Characterizing Indirect Speech Acts

(1)

Characterizing Indirect S p e e c h A c t s 1

Gretchen P. Brown

Computer Corporation of America

575 Technology Square

Cambridge, Massachusetts 02139

This paper presents the core of a descriptive theory of indirect speech acts, i.e. utterances in which one speech act form is used to realize another, different, speech act. The proposed characterization of indirect speech acts is based on principles of goal formation, viewed in the context of a general structural model of action. The model of action is used to develop rules that characterize a large number of indirect speech act forms. Computational implications of the theory are discussed.

1. Introduction

In r e c e n t years, a considerable a m o u n t of attention has b e e n d e v o t e d to the topic of indirect speech acts, i.e. utterances in which one speech act f o r m is used to realize another, different, speech act. A simple example of an indirect speech act is the question form 1.1 u t t e r e d with the intent to c o n v e y a request to close the door.

1.1 Can you close the d o o r ?

Despite the volume of work that has b e e n done o n indirect s p e e c h acts, f u n d a m e n t a l questions r e m a i n unanswered. We still lack a complete answer to even the basic question of what forms can realize a given speech act. T w o properties of the problem have made the search for a c o m p l e t e t h e o r y of indirect f o r m s particularly difficult:

1 . . S h e e r numbers: T h e r e are a c o n s i d e r a b l e n u m b e r of different speech acts, and m a n y have a wide selection of possible indirect realizations. A t h e o r y must be quite general to take these into account.

2. Variety: Indirect speech act forms range f r o m highly c o n v e n t i o n a l i z e d to a p p a r e n t l y free forms. It appears that no single, simple set of generalizations can adequately capture the complexity of indirect speech acts.

1 T h e r e s e a r c h for this p a p e r w a s carried o u t while t h e a u t h o r w a s on t h e s t a f f of t h e L a b o r a t o r y for C o m p u t e r Science at the M a s s a c h u s e t t s I n s t i t u t e of T e c h n o l o g y . T h e r e s e a r c h was s u p p o r t e d b y t h e A d v a n c e d R e s e a r c h P r o j e c t s A g e n c y o f the D e p a r t m e n t of D e f e n s e a n d was m o n i t o r e d by t h e Office of N a v a l R e s e a r c h u n d e r C o n t r a c t N u m b e r N 0 0 0 1 4 - 7 5 - C - 0 6 6 1 .

It is the claim of this p a p e r that previous inv~estigations of indirect speech acts ( a b b r e v i a t e d ISAs 2) have b e e n h a m p e r e d by inadequate semantic theories. This study takes as primary the central t e n e t of speech act t h e o r y that language is action (Austin [2]) and brings to b e a r some of the perspectives on the r e p r e s e n t a t i o n of actions d e v e l o p e d in the course of Artificial Intelligence research. A c c o r d i n g l y , principles of goal f o r m a t i o n are discussed in the c o n t e x t of a general structural model of action. T h e model of action is used to develop rules that characterize a large n u m b e r of indirect speech act forms.

T h e focus of this investigation is o n the d e v e l o p - ment of a descriptive t h e o r y of ISAs. A c c o u n t i n g for the diversity of ISAs is an i m p o r t a n t goal, but I see the f o r m u l a t i o n of a solid and c o m p l e t e descriptive t h e o r y as a necessary prerequisite to an e x p l a n a t o r y theory. This is not to say that e x p l a n a t i o n can be totally d e c o u p l e d f r o m description, and, in fact, the use of the general m o d e l o f actions to derive ISA forms has significant e x p l a n a t o r y potential. T o fully a c c o u n t for d i f f e r e n c e s in ISA f o r m s , h o w e v e r , we must have a good c h a r a c t e r i z a t i o n of what these differences are.

While the claims that will be made in this p a p e r stop at a (partial) descriptive t h e o r y of ISAs, the un- derlying motivations do not. C o m p u t a t i o n a l consider- ations have played a significant role in the develop- m e n t of the ISA categorization. T h e work p r e s e n t e d here grew out of the i m p l e m e n t a t i o n e f f o r t r e p o r t e d in

2 It is helpful to p r o n o u n c e I S A as initials to avoid c o n f u s i o n with IS-A, t h e n a m e u s e d c o m m o n l y in t h e Artificial Intelligence literature for a hierarchical s e m a n t i c relationship.

Copyright 1980 by t h e A s s o c i a t i o n for C o m p u t a t i o n a l Linguistics. P e r m i s s i o n to c o p y w i t h o u t fee all or part of this material is g r a n t e d p r o v i d e d t h a t t h e copies are n o t m a d e for direct c o m m e r c i a l a d v a n t a g e a n d t h e Journal r e f e r e n c e a n d this c o p y r i g h t notice are i n c l u d e d o n t h e first page. T o c o p y o t h e r w i s e , or to republish, requires a fee a n d / o r specific p e r m i s s i o n .

0 3 6 2 - 6 1 3 X / 8 0 / 0 3 0 1 5 0 - 1 7 5 0 1 . 0 0

(2)

Gretchen P. Brown Characterizing Indirect Speech Acts

[3]. The actual behavior of the system was limited (internal manipulations for a t w e n t y turn sample dialogue), but the process model implemented was relatively sophisticated. A further e x p a n d e d and refined version of this model is p r e s e n t e d in [4].

In viewing the characterization of ISAs as a c o m p u - tational problem, the central premise has b e e n that the p h e n o m e n o n of ISAs is too c o m p l e x to admit to a single u n i f o r m c o m p u t a t i o n a l t r e a t m e n t . T h e two stumbling blocks to a descriptive t h e o r y -- the n u m b e r and variety of ISA forms -- are doubly troubling when the t h e o r y is to have a c o m p u t a t i o n a l application. Some means must be f o u n d to divide the class of ISAs i n t o subclasses which have their own specialized representations and processing strategies. The d e v e l o p m e n t of the descriptive t h e o r y of ISAs p r e s e n t e d has b e e n a f f e c t e d in various ways, subtle and not so subtle, b y this c o m p u t a t i o n a l hypothesis. T h e p r o p e r level of representation of ISA rules has b e e n of primary con- cern, as has the identification of classes of ISAs according to the complexity of their derivations.

