LINGUIST List 6.49

Mon 16 Jan 1995

Qs: NLP Projects; Feature agreement; March hares; Loans

Editor for this issue: <>

Directory

1. Phil Bralich, Natural Language Projects
2. , Query concerning syntactic feature agreement in control
3. CAVEMAN -- San Bernardino, Calif. USA, Q: Mad as a March hare
4. Jorge Baquero, Foreign Loan Words

I am trying to find the names and addreses of indivduals, departments,
and companies that have or are developing Natural Language Processors.
Ideally, I would like to both receive the parser and communicate with
the developer. Any assistance you can give me on this matter would be
greatly appreciated.

I can be reached at:

bralichuhccux.uhunix.Hawaii.edu

Sincerely,

Philip Bralich, Ph.D.

 I have a question about syntactic feature agreement in controlled
clauses that I'm hoping someone can help me answer. It is often
assumed that agreement between syntactic features should be enforced
by employing unification. For instance, this correctly accounts for
the ungrammaticality of (1); although "sheep" is both singular and
plural, it cannot unify with both the number of relative verb "is"
(singular) and the number of the matrix verb "are" (plural).

 (1) * The sheep that is ready are there.
 {sg,pl} {sg} {pl}

 In his ACL-90 paper, Bob Ingria argues that formal agreement among
"purely syntactic" features should not be enforced by unification, but
instead using a non-distinctness check. (Number is not treated as a
purely syntactic feature given its ramifications for semantics.) He
argues his point with a range of data from a number of languages; for
instance one set of examples is based on free relative clauses in
German. These clauses require that the relative pronoun agrees in
Case both with the position of the relative clause as a whole and also
with the gap that the relative pronoun fills. Yet, sentences like (2)
(Ingria's (7)) are grammatical:

 (2) Ich habe gegessen was noch ubrig war.
 I have eaten what still left was
 ACC NOM/ACC NOM
 `I ate what was left'

Here, a non-distinctness check (as if unifying without saving the
results) predicts the acceptability of (2), whereas unification
predicts ungrammaticality. Ingria gives several other examples,
including the `quirky case' examples for coordination in Icelandic
(Zaenen and Karttunen, 1984) and German (Pullum and Zwicky, 1986).

 My question has to do with what the facts are concerning control.
For instance, consider a language where a matrix subject that also
serves as the subject of a controlled clause has to agree in Case
both with the matrix verb and the verb in the controlled clause. If
the subject is compatible with two different cases, the matrix verb is
compatible with one of these, and the verb in the controlled clause
is compatible with the other, is the resulting sentence grammatical?
That is, does control behave like the quirky case constructions,
requiring only a non-distinctness check for syntactic feature
agreement? Languages like Icelandic, which have been argued to
involve anaphoric control, shouldn't be considered.

 Thanks in advance,

 -- Andy Kehler
 kehlerdas.harvard.edu

Does anybody know the origins of the phrase "mad as a March hare?" What is
it about March hares that make them mad? Does it refer to a hare in the
month of March, or is a March hare a specific type of hare?

Chuck Coker
CJCokerCSUPomona.Edu

I am working on English loan words that are used in Spanish. I would
appreciate it if somebody could give me some hints about where I can find
references about loan words.

Marcela

 Marcela Acosta Garces
 please reply to me at: jbaqueropuce.edu.ec
 Facultad de Linguistica
 Pontificia Universidad Catolica del Ecuador
 Quito, Ecuador (South America)