The aim of this project is to analyze Swedish polarity items (such as någonsin ‘ever’ or ett dugg ‘a
bit’) from a lexical semantic perspective. Polarity items (PIs) can be characterized as expressions
with limited syntactic distribution, sensitive to some semantic/pragmatic property of the clause in
which they occur. However, there is no consensus regarding the nature of polarity sensitivity. In
Brandtler (2010), I argue that the distribution of PIs in Swedish is dependent on the evaluative
status of the hosting clause. Supposing this hypothesis is correct, the next natural step is to shift
focus from the hosting environment to the hosted item. That is, what lexical property makes
certain items sensitive to evaluability?
Most PIs refer to scalar notions of amount, degree or intensity; this holds for both positive
and negative PIs (e.g. Swedish ett dugg ‘a bit’, or till och med / ens ‘positively/negatively even’).
Furthermore, negative PIs often function as emphatic focus markers, as in the sentence I have
never ever seen him before. However, it is not immediately obvious how (or if) the lexical property
of inducing scalar and/or focus readings relates to the notion of evaluability. The main purpose of
this project is to establish whether such a connection exists. If this is proven to be the case, the
Evaluability Hypothesis would be the first theory of polarity item licensing that takes into
account both the licensing environment and the sensitive item.