Bracketing paradoxes and particle verbs: a late adjunction analysis

It is well known that words like unhappier give rise to bracketing paradoxes (Pesetsky 1979, 1985; Kiparsky 1982; Sproat 1992; Lieber 1992; Hoeksema 1987, among many others): their phonological structure seems to be in conflict with their semantics. This paper will propose that the solution to this puzzle rests on the following generalization: all morphological bracketing paradoxes must involve a morphological adjunct. It is argued here that all morphological bracketing paradoxes involve one morpheme whose contribution to the word involves no projection of features.