The purpose of this paper is twofold. One is to present an analysis of the adjunct on with an adversative effect as in (1) in terms of the representation of the conceptual structure:
(1)a. John's car broke down on him.
b. John's savings were wiped out on him.
It is proposed that sentences (la) and (1b) are represented as (2a) and (2b), respectively:
(2)a. [_<Event> INCH([_<State> BE([_<Event> JOHN_<i>'S CAR BREAK DOWN], [_<Place> ON([_<Thing> JOHN_<i>])])])]
b. [_<Event> INCH([_<State> BE([_<Event> JOHN_<i>'S SAVINGS BE WIPED OUT], [_<Place> ON([_<Thing> JOHN_<i>])])])]
The other is to show that the approach employed in the analysis of adversative on can be applied to the semantic description of three other classes of adjuncts.
Section 1 is devoted to a brief survey of the framework upon which the present study is based. The framework basically follows Jackendoff's model(1976, 1983). Section 2 provides evidence that argues for structures (2a) and (2b). The main focus is placed upon the presence of the function INCH-BE in (2) and the relation of adversative on to the 'experiential' have-sentences such as (3):
(3)a. John had his car break down.
b. John had his savings wiped out.
Section 3 argues that in parallel with the on-adjunct construction, the constructions of time and place adjuncts require BE and that of resultatives CAUSE, as the superordinate function including the main clause as its embedded structure, as in the following:
(4)a. In Paris John went to the Eiffel Tower. k b. [_<State> BE_<Posit>([_<Event> JOHN GO TO EIFFEL TOWER], [_<Place> IN ([_<Place> PARIS])])]
(5)a. Bill met Mary in 1964.
b. [_<State> BE_<tem>([_<Event> BILL MEET MARY], [_<Place> IN([_<Time> 1964])])]
(6)a. John hit the ball into the field.
b. [_<Event> CAUSE([_<Event> JOHN HIT BALL], [_<Event> GO([_<Thing> BALL], [_<Path> TO([_<Place> IN([_<Thing> FIELD])])])])]
If the present analysis is valid, it follows that the constructions of these adjuncts form on the semantic representation a single hierarchical structure, which clearly contrasts with Jackendoff's treatment of the modification structure and that it enables us to obtain the kinds of generalization that we otherwise could not.