picture, that appears to
support the enclave hypothesis.  Interestingly, the
other reference I quoted (Bickell) also includes a
map of his synthesis index, based on N = around 130
cases rather than 199, and the index values seem at
first glance to have changed a bit between versions
of the map.

What concerns me here is to be sure that what
Bickell is mapping actually represents an actual
phenomenon and not a mere statistical artefact.
Is his measure stable? objective?  And if it is these
things, what does it explain, and under what
assumptions?

Further, the construction of the index as a sum
of two different counts makes me wonder: does
it confound two different phenomena, which might
have more explanatory power in a two-dimensional
model rather than a one-dimensional one?

Jim Henry also voiced other concerns I have when
he wrote:
[snip]
> > Typological Enclaves, at:
> > http://www.uni-leipzig.de/~autotyp/download/enclaves@ALT5-2003BB-JN.pdf
> > and another (found when I added "linguistics" to the
> > search terms) was Bickell, The Autotyp Research
> > Program, at:
> > http://www.uni-leipzig.de/~autotyp/download/ltrc02.pdf
> >
> > Bickell's definition is:
> > SYN = Nmax(categories) + Nmax(formatives)
> > (page 159 of the latter reference), and he shows a
> > map of its distribution for N (languages?) = 199 (page
> > 8 of the former reference) with values of SYN from
> > 0 to 28.
>
> I've read a fair part of that PDF now, and I'm still not sure
> what Bickell & Nicholls mean by their synthesis index.
> It seems like they might intend  a count of how many
> verbal categories are marked by inflexion in a given
> language rather than separate words, but I'm not sure.
> Does anyone else know what this
> "SYN = N (categories) + N (formatives)"
> means?
The terms "category" and "formative" certainly
need a clear definition, which I haven't seen yet
either.

Assuming that the Bickell synthesis index does
actually reflect a real phenomenon that we can
sum up qualitatively in the contrast between an
"area standard" and an "enclave", I think it an
important task to explain the phenomenon.  The
simplest hypothesis that I can think of is one we
might call "the trade hypothesis" - that the degree
of language feature change (reflected in the BSI
- synthesis index) is a direct consequence of the
degree of interaction of one group with another.
In those terms, the "Pan-Eurasian area's" enclaves
would have arisen by the relative isolation of those
groups from the other groups.  If I'm right, this
isolation should also be reflected in other macro
linguistic features besides the BSI, such as the
agglutinativity index AI.  Anyone got a map of the
AI?

Then, too, I'd explain the existence of enclaves
and standards in conlangs also in trade terms,
albeit the trade is rather more virtual than physical
(*), mediated by the internet and very largely
between individuals and small groups rather than
between entire merchant classes of entire cultures,
and thus pre-eminently one of ideas rather than
things (*).

(* Conlang conferences and t-shirts aside!)

Such is my attempt at a "sociological" forecast of
conlang evolution ... without, at this stage, any
reference to class structure ;-)

Regards,
Yahya

--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.385 / Virus Database: 268.3.5/301 - Release Date: 4/4/06