Some times ago, I came up to you with the following request:
>
> Can you inform us about statistical packages (freeware, others)
> you are used to work with in your corpus linguistic research
> (for the purpose of building intercorrelation matrices,
> doing factor analysis, cluster analysis etc.) and where we can obtain
> them (required platforms: UNIX, Windows)?
>
> Thanks is advance,
>
I would like to thank all of you who gave us a prompt and usefull answer. Below
you will find the list of the anwers given.
Franck Bodmer
Institut fuer deutsche Sprache - Mannheim
=================================================
1)
Dr. Andrew Wilson
University Centre for Computer Corpus Research on Language
Department of Linguistics and Modern English Language
Lancaster University
I use MVSP for calculating intercorrelation matrices and performing
cluster analyses.
MVSP is a DOS program, but I've a feeling there *may* also now be
a version for Windows (I'm not sure about this, though).
You can check out the details on the WWW at:
http://www.kovcomp.co.uk/kcssoft.html#mvsp
MVSP is inexpensive, fairly easy to use (*much* easier than programs
such as Clustan), and you can download and try out a shareware version
before buying a licence.
I've been using it for a couple of years or so now and have been
quite happy with it.
----------------------------------------------------------------------------
2)
Marco A E Rocha
University of Sussex
School of Cognitive and Computing Sciences
Falmer, Brighton
BN1 9QH - U.K.
I have been using SPSS for Windows on a UNIX platform with a X-Windows
interface. I am pretty happy with it and the package can handle the sort of
analysis you require, but it does not come free, I'm afraid. Best.
SPSS-PC+ is very similar as far as operation goes. As
the name implies, SPSS-PC+ is a version for PCs. I run the version for Windows
on Solaris in a Sun machine in my school. It is roughly the same program
adapted for use in a Sun with Unix - Solaris is the version of Unix for Sun
machines. My version is called SPSS for Unix, Release 6.1 (Solaris 2.3).
I am being a bit fastidious, because it is the same package adapted to
different machines.
--------------------------------------------------------------------------------
3)
Dagmar Barth
Universitaet GH Paderborn
FB 3: Sprach- und Literaturwissenschaften
Anglistik/ Sprachwissenschaft
If that is, what you are looking for you could try with SPSS-PC+ or
any other version of SPSS. The versions are running on DOS and
WINDOWS. For complete beginners, I believe, the handbook is at
times not too useful without an introductory coursebook to statistics.
--------------------------------------------------------------------------------
4)
Stefan Langer
MicroCentre, University of Dundee
Hi,
Einen guten Ueberblick finden Sie auf der folgenden Webseite:
http://www.ruf.rice.edu/~barlow/corpus.html#Software
--------------------------------------------------------------------------------
5)
Kevin Russell
Linguistics, University of Manitoba
The statistical package I like best is Xlisp-Stat. We've got
Minitab 10 on our computers, too, and I'll still use it sometimes, but I
much prefer Xlisp-Stat.
I found it almost by accident this summer. I was getting very
frustrated with having to enter long, repetitive, slightly different
strings of commands because Minitab's macros weren't powerful enough to
automate them. It would have been complex and costly to get in on our
campus' site-licences for either SPSS or SAS, and I haven't had great
experiences with them either. (Sure, they're wonderful if you want one of
the 10,000 things they can do, but if you want the 10,001st it's like
pulling teeth.)
You can think of Xlisp-Stat either as a decent version of Lisp with a
ton of statistical functions built-in, or as a statistical package with a
powerful and seamlessly integrated macro facility (assuming you can
tolerate Lisp :-). It also seems to have a fairly active community of
users. If the kinds of analyses you want aren't already built-in, there's
a good chance that somebody else has already written some add-on
procedures for them, or at least done *most* of the background work for
you. (Hey, I've even used the time-series analysis package to give me
crude spectrograms when I'm away from the phonetics lab!)
The biggest plus: it's free!
A couple of drawbacks: the graphics aren't great. They're perfect
for exploratory analysis, but if you want a few professional-looking
diagrams for a publication or some presentation slides, you'd be better
off moving to a dedicated graphing program. The input/output facilities
are a little awkward, but I've been able to live with them.
The web page you can get more information (or the actual program)
from is
http://stat.umn.edu/~rcode/xlispstat-resources.html
The page also has links to archives of extra functions, tutorials, and
people already using Xlisp-Stat.