grammaticalized discourse functions and information structure

[Martin Salzmann]

Dear all,

Has there been any recent work on the relationship between the
grammaticalized discourse functions and the information structural
categories of topic and focus? I'm only aware of work dealing with one
of these dimensions (such as Bresnan/Mchombo on DF and Choi (scrambling
and IS) on IS. An example: According to Bresnan (1994), locative
inversion in English and Chichewa differ e.g. with regard to the
structural position of the preposed locatives. Spec- of-IP in Chichewa
vs. IP-adjoined in English. They also differ as to their discourse
functions, the loc being a sub in Chichewa but a top in English. While I
agree on this analysis, I think that on some level of grammatical
structure, the discourse functional similarities of the locatives should
be captured - nobody wil dispute this, I guess,  they are BOTH topics,
after all, though only one of them is grammaticalized. The level of
information hinted at is, of course, IS. Would it be possible, to
eliminate the grammaticalized DFs completely and derive those cases
where discourse categories are in some way grammatically marked from a
more finegrained theory of IS?  For instance, I would speculate that
those topics that are topicalized in English differ information
structurally from those that are not. I'd appreciate any suggestions or
pointers to a discussion of this issue.
Back to the grammaticalized discourse functions: I still do not really
grasp what counts as grammaticalized. Cases where participants occupy
special phrase structural positions or those involving relative and
wh-pronouns are rather clear. But apart from those ... Is there a clear
defined notion of grammaticalized df? Again,  I'd be grateful for a
hint.

Best,


Martin Salzmann


University of Zurich