[997] in tlhIngan-Hol
Re: verbs in compounds (was: Re: epithets (taHqeq))
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Wed Jun 16 21:30:59 1993
Errors-To: tlhIngan-Hol-request@village.boston.ma.us
Errors-To: tlhIngan-Hol-request@village.boston.ma.us
Errors-To: tlhIngan-Hol-request@village.boston.ma.us
Errors-To: tlhIngan-Hol-request@village.boston.ma.us
Errors-To: tlhIngan-Hol-request@village.boston.ma.us
Errors-To: tlhIngan-Hol-request@village.boston.ma.us
Errors-To: tlhIngan-Hol-request@village.boston.ma.us
Reply-To: "Klingon Language List" <tlhIngan-Hol@village.boston.ma.us>
From: Ken_Beesley.PARC@xerox.com
To: "Klingon Language List" <tlhIngan-Hol@village.boston.ma.us>
Date: Wed, 16 Jun 1993 12:07:19 PDT
In-Reply-To: "krankor@codex.prds.cdx.mot:com:Xerox's message of Tue, 15 Jun 199
I see that Krankor has already responded to Mr. Appleyard on the issue of VN
compounds. Just a couple of comments:
Krankor:
>>The only thing obvious about it is that that
is what one would get if one assumed that Klingon syntax for doing
such things were identical to English.<<
Krankor is on the right track here. We must constantly guard against
reflecting English (or our native language) assumptions onto Klingon. There
are, by most counts, 3500 languages spoken by sentients on Earth, and there is
tremendous variety in the way languages "do" things. Klingon may be even
stranger than most.
The term "syntax" is used by linguists to refer to the rules for combining
words into phrases and sentences. The structure or "shape" of words is the
study of morphology. Within morphology, "morphotactic" rules control the ways
that words can be constructed out of parts called "morphemes." When we point
out, respectfully and helpfully I hope, that Mr. Appleyard's VV and VN
solutions are invalid, it is because they are not allowed by the know
morphotactic rules for Klingon.
Krankor:
>>I *certainly* can't see any justification for building such [VV, VN]
syntax [morphotactics] into an analyser, which, by definition is expected to
adhere
strictly to the rules of the language. (Or perhaps I'm
misunderstanding the point of an analyser; my only knowlege of them
is what I've picked up from Ken).<<
Every language has it's own morphotactic rules, some of them absolutely amazing
from the English point of view. Out of an incredible variety of morphotactic
constructions that are known to exist in human languages and therefore are
possible, any particular language allows only its own small set of
constructions. The standard goals in writing a morphological analyzer are, for
a particular language,
1. To accept all valid words, and
2. To reject all invalid words
If an analyzer fails to accept valid words, then it is "underrecognizing."
Conversely, if it accepts invalid words or returns invalid pseudo-analyses for
words, then it is "overrecognizing." No morphological analyzer for a natural
language will ever be perfect; like any scientific theory you just keep
working away at it to get an increasingly accurate match of theory and data.
Ken Beesley
