Situations and Speech Acts

Toward a Formal Semantics of Discourse

David A. Evans, David A. Evans

• Publisher: Routledge
• Year: 2016
• ISBN:
  9781315401768
• Edition: 1
• Topic: Lingue e linguistica
• Subtopic: Linguistica

First published in 1985, this book aims to develop an approach to speech acts that has the virtue of being straight-forward, explicit, formal and flexible enough to accommodate many of the more general problems of interactive verbal communication. The first chapter introduces situation semantics with the second addressing the assumptions implied by the problem of representing speaker intentionality. The third chapter presents a streamlined theory of speech acts and the fourth tests the predictions of the theory in several hypothetical discourse situations. A summary and suggestions for further research is provided in chapter five, and appendices facilitate reference to key concepts.

Chapter 1 Situations and Speech Acts

1.0. Introduction

This chapter has four goals. First, an apparatus for describing situations. viz. a situation semantics á Ia Barwise and Perry (1981a), is presented, along with a brief discussion of some of the motivation for using situations as the basis of meaning representation. and some of the theoretical consequences of such an orientation. The notion of situation-type is explicated, accompanied by formal definitions, and forms the basis of the semantics of discourse developed later. Second, it is argued that discourse constitutes a distinguished situation, a discourse situation, which has properties that affect linguistic interpretation at all levels. Third, speech acts are described in terms of their effects on discourse situations, and it is argued that current theories treating speech acts are inadequate to deal with the variety of speech act phenomena encountered in actual discourse. Finally, desiderata for a formal theory of speech acts are presented.

1.1. Situations and Situation-Types

Before presenting the brief introduction to situation semantics that this chapter requires, it is worth considering some of the factors that motivate the use of situations as distinguished descriptive entities. To do this, it is useful to consider, further, the difference between a scene and a situation.

When we look out upon the world, a scene presents itself to our view. The scene consists of objects with properties, standing in relationships. The situation of which the scene is a part, contains us, the observers, and all that is observed.

But the situation contains much more than what we observe. As though obeying the Law of Plenitude,¹ it contains all the objects that are there, and all their properties and relationships, not just the ones on which we fix our attention. The scene we view represents a small subset of the elements of the situation and is, in a sense, an artifact of the observation process: we cannot attend to all the elements of a situation, but only to a handful, in the time the scene is accessible to us. This limitation is a limitation in our cognitive capacity, not a limitation inherent in situations under observation (at least down to the atomic level²).

Now, a scene is part of not just one situation, but of an infinite number of situations. This multitude of situations reflects the infinite number of possible partitionings of the space-time environment in which the scene is set. Still, the facts of the scene are not etiolated by this plethora of embedding possibilities: what is true of the scene is true regardless of the situation containing it. This feature of scenes gives them a powerful ability to characterize the superset situations of which they are parts. And it is this feature that is exploited in situation semantics.

The object of situation semantics is to provide a formal characterization of situations, taking objects, properties, and relations as primitives. One of the most useful means of describing situations is through situation-types, which can be thought of, intuitively, as minimal scenes: highly limited information about situations. The situation-type in the semantic theory can be regarded as the analog of the activated network in psychological theory, and is designed to capture certain facts of cognitive processing.

When we observe a situation, or think about it, or talk about it, we are forced to choose some features and exclude others. If we look out into a yard and fix our attention on a nearby tree. the surrounding shrubs, grass, and objects of all kinds wash into a blur. We are aware of the tree, its leaves. its shape, its color, perhaps, but of only a handful of the many relationships the tree has to its environment. Nevertheless, the features of the tree that we bring into focus are true, independent of the many other ways the tree and its environment might be related. And what we know about the tree from our limited observation provides us with accurate information about some aspects of all the situations of which the tree is a part.

In situation semantics, we define a situation-type to be a partial function characterizing various kinds of situations. More precisely, if we let A represent the set of objects, { a₀, a₁, a₂, ... }, and R the set of sets of n-place relations, { R⁰, R¹, R², ... }. with R⁰ being the set of 0-place relations (one of which might give, for example, the state of it raining); R¹, the set of properties, or 1-place relations (including, for example, the property of being green); R², the set of 2-place relations (including, for example, the relation of one person hitting another person); etc.; then we can define a situation-type, s, to be a partial function that returns the values 0, 1, or undefined for any relation and set of objects depending on whether that relation holds of that set of objects for the situations of which s is the type. We write:

(1)	s(r^n, a1, ... , an) = l	iff a1, ... , an are in the relation rn in the set of situations given by s,
	s(r^n, a1, ..., an) = 0	iff a1, ... , an are not in the relation r^n in the set of situations given by s, and otherwise
	s(r^n, a1, ... , an) = undefined.

