Bošković (2004) argued that a quantifier cannot float in θ-positions in English, German and other several languages. In this paper, I strengthen this generalization by offering an independent set of arguments from Korean and Japanese. It will be shown that floating numeral quantifiers in these languages pattern precisely like the quantifiers in other languages, so that they only occur in non-θ-positions. Notwithstanding these gratifying results, several gaps in the generalization remain to be explained. In this paper, I argue that these and other related questions are adequately resolvable under the DP Split Hypothesis, proposed by Takahashi and Hulsey 2008 (see Sportiche 2005 for a similar argument), the essence of which is that A-moved subjects need not have a full set of DP in θ-positions. More specifically, an FQ cannot occur in θ-positions because there is no complete DP to which it can be merged in θ-positions. It can only occur in non-θ-positions when a full-fledged DP is available.

This paper examines behaviors of floating quantifiers in English and Korean, and provides analyses for them. These two languages have ordinary quantifiers and numeral quantifiers. Their syntactic behaviors are a little different in two languages, and those differences also make differences in their meaning. When they do not float, they can have both cardinal and presuppositional reading. When they float, however, they have only cardinal reading. For their analyses, this paper adopts Heim`s tripartite structure and Diesing`s Mapping Hypothesis. In those analyses, when QPs are mapped into restrictive clause, they have presuppositional reading. But, when they are mapped into nuclear scope, they have cardinal reading. In sum, this paper provides some theoretical accounts for ordinary and numeral quantifiers of English and Korean.