1 Inherent Focus on Wh -

In the first part of the paper, I propose a framework for focus and question interpretation. While both serve to introcude alternatives, I offer reasons that the two kinds of alternatives arise on different levels of semantic interpretation. The resulting approach can correctly account for all possible constellations of focus and questions, including intervention effects, without the use of type shifts In the second part of the paper, I will investigate data which suggest that the presence/absence of focus features correlate with a question/indefinite interpretation of wh-words. I propose that the semantic closeness is motivated diachronically. Importantly, type shifts between question alternatives and properties are possible steps in diachrony, but not part of the productive part of semantic interpretation. PART I: Questions and Focus 1. Questions and focus: some combined models The first part of the paper offers a simple and concise account that integrates focus interpretation in the tradition of Rooth (1985), and question interpretation in the tradition of Hamblin (1973). While it is more than obvious that the two accounts should be combined (Rooth 1985, 1992 and many followers), specific proposals for a common syntax/semantics/pragmatics interface do not abound. In this first section, I will list some proposals and discuss why a genuine overarching analysis is still missing. 1.1 Rooth and Karttunen One elegant and troublefree way to combine focus and questions consists in taking a (Neo-) Karttunen computation of question semantics (= sets of possible answers) together with a Rooth type derivation of focus alternatives. The details of the derivation, specifically the question interpretation, have been explored in many papers about question semantics, e.g. Heim 1994, Beck 1996, von Stechow 1991 and others, and I will not recapitulate the technicalities here. However, those who really believe in the power of alternative semantics will not be satisified with this solution, because the crucial sets of propositions are computed in two entirely different fashions. The Rooth-Karttunen combination offers no clue whether question formation and focussing share semantic/pragmatic features, or whether they just lead to the same semantic objects at the end of the day, by sheer accident. 1.2 Beck 2006 Beck 2006 develops a very elaborate multi-dimensional Karttunen account where question alternatives and focus alternatives are modelled as sets of assignments for alternativeintroducing variables. The account computes (a) a parametrized property, (b) a set of possible value assignments to the parameters and (c) possibly, one ‘true’ instantiation of the parameter 1 Karttunen’s (1977) requirement that the meaning of a question be the set of true answers is usually dismissed, for reasons that would lead us too far here.