Two new papers by Michael Wagner have appeared.  Congratulations!

Contrastive topics decomposed. Semantics & Pragmatics 5.8 (2012): 1–54.
The analysis of contrastive topics introduced in Büring 1997b and further developed in Büring 2003 relies on distinguishing two types of constituents that introduce alternatives: the sentence focus, which is marked by a FOC feature, and the contrastive topic, which is marked by a CT feature. A non-compositional rule of interpretation that refers to these features is used to derive a topic semantic value, a nested set of sets of propositions. This paper presents evidence for a correlation between the restrictive syntax of nested focus operators and the syntax of contrastive topics, a correlation which is unexpected under this analysis. A compositional analysis is proposed that only makes use of the flatter focus semantic values introduced by focus operators. The analysis aims at integrating insights from the original analysis while at the same time capturing the observed  syntactic restrictions.
A givenness illusion. Language and Cognitive Processes 27.10 (2012): 1433–1458.
Constituents that encode information that is salient in the discourse or ‘given’ are oftenprosodically reduced and remain unaccented. What is given and new is usually defined at the level of meaning: given expressions are those that refer to salient referents or predicates that have been made salient by the previous discourse. This paper presents evidence from two production studies that sometimes, a constituent that semantically should be contrastive, and hence accentable, is treated prosodically as if it was given, and placing an accent on it is consistently avoided–an illusory case of givenness. This effect can be explained by assuming that givenness is not only evaluated in terms of semantic content, but also at the phonological level. Prosodically marking a semantic contrast requires the presence of a phonological contrast. This effect thus provides evidence that the notion of ‘‘antecedent’’ relevant for prosodic givenness-marking needs to include reference to linguistic form, and not just to referential meaning.