LINGUIST List 7.518

Tue Apr 9 1996

Sum: Statistical methods

Editor for this issue: T. Daniel Seely <dseelyemunix.emich.edu>


Directory

  1. XU LUOMAI, Statistical methods

Message 1: Statistical methods

Date: Mon, 08 Apr 1996 10:23:05 +0800
From: XU LUOMAI <itxulmscut.edu.cn>
Subject: Statistical methods
Several weeks ago, I posted a request for some statistical methods to
analyze matrices derived from some copora of written English by the native
speakers and the Chinese EFL learners. So far, I have received only 2
responses that offered some help.

1. alicesungarnet.berkeley.edu :
There is another improved and better program coming out: STATISTICA. It has
the function you specified: multiple regression, general non-linear analysis,
stepwise discriminant analysis, log-linear, ANOVA, and so on. If you can get
on World Wide Web, their address is: http://www.statsoftinc.com. Or you can
give my your mailing address so that I can copy the brief intro to the
product, and you can have a look.
The company's name and phone/fax numbers:

StatSoft
2325 E. 13th St.
Tulsa, OK 74104
USA
Phone: 918-583-4149
Fax: 918-583-4376

It has offices in London, Paris, ... and Taiwan. The phone number for the
Taiwan representatives:
Intelligent Integration Corp: 2-759-1791, fax: 2-759-1790

I think that you can contact the company directly. 

2. Prof. Gui Shichun from my own university (email:
itscguiletterbox.scut.edu.cn) suggested that we use entropy-based
redundancy scores. The formula of entropy can be found in DATA, MODELS AND
STATISTICAL ANALYSIS by R.A.Cooper & A.J.Weekes, p.351, and the redundancy
score R is obtained by the formula

	R=H/ln(N)

where H is the entropy, N is the number of cases.

We found some encouraging preliminary results: a)Both cluster analysis and
redundancy scores show that the use of certain tag-pairs distinguish the
Chinese EFL learners from the native English speakers. The frequency of
these tag-pairs often reflects the differences between Chinese and English.
b) There is strong relationship of linear regression between the non-zero
frequency scores and the sample sizes.
Xu Luomai
English Department
Guangdong University of Foreign Studies
Guangzhou 510420
P.R. China
Tel. (020)86656476
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue