The status of P-NP constructions in HPSG
Ineke Schuurman
Centrum voor Computerlinguïstiek
K.U.Leuven
ineke.schuurman@ccl.kuleuven.ac.be
In Verspoor 1996 it is argued that some P-NP constructions (the for-dative
PPs and soem to-dative ones) are neither complements nor adjuncts but
should be treated as pseudo-complements instead. Syntactically such
pseudo-complements behave as complements, semantically as
adjuncts. However, their scope is not the sentence (i.e. the verb plus its
semantic complements) but something smaller: "the pseudo-complement
preposition specifies a relation between an element within the semantics of
the verb it modifies and the object of the preposition" (i.c., p.239)
1a) Adam sang a song
1b) Adam sang a song for his children
2a) John baked a cake
2b) John baked a cake for his mother
3a) Sam sent a letter
3b) Sam sent a letter to Bill
In the b-sentences above pseudo-complements are involved.
Some questions arise:
- What about Dutch?
- What happens when the tests are used with other kinds of P-NP constructions, for example temporal ones?
- How does this approach relate to an approach in which prepositions are considered to be either substantive or functional (cf. van Eynde)?