Affectivity reconsidered

When Klima first proposed the feature [+Affective] to explain the licensing properties of negatives, interrogatives, and conditionals, it was really just a convenient way of labeling polarity triggers. In creating the label, however, Klima effectively advanced a hypothesis that polarity triggers as a group constitute a natural class in the grammar of English, and by implication in Universal Grammar. The history of polarity studies is in large part an effort to flesh out the intuition behind this hypothesis.

The history having proceeded for over forty years now, it is not too soon to reconsider the hypothesis and the intuition behind it. The assumption that polarity licensing in general can be explained by any single mechanism has already been largely abandoned: theorists who agree on little else – Linebarger (1980; 1987), Progovac (1994), van der Wouden (1997), Giannakidou (2006) – agree in positing distinct licensing mechanisms to account for the distinct sensitivities of different types of polarity items. One might go further, however, and wonder whether the sensitivities of any single polarity item can be adequately captured in terms of a single grammatical feature in the first place. The question is, is polarity licensing in fact a property of some class of linguistic representations (whether syntactic or semantic or of some other sort)? And is it sensible to define affective contexts in terms of necessary and sufficient (or even just necessary) conditions on linguistic structures?