LINGUIST List 4.936

Thu 11 Nov 1993

Qs: Corpora, Spelling Sensitive Rules, Parsing

Editor for this issue: <>

Directory

1. Jeff Tennant, Query: Electronic Texts in French
2. , Q: Phonological or Versification Rules Sensitive to Spelling
3. Kripa Sundar, Q: island-driven parsing ...
4. M Perkins, Corpora of Disordered Language

A colleague in my Department is interested in locating sources of
electronic texts in French, to be used for teaching purposes. He would
appreciate any information on ways of accessing such material: literary
texts, press articles, and just about anything else, provided it is in
French. Please sent replies to:

gholmesuwovax.uwo.ca
Glynn Holmes
Department of French
University of Western Ontario
London Ontario, Canada
N6A 3K7

Thanks.

Jeff Tennant

I am trying to collect examples of phonological rules or rules
of versification that are sensitive to spelling. The sort of
thing I have in mind is illustrated by the following:

(a) In Polish noun stems ending in /w/ alternate with /l/ before
certain suffixes, e.g., /stuw/ 'table'(nom.) but /stole/ 'table (loc.)'.
However, some speakers do not use such forms unless the word in
spelled with the "slashed l". Thus, words of foreign origin incl.
proper names do alternate if spelled with "slashed l" (this includes
words from languages using the Cyrillic alphabet as well as Mongolian)
but do not if spelled with "w" or "u". In the latter case, these
words have no corresponding forms at all, i.e., there are gaps in
the paradigm.

(b) In French, words ending in final orthographic "e" are not equivalent
in terms of versification to homophonous words without the "e". For
example, the two do not rhyme, and in certain cases the "e" counts
as a syllable. Now, it can easily be shown that this rule refers to
the spelling (rather than to so-called underlying representations)
since there are examples where no phonological or morphological evidence
could have determined the alleged difference in underlying representations
(as has been proposed over the years by numerous generative phonologists),
e.g., 'foie' "liver" and 'foi' "faith".

i am looking for references on island-driven parsing.
i have only come across references from 1983 or before
(WA Woods and related papers). the names i've come across
so far include John Carroll and Manfred Gehrke.

i am hoping to find current or recent work on this and
related techniques to handle the parsing of malformed
or noise-prone input.
thank you.

peace,
 --kripA

I'm trying to track down machine-readable corpora of disordered language -
in particular that of brain-damaged adults (ie aphasia, right brain
damage, head injury etc).
I'm aware of the Bates 'CAP' CHILDES corpus, and the Edwards/Garman
aphasic corpus at Reading University, UK, but that's all I know of for
adult language.
At Sheffield University, UK, we are developing an extendable machine-readable
corpus of aphasic discourse which is currently in the pilot stage.We'd
like to be in touch with anyone doing similar work.

Mick Perkins
Speech Science Unit
Sheffield University, UK