Date: Tue Aug 01 2000 - 21:22:36 EDT
In a message dated 8/1/00 3:16:39 PM Mountain Daylight Time,
One conclusion that came out of this long term but very casual investigation
was that as of about1990 we were a long way from being able to produce an
adequate computational model of the semantic structure of any natural
language. My most recent reading in R.E. Longacre* and Simon Dik** has not
changed my mind on this topic.>>
I'm far from claiming that you could have an adequate computational model of
semantic structure! But on very unambitious terms, I know that the formal
linguistic criteria for markers of prominence and marked grammatical features
could be entered in a computer and traced through a discourse. On the other
hand, I'm just as sure that I can track them fairly well without. It is
helpful and interesting to see print-outs, and there is the occasional
pattern that a computer picks up that I miss. So, I've gotten an insight or
two from a print-out, though most that I have seen deal with topic and
semantic domains. I'm convinced myself that the computer will yield only so
much, but on the other hand, computational statistics can be a strong
confirmation and affirmation to some claims.
<< I am all in favor of looking at the formal features of a text but I don't
think that the semantic structure of a text is tied down hard and fast to
these formal features. This is where I leave the computational people
behind. I do not lay awake nights worrying that my analysis of a text might
be subjective. I am perfectly willing to live with subjectivity.
Yes, I think the objective would be to not let subjectivity or any single
consideration (as often happens in content analysis) to overrule the formal
patterns and signposts in the text--so the best starting point is the formal
analysis, and then the interpretation follows.
Any system that did not include interpretation would be a sham as a model.
It is the major role of the receptor.
PhD Student, Roehampton
B-Greek home page: http://metalab.unc.edu/bgreek
You are currently subscribed to b-greek as: [email@example.com]
To unsubscribe, forward this message to leave-b-greek-327Q@franklin.oit.unc.edu
To subscribe, send a message to firstname.lastname@example.org
This archive was generated by hypermail 2.1.4 : Sat Apr 20 2002 - 15:36:33 EDT