The event structure of perception verbs (review)

This book is about much more than perception verbs. It ranges over causation, Aktionsarten, evidentiality, epistemic modality, polysemy, raising and control, the semantics of ditransitives, and further sundry topics at the syntax-semantics interface. These are discussed within the framework of Hudson’s (2007) word grammar (WG), though with ample reference to their treatment in other theoretical approaches.

Perception verbs have some unusual properties. Although see and hear can be used in transitive clauses, they lack some of the hallmarks of transitivity, such as an agentive subject and an affected patient, referring instead to a relation between an experiencer and a percept, with the ‘responsibility’ for the event being distributed over the perceptual abilities of the experiencer and the perceptible properties of the percept. When used in the present tense (in their basic perceptual senses), the verbs tend to be infelicitous with both progressive and nonprogressive aspect, typically requiring the use of the modal can. Even this use of can, though, is odd. It does not refer to ability as such, even less to permission, but to the actualization of the ability. There are also some puzzling passivization facts.

1. 1.

   1. a. I saw the dog cross the road.

   2. b. *The dog was seen cross the road.

2. 2.

   1. a. *I saw the dog to cross the road.

   2. b. The dog was seen to cross the road.

3. 3.

   1. a. I saw the dog crossing the road.

   2. b. The dog was seen crossing the road

Why is 1b unacceptable? What is wrong with 2a? What is the meaning difference between 1a and 3a? Do the examples exhibit the same sense of see?

Ch. 1, ‘Introduction’, distinguishes three classes of perception verbs: the hear-class (see, hear), where the subject refers to an experiencer; the listen-class (look, listen), the verbs of which take an agentive subject; and the sound-class (look, sound), where the subject designates the percept. (Feel, taste, and smell are, of course, polysemous with respect to these categories.) One aim of the book is to elucidate the relations between the three classes.

Ch. 2, ‘Word grammar’, consists of a tutorial on WG, the theoretical model adopted in the book. Especially highlighted is the WG emphasis on a language as a conceptual network, structured by relations of various kinds, one of the most important of which is the Isa (‘is an instance of’) relation.

Ch. 3, ‘Causation and relations between events: An introduction to WG semantics’, is mainly concerned with an analysis of causation. (The relevance of the account to perception verbs becomes apparent in later chapters.)As G documents, causation is a complex topic, subject to many different definitions and analyses. G sees as a necessary (though not sufficient) condition on causality the presence of two events, the second of which, event2, is regarded as the result of the first, event1. The specifics of event1 are often not stated. If I open the window, it is not made explicit exactly what it is that I do; the outcome is however clear, namely, that the window opened. As a matter of fact, intransitive open, as in The window opened, is also taken to contain a result component: event1 is the process of the window opening; event2 is the state of the window being open.

A result is entailed. I broke the window but it didn’t break is a contradiction; likewise with I gave her a present but she didn’t receive it. It would not, however, be a contradiction to state that I baked her a cake but she didn’t receive it (perhaps the dog ate it first). This example illustrates the relation purpose. Purpose and result are brought under the more general relation outcome.

While result is a necessary ingredient of causation, it is not sufficient. The window opened contains a result relation, but the sentence is not causative. Causation requires a force-dynamic component, whereby an initiator (Talmy’s (1988) antagonist) exerts force on an endpoint (Talmy’s agonist). One conclusion is that, contrary to...