10 - Event structure

Summary

INTRODUCTION

Abstracting from its noun phrases, a clause is semantically a predicate of events, (1). Call this the main event of the clause.

1. (1) λe[P(e)]

P might relate the main event to others, as in (2), which cashes out P as a predicate of events that bear R to some other event. R might be Cause, for example.

1. (2) λe[∃e′[R(e)′](e)]

A semantic analysis like this is called an event structure. The events to which the main event is related are called subevents of the clause. Sometimes subevents might be parts of the main event (Schein 1993, Rothstein 2004).

An event structure is complex when it includes more than one predicate of events. When a clause with a single audible predicate is given a complex event structure, this is a claim of semantic decomposition. I assume by default that decomposition is strict, in the sense of Chapter 2, but will sometimes consider alternatives.

This chapter is about a much-used class of event-structural decompositions that derive from sources including Lakoff (1965, 1971), McCawley (1971), Ross (1972) and Dowty (1972). These works advance analyses like (3–6), with (b) giving the meaning of (a).

1. (3) a The glass is hard.

2. b Hard(the glass)

3. (4) a Floyd viewed the glass.

4. b Do_p 〈Floyd, View (Floyd, the glass)〉

5. (5) a The glass hardened.

6. b Become _p 〈Hard(the glass)〉

7. (6) a Floyd hardened the glass.

8. b Cause_p ∃φ[Do_p 〈Floyd, φ〉], Become_p 〈Hard(the glass)〉〉

Here Do_p, Become_p and Cause_p are propositional operators. Event structures are often written in this format. Yet they are usually talked about as relations over events, and we can rewrite them explicitly in those terms. One possible revision is given in (7–10), roughly following Parsons (1990). For expository clarity I mark the main event as e₁ and leave it bound only by a lambda; the meaning of the complete sentence would have this variable existentially quantified.