Suppose. for example, that we are looking at a room containing, along with its unexceptional furnishings, a red balloon and a green telephone. If we let a₀ designate the room, a₁, the balloon, and a₂, the telephone, and assume that contain is a 2-place relation, and that be-red and be-green are properties (1-place relations), then the following situation-types can be given:³

• (2) s(contain, a₀, a₁) = 1
  s(contain, a₀, a₂) = 1
  s(be-red, a₁) = 1
  s(be-red, a₂) = 0

This captures some of the (partial) information we have about the situations that have this room as a part. Of course, we could list many other facts about these situations, including the following:

• (3) s(contain, a₁, a₀) = 0
  s(he-green, a₁) = 0
  s(be-green, a₀) = undefined

The virtue of the situation-type is that it enables us to encode accurate information about situations without requiring that we know everything about the situations. And there seems to be in situation-types a natural way of representing the meanings of sentences.

As argued in Barwise and Perry (1981b), there is an essential difference in sentences of the following type that cannot be captured in post-Fregean approaches to semantics that take the interpretation (=reference) of a sentence to be its truth-value:

• (4) Aristotle said that the meaning of a sentence is proportional to its length.
• (5) Cicero thought that silence is golden.

Though both (4) and (5) are false, their subject matter is very different. Furthermore, their embedded sentences contribute to the interpretation of the whole by representing some situation that we must understand if we are to determine the attitude being expressed and, ultimately, the truth conditions on the whole. We would want a semantics of English to have a way of representing the differences in (4) and (5) explicitly, while offering us the possibility of evaluating the statements made as equally false. This is possible if we let these sentences determine situation-types, since the situation-types associated with (4) and (5) will be quite distinct.

These examples underscore two orientations lurking in situation semantics which should be made explicit. The first is philosophical; the second, psychological. The philosophical commitment situation semantics makes is to a strong realism. There is no distinction between the objects of situation-types (partial functions) and the objects of situations (slices of space-time). In the psychological realm, this commitment is translated into an antiphenomenalism. When we see a tree, it is the tree itself that we see, not some pattern of sensations.

In terms of linguistics, this leaves us in a very interesting position. Since words are the names of objects and relationships, and since sentences designate situation-types, which are partial characterizations of situations, the object of inquiry in linguistics should be nothing less than the situations of the actual universe. Grammatical rules can be regarded as a means of encoding relationships, so a failure of grammar results in ill-formedness not so much because of language rule violation as because it creates a situation-type for which there is no corresponding situation in our experience.

There is, of course, an clement of Whorfianism in this position, with attendant chicken-and-egg problems. Does the situational structure of the world constrain cognition, or does some cognitive predisposition affect and limit our perception of the structure of the world? Do language rules reflect the situational environments in which they develop, or does a tradition of arbitrary grammatical form become associated with some possible situations in the world and not others, inhibiting our ability to view the world ingenuously?

Situation semantics does not address these issues directly; it is, after all, a descriptive formalism. Indeed, it leaves the question open while accommodating linguistic determinism.

For example, the situation-type

(6) s(want, Brynn, the-ball) = 1

gives just those situations in which the person designated by Brynn (i.e., Brynn) is in a want relationship to the person or object designated by the ball (i.e., the-ball). Whether or not there are, in fact, situations that have this type is an empirical matter. Conventionally, in situation semantics, the problem is whether or not S, the set of situations having s as their type, is contained in W, the set of situations that obtain in the world. If S ꜃ W, then (6) gives the type of a set of real situations; if not, then (6) gives the type of a set of imaginary situations.

The linguistic problem raised by this example is how the interpretation of the use of an utterance (or inscription) of the form

(7) Brynn wants the ball.

can have a situation-type such as (6) as a part. Consider just the lexical issues. Brynn must be associated with a particular individual, Brynn; the ball must be associated with a particular individual or object, the-ball; and want must be associated with. the relation, want (which we might assume, here, gives the relationship normally understood as desiring to have or hold for a limited time, rather than other relationships subsumed under the English word want, such as lacking, or needing, or desiring to eat, etc.)

But this, of course, overly simplifies the matter. If (7) is a real utterance. then it is issued by someone. to someone. in the service of some intention. Its interpretation might involve distinguishing the word Brynn from the individual Brynn, if, for example, it is produced as part of an exercise in elocution. Or, its interpretation might involve distinguishing the state of a fact about the world, as encoded in (6). from the statement of a reason for a particular action, say, moving to get the ball; or from a request to an addressee to get the ball. Or, its interpretation might involve recognizing that Rrynn is a certain kind of ind...