Re: synthesis index (was: Of of)

On 4/5/06, Yahya Abdal-Aziz <yahya@...> wrote:
>
> On Tue, 4 Apr 2006 Peter Bleackley wrote:
> >
> >
> > By synthesis index I mean "Average number of morphemes per word".
> > For this
> > definition, 1 is an entirely isolating language, and I think that 8 would
> > definitely be polysynthetic. 28 would be frightening.
> >
> Wouldn't it, though!
>
> A post on Zompist pointed to a message
> on CONLANG by Dirk Elzinga, discussing
> Greenberg's usages, which help somewhat
> to distinguish agglutination from synthesis:
> http://listserv.brown.edu/archives/cgi-bin/wa?A2=ind0309C&L=conlang&P=R27417
>
> The synthesis index is there taken to be
> the average number of morphemes per word.
>
> Interestingly, Dirk wrote that he decides
> "impressionistically" whether to call a
> language "polysynthetic".  While I understand
> where he's coming from, linguistics as science
> does need objective measures.  The Greenberg
> measure is obviously subject to where the
> orthography chooses to place word boundaries.
> I wonder if there's a better?
I wonder, too. Since the Greenberg method uses a continuous scale (or
set of continuous scales), it isn't possible to find a principled
point at which a language is no longer merely "synthetic" but has
attained the status of "polysynthetic". Greenberg suggests that a
language scoring 3.00 or above on the synthesis index (average number
of morphs per word) be called "polysynthetic". But a language with a
score of 3.00 or above could very well be missing some of the
morphosyntactic features that are usually associated with
polysynthesis such as  head-marking morphology, object incorporation,
an open class of bound morphemes, object and subject agreement, etc.

As for word boundaries, speakers of a language are generally in
agreement as to what constitutes a "word" of their language. If a
Greenbergian typological examination is being done using texts that
have been "vetted" by native speakers, then I think the results can be
relied upon.

Dirk