• Next message: Ken Steele: "The Bed's Still Second-Best (41)"
• Previous message: Ken Steele: "Etexts on SHAKSPER (52)"
• Messages sorted by: [ date ] [ thread ] [ subject ]

Shakespeare Electronic Conference, Vol. 1, No. 42. Friday, 31 Aug 1990.
 
 
(1) Date: Fri, 31 Aug 90 12:33:23 EDT (7 lines)
      From: "Michael S. Hart" <HART@UIUCVMD>
      Subject: Re: SHK 1.0033 WordCruncher Riverside Shakespeare (156)
 
(2) Date: Fri, 31 Aug 90 17:20:48 EDT (48 lines)
      From: Ken Steele <KSTEELE@vm.epas.utoronto.ca>
      Subject: The Value of Non-Hierarchical Coding
 
(1) --------------------------------------------------------------------
Date: Fri, 31 Aug 90 12:33:23 EDT
From: "Michael S. Hart" <HART@UIUCVMD>
Subject: 1.0033 WordCruncher Riverside Shakespeare (156)
Comment: Re: SHK 1.0033 WordCruncher Riverside Shakespeare (156)
 
As for collecting all references to Juliet's lines, why not creat a macro
to search for Juliet as speaker, then export her portions to a file. The
file could then be searched for "love" as I recall the example was. mh
(2) --------------------------------------------------------------52----
Date: Fri, 31 Aug 90 17:20:48 EDT
From: Ken Steele <KSTEELE@vm.epas.utoronto.ca>
Subject: The Value of Non-Hierarchical Coding
 
 
As Michael Hart suggests, there are makeshift solutions to any individual
problem. If my desire was to detect all occurrences of "love" in the
dialogue of Juliet, I could indeed create a database input text which
consisted solely of her lines, index it (with TACT or WordCruncher), and
then perform whatever operations upon it I wished.
 
However, if I wished to compare Juliet's use of "love" with Romeo's
use of it, or Juliet's use with all other characters' use, this solution
would be extremely trying -- rather than a single distribution screen,
one would be faced with multiple databases and very unhelpful results.
 
Furthermore, this presumes that all I wish to do with the texts is
explore this issue of Juliet's dialogue. The remarkable power of
computer-assisted research comes from its spontaneity -- the researcher
quite simply need not know what s/he is looking for until a pattern
emerges from the evidence. It might not be the speeches of Juliet at
all, but the spelling preferences of a single compositor; it might be
that the word "love" does not appear at all in a particular gathering;
it might be that love is always discussed in verse... (you get the
idea).
 
Using TACT, for instance, it is possible to evaluate the evidence in
a single textbase according to any number of variables, because it is
not limited to three rigidly hierarchical levels of codes. The user
can type in a word, list of words, or presaved category of words, such
as, for example, the language of Petrarchan love. One could then view
the results of this search in a given scene, play, or genre in a one-line
index screen, Keyword in Context screen, or (my favourite) a histogram.
TACT's graph (called the Distribution Display) could depict the occurrences
of the Petrarchan language according to Act, Scene, and Line. With a
few keystrokes, the display would shift to a graph of occurrences by
speaker, or by prose/verse, or by signature, compositor, gathering,
and so on.
 
Just like WordCruncher, TACT allows the researcher to
*discover* new patterns, not simply to demonstrate anticipated ones --
but TACT (and apparently the eventual upgrade of WordCruncher) will
allow the use of multiple, non-hierarchical tags, which greatly increases
the utility and user-friendliness of the software, and advances by an
order of magnitude the possibility for original discovery in research.
 
                                  Ken Steele
                                  University of Toronto

• Next message: Ken Steele: "The Bed's Still Second-Best (41)"
• Previous message: Ken Steele: "Etexts on SHAKSPER (52)"
• Messages sorted by: [ date ] [ thread ] [ subject ]