[1] From: Paul Schaffner <pfs@umich.edu> (37)
Subject: Vagueness in dictionaries
[2] From: "Robert S. Tannenbaum" <rst@service1.uky.edu> (21)
Subject: Fuzz
--[1]------------------------------------------------------------------
Date: Wed, 15 Nov 1995 04:59:59 -0500 (EST)
From: Paul Schaffner <pfs@umich.edu>
Subject: Vagueness in dictionaries
I was interested to see Russon Wooldridge's comments on retrospective
tagging of dictionaries. I have been engaged recently in adapting the
TEI dtd-for-print-dictionaries to the retrospective encoding of the
Middle English Dictionary. The MED is remarkably consistent and precise
for a work of its size and longevity, much more so than the OED certainly:
but, inadvertent ambiguities aside, any attempt at rigorous tagging
soon reveals deliberate ambiguities, often adopted as "fudges": means
to conceal ignorance. Two obvious examples:
(1) The MED generally supplies two dates for each work quoted: the
manuscript date and the composition date. Consistent electronic searching
by date requires that this be carried through for every work. But
in many cases, MED lists only a composition date. When it does so, it
may be implying:
(a) That the MS date and composition date are identical or nearly so.
(b) That the MS date is wholly unknown, but presumed contemporary with
the document.
(c) That the MS date is wholly unknown, period.
A similar set of implications applies when only a manuscript date is given.
(2) Compounds and phrases are cited without morphological tagging.
"Domes dai" is cited as such (at least I suppose so; I haven't looked),
but does this mean "dooms day" or "doom's day" or "dooms' day"? In many
cases, the answer is not obvious, and was probably not obvious even to the
original ME speaker. Yet TEI would like us to supply it....
I won't even mention the ambiguities of phonological (non-)representation.
Of course, the process of writing a dictionary is in itself a kind of
text encoding, a valiant if vain attempt to disambiguate natural language.
Any lexicographer can attest that much of the lexicographic task involves
either a palpably arbitrary disambiguation or a palpably subjective
organization by affiliation of ambiguity (that is, grouping examples
together not as sharing an unambiguous meaning "such-and-such," but as
sharing a particular species of ambiguity!) Try to define "truth" in
the Gospel of John, Chaucer, and Piers Plowman and you'll see what I
mean.
<plug type=shameless>I expect to be talking about some of this in more
detail in next year's Kalamazoo session on glossarial concordances.</plug>
Paul Schaffner (pfs@umich.edu)
o Middle English Dictionary, University of Michigan
o Humanities Text Initiative, University of Michigan Libraries
--[2]------------------------------------------------------------------
Date: Wed, 15 Nov 1995 09:11:20 -0400
From: "Robert S. Tannenbaum" <rst@service1.uky.edu>
Subject: Fuzz
Willard,
I have been lurking on Humanist for some time and enjoying it immensely.
Thanks for the enormous effort that you expend in maintaining this superb
list. And thanks, too, for introducing the wonderful discussion of
vagueness.
In 9.295, you make the assertion that "The perfect language is, I suppose,
one in which there is no fuzz, the Adamic speech in which the name of a
thing is its nature." I disagree. Such a language would be barren. How
could there be any poetry in such a language? I have always reveled in
the joys of language used in such a way that words convey a number of
meanings and shades of meaning simultaneously, challenging the
listener/reader to become totally engaged in the multiple levels and
implications thus communicated. I seek precision and no ambiguity in a
budget spreadsheet, in a computer program, but not in a poem or a work of
art.
Keep up the great work.
Bob
Robert S. Tannenbaum, Ed.D. 606 / 257 - 2900 office
Director, Academic Computing Services 606 / 323 - 1978 fax
128 McVey Hall rst@pop.uky.edu
University of Kentucky
Lexington, KY 40506-0045