Various discussions in the literature (e.g. Levin and Rappaport Hovav 1991) treat the formation of resultatives as a
“lexical rule”that changes and/or adds totheargumentstructureoftheverb. Thus in additiontotheregular verbcook
NP‘cause NP tobecome cooked’, there is also“inthelexicon”a further verbcook NPAP‘cause NP tobecome AP by
cooking (with it)’. As usual, to say this is“in the lexicon”in the sense“stored in long-ter m me mory”makes the otiose
clai mthat every appropriate verb carries around yet another extra argu ment structure fra me. The present approach
(advocated also by Goldberg 1995) and Jackendoff 1990a) is to say that the lexicon contains an idiomatic resultative
construction, and the meaning of (24) is the product of freely composing this construction with four semantic
arguments, expressed by the subject, the verb, the object, and the AP. The constructionitself comprises nothing but a
syntactic frame associated with a meaning—yet another violation of the Head Constraint.
(25) [VPv np ap],‘cause NP to become AP by V-ing ((with) it)’This lexical ite mhas no phonology. It is just a pairing of a syntactic and se mantic structure. However,this should not
be too disturbing. We already know of the existence of other such“defective”lexical items from the discussion in
section 5.7:helloandouchlack syntax, and do-supportdoand expletiveitlack semantics. So in principlethereshould be
nothing objectionable about a lexical ite mthat lacks phonology. (A more detailed for malization of the resultative
appears in section 12.2.)
To su mup, in each of these cases—theone's head offconstruction, thewayconstruction, thetime awayconstruction, and
the resultative—we have come to regard the construction not as a lexical rule that creates new verb argument
structures“in the lexicon,”but rather as a lexical ite min its own right that undergoes free co mbination with verbs.
What makes these cases unusual is thattheverbdoes notdetermine theVP's syntacticargumentstructure.Rather,the
construction itself determines the VP's syntactic argument structure, in violationof the Head Constraint, and the verb
satisfies a free position in the construction.
We have still omitted one combination of specified and free constituents: a specified verb with free positions to its
right. But thisis preciselythecase of a verb thatstrictlysubcategorizes its syntacticarguments. For example, as we saw
insection5.9.2,expressrequiresanNPargument,eventhoughsemanticallya clausewouldbeequallyappropriate(26b):