LINGUIST List 4.267

Tue 13 Apr 1993

Qs: Jacaltec, generation and LFG, frequency differences

Editor for this issue: <>

Directory

1. Michael Newman, Jacaltec
2. Charlotte Webb, Frequency differences
3. John Hughes, Lexical-Functional Grammars for generation

I am posting this for someone who is not on this list. He would like to get
into contact with anyone who has done research on the syntax of the Mayan
language, Jacaltec. Send responses to Francisco Ordonez, ORDGCCUNYVM.CUNY.EDU.
Thanks

I am doing research which uses Kucera and Francis' Computational Analysis
of Present-Day American English published in 1967. They published a
corpus of more than 1 million words with a frequency count, as well as
other summaries.

Matsumoto in an article claimed that for words in this corpus which
differed in number by 100 (e.g. 46 vs 150), this frequencyt difference
was a statistically significant difference.

Does anyone know if Matsumoto is correct? Specifically our experiment
has to do with words which are identical with respect to a particular
phonemic contrast, e.g. /f/ "fan" vs "fin" but which differ in
frequency of occurrence. We constructed minimal pair lists using the
values in Kucera and Francis and Matsumoto's claim that a difference of
100 is a significant difference. Now, having done some chi sqare analyses
on some of our word pairs, it appears that Matsumoto's claim is probably
incorrect.

Please send replies to me directly and I will summarize and publish the
responses.

Thank you.

 cwebbsciences.sdsu.edu

Charlotte Webb
Dept. of Linguistics and Oriental Languages
San Diego State University
San Diego, CA 92182
(619) 594-1910

cwebbsciences.sdsu.edu

Could anyone here point me to work done in natural language generation using
lexical-functional grammars? Also, does anyone know where I might find a fairly
large, detailed grammar of LFG for English (not necessarily implemented in any
computer laguage) to inspect as a model for such an enterprise?

Please respond by e-mail. Thanks.

John
hughescis.udel.edu