Do We Wanna (of Hafta) Have Empty Categories?

This paper revisits the relevance of wanna contraction for the existence of empty categories. An analysis of wanna contraction is proposed under which empty categories have a role in blocking contraction. It is then shown that alternative analyses (subject sharing, local c-command, and morpholexical analyses) are inadequate. It is suggested that empty categories are a last resort, which can only appear in non-subject LDD constructions because these have to be licensed by inside-out functional uncertainty equations, and the empty categories are needed as the c-structure positions to which these equations are annotated.