These syntactic and semantic tests for investigating the status of the dative PPs have given conflicting evidence, making it difficult to convincingly establish PPs as either complements or adjuncts. We can, however, identify the following properties:
The most influential argument for the treatment of certain dative PPs as adjuncts rather than complements is that they seem to have a constant semantic contribution across all applications. The fact that these PPs often appear to behave as complements syntactically is overshadowed by the semantic generalisations which can be made by treating them as adjuncts. It does not seem to make sense to treat these PPs as idiosyncratically contributing semantic information to the heads they modify (or not contributing any semantics) when this semantic contribution is so consistent and predictable across word classes. A way of acknowledging their independent semantic contribution while accounting for the syntactic facts is needed. To do this, I appeal to Gawron's (1986) arguments that prepositions have consistent lexical content but that they can modify a verbal head in different ways. I will, however, classify the prepositions according to both syntactic and semantic properties while Gawron's classification is purely semantic.
All of the dative PPs are instances of something which I will call a pseudo-complement. This is an element which often behaves syntactically as a complement but which behaves semantically as an adjunct. The information expressed by the pseudo-complement cannot always be logically inferred from the use of the verb, but is somehow ``closer'' to the meaning expressed by the verb than true adjuncts (in Gawron's terms, they are internal predicators). This idea will be developed further in the section which follows.