Inversion Structures: some puzzles of reconstruction

Abstract

Inversion structures present a singular problem: arguments and sentential operators of the inverted phrase seem to c-command “out of” the phrase into the sentence remnant. This phenomenon is often referred to as “connectedness”. This paper points out that the class of inversion structures that exhibit connectedness is larger than just “inverse” copular sentences and pseudoclefts; it includes experiencer predicates with “extraposed” clausal theme arguments and also SOV language sentences with post-verbal elements. It then examines some of the solutions proposed for connectedness, in particular a solution in terms of an ?-reduction rule proposed by Heycock & Kroch (1999). It notes a hitherto-unnoticed problem with the last mentioned solution, namely that the rule works only when the variable of the pseudocleft is at the lower end of the clefted phrase. The paper then suggests that the interpretive mechanism of LF is in fact extremely simple: it appends (by adjunction) the sentence remnant at the bottom of the inverted phrase. This solution yields the correct results for all the inversion data.