[296] in tlhIngan-Hol
Transitive distinction, etc.
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Sun May 3 10:20:38 1992
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>
Cc: Ken_Beesley.PARC@xerox.com
Date: Sat, 2 May 1992 19:54:00 PDT
Thanks to Krankor for the response to my proposal (that -moH be treated as a
transitivizer that can recategorize a verb, thereby allowing suffixes like
-'egh to come after it). I eagerly await authoritative clarification on this
and any number of other holes in the description (I'm preparing a list).
Krankor, you seem to have some kind of access to Marc Okrand. Does he answer
questions on Klingon? Is he preparing a follow-up description? Does he think
that we're all nutters?
On the transitive/intransitive distinction.
I just joined the Klingon list, and I missed any previous discussion on this
topic; so I apologize in advance if I rehash some old material. Perhaps some
kind soul could send me copies of previous messages.
From my perspective (computational linguistics) it's pretty clear to me that
there IS a distinction between transitive and intransitive verbs in Klingon.
When I found The Klingon Dictionary, it amused me a great deal, and I
immediately set out to see if I could write an automatic morphological analyzer
for Klingon words. I've helped to write such morphological analyzers for
several human languages, including Arabic and Aymara, which like Klingon is
agglutinating. [Aymara verbs have 20 different suffix classes (no prefixes),
compared to Klingon's 9 + prefix; Aymara nouns have 13 different suffix
classes, compared to Klingon's 5.] The goal for morphological analyzers is to
1) accept all well-formed words, returning a morphological breakdown, and 2)
reject all ill-formed or ungrammatical words. In order for the program to
reject a large class of ill-formed Klingon words, it has to know what is (or
could be) transitive and what is intransitive. Three example types come to
mind:
1. SubjObj vs. Subj prefixes
The Subj prefixes (those that indicate no object or an unspecified object) can
apparently attach to all or almost all verbs. (Exceptions might include
impersonal verbs like SIS "to rain"; we'll have to wait for some examples
involving the weather.) So you can apparently say
jIHoH I-(no object/unspecified object) kill
jIQong I-(no object/unspecified object) sleep
HoH, I would say, is a transitive verb in Klingon. And Qong is one of the few
intransitive verbs. HoH, of course, can also co-occur with SubjObj prefixes:
vIHoHta' I-him/etc. kill perfective-intentional
But it would seem (from all available examples that I can find) that certain
intransitives (a relatively small group) like Qong can appear only with the
Subj prefixes, not the SubjObj prefixes (unless the intransitive verb is
transitivized with -moH). So the following would be "ill-formed" or
"ungrammatical" even though they conform to the superficial rules
*vIQongta' *I-him slept
*vIQongpu' *I-him slept
But if Qong is transitivized, with -moH, the resulting words are just fine.
vIQongmoHta' I-him/etc. sleep causative perfective-intentional
vIQongmoHpu' I-him/etc. sleep causative perfective
To accept vIHoHta' and vIQongmoHta' while rejecting *vIQongta' and *vIQongpu'
the program simply has to be able to tell when it's dealing with a transitive
(or transitivized) verb. In the computational linguistics world, that boils
down to a transitive/intransitive distinction in the language, which must be
coded somehow in the on-line lexicon used by the program. The current program
does have that distinction, and it correctly accepts and rejects the examples
as shown above. The fact that the printed dictionary does not make the
distinction is beside the point. Printed dictionaries for all languages are
notorious for omitting linguistically real and computationally necessary
distinctions, relying to a large extent on the intuitions of the user. It is
not unusual in a human language to need to subcategorize verbs in dozens of
different ways, including transitives, intransitives, bitransitives,
ditransitives, different conjugation paradigms, etc. It would be highly
unusual if that were not true for Klingon as well.
2. Reflexive 'egh.
Similarly, I think that any reasonable morphological analyzer for Klingon
should reject *Qong'eghta' and *Qong'eghpu' (*he/she/it/they sleep -self
perfective) while accepting HoH'eghpu' and HoH'eghta' and also Qong'eghmoHta'
and Qong'eghmoHpu' (or, by my earlier suggestion, perhaps ?QongmoH'eghta' and
?QongmoH'eghpu'). Again, what the program needs to know is that Qong is
intransitive. (The current program needs to be fixed to accept Qong'eghmoHta'
or ?QongmoH'eghta'--I can do it either way. I'll probably just create two
versions for now and wait for Okrand to pontificate.)
3. Impersonal subject -lu'
I would again argue that -lu' is incompatible with intransitive verbs like
Qong, unless those verbs are transitivized with -moH. The program needs the
distinction in order to reject examples like the following
*vIQonglu'ta' me sleep someone perfective-intentional
(*Someone slept me. or *I was slept.)
while the following are well-formed.
vIQongmoHlu'ta' (Somebody caused me to sleep)
vIHoHlu'ta' (Somebody killed me)
The program correctly handles these examples.
Having said all this, I would add a caution that we cannot transplant our
English intuitions about transitivity into Klingon. Klingon transitivity has to
be assigned on its own terms. It would seem that any word that can conceivably
take an object should be considered transitive or potentially transitive in
Klingon. Also, some copula-like verbs like moj (become) are obviously
transitive in Klingon (i.e. they take the SubjObj prefixes).
p. 22
yaS vImojpu' (officer I-it became)
Feedback is really appreciated. It gives me new words to run through my
program, like Mark's yItay'eghmoH, hopefully making it better and defining the
dark areas.
Ken Beesley
beesley.parc@xerox.com
