Re: Agglutinativity Index (was: Re: What's a good isolating language to look at)

On 12/8/05, Jim Henry <jimhenry1973@...> wrote:
> On 12/8/05, Thomas Hart Chappell <tomhchappell@...> wrote:
> > --- In conlang@yahoogroups.com, Jim Henry <jimhenry1973@G...> wrote:
>
> > > John C. Wells, in _Lingvistikaj Aspektoj de Esperanto_,***
> > > quotes the Greenberg article and calculates indexes
> > > of synthesis and agglutinativity for Esperanto.
>
> > Are these ratios and averages "by type" or "by token"?
> >
> > That is, for the agglutinativity index, do you count each morpheme
> > just once, no matter how many times it occurs in the text; or do you
> > weight more-frequently-used morphemes with more weight?
>
> > (For the synthesis index, the question would be, do you count each
> > _word_ just once, or weight more-frequently-used _words_ with more
> > weight?)
>
> I just re-read the passage in Wells 1989 and can't find anything
> explicit one way or another.  Perhaps someone with
> access to the original Greenberg 1960 article can answer this.
Greenberg used passages of 100 words to calculate the value of various
indices, including the synthesis index. So it seems to be tokens
rather than types. This was my procedure as well; for Shoshoni the
passage was the story "Coyote and Mouse" (from Crum and Dayley's
Western Shoshoni Grammar), which ran to 500 words. For Miapimoquitch
the passage was the story "Eye Juggler", which ran to 58 words
(obviously not really a large enough sample to get a reliable number).

Greenberg does give sources for the texts; the Eskimo text (Henrik
asked about it) comes from Thalbitzer's contribution to the _Handbook
of American Indian Languages, Part I_, edited by Franz Boas (1911). I
don't know which variety Thalbitzer examined.

> (I just realized no one has cited the Greenberg article's title etc
> in full; Wells gives it as
>
> A quantitative approach to the morphological typology
> of languages.  _Int. J. American Linguistics_  26.3.178-194.
That's the one.

Dirk
--
Gmail Warning: Watch the reply-to!