LINGUIST List 5.2

Sat 1 Jan 1994

Sum: Regular Expressions; Aristotle; li and discontinuity

Editor for this issue: <>

Directory

1. , Addendum to Summary on Regular Expressions
2. , Summary: Aristotle on cross-classification
3. , Summary: Russian _li_ and discontinuity

Thanks to Gilbert Harman for pointing out that Chomsky (in Chomsky
and Miller 1963 and elsewhere) seems to have hit upon the notation
A -> B^n (^ means the n is superscripted), which was called a
'rule schema'. This idea appears to have been totally independent
of what was going on in the programming language community.
Interestingly, in Chomsky and Miller the idea, after being broached,
is then dismissed as a hopeless attempt to salvage the supposedly
unsalvageable phrase structure model of grammar. However, in
Aspects, Chomsky appears to take this proposal seriously, as a way
of extending the descriptive power of the phrase structure component
of a transformational grammar.

It thus seems that the idea of allowing regular expressions--or
something equivalent--on the right hand sides of CF rules arose
independently several times in the early and mid sixties.

It also occurs to me that it should be pointed out that the
credit for allowing nulls on the right hand sides of CF productions
appears to go to Bar-Hillel and his coworkers, although again it
is not clear to me that there were not others who had the same
idea.

I recently posted a query regarding Aristotle's apparent
position that there is no such thing as crossclassification.
I received a number of responses, none of them absolutely
decisive. Perhaps crucially, Julius Moravcsik (who is one
of the world's authorities on good ole Aristo, I gather) tells
me that he views Aristotle as having vacillated on this point.

I guess a better question should have been not what Aristotle
really thought but whether the apparent reluctance of grammarians
to recognize crossclassification among grammatical categories
such as parts of speech over many centuries was due to the wya
they UNDERSTOOD Aristotle (whatever he may ACTUALLY have thought)
or to some other factor.

Of course, as Dick Hudson reminded us, there is room for
disagreement about particular cases, as to whether they involve
crossclassification. I might point out for example that the
affix-hopping phenomena, which Chomsky regarded as involving
discontinuity, have more recently been discussed as cases of
crossclassification (e.g. in the GPSG analyses). In general,
as I think was pointed out in the MR and Kac paper "The conceot
of phrase structure" in L&P a few years back, there seems to
be a trade-off between discontinuity and crossclassification.

Alexis Manaster Ramer
Professor of Computer Science
Member of the Linguistics Program
Member of the Executive Committee, Center for Peace & Conflict Studies
Wayne State University

I have received numerous references to recent treatments of
the Russian (and other Slavic) _li_, the clitic marking yes/no
questions, but no one mentioned any analyses involving discontinuity.

This is somewhat puzzling to me. If I can naturally analyze
Polish yes/no questions beginning with _czy_ as having _czy_
as an operator on the whole rest of the clause, e.g.,
[czy [Jan idzie]] 'Is John going?' with the structure as indicated
by the brackets, then it would seem logical that Russian
_Idyot-li Ivan_ could be viewed as having the structure where
[idyot Ivan] is a constituent which wraps itself around the
operator _li_.

In the days when people talked a lot about discontinuity (i.e.
in the 1940's and 50's), there were analyses of English sentences
like _Is John going?_ as involving discontinuity, with [is going]
wrapped around [John], incidentally (Nida, Hockett).