gimbles NP 
for NP 
can be paraphrased as,
``NP acts on NP in order for NP to have NP ''. The
contribution of the for-PP can be identified as the ``in
order for NP to have NP '' portion of the paraphrase. In the
notation of verspoor:94, this is represented as a
HAVE event related by a for_to subordinating
relation to the main event expressed by the semantics of
gimble.
jack:90 argues that this HAVE event doesn't
properly capture the semantics contributed by the for-PP. He
claims that the event is rather forced when applied to certain
verbs. For example, when John sings a song for Mary, in what
sense does Mary have the song? Jackendoff therefore suggests
that the contribution of the for-PP is better described as
indicating that the object of the preposition (NP ) is intended to
benefit from
the action of the subject (NP ). The event embedded by the
for_to relation would more appropriately be as in
(3.56b) rather than Pinker's proposal of (3.56a). This event
represents ``NP affects NP positively'', or in other words,
``NP is intended to benefit from the actions of NP ''.
[Pinker's proposal]
[Jackendoff's proposal]
However, Jackendoff's proposal also does not seem to accurately
capture the interpretation associated with the PP in this form. The
benefit represented in his form is indirect -- since NP does not
act upon NP directly, what actually is intended to benefit NP
remains unclear. In fact, it seems that what is intended to benefit
NP directly is NP , the object upon which NP acts in order
to benefit NP . Thus it seems more accurate to represent the
semantics of the -dative as a relationship between the referent of
NP and the referent of NP . Incorporating a
benefit_rel relation from the representation developed in
Chapter 2, I propose that the semantics associated
with the -dative preposition should be as indicated in d53.
It is clear that this representation involves a pseudo-complement interpretation since the act argument of the benefit_rel relation is also a semantic argument of the verb. In addition, the treatment of the pseudo-complement modification will include incorporating the subordinated for_to relation directly into the internal semantics expressed by the verb rather than the external semantics.
In contrast, the -adjunct preposition (as in John ran a marathon for Mary) adds the semantic content in d69 to the representation of the full situation (to the external semantics). Its definition specifies that the influencing argument of the benefit_rel corresponds to the entire situation expressed in the sentence. The object of the preposition is therefore affected positively by the event (referred to via the sit-ind, situation index), rather than by a particular semantic element within the event representation.
Note that although Jackendoff (1990:195) suggests that the for-PP can be given precisely such an event interpretation, he provides no formal mechanism for doing so, or for distinguishing between the two possible interpretations of the for-PP. In Jackendoff's approach, the two different readings of the for-dative form must fall out of a single representation (that in (3.56b)), which fails to adequately reflect either reading and does not account for the identity of the interpretation of the double object form with only one of these readings (that in d53).
The distinction that pseudo-complements pick out a semantic argument from within the verb semantics while adjuncts incorporate the event expressed in the sentence as an argument in the relation they express is thus formalised in the semantics of the two forms associated with for. These will, upon formalisation, actually be captured by a single lexical entry for the preposition which can interact with the modified verb phrase in multiple ways. This interaction and the difference in where the semantic contribution is integrated with respect to the verb semantics -- that pseudo-complements contribute to the internal semantics of the verb (the verbal relation) while adjuncts contribute to the full situation expressed by a sentence (the external semantics) -- will be discussed in more detail in Section 3.5 and handled by the lexical rules which will be introduced there.
We will see that the double object form of the dative sentences is only possible on the pseudo-complement interpretation. This will account for the inability of some verbs to exhibit dative alternation, as shown in d103. In d103a only the adjunct interpretation of the PP is possible and hence the double object form is infelicitous. This syntactic structure will be licensed by a lexical rule which will constrain its interpretation to internal predication.
John extinguished the light for Mary. *John extinguished Mary the light.