Section 2 introduces some of the issues that have been raised a b o u t ISAs and Section 3 lays the ground- work for the a p p r o a c h t a k e n here. Section 4 t h e n presents a set of general rules that handle a large number of ISA forms. The rules in that section are proposed as the core of a descriptive t h e o r y of ISAs. Is- sues surrounding the application of the rules are ad- dressed in Section 5. Finally, Section 6 discusses some of the implications of the theory, with comparison to recent computational work.

2. P r e v i o u s A p p r o a c h e s

Two approaches to the characterization of indirect speech acts have b e e n particularly influential for b o t h computational and traditional linguists: the views proposed by G o r d o n and L a k o f f and by Searle. Since the rules p r e s e n t e d in this p a p e r c o m b i n e p r o p e r t i e s of each approach, we start with a brief description of each.

We consider first the a p p r o a c h t a k e n b y G o r d o n and L a k o f f [12]. C o n c e n t r a t i n g primarily on request, G o r d o n and L a k o f f propose a set of what they call sincerity conditions and t h e n give a single p o w e r f u l rule to account for the different ways that a request can be framed. T h e y say that to make a sincere request a s p e a k e r must, first, want the action d o n e , second, believe that the h e a r e r can do the action, third, believe that the hearer wants to do the action, and, fourth, believe that the h e a r e r would not do the action unless asked to. The first of these sincerity conditions is called speaker-based and the remaining three are called hearer-based. T h e rule given is:

One can c o n v e y a request by

(a) asserting a speaker-based sincerity condition or

(b) questioning a h e a r e r - b a s e d sincerity condition.

This f o r m u l a t i o n is attractive because it is so elegant and simple, but it is also, as the authors are the first to observe, only a preliminary answer. T h e conditions associated with request are incomplete, since t h e y lack any m e n t i o n of obligation relationships; these are discussed below in Section 3.3. M o r e problematic is the lack of detailed guidelines for e x t e n d i n g the t h e o r y b e y o n d requests.

A second major a p p r o a c h to ISA regularities is that of Searle. Searle presents a more complete a c c o u n t of ISAs, proposing generalizations associated with the five major classes of speech act defined in [26]. In [25] he lists four generalizations for directives and five others for commissives. T h e generalizations are differ- entiated according to the parts of the speech act identified in [24], i.e. p r o p o s i t i o n a l c o n t e n t conditions, sincerity conditions, and p r e p a r a t o r y conditions. ( G o r d o n and L a k o f f ' s sincerity conditions, in contrast, seem to be an amalgam of Searle's sincerity and pre- p a r a t o r y conditions.)

Searle's contribution is a valuable one, in that he has s u c c e e d e d in a c c o u n t i n g for a b r o a d range of speech acts. At the same time, Searle's generalizations can be questioned on the c o u n t that they are too specific. Generalizations are stated in terms of types of p r e p a r a t o r y conditions, rather than in terms of prepar- a t o r y conditions as a whole. A more serious problem is the relegation of the notion of speaker- and h e a r e r - b a s e d conditions to an informal role, as o p p o s e d to giving it an explicit place in the theory.

The t h e o r y p r o p o s e d in this p a p e r is b o t h a synthe- sis and a generalization of the two approaches. R a t h e r than derive ISA forms from a single set of conditions associated with the speech act, as do G o r d o n aild L a - koff, I follow Searle in looking for important classes of ISA forms based on different parts of the speech act. T h e t h e o r y p r e s e n t e d goes a step further, h o w e v e r , looking b e y o n d the structure of individual speech acts to derive ISA forms f r o m v e r y general principles of goal formation.

3. P r e l i m i n a r i e s

We first introduce the model of actions o n which the ISA rules will be based, and Section 3.2 looks at speech acts from the perspective of this model. Sec- tion 3.3 then discusses the request speech act as a basis for examples used t h r o u g h o u t the paper.

3.1. A n O u t l i n e of t h e S t r u c t u r e of A c t i o n s

(3)

G r e t c h e n P. Brown Characterizing Indirect Speech Acts

within the larger model, since s t r u c t u r a l i n f o r m a t i o n can be used in a n u m b e r of different ways. This subsection gives a very general t r e a t m e n t of actions, just enough to support the ISA rules proposed. T h e ac- c o u n t of actions is t a k e n from the O W L - I representation scheme (Szolovits et al. [27] and B r o w n [3,6]), 3 and it has c o u n t e r p a r t s in work by Bruce [7], Schank and Abelson [23], G r o s z [14], and M o o r e , Levin, and M a n n [18,19]. Some of these approaches differ in the type of action modelled, and all of them differ in the details, b u t each of the a p p r o a c h e s is o p e n to the t r e a t m e n t of action representations as general knowledge. Thus, action r e p r e s e n t a t i o n s are not m e r e l y programs for doing something, they are also knowledge structures that may be used by o t h e r processes.

We start with the notion of a method, a r e p r e s e n t a - tion of an action. M e t h o d s have three main parts: a h e a d e r , a r g u m e n t specifications, and a p r o c e d u r a l body. The h e a d e r is the m e t h o d ' s unique name. Argu- ment specifications, organized by semantic cases, are used for type checking of inputs to the m e t h o d (input cases) or to specify the form of results (output cases). T h e p r o c e d u r a l b o d y is divided into two parts: (optional) prerequisites and procedure steps.

N o t e that input cases are associated with methods, not surface English verbs. Input case specifications give constraints on the participants in the method, the materials used, objects manipulated, etc. (A suggested set of semantic input cases derived by William A. Mar- tin can be f o u n d in [4].) An important type of input case constraint, the precondition, is discussed in the next subsection.

Besides input case specifications, we said that methods may have associated o u t p u t case specifications, i.e. specifications of results. O n e i m p o r t a n t n o t i o n here is that of principal result, which is the main result of the m e t h o d and, typically, the reason that the me- t h o d is undertaken. F o r example, the action c o n v e y e d in "Paint the block red" has as principal result that the block is red. T h e paint brush may also end up red, but this is not the principal result.

Turning to the m e t h o d ' s procedural b o d y , we need to know that p r o c e d u r e steps may c o r r e s p o n d to sub- actions, i.e. they may be used as calls to o t h e r methods. B e y o n d this, p r o c e d u r e steps have a good deal of interesting structure, discussion o f which is not necessary for the purposes of this paper.

As for prerequisites, the ones that are of interest here are states. A stative prerequisite of an action is a condition that must obtain b e f o r e that action is carried out. If the condition does not hold, then one must

3 0 W L - I was developed by William A. Martin, Lowell Haw- kinson, William Long, Alexander Sunguroff, William Swartout, Peter Szolovits, and the author. OWL has continued to develop since that time.

bring it a b o u t b e f o r e carrying out the action. An example of a prerequisite is the r e q u i r e m e n t that an ele- m e n t a r y course of study be c o m p l e t e d b e f o r e a more advanced one is undertaken.

3.2. The Model of Actions Applied to Speech Acts

Speech acts, because they are actions, can be repre- sented by methods. Speech act r e p r e s e n t a t i o n s there- fore have semantic input cases, which typically include cases f o r the participants in the c o n v e r s a t i o n and a case for what Searle calls the propositional content condition of the s p e e c h act ( v e r y roughly, what the speech act is " a b o u t " ) [24]. A m o n g the constraints on these input cases are preconditions. Preconditions are constraints o n the beliefs, desires, or o t h e r intentions of the agent of the m e t h o d (the participant responsible for the action) that should be satisfied be- f o r e the s p e e c h act gets u n d e r w a y . P r e c o n d i t i o n s d i f f e r f r o m prerequisites in that a failure to satisfy preconditions typically means that a m e t h o d is elimi- n a t e d from consideration as a possible plan; a prerequisite that is not satisfied merely adds extra steps to be p e r f o r m e d . Preconditions will play an i m p o r t a n t role in the framing of ISAs; a sample set is given in the next subsection.

A c o n c e p t that will be useful in talking a b o u t ISAs is the intended effect. T h e i n t e n d e d effects of speech acts are those effects that P1 (the agent of the speech act) intends to have on P2. T h e most i m p o r t a n t of these effects will be called the principal intended effect. F o r request, the principal i n t e n d e d e f f e c t is that P2 take responsibility for carrying out some action. F o r offer, it is that P2 accept the offer. " A c c e p t " here includes not only a verbal acceptance, but also that P2 p e r f o r m some action that c o m p l e m e n t s P l ' s offer, e.g., P2 takes f o o d that is offered. T h e notion of i n t e n d e d e f f e c t comes f r o m V e r s c h u e r e n [28], but it has b e e n a d a p t e d somewhat. In particular, for uniformity, in- t e n d e d effects will be restricted to be actions only, not states. F o r example, the principal i n t e n d e d e f f e c t for state is that P2 come to know (as o p p o s e d to just know) that P1 believes something to be a fact.

I n t e n d e d effects and principal i n t e n d e d effects can be related in a straightforward way to methods. In- t e n d e d effects are actions precipitating certain m e t h o d results (i.e. i n t e n d e d e f f e c t s are the direct causes); principal i n t e n d e d e f f e c t s are actions precipitating certain principal results. T h e results and principal results are not necessarily associated with the speech act m e t h o d but are instead associated with higher level methods that include b o t h the speech act and its pro- totypical linguistic and nonlinguistic responses.

Once speech acts have b e e n set within the action r e p r e s e n t a t i o n , we can define ISAs m o r e closely to delimit the p h e n o m e n a of interest. Speech acts con- v e y e d b y ISA forms are derivable f r o m parts of, or

(4)

G r e t c h e n P. Brown Characl:erizing Indirect Speech Acts

conditions associated with, the c o n v e y e d speech act(s). O t h e r implications of an utterance may arise from a particular c o - o c c u r r e n c e of steps within larger patterns of dialogue, but these implications will not be considered to be c o n v e y e d speech acts. A very simple example of such an implication comes f r o m a c o m p u t e r console session e n v i r o n m e n t , w h e r e some users type " T h a n k y o u " in a place where o t h e r s type " T h a n k y o u " followed b y " G o o d - b y e " . When " T h a n k y o u " occurs alone in such a situation, it will not be considered to be an indirect closing. Instead, the closing is seen as an optional step, which may be omitted in the presence of utterances that uniquely identify the place in the dialogue. U t t e r a n c e s that imply omitted steps do so based on relationships at a higher level of dialogue structure than individual speech acts.

3.3. P r e c o n d i t i o n s for Requests

Although the rules p r e s e n t e d in the next section are intended to apply to speech acts in general, examples will be drawn primarily from request forms. Since the ISA rules d e p e n d in part on the p r e c o n d i t i o n s of a speech act, the preconditions of the speech act request are discussed here; preconditions for ask, state, offer, and suggest appear in the Appendix. 4 In these p r e c o n - ditions and t h r o u g h o u t the paper, P1 is the originator of the utterance (or written message) and P2 is the receiver. If s u b s e q u e n t related u t t e r a n c e s are discussed, then P1 and P2 continue to refer to the same participants. Consider, then, the following preconditions for request:

I. P1 wants P2 to take responsibility for carrying out the action.

II. P1 believes that P2 can take responsibility for carrying out the action.

III. P1 believes that P2 is willing to take responsibility for carrying out the action.

IV. P1 believes that P2 is obligated to P1 (and possibly to others) to take responsibility for carrying out the action.

T o clarify the terms used in the p r e c o n d i t i o n s , I will outline some of the relationships that should be c a p t u r e d in a semantic n e t w o r k r e p r e s e n t a t i o n of them.

