4.0560 Rs: Grammar & Syntax Models; Pedagogical Parsers (2/34)

Elaine Brennan & Allen Renear (EDITORS@BROWNVM.BITNET)
Wed, 3 Oct 90 22:46:15 EDT

Humanist Discussion Group, Vol. 4, No. 0560. Wednesday, 3 Oct 1990.


(1) Date: Wed, 3 Oct 90 09:16:07 MDT (20 lines)
From: koontz@alpha.bldr.nist.gov (John E. Koontz)
Subject: Re: 4.0554 Qs: Grammar & Syntax Analysis ...

(2) Date: Wed, 03 Oct 90 16:43:40 EST (14 lines)
From: "Dana Paramskas, U. of Guelph" <LNGDANAP@VM.UOGUELPH.CA>
Subject: English Pedagogical Parser

(1) --------------------------------------------------------------------
Date: Wed, 3 Oct 90 09:16:07 MDT
From: koontz@alpha.bldr.nist.gov (John E. Koontz)
Subject: Re: 4.0554 Qs: Grammar & Syntax Analysis; Spoken German; ...

In regard to Andrew Oliver's query, for a program that sets up models of
syntactic structure and grammatical structure at the phrase level, here's
what I would suggest, if I understood his question properly.

1) Use SIL's IT (Interlinear Text) program to tag the text interlinearly
with part of speech and/or phrase structure information.

2) Investigate the patterns that appear in the tagging fields (lines) with
something like a concordance program or perhaps something like TALLY -
a program that I am not familiar with, which allows one to investigate
patterns of appearance of particular strings.

Note that it would be desirable to have a concordance program that could
be instructed to restrict its attention to particular fields, i.e., the
tagging fields, not the original text field.

(2) --------------------------------------------------------------22----
Date: Wed, 03 Oct 90 16:43:40 EST
From: "Dana Paramskas, U. of Guelph" <LNGDANAP@VM.UOGUELPH.CA>
Subject: English Pedagogical Parser

Sometime ago, a HUMANIST member asked about parsers capable of dealing
with ill-formed input. Don Loritz of Georgetown has written one and
is willing to distribute it free to anyone interested. He writes:
"We achieved 90% accuracy with *Ms. Plurabelle* among our population
of deaf college-prep students at Gallaudet this past spring..."

He has an e-mail address, but it works erratically: loritz@guvax.georgetown.edu
Better to write: 1313 North Illinois Street, Arlington, VA 22205.

I've seen earlier verions of this program and was very impressed.