[300] in tlhIngan-Hol
Re: Causatives
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Tue May 5 19:25:14 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: tlhIngan-Hol@village.boston.ma.us, Ken_Beesley.PARC@xerox.com
Date: Tue, 5 May 1992 15:05:20 PDT
In-Reply-To: "shoulson@ctr.columbia:edu:Xerox's message of Tue, 5 May 1992 06:5
Mark,
Thanks for the response. I was aware that QaghHommeyHeylIjmo' is from the
book. I ran all the book examples a few weeks ago. I wish I had more test
data, like an extended text of some sort. I'm still impressed by your heroic
verb. Don't worry too much about whether a word is likely, silly, or "makes a
lot of sense." In formal linguistics, there's a fairly well-accepted
distinction between Grammatical (or "well-formed") on one hand vs.
Ungrammatical (or ill-formed) on the other; but Grammatical constructions can
still be odd, silly, or senseless. The most famous example is from Noam
Chomsky, who argued that the following sentence is grammatical in English
Colorless green ideas sleep furiously.
That is, you have a noun, adjectives, a verb and an adverb combined together in
legal orders to form legal phrases, and, ultimately, a legal or grammatical
sentence. Chomsky (and most other formal linguists) would expect to be able to
parse and generate such sentences using their syntactic grammars. On the other
side, the following sentence would be ungrammatical:
furiously colorless sleep ideas green
This is just a "word salad." It would be the job of any syntactic parser to
reject such a sentence. It's not just odd or silly--it's just downright
ill-formed. Some describe the reaction to ungrammaticality as a bad feeling in
the pit of the stomach. similarly,
John is tall Mary.
gives you that same pit-of-the-stomach feeling because "is-tall" is a one-place
predicate (corresponding to an intransitive verb in Klingon).
In morphology (word construction) the principle is much the same. The word
"ungrammaticality" can be analyzed roughly as un+grammatical+ly, consisting of
three morphemes in a legal combination and order. But the following word,
consisting of the same three morphemes, is ill-formed and should be rejected by
any morphology program.
ly+un+grammatical
But you're pretty much forced to allow words like "ungraciousnesses" even if
you can't think of a good way to use it offhand.
In Klingon, as in other languages, the only immediately tractable morphological
task is to identify all possible, grammatical readings for each word. The
ambiguity is definitely there, and sometimes you get silly or wildly unlikely
readings, like "mistake-bones". But maybe the Klingons do have bones used for
divination or something, and they COULD be called mistake-bones. Who knows?
But it's really beyond the scope of these little morphology programs to decide
what makes sense.
But it remains the job of a morphology program to reject words that are just
downright ill-formed. So running such a program on your own text will
1) point out possible ambiguities
2) serve as a kind of spell-checker (all ill-formed words are rejected, and
typos, if they succeed, will give you a reading other than the one you
intended)
3) serve (for Klingon especially) as a kind of "grammar checker" because the
language does so much within the morphology
By running someone else's Klingon text through the program, it serves as a kind
of automatic analysis and dictionary lookup system, listing and spelling-out
possible readings. Crude English glosses are printed out for each morpheme,
providing a very rough English translation. It sure beats flipping through the
separated lexicons in the book.
Maybe someday it will be interesting to write a syntactic grammar for Klingon.
That would take the output from the morphological analyzer, which will be
highly ambiguous for many words, and (hopefully) weed out many of the readings
by looking at words in syntactic context. This is often called
"disambiguation".
Compounding On the Fly
In linguistic jargon, the question is whether Klingon compounding is
"productive" or not, where productive means that speakers can make up new
compounds on the fly. German, for example, has very productive noun-noun
compounding. English compounding is less productive, but still fairly
flexible. The fact that Okrand describes the process of noun-noun compounding,
with a limit on the number of compound elements, strongly implies productive
compounding to me. The other possibility is that words like jolpa' were simply
"coined" words that somebody made up at some point and then caught on (became
"entrenched", as you say)-- and they would simply be added to the lexicon. To
some extent, I can order things in my program to force the analysis to pursue
non-compounding solutions first. The practical problem for me is whether
jolpa' counts as one or two elements when it's used in a compound. I'm faced
with these little decisions at every step of my morphological description.
I think you're quite right about ?ropmeyyaH being unlikely. I haven't allowed
it. But I rather like ?ropHomyaH so I did allow it. Obviously I'm just
guessing. Because the morphological formalism I use forces you to make such
decisions, linguists often write them as an exercise to discover the holes in
their descriptions.
Where did you track down Okrand? Does he entertain questions?
Ken
