Conclusions

  In spite of my criticisms of the implementational details of Wechsler's () approach to resultatives, in particular of the requirement of extra lexical entries specifically for the purpose of capturing resultative behaviour and the lack of independent justification for the specification of a background:telos value, Wechsler's intuitions about the semantic constraints on the different resultative constructions do go much further towards capturing the apparent semantic restrictions on this construction than the Levin and Rappaport Hovav explanation (in fact this is unsurprising given that focus on syntactic variations rather than semantic restrictions). This seems to be particularly true for the case of control resultatives in which the resultative phrase is predicated of the direct object of a transitive verb. Consider the sentence in w3b, repeated here as w11a, the further variations of the sentence in w11b (Wechsler , p. 2, (6c)), and sentences w12. In each case, the acceptable resultative phrases do seem to make specific some canonical goal of the activity.

 

John hammered the metal flat/*safe/*red. John hammered the metal into a triangle/smooth/shiny/into the ground.

 

John chiseled the ice smooth/into a bird/onto the floor. John chiseled the ice *cold/*shiny.

Wechsler's presentation of the treatment of resultative constructions in terms of the telos feature implies that there is a semantic generalisation about the inherent goal-oriented nature of certain events which is derivable from general world knowledge. I have shown in the preceding sections that this generalisation cannot be isolated from the syntactic construction for which it is an issue, and that the semantic restrictedness of the construction cannot be accounted for entirely in terms of this generalisation.

Rather than assuming that these verbs specify a highly specific canonical result (the examples above and in Section 4.4.2 show clearly that this canonical result cannot simply be required to be a state as this is not restrictive enough) which cannot independently be motivated, then, it seems to make more sense to argue that the semantic restrictedness hinges in part on the requirement of establishing a causal relation between the main event expressed in the sentence and the result state. This enables us to account for sentences such as w11-w12, and would also allow us to account for situations in which pragmatics could ameliorate a particular sentence. For example, in a context such as w13, the sentence w13c, infelicitous in a context-neutral environment, seems to be completely felicitous. Here the context allows us to establish a natural causal relation between hammering and the metal being safe. I will pursue this idea further in Section 4.7, in light of the approach suggested by Construction Grammar (Goldberg 1995).

 

The slide at the park had a section which had come loose. Several children had hurt themselves on the protruding edge. In order to prevent further injuries, John hammered the metal safe.