LINGUIST List 4.136

Sat 27 Feb 1993

Disc: Subject/Object Asymmetry

Editor for this issue: <>

Directory

1. , Re: 4.126 Subject/Object Asymmetry
2. , 4.126 Subject/Object Asymmetry

Message 1: Re: 4.126 Subject/Object Asymmetry

Thanks to Joe Stemberger for clarifying my concern about linguistic modes
of argumentation. His case (2) whereby 'the hypothesis doesn't itself
make a prediction about the data, but, in combination with another assumption,
it might' is indeed a recurrent type in linguistic theory, and therefore
deserves such scrutiny. It appears to involve ABDUCTIVE, as opposed to
inductive or deductive reasoning. As explained in Eco's 'Semiotics and the
Philosophy of Language (p.40f), abduction involves taking a given result
(e.g. most idioms consist of verb + object) and a hypothesized rule
(e.g. verb + object forms a constituent), then reasoning that the result
COULD be a case of the rule. The rule is then seen as explaining the result.
When linguistic theories are said to explain a variety of phenomena, the
reasoning is often of this kind.
 Eco points out that abductive thinking plays an important role in creative
thinking and thereby in scientific advances, for example Kepler's theory
of the elliptical orbit of planets which was later supported by independent
evidence. The problem, of course, is that the reasoning is fallible in that
the Result may not be a Case of the Rule at all. Linguistic hypotheses
arrived at by this kind of reasoning are thus vulnerable to epistemological
critiques from skeptics like Manaster-Ramer! The challenge is thus to define
what counts as independent evidence to confirm or refute the 'leap of faith'
which is necessary in an abductive argument.
 This metatheoretical discussion is clearly important in that it has
consequences for methodology, e.g what kinds of data to look for,
as well as for the defense of our field (see JAckendoff's 'Topic...comment'
column, 'Why are they saying these things about us?' (NLLT 1987?)).
Steve Matthews (regularly under fire from metalinguistic skeptics
at U. of Hongkong)

The postings by Pesetsky and Stemberger on Feb. 23 in response
to my earlier comments about the proper response to the falsification
of claims in linguistics say a number of wise things, but perhaps
because I did not make the issues as clear as I should have, they
do not seem to address the central concerns.

Not to dwell on these things too much, I am concerned about two
points:

(a) In a self-respecting field of scholarship, it is better if
the falsification of a widely-made claim is openly acknowledged
than if it is not (thus, linguistics a la McKaughan is a not
just kinder, gentler but better scholarship than linguistics as
too often practiced).

(b) Although I think Joe Stemberger hints at this, too, it needs
to be made crystal clear that linguistics (especially perhaps
theoretical linguistics) is full of instances where something
is claimed to be an essential or at least a very important test
for certain kinds of hypotheses but in fact it turns out that,
strictly speaking, the hypothesis is UNFALSIFIABLE (which is,
of course, what allows us to be perfectly happy when we find
out that the test does not work as predicted). I would like
to see a statement of the theories concerning subject idioms
which IS clearly falsifiable, so that I can see once and for all
whether the examples I have falsify it, for example. As things
stand, I am not sure what the situation is (much as Chomsky
pointed out a few years ago in a different context that he
was not sure whether the existence of transformations was
an "empirical" issue).