The internal semantic node believe has a superclass relationship to semantic nodes for idea-holding concepts, e.g., thinking, knowing, assuming, and h y p o t h e s - izing. The different specializations (i.e. subclasses) of believe differ according to the strength of c o m m i t m e n t to the belief. In addition, they differ according to

4 In the i n t e r e s t s of readability, p r e c o n d i t i o n s a n d ISA rules are p r e s e n t e d in this p a p e r informally. T h e m o d e l of a c t i o n s a n d rules h a v e b e e n r e p r e s e n t e d in O W L - I , w h i c h implies a n u m b e r of c o m m i t m e n t s , m a n y s h a r e d by o t h e r r e p r e s e n t a t i o n s c h e m e s of the late seventies. T h e s e c o m m i t m e n t s are d i s c u s s e d f u r t h e r in Sect. 5.

w h e t h e r the belief is o p e n to c o n f i r m a t i o n against some e x t e r n a l reality (i.e., facts), will e v e n t u a l l y be open to c o n f i r m a t i o n (i.e., guesses and predictions), or is generally considered to be a m a t t e r of taste (i.e., opinions). The link b e t w e e n the various specializations of believe is the fact that beliefs can be partially supported by evidence, w h e t h e r or not complete con- firmation of the beliefs is ultimately possible. This excludes idea-holding actions such as dreaming.

In precondition I, the internal node want has a superclass relationship to semantic n o d e s for all goal- holding concepts, e.g., desiring and hoping. " T a k e responsibility" is used in the preconditions to permit subcontracting. W h e t h e r P2 does all the action steps or not, P2 still remains responsible to P1 for the re- suits.

In precondition II, " c a n " is m e a n t to c o n v e y the general notion of e n a b l e m e n t for actions. One specialization of the semantic node can is may, enable- ment through permission. The internal r e p r e s e n t a t i o n for " c a n " is discussed f u r t h e r in Section 5.

In the third precondition, the internal node for " b e willing" has a superclass in c o m m o n with want (perhaps called " b e inclined") but it differs in that if P2 is willing to do action A1, he or she is not disin- clined to do it. T h a t is to say, P2 does not necessarily have A1 as a goal, but P2 has no conflicting goals which, when weighed against A1, result in a decision against adopting A1 as a goal. P r e c o n d i t i o n III is w o r d e d " P 2 is willing t o " rather than " P 2 wants to," because P2 will not necessarily already have the action requested as a goal at the time that P1 makes the request.

(5)

Gretchen P. Brown Characterizing Indirect Speech Acts

version is the injunction to help s o m e o n e in an emer- gency. N o t e that this obligation is not absolute (nor are role or authority obligations), and it may be over- ridden b y o t h e r obligations. A point w o r t h mentioning is that my notion of obligation includes the notion of P l ' s right to invoke the obligation. (See [16], p. 78.) A n obligation is seen as a t h r e e - p l a c e relationship b e t w e e n P1, P2, and the thing that P2 is obliged to do. N o t e that the specific persons P1 and P2 n e e d not be n a m e d explicitly in the obligation. F o r example, the obligation to drive carefully m a y be an obligation to society in general, and h e n c e to any individual P1 b y inclusion in the larger set. G i v e n this formulation, P1 has the right to invoke the obligation to drive carefully because P1 is one of the parties to the obligation, even if P1 is not n a m e d explicitly.5

Philosophical c o n t r o v e r s y surrounds several of these terms, and a complete and detailed definition for any of them is a research project in itself. T h e com- ments o n the terms used in the p r e c o n d i t i o n s are sketchy, but the intent of the c o m m e n t s is to give the r e a d e r enough i n f o r m a t i o n to evaluate the a p p r o a c h to ISA c h a r a c t e r i z a t i o n p r o p o s e d in this paper.

4. A Set of General Rules for Indirect Speech Acts

If one looks carefully at a varied group of indirect s p e e c h acts, an outline of a c o m m o n sense view of rational b e h a v i o r begins to emerge. This c o m m o n sense view can be used as a c o n c e p t u a l organization of ISA forms, an a p p r o a c h t a k e n here in the p r e s e n t a t i o n of a set of general rules for ISAs.

4.1. S o m e Basic O b s e r v a t i o n s

We start with some very basic observations, none of which should seem particularly remarkable since the p h e n o m e n a involved lie just b e l o w the surface, and sometimes right at the surface, of English (and o t h e r languages as well).

Strategy 1. If y o u believe that a p r o p o s i t i o n holds, y o u can tell someone.

Strategy 2. If y o u want to k n o w w h e t h e r a proposition holds, y o u can ask s o m e o n e if it holds.

Strategy 3. If y o u w a n t to k n o w w h e t h e r a proposition holds, y o u can ask s o m e o n e if it does not hold.

T h e s e t h r e e c o m m u n i c a t i o n strategies are e x t e n s i o n s of the observations made b y Searle and built into G o r - d o n and L a k o f f ' s rule for requests. " C a n " is used a b o v e to indicate that o t h e r o p t i o n s do, o f course, exist; these are merely the options of interest for ISAs.

5 R e m i n d e r s are o n e class of u t t e r a n c e in w h i c h P1 d o e s h a v e t h e r i g h t to i n v o k e an o b l i g a t i o n w i t h o u t b e i n g a p a r t y to it. This is n o t n e c e s s a r i l y a p r o b l e m for requests, h o w e v e r , b e c a u s e r e m i n d e r s c a n be t r e a t e d as s e p a r a t e s p e e c h acts.

T h e three strategies can be a u g m e n t e d b y what will be called here the better-knowledge principle: if b o t h y o u and a conversational p a r t n e r have a degree of knowledge a b o u t a proposition, the decision w h e t h e r to tell what y o u k n o w (or think) or ask what the o t h e r per- son knows (or thinks) can be made based on which participant has the b e t t e r knowledge of the proposition.

