5 - The middle-passive voice: evidence from Albanian

Summary

Fifty years ago, Chomsky (1957: §5.4) argued that passive sentences should be excluded from phrase structure grammar and introduced instead by a transformational rule applying to active sentences. This was because introducing passives through rewriting rules would mean doubling the selectional restrictions independently imposed on actives, while a transformational rule would allow them to be stated only once. Chomsky (1965: 103–4) provides what has remained the standard conceptualization of this transformational process by proposing that

the Manner Adverbial should have as one of its realizations a ‘dummy element’ signifying that the passive transformation must obligatorily apply. That is, we … may formulate the passive transformation … with an elementary transformation that substitutes the first NP for the dummy element passive and places the second NP in the position of the first NP.

In current practice, the by-phrase is independently generated by Merge; but the analysis whereby passive is defined by ‘substitution’ of an internal argument for the EPP position (second or internal Merge) remains at the core of generative transformational grammar.

In this chapter we propose to evaluate this analysis in the light of data from Albanian, which provides two separate and complementary phenomena of interest. On the one hand, the passive (i.e. promotion of the internal argument to the EPP position with the external argument independently interpreted) has the same lexicalization as the reflexive, the anticausative and the impersonal. The question then is whether all of these different interpretations are associated with the same underlying syntax.