The talk analyses the grammar of event nominalisations, specifically problems
attending the realisation of DP arguments of nominalised verbs (the reading of books).
We focus mainly on English of-phrases, with side glances at German genitives.
I begin by trying to sort out what is right and wrong about the two main
approaches used in lexicalist and syntactic circles. Most current syntactic approaches
assume that affixes select (extended) verbal projections which contain (traces of) the
of-marked DP and that the nominalised verb unites with the affix using head
movement; (1) is a simple example.
(1) [NP read–ing [VP read [DP (of) the book]] (syntactic head movement)
Lexicalist approaches assume that of-marked DPs are inserted after affixation, cf. (2).
This entails that of-phrases are arguments of nouns and that affixes combine with
verbs with open arguments. I argue that this form of long-distance argument
realisation is correct, but also provide arguments that some nominalisations are formed
by syntactic head movement.
(2) [NP [N° [V° read] –ing] [DP/PP of the book]] (lexicalist-style structure)
I then give a syntactic analysis of of-phrases which reflects the following points:
- Of-phrases realising arguments of nominalised verbs are introduced by a device
whose primary function is to introduce arguments of nouns. Argument-realising
and possessive of are given a unified treatment, a conclusion supported by the
fact that both are unique in displaying a previously unstudied expletive/weak
use of the which displays no uniqueness or familiarity effects.
- (Non-extraposed) of-phrases appear to the left of other dependents in
nominalisations, which, coupled with other facts, suggests that N moves in front
of of.
- The possibilities for realising arguments as pre- or postnominal genitives (e.g.
John’sAgent shooting of BillPatient) is derived from principles affecting the
interpretation of simplex nominals (e.g. John’sCreator/Owner picture of BillDepictee).

I then turn to nominalisations without arguments. I suggest a preliminary analysis for a
little-known kind of exception to Grimshaw’s claim that obligatory patient/themes
arguments of V remain obligatory in eventive nominalisations (the device was faulty
despite frequent maintenance/checking by experts vs. *they maintain/check). The
behaviour of so-called ‘result’ nominals (the examination is on the table) and simple
event nominals (the trials lasted two hours) is analysed in parallel to similarly
interpreted compounds.
Finally I turn to the consequences of my proposal for secondary predicate
constructions. My approach explains the absence of of-marking in ECM and small
clause contexts straightforwardly (*the believing of John (to be) untrustworthy). It
commits me to a complex predicate analysis of resultatives, particle and prefix verbs
that undergo such nominalisation (the pressing of metal flat, the voting out of the
party). I present empirical evidence for this from non-argument-supporting complex
verbs (a rework/bailout would be a good idea).