The interpretation of focus-sensitive operators such as ‘only’ and ‘even’ depends on the presence of a focused constituent in their scope. I document the complex conditions under which focus-sensitive operators are able to associate with a focused constituent which has moved out of their scope. In particular, I concentrate on the ability of English ‘even’ but not ‘only’ to associate “backwards” in this configuration. I propose a theory based on the Copy Theory of movement which predicts the attested patterns of backwards association. When an operator gives the appearance of associating backwards, it is in fact associating with focus in the lower copy of the movement chain, within its scope. Differences in the availability of backwards association between ‘only,’ ‘even,’ and ‘also’ are shown to follow from independent semantic differences. The proposal explains a range of constraints on patterns of focus association, and more generally contributes to our understanding of the interaction of the syntactic operation of movement with the semantic and information-structural notion of focus.

One consequence of this work is a new argument against the so-called “scope theory” of ‘even’: the idea that ‘even’ can be interpreted at LF higher than in its pronounced position. In particular, I show that my copy-theoretic analysis for backwards association with ‘even’ defuses Nakanishi 2012’s argument for the scope theory from patterns of antecedent-contained deletion. In fact, a closer look at examples of Nakanishi’s form forms a new argument against the scope theory.

Return to all projects