4.0187 Text Analysis (1/30)

Elaine Brennan & Allen Renear (EDITORS@BROWNVM.BITNET)
Thu, 14 Jun 90 17:46:46 EDT

Humanist Discussion Group, Vol. 4, No. 0187. Thursday, 14 Jun 1990.

Date: Wed, 6 13 22:29:45 MDT
From: David Miall <DMIALL@UALTAVM>
Subject: Text Analysis


One step beyond either concordances or marked-up text is to use
the Z-score: this does give some help in the case of a text that
may be not too familiar. The latest version of TACT (1.2), just
released by the Centre for Computing in the Humanities at
Toronto, now includes a neat Z-score facility. TACT, in case you
haven't yet seen it, offers standard concordancing (interactively)
plus several related features, and now in my view outdoes
WordCruncher by some distance.

In looking at an unfamiliar text, assuming you have it indexed in
TACT, you might start by looking for substantive words that occur
with a high frequency (TACT provides a frequency command), then
run the Z-score on each of the words. This will bring up a list of
words occurring within a span of (say) 5 words on each side of the
target word, ordered by Z-score, ie., showing the degree of significance
of each word as a collocate. Such a list begins to get from the level
of words up to that of themes and ideas, and is a useful discovery tool
for mapping the semantic shape of a text. If you then want to follow
up a given collocation, TACT lets you jump direct from the Z-score list
to the text itself. With this facility and some other features, TACT
seems to me provide a good tool for use in the classroom as well as for
research, and it is of course very cheap: currently $30 CAN for disk and
manual (and you can make as many copies as you like). It runs on IBMs
and clones. Texts with COCOA or WordCruncher codes already in them can
be indexed by TACT with no problem, by the way: it's remarkably flexible.

Regards, David Miall
University of Alberta