Re: Terminology defs

On Mon, 13 Sep 1999, Bryan Maloney wrote:

> On Mon, 13 Sep 1999, Tom Wier wrote:
>
> > (Also -- there's no particular reason to single out linguistic "rules" as
> > somehow methodologically inferior to those of the physical sciences.
>
> "Rule" implies a legislator--it's teleological.  In biology, we've
> stopped using "law" and "rule" whenever possible.  Instead, we refer to
> "trait" or "characteristic", far less teleological.  Unless one is
> referring specifically to a system wherein there *is* a "legislator" of
> some sort, "rule" or "law" is unnecessarily teleological and proscribptive
> language.  "Trait" does not carry such baggage.
I use the term "rule" to mean roughly "a formal statement expressing a
systematic pattern or regularity found in language data".  If I were
to substitute the word "trait" in this case, I think I would have
trouble distinguishing the formal statement of a linguistic phenomenon
from the phenomenon itself.

Of course, the word "rule" has its own ambiguities - at least in
generative linguistics, where a rule can be either a formal statement
in a linguistic analysis, OR the (set of) instruction(s) within one's
mental grammar which that formal statement supposedly expresses/captures.
This dual use of the term "rule" goes back ultimately to Structuralism,
I think (although the Structuralists obviously did not speak in terms
of mental grammar per se).

One recent sub-theory of generative linguistics, namely Optimality
Theory, has actually proposed the elimination of rules altogether, on
the grounds that they are not psychologically real(istic).  In place of
rules, Optimality Theory proposes a set of ranked constraints, which
interact to produce the patterns we see in language.  The constraints
say things like "avoid structure X" or "avoid configuration Y", and the
rankings determine which of these constraints are given priority over
others.  It's an interesting way of looking at things, although I'm not
entirely convinced that it's anything more than a notational variant of
traditional rule-based approaches to linguistic analysis.

Matt.