Moving f r o m i n f o r m a t i o n exchanges to actions in general, we can i d e n t i f y some basic f a c t o r s in the process of u n d e r t a k i n g an a c t i o n (i.e. a d o p t i n g the a c t i o n as a goal, n o t necessarily with the i n t e n t of being the agent yourself).

1. One should only u n d e r t a k e actions that are necessary.

2. One should only u n d e r t a k e actions for which some desirable result or results can be expected.

3. One should only u n d e r t a k e actions that one expects to be possible.

T h e s e three maxims, which will be r e f e r r e d to as the maxims o f Necessity, Desirability, and Possibility, summarize factors that should be weighed in goal formation, the process of deciding to a d o p t some action as a goal. Necessity, desirability, and possibility of actions are not necessarily, of course, evaluated inde- p e n d e n t l y , b u t the maxims a b s t r a c t a w a y f r o m the actual weighing p r o c e d u r e . I n t e r p r e t a t i o n of these maxims is i n t e n d e d to be quite broad. " N e c e s s i t y " is assumed here to include obligations, and "possibility" is assumed to include having permission.

R e a d e r s familiar with the classic work of Grice on conversational implicature [13] will recognize the ap- p r o a c h that is being taken. Grice suggests four categories of maxims that are applicable to linguistic actions but which have analogues in o t h e r types of actions. T h e maxims given here are applicable to actions in general b u t apply to speech acts as a special case. The Maxim of Necessity above has a partial c o u n t e r - part in Grice's c a t e g o r y of Quantity. T h e o t h e r two maxims have no direct c o u n t e r p a r t s , and t h e y suggest extensions to Grice's framework.

G i v e n these basic observations a b o u t c o m m u n i c a - tion and action in general, the question is h o w they should be i n c o r p o r a t e d into a theory. O n e possible a p p r o a c h is to r e p r e s e n t the observations at essentially the level of generality given, t h e n derive ISA forms b y a u n i f o r m inference process. H e r e , in contrast, the observations will be used as a c o n c e p t u a l organization and as a guide to rule specification. T h e resulting rules will be more specialized, but they will be at a level closer to the ISA forms that they describe.

T h e m o t i v a t i o n for the choice of this a p p r o a c h can b e t t e r be described after the rules have b e e n present- ed. A c c o r d i n g l y , the rest of this section discusses

(6)

rules associated with the maxims of Necessity, Desira- bility, and Possibility.

4.2. Rules Related to t h e M a x i m of N e c e s s i t y

T h e Maxim of Necessity says that one should act only w h e n necessary, avoiding e x t r a n e o u s actions. T h e following rules a c c o u n t for speech act forms related to this maxim:

P 1 c a n c o n v e y a s p e e c h a c t i n d i r e c t l y b y --

Rule N E C E S S A R Y - A S S E R T

-- a s s e r t i n g that the intended speech act is necessary

e.g., the request "I have to ask y o u to shut the d o o r . "

Rule N E C E S S A R Y - A S K

-- a s k i n g w h e t h e r the intended speech act is necessary

e.g., the r e q u e s t " D o I need to ask y o u to shut the d o o r ? "

Rule E Q U I - A S K

-- a s k i n g w h e t h e r an equivalent s p e e c h act (i.e., one with the same principal i n t e n d e d e f f e c t ) has already b e e n p e r f o r m e d

e.g., the r e q u e s t " D i d a n y o n e ask y o u to take out the garbage?"

Rule F U T U R E - E F F E C T - A S K

-- a s k i n g w h e t h e r the principal intended e f f e c t can be expected to occur without the speech act

e.g., the requests:

" A r e you planning to take out the garbage?" " A r e y o u going to take out the garbage?"

Rule P A S T - E F F E C T - A S K

-- a s k i n g w h e t h e r the principal intended e f f e c t of the speech act has already occurred

e.g., the requests:

"Did you take out the garbage?" " H a v e you t a k e n out the garbage?"

and, using additional rules (see Section 5), "Is the garbage o u t ? "

" A s s e r t " is used in these and subsequent rules to in- elude the speech acts of stating a fact and giving an opinion, i.e., those speech acts that Searle calls repre- sentatives [26].

The "necessity" rules exemplify the three c o m m u - nication strategies listed at the beginning of this section. N E C E S S A R Y - A S S E R T exemplifies the first strategy, in which P1 tells what he or she knows a b o u t the necessity of the speech act. In N E C E S S A R Y - ASK, P1 asks w h e t h e r the speech act is n e c e s s a r y (strategy 2), and in the last t h r e e rules, P1 asks w h e t h e r the s p e e c h act is u n n e c e s s a r y ( s t r a t e g y 3). N o t e that there is no rule for the explicit version of the third strategy, e.g. for the form "Is it u n n e c e s s a r y f o r me to < s p e e c h a c t > ? " This f o r m is practically

incomprehensible as a way to carry out the speech act, even though the reasoning involved is c o m p a r a b l e to that for the E Q U I - A S K form. Perhaps this gap re- fleets a p r e f e r e n c e for more specialized forms. T h e t h r e e s t r a t e g y - 3 rules f o r necessity, which are m o r e specific, supersede the explicit "Is it u n n e c e s s a r y . . . " form.

G o r d o n and L a k o f f use a condition analogous to the F U T U R E - E F F E C T - A S K rule to a c c o u n t for the "Will y o u < a c t i o n > ? " r e q u e s t form. In this interpretation, the form asks if the r e q u e s t is unnecessary because P2 was going to p e r f o r m the desired action any- way. While this a p p r o a c h is plausible on the face of it, some uses of the "will" f o r m are not motivated b y questions of the necessity of the action. Consider, for example, 4.1:

4.1 Will y o u accept a ride to the airport?

One can view example 4.1 as P1 asking P2 w h e t h e r the o u t c o m e of an o f f e r by P1 will be successful (i.e., acceptance). This example can t h e r e f o r e be a c c o u n t e d for b y the Maxim of Possibility; "will" forms are discussed f u r t h e r in Section 4.4.

