Re: VMs: left & right word entropy

[Jacques Guy <jguy@xxxxxxxxxxxxxxxx>]

24/12/2003 11:49:39 AM, Gabriel Landini <G.Landini@xxxxxxxxxx> wrote:


>It is difficult for me to argue because I do not know latin, but can the word 
>"et" appear a at the end of a sentence?

Not to my knowledge.

>And in Japanese the "desu ka" whould be appearing in paire quite often.

And so "desu ne", and "arimasu ka" and "arimasu ne" and "ikimashoo ka".
And not only verbs but nouns: "ka" also means "or".
 
>Same with "xxxx no ue ni" (on top of) but never in reverse.

"Ni" can be preceded by just about any noun as it is
a locative ("ubi") or a dative. "Ue" is a noun, anyway,
as is "shita" (under), "naka" (inside), "ushiro" (behind),
etc.

>I know that Japanese is quite rigid in this sense, what about 
>Chinese?

Like all languages I can think of, there are words in Chinese
which can only be preceded by certain classes of words
(nouns for instance). But this is quite unlike English 
"am" which calls for ONE particular word: "I"--a class (
set) of one member. The only Chinese words I can think of
which call for a small set are classifiers, which are
inserted between numeral and noun, e.g. yi (one) zuo4 
(classifier) qiang2 (wall), liang3 (two) ben3 (class.)
shu1 (book), and so on.  However, they have many
homophones. Zuo4, for instance, is also "to do" and
"to sit" (which are written with different characters).

>
>> It MIGHT tell us that word-order is free. 
>Ok, I will change the question then. Is the right-left entropy a good measure 
>of order in the text?

Somewhere in the archives you will find the 
translation of an algorithm that claims to
separate continuous text into phrases, words
and morphemes. It makes use of a similar 
measure.

>Has this been done in many other texts?

I don't know.





______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list