People have talked about syntactic alternations as a) meaning-equivalent variants with choice entirely driven by processing pressures or b) different variants being associated with categorically different meanings --> we propose a tradeoff between the two: Gradient Alternation Hypothesis. (But probably start out more generally, ie point to the many-to-many mapping problem of forms to meanings and people's presumed desire to both maximize communicative success -- minimizing distance between their own and their interlocutor's representation of the world through language --and minimize effort).
Mark Myslin and Roger currently have their meaning-processing-pressure-tradeoff paper under review (I think), and they frame it all as an explaining away effect. I think we'll take a slightly different (broader, I think) take but should point to those results. In some ways, they're doing the opposite of what we're doing: given two forms that either do or don't violate processing pressures, how big are the associated meaning differences? What we're doing is saying: given forms that are independently rated as having more or less "extreme" meanings, how big is the effect of production pressures? But these approaches are obviously compatible and stem from the same general idea.
All the work for "some" vs "some of DT" is done; this will just be a matter of finding all the relevant code and files again. The question is whether we want to add the same sort of work on "all", looking specifically at what governs the choice between:
-
John ate all of the cookies.
-
John ate all the cookies.
-
John ate all cookies.
The "all" cases are interestingly different from the "some" cases, because there the ones we're looking at are only (4) and (6) with (5) ungrammatical:
-
John ate some of the cookies.
-
- John ate some the cookies.
-
John ate some cookies.
In the "all" case, it seems like we'd want to focus on the difference between (1) and (2), which, in contrast to the some cases, are meaning-equivalent (or so it seems), but leave out (3), the case that's analogous to the comparison case for "some". Reason: in that case it seems like quantification happens over kinds. I.e., for all kinds of cookies, it is the case that John ate (some of) them. Though it even sounds sort of weird, like you'd want it to be more like a law. Better, for example, are "John has eaten all cookies" or "John eats all cookies".
Not sure what to think about this (3) case, but it would be interesting to run our analysis on (1) vs (2).
The prediction: given our hypothesis that...
-
the choice of the partitive is the result of a tradeoffs between meaning and processing pressures
-
and there is no discernible meaning difference between (1) and (2)
...we should see that the processing pressures (including UID) have a greater effect on the choice between (1) and (2) than on the choice between (4) and (6).
Does this sound right? If so, then maybe a potential role for the (3) type sentences would be as a meaning-difference control where we expect the choice to not be as strongly governed by processing pressures. But then we'd have to get independent estimates of meaning differences for those three cases. Maybe with the same test we applied to the "some" cases, i.e. the leave-out-the-quantifier test? But more likely it would have to be something more clearly related to the actual meaning component that is contributed, e.g. by asking participants about kind quantification somehow.