Novel Sentences

[Carl Alphonce]

I have been lurking in the background, reading the discussion with
interest.  I am afraid I don't quite see what the real issue here is.

If I recall correctly, the discussion started with a question regarding
the claim that people can produce novel sentences.  Since then it seems
to have moved on to a discussion of whether or not the set of sentences
in a language is finite or infinite.

I don't think it can be disputed that the number of sentences uttered
by a person, or uttered by all persons in all of history is finite.
The number of speakers is finite.  Each speaker has a finite lifespan.
No sentence is of infinite length.  An so on.

At any point in time it is (theoretically) possible to construct the
(finite) set of all natural language sentences every uttered by anyone
in any language.

The question of the nonfiniteness of such a set comes into play when
you want to construct a theory to account for those sentences.  By
assuming that the set is in principle infinite, you can use recursive
rules to capture significant generalizations about the structure of
the sentences in the set.  If you insist that the set is finite, then
recursive rules cannot be permitted (without ad-hoc constraints on
their applicability).

Because people do not in fact use or comprehend sentences of arbitrary
length, it is not enough to have only a theory which permits
arbitrarily large sentences.  But this is what the competence
performance distinction is all about.  A competence theory is about
our idealized capacity for language, while a performance theory can be
viewed as constraints of a non-grammatical nature which limit what we
are able to produce and comprehend.

These are abstractions that we use when investigating language.  These
abstractions happen to be very useful, but they are not themselves
fact.  Other abstractions yield theories with different properties,
empirical coverage, and predictive power.

Perhaps I am just being obtuse, but what is the real issue that people
are discussing?

Carl

--
Carl Alphonce                     / email: alphonce at cs.ubc.ca
Department of Computer Science   / phone: (604) 822-8572
University of British Columbia  / FAX:   (604) 822-5485
Vancouver, BC, CANADA, V6T 1Z4 / http://www.cs.ubc.ca/spider/alphonce/home