question on Preferred Arg. Str.

[Östen Dahl]

Jack Du Bois says: 
 
"Regarding the specific issue of reducing the number of Preferred Argument
Structure constraints (e.g. from 4 to 2), this has long been an obvious and
tempting target for linguists, given our interest in economy. But there’s
evidence that the constraints cannot be collapsed. Probably the most
extensive statistical argumentation has been offered by Arnold (2003) for
Mapudungun (see reference below). She shows that the One Lexical Argument
Constraint is not reducible to the Avoid Lexical A Constraint, nor is it
reducible to accessibility effects. (Related arguments have been given by
Goldberg 2004.)"
 
The question that Arnold discusses seems to be whether differences in
salience between intransitive subjects (S's) and transitive subjects (A's)
can explain why lexical arguments are used more for S than for A. However,
she does not, as far as I can see, address the question if the propensity
for lexical expression of A's differs depending on whether the direct object
is expressed lexically or not, as would be predicted if there is a One
Lexical Argument Constraint which is independent of other factors. The data
that I quoted in my previous posting indicate that there may be no such
difference between transitive clauses with lexical objects and non-lexical
objects. But maybe there is contradictory evidence somewhere else?

- Östen Dahl