Finally, n o t e that P1 is p e r m i t t e d to use an ISA only when P1 can r e a s o n a b l y e x p e c t P2 to decipher P l ' s intent, i.e., to recognize the indirection. N e i t h e r the " n e c e s s i t y " rules nor any of the o t h e r rules to be p r e s e n t e d here, however, include this information. It appears that this constraint is part of a more general c o n s t r a i n t that P1 avoid ambiguity. T h a t is, P1 is obligated -- to the best of his or her ability -- to frame a n y utterance (ISA or not) in such a way that P2 can u n d e r s t a n d the message that P1 i n t e n d e d to convey. See Grice [13] for discussion of an " a v o i d ambiguity" maxim.

4.3. Rules Related to t h e M a x i m of Desirability

Next we come to the M a x i m of Desirability, which says that one should initiate actions for which some desirable result or results can be e x p e c t e d and avoid actions for which an undesirable result or results can be expected. Related to this maxim, we have the following ISA rules:

P 1 c a n c o n v e y a s p e e c h a c t i n d i r e c t l y b y --

Rule D E S I R A B L E - A S S E R T

-- a s s e r t i n g that some desirable result or results can be e x p e c t e d or some undesirable result or results can be avoided for some i n t e n d e d e f f e c t o f the speech act.

e.g., the r e q u e s t "I'11 be happier when y o u sub- stantiate those figures."

(7)

Rule D E S I R A B L E - A S K

-- a s k i n g w h e t h e r s o m e desirable result or results c a n be e x p e c t e d or w h e t h e r s o m e undesirable result or results can be a v o i d e d f o r some i n t e n d e d e f f e c t of the speech act.

e.g., the request "Will m o r e light c o m e in if y o u m o v e it a little to the right?"

Rule U N D E S I R A B L E - A S K

-- a s k i n g w h e t h e r s o m e undesirable result or results c a n be e x p e c t e d f r o m the i n t e n d e d s p e e c h act.

e.g., the request "Will y o u be o f f e n d e d if I ask y o u to loan m e s o m e m o n e y ? "

F o r the first t w o rules, n o t e t h a t the i n t e n d e d effect n e e d not be an i m m e d i a t e result of the speech act; it m a y be several times r e m o v e d in the causal chain. Similarly, the desirable result n e e d not be an i m m e d i - ate result of the i n t e n d e d effect.

T h e " d e s i r a b i l i t y " rules e x e m p l i f y the three linguistic strategies listed at the b e g i n n i n g of this section. Again, as f o r the s t r a t e g y - 3 " n e c e s s i t y " rules, D E S I R A B L E - A S S E R T a n d D E S I R A B L E - A S K do not include the m o s t general possible forms. F o r e x a m p l e , no rule has b e e n given to permit e x a m p l e 4.2 to be i n t e r p r e t e d as an indirect request t h a t P2 be quiet.

4.2 I will be h a p p i e r if I ask y o u to be quiet.

W h e r e a s D E S I R A B L E - A S S E R T is f r a m e d in t e r m s of an i n t e n d e d e f f e c t of the s p e e c h act, e x a m p l e 4.2 r e f - ers to the s p e e c h act explicitly. T h e s a m e hypothesis applies f o r this gap: the m o r e s p e c i a l i z e d D E S I R A B L E - A S S E R T has displaced the explicit, and m o r e general, f o r m exemplified b y 4.2.

4.4. Rules Related to t h e M a x i m of Possibility

T h e third m a x i m p r o p o s e d was the M a x i m of Possi- bility: one should only initiate actions that o n e e x p e c t s to be possible. This m e a n s that a s p e e c h act should only be initiated when:

1. P1 has the a p p r o p r i a t e a u t h o r i t y or p e r m i s - sion for the speech act; and

2. it a p p e a r s likely t h a t the specific p r e c o n d i - tions a s s o c i a t e d with the a c t i o n ' s m e t h o d c a n be satisfied.

O n l y the second case, preconditions, will be consid- e r e d here. T h e I S A f o r m s derived f r o m the first case all s e e m to b e l o n g to a class that F r a s e r has called h e d g e d performatives, a n d which are well a c c o u n t e d f o r in [11].

T h e a p p r o a c h t a k e n f o r ISAs b a s e d on p r e c o n d i - tions will be to distinguish three classes of p r e c o n d i - tion a n d f o r m u l a t e six rules using the classes distin- guished. T h e classes will be b a s e d o n the b e t t e r - k n o w l e d g e principle f r o m the beginning of this section, specialized in t e r m s of preconditions. T h e classes of p r e c o n d i t i o n are as follows:

1. P l - b a s e d p r e c o n d i t i o n s

H e r e P1 has i n h e r e n t l y b e t t e r k n o w l e d g e of w h e t h e r or n o t the t o p i c o f the p r e c o n d i t i o n holds. T h e topic of p r e c o n d i t i o n s t h a t b e g i n h e r e with "P1 believes t h a t P 2 " is c o n s i d e r e d to be the direct o b j e c t of the initial " b e l i e v e , " i.e., the rest of the precondition. F o r o t h e r p r e c o n - ditions, the topic is the entire p a t t e r n . P r e c o n d i - tions t h a t are P l - b a s e d r e p r e s e n t i n t e n t i o n a l states of P1, i.e., beliefs, intentions, wants, desires, a n d d e g r e e s of willingness. A n e x a m p l e is request I, P1 w a n t s P2 to t a k e responsibility f o r carrying out the action.

2. P 2 - b a s e d p r e c o n d i t i o n s

H e r e P2 has i n h e r e n t l y b e t t e r k n o w l e d g e of w h e t h e r or n o t the t o p i c of t h e p r e c o n d i t i o n holds. P r e c o n d i t i o n s t h a t fit this c a t e g o r y include P l ' s beliefs a b o u t P 2 ' s intentional states. 6 A n e x a m p l e of a P 2 - b a s e d p r e c o n d i t i o n is request III, P1 believes t h a t P2 is willing to t a k e responsibility for carrying out the action.

