Abstract Broccias

Arguments in resultative constructions: From asymmetric to non-inheriting resultatives (via metonymy)

One of the dimensions of variation in the analysis of so-called resultative constructions (RCs) is what distributional relation obtains between the arguments of the RC and the arguments (either optional or obligatory) of the RC verb. Goldberg and Jackendoff (2004), henceforth GJ, propose Full Argument Realization (FAR): “All the arguments obligatorily licensed by the verb and all the syntactic arguments licensed by the construction must be simultaneously realized in the syntax, sharing syntactic positions if necessary […]” (p.547). Crucially, they consider an argument as “obligatorily licensed by a verb [iff] an expression involving the verb in active simple past tense without the argument is ill-formed” (p.548).

FAR correctly accounts for the impossibility of e.g. (1) *The bears frightened the campground empty, from Levin and Rappaport Hovav (1995), henceforth LRH, with the intended meaning of “The bears frightened the hikers and, as a result, the hikers left the campground”. LRH themselves explain the impossibility of (1) by showing that unsubcategorised objects of transitive verbs cannot appear in RCs. The only exception they note involves wash-verbs (e.g. (2) She washed the soap out of her eyes, cf. *She washed the soap). However, it can be shown that the lack of inheritance of obligatory verbal objects as constructional objects (i.e. objects in the RC) is not limited to wash-verbs:

(3) She frightened the Ten Commandments *(into her children).

(4) She kicked a hole *(in the door).

(3) and (4) examples, as well as (2), do contain the verbal object (e.g. her children, cf. She frightened her children) but such an object appears in the resultative phrase slot, i.e. the verb – used independently of the RC – and the RC are asymmetric with respect to the subcategorised object (hence, such RCs are termed asymmetric resultatives here).

Whereas LRH’s model cannot handle cases like (3)-(4), FAR can since it does not specify where (i.e. through which constructional argument) the subcategorised object must be realised. At closer inspection, however, GJ’s approach also turns out to be problematic. Firstly, it does not explain what principles, if any, account for the positioning of verbal arguments in the RC (e.g. why her children appears in the resultative phrase slot in (3)). Secondly, and even more importantly, obligatorily subcategorised objects are not always realised in RCs (these are called non-inheriting RCs here):

(4) She cut herself from her family. (cf. She cut the ties with her family.)

(5) Cole {headed/nodded} Chelsea in front. (i.e. Cole scored a goal by hitting the ball with his head so that his team, Chelsea, went one up against their opponents.)

(4) is similar to an asymmetric RC but shows that metonymy can affect the coding of subcategorised objects in the RC (i.e. family stands for ties with her family). (5) shows complete lack of inheritance of the obligatory, subcategorised object ball (cf. Cole {headed/nodded} *(the ball)), contrary to what FAR predicts. Finally, FAR may be viewed as problematic also because it is not explained where GJ’s test for obligatory argumenthood comes from.

In view of these problems, I will propose an alternative approach to asymmetric and non-inheriting RCs which does not rely on the notion of obligatory argumenthood but rather claims that (a) the distribution of arguments in the RC depends on the conceptual feasibility of the scenario evoked by the “change complex” (i.e. what is termed “small clause” in formal approaches) and (b) unsubcategorised objects in the RC are allowed iff the comceptualiser can establish tight conceptual links between the verbal event and the change event (i.e. the event coded by the change complex).