Conclusions

  Pseudo-complements are elements very close in nature to true verbal complements. They specify a particular relation between a semantic argument of the verb and the object of the preposition. Their semantics can, however, be treated by the same mechanisms as true adjuncts. Pseudo-complements and adjuncts share the property of specifying a consistent, contentful, and identifiable relation which can be applied across modified heads.

Specification of the type of modification which an individual PP may provide with respect to a head must occur lexically. Three types of PPs will be allowed for in the treatment of PPs to be introduced in Section 3.5 (in particular, in the sort hierarchy) -- PPs which can only behave as pseudo-complements, PPs which can only behave as adjuncts, and PPs which are ambiguous between the two.

Prepositions which are strict complements will not be represented as a distinct semantic type in the sort hierarchy. Complement PPs which mark a verbal argument according to the normal interpretation of the head preposition can be treated semantically as pseudo-complements: their semantics, marking a verbal argument, is simply integrated into the verbal relation directly in the lexical entry through unification, rather than through an external combination process as will be defined for standard pseudo-complementation. Thus there will be no semantic (type) distinction between the -PP complements of give and kick, for example, although they will be licensed in different ways (for give in the lexical entry and for kick through the mechanisms of pseudo-complementation).

The type of a particular preposition must then be specified in the lexical entry of the preposition. This type will be used as a criteria for determining how semantic integration between the semantics of the PP containing the preposition and the semantics of the modified verb is to occur.