3. U n m a r k e d p r e c o n d i t i o n s

F o r these preconditions, d e t e r m i n a t i o n of which p a r t i c i p a n t has b e t t e r k n o w l e d g e of the p r e c o n - dition d e p e n d s on p r o p e r t i e s of the c o n t e x t or its particular s p e e c h act. E x a m p l e s are request II a n d IV.

U s i n g t h e s e p r e c o n d i t i o n t y p e s , we c a n c o n s t r u c t the following six rules for I S A forms.

P1 can convey a speech act indirectly by --

Rule P 1 - A S S E R T :

-- asserting a P l - b a s e d p r e c o n d i t i o n o f the s p e e c h act; e.g.,

"I w a n t y o u to w a t e r the p l a n t s . " (request I) "I h o p e y o u will use c o m m o n s e n s e . " (request I)

Rule P 2 - A S K :

-- an a s k of the topic of a P 2 - b a s e d p r e c o n d i - tion of the s p e e c h act.

e.g., " D o y o u w a n t to shut the d o o r ? " (request I I I )

Rule U N M A R K E D - A S K :

-- an a s k of the topic of an u n m a r k e d p r e c o n d i - tion of the s p e e c h act.

This rule applies in a c o n t e x t w h e r e P1 believes P2 has b e t t e r k n o w l e d g e of the condition in the p r e c o n d i t i o n topic.

e.g., " I s it y o u r t u r n to do the d i s h e s ? " (request I V )

6 A n e x c e p t i o n is the d e g r e e of k n o w l e d g e of facts, w h i c h will be c l a s s i f i e d as an U n m a r k e d c o n d i t i o n . T h e r e are c a s e s in w h i c h P1 is a s s u m e d to h a v e b e t t e r k n o w l e d g e a b o u t w h a t P2 k n o w s or d o e s n o t k n o w t h a n P2. Such an a s s u m p t i o n u n d e r l i e s t h e use of t h e f o r m " Y o u d o n ' t k n o w < f a c t > " as a w a y to s t a t e t h e fact.

(8)

Rule U N M A R K E D - A S S E R T :

-- asserting the topic of an u n m a r k e d p r e c o n d i - tion of the speech act.

This rule applies in a c o n t e x t w h e r e P1 believes P1 has b e t t e r k n o w l e d g e of the condition in the precondition topic.

e.g., " I t ' s y o u r t u r n to do the dishes." (request I V )

Rule C O M P O S I T E - R E Q U E S T :

-- a request f o r m of an action that is a goal of P1.

This rule is applicable only w h e n the speech act has preconditions that are exact m a t c h e s or specializations of the f o u r preconditions of request.

e.g., " T a k e a c o o k i e . " (offer I V - V I I , in A p - pendix)

Rule C O M P O S I T E - A S K :

-- an ask a b o u t w h e t h e r P2 will take responsibility f o r carrying out an action t h a t is a goal of P1.

This rule is applicable only w h e n the speech act has preconditions that are exact m a t c h e s or specializations of the four preconditions of request.

e.g., "Will y o u accept a ride to the a i r p o r t ? " (offer I V - V I I , in A p p e n d i x )

T h e rules as written do not a c c o u n t for differences in tense and mood. T h a t is, the U N M A R K E D - A S K rule accounts f o r e x a m p l e 4.3 but not 4.4 and 4.5.

4.3 Are you able to drive Sarah to school? 4.4 Will you be able to drive Sarah to school? 4.5 Would y o u be able to drive Sarah to school?

E x a m p l e s 4.4 a n d 4.5 c a n be h a n d l e d as legitimate requests if we e x t e n d the rules to a c c o u n t for a wider range of tense and m o o d behavior. See [4] f o r sugges- tions.

T h e " p o s s i b i l i t y " rules given also do n o t derive " n o t " forms, i.e. strategy-3 rules related to w h e t h e r an action is impossible.

4.6 Shouldn't y o u shut the d o o r ? 4.7 C a n ' t y o u shut the d o o r ?

N o t e , h o w e v e r , t h a t the rules given c a n be used as p a t t e r n s for producing rules that a c c o u n t for e x a m p l e s 4.6 and 4.7. A n y of the rules a b o v e that involve an ask h a v e rule c o u n t e r p a r t s with not inserted a f t e r the ask. 7

In t e r m s of specific rules, U N M A R K E D - A S S E R T m a y seem odd w h e n applied to preconditions involving

7 This is a change from [4], where the "not" forms were seen as realizations of a different speech act, with different preconditions. The "not" forms are now seen as requests with a particular set of connotations. The motive for the change is to make the "possibility" rules consistent with the rules for the maxims of Ne- cessity and Desirability by allowing the strategy of questioning whether a condition does not hold.

capability, producing indirect requests such as e x a m p l e 4.8.

4.8 Y o u can o p e n the door.

Such f o r m s do occur, h o w e v e r , particularly in requests to children where there m a y be s o m e question of the child's capability to p e r f o r m the action requested.

C O M P O S I T E - R E Q U E S T a n d C O M P O S I T E - A S K differ m o s t f r o m rules in p r e v i o u s t h e o r i e s b e c a u s e t h e y are b a s e d on g r o u p s of p r e c o n d i t i o n s . T h e C O M P O S I T E - A S K rule is of special interest. In S e a r l e ' s s c h e m e , the v e r y c o m m o n "Will y o u < a c t i o n > ? " f o r m is d e r i v e d f r o m the p r o p o s i t i o n a l c o n t e n t condition of directives (the class that includes request). This a p p r o a c h seems to p r o d u c e the c o r r e c t forms, b u t it is basically a structural account, w i t h o u t s t r o n g s e m a n t i c m o t i v a t i o n . I n s t e a d , the a p p r o a c h t a k e n here is to a p p e a l to the M a x i m of Possibility. T h e a p p e a r a n c e of the four request preconditions in a set of p r e c o n d i t i o n s indicates an action that P1 w a n t s done. We can think of a speech act with this p r e c o n - dition subset as an act with a c o m p o n e n t request. By using a "will" f o r m to p e r f o r m the s p e e c h act, e.g. the offer e x a m p l e 4.1, P1 is asking a question a b o u t h o w P2 will r e s p o n d to the offer, i.e., w h e t h e r the c o m p o - n e n t request will h a v e a s a t i s f a c t o r y response. W h e n the speech act is itself a request, t h e n the question in the "will" f o r m is w h e t h e r the speech act as a whole will h a v e a s a t i s f a c t o r y response.

