Re: Quantitative Methods in Linguistics

From: Richard Wordingham
Message: 48158
Date: 2007-04-01

--- In cybalist@yahoogroups.com, "mkelkar2003" <swatimkelkar@...> wrote:
> Three dimensional map in Fig 6.8 of the link below. Germanic falls by
> the wayside.
>
http://linguistics.berkeley.edu/~kjohnson/quantitative/historical/historical.pdf

Figure 6.3 is easier to read for the details.

Isn't this tree based on comparisons of spellings? I strongly suspect
Grimm's law has a lot to do with the distinctiveness of Germanic.

Notice also that NW Germanic (East Germanic is not represented) has
its deepest difference between English and the rest. This phenetic
comparison, not phylogenetic.

Try this thought-experiment. Cypher a language by shifting each vowel
to the next in alphabetic order, and each consonant to the next in
alphabetic order. (The spelling comparison is aware of the difference
consonants and vowels - I wonder how that was handled for Welsh!)
Now, cladistically, the cyphered language and the original should
always appear in the same group. But of you added the cyphered
language, I rather suspect that the cyphered language would appear in
a branch of its own coordinate with the rest of Indo-European!

We have no reason to think that Johnson is unaware of this. He does
caution that the spelling method may be sensitive to orthographic
conventions.

I wonder where French written according to the non-Gallicising
orthographic conventions of Haitian Creole and Pig Latin would show up
in such analyses.

Richard.