It has been assumed not only by linguists but also by philosophers and psychologists that causation is a function over two events. Using the semantic function CAUSE as a common element underlying all the verbs which imply causation, we can formulate this assumption as in the following:
(1) CAUSE(E1, E2) (E=EVENT)
The purpose of the present paper is to discuss whether or not the above assumption is valid in terms of linguistic semantic representation.
Section 1 gives an outline of the analyses based upon the assumption. It deals with the treatments made by scholars from various fields, such as Davidson (1967), Vendler (1967), Miller & Johnson-Laird (1976), Dowty (1972) and Schank(1973). Section 2 is concerned with how to define the event in linguistic terms. Two tests are proposed which distinguish event expressions from non-event ones. Section 3 discusses the issue of whether or not the first argument of CAUSE must always be an event and that of whether or not the "by"-phrase as in 'John made Mary leave the party by telling dirty jokes' fills in the action structure of the first event argument of CAUSE. It follows from the discussion that while the second argument of CAUSE is explicitly an [EVENT], its first argument can take a [THING] or a [STATE] as well and that the means expression cannot be taken to be part of the first argument of the CAUSE function; it is simply a restrictive modifier of the CAUSE function. In Section 4 I argue against an attempt to reinterpret as an event a subject NP such as 'Cleopatra's nose' in 'Cleopatra's nose made the course of history change' by recovering what is missing from the context. It is shown that such an analysis will admit indefinitely many events between a cause and its effect and will thus lead to a reductio ad absurdum of the proposed analysis, as pointed out in Fodor et al.(1980).