T h e distinction b e t w e e n P l - b a s e d , P 2 - b a s e d , a n d U n m a r k e d p r e c o n d i t i o n s is p r o b a b l y u n c o n t r o v e r s i a l ; the question is w h e t h e r the categories should be given a p r i m a r y place in the theory. T h e r e a s o n that the b e t t e r - k n o w l e d g e split has b e e n given a central place is that there is t h e n a distinction b e t w e e n k n o w l e d g e a b o u t a p r e c o n d i t i o n t h a t is i n d e p e n d e n t of c o n t e x t (P1- and P 2 - b a s e d ) and that which is not ( U n m a r k e d ) . I n s t e a d of deriving the invariant k n o w l e d g e f r o m first principles e a c h time, it is " p r e c o m p i l e d " into the rules. This choice reflects an a p p r o a c h t h a t will be discussed in Section 6.

4.5. The S c o p e o f t h e Rules

(9)

general, force shift seems to be used to give P1 the appearance of greater benevolence or to save face for P2.

What these phenomena have in common, I think, is their "second order" nature. All can be seen as pre- supposing a set of rules and then deviating from them. I expect these phenomena to be modelled by the mechanisms for rule application, not accounted for by individual rules alone. Since such mechanisms could be expected to build on, and interact with, the "first order" rule application mechanism, these phenomena have been considered beyond the scope of the current investigation.

The rules in this section are proposed to hold for speech acts in general. ISAs are not grounded solely in individual speech acts, as for Gordon and Lakoff, or even in types of speech acts as for Searle. Instead, they are related to a broader view of rational action analogous to that expounded by Grice. Speech acts, because they are actions, do have structural compo- nents that play an important part in the derivation of ISAs. The driving force behind ISAs, however, is the process of goal formation, i.e. the process of deciding whether to adopt a speech act as a goal. This process is reflected in the three maxims that were used as a conceptual organization for the presentation. This emphasis on the goal formation process is closely related to the work of Allen, Cohen, and Perrault [1,8,21]. The similarities and differences between the two approaches are discussed in Section 6.

5. Relating Utterances to the General Rules

The general ISA rules in the last section were illus- trated with English sentences, but nothing has as yet been said about the correspondence between particular utterances and rules. This section discusses in broad terms the nature of the correspondence, focusing on differences in complexity. Because the topic is difficult to present in a neutral way, it will be approached from the point of view of language recognition, i.e., matching utterances against rules. Much of what is said, however, is relevant for generation as well. Note that discussion in this section is restricted to the issue of proposing correct matches; issues related to choos- ing between alternative interpretations of an utterance (i.e., alternative matches) are deliberately avoided. (See, however, [4]).

5.1. Levels of Matching Complexity

A n y discussion of matching rests on a set of assumptions about the representations involved. We briefly outline some of the assumptions made here, starting with a distinction between two levels of representation: surface and internal.

Each utterance is expected to have (at least) a surface representation and an internal representation. Internal representations are also used for action structures, including preconditions, and for ISA rules and patterns. Internal representations are organized in a knowledge base according to a semantic network for- malism. Surface representations closely reflect the surface form of an utterance, and only those distinc- tions forced by the parsing process are made. Thus, noun group references not needed by the parser may remain unresolved (e.g., "I saw him"). Choices among systematically ambiguous relationships of con- stituents and choices among ambiguous word senses also need not be made unless they are forced. ISA forms are preserved; e.g., "Can you close the door?" would have a surface representation that records its interrogative nature and that contains a surface item corresponding to "can."

An important implication of these attributes is that surface representation draws from a different vocabu- lary of semantic items than internal representation. For example, the surface item "believe" used in repre- senting "I believe you're fight" is related to, but is not the same as, the internal item "believe" that corre- sponds to the general idea-holding concept from Sec- tion 3.3. Surface items do, however, have associated internal level definitions which specify the ways that they can be translated into internal level representations. These definitions include various potential translations; context is typically called on in each individual case to choose among alternatives and to specify details.

The problem for ISA matching, then, is to relate the definitions of items in the surface representation of an ISA to ISA patterns. The ISA patterns are produced by applying the general rules from the last section to the method representations of particular speech acts. Matching will be discussed using as an example the pattern produced by applying rule UNMARKED- ASK to request precondition II:

P1 asks whether P2 can take responsibility

for carrying out action A.

Consider, then, the following examples interpreted as indirect requests.

5.1 Can you close the door? 5.2 Are you able to close the door? 5.3 Are you permitted to close the door? 5.4 Can you please close the door?

5.5 Will you be home in time to walk the dog? 5.6 Have you got a hammer to put up that hook? 5.7 Must you smoke?

5.8 Can you reach the salt? 5.9 It's cold in here.

(10)

Gretchen P. Brown Characterizing Indirect Speech Acts

Examples 5.1 to 5.3 can be handled by a set of general purpose matching rules that reflect hierarchical relationships in the k n o w l e d g e base. T h e " c a n " in example 5.1 matches can in the a b o v e ISA p a t t e r n , since we can expect surface " c a n " to have internal can as a major part of its associated definition. ( O t h e r c o m p o n e n t s in the definition might include the c o n n o - tations of the lexical item.) Similarly, "are y o u able" in example 5.2 is an exact match, since we can also e x p e c t its associated definition to c o n t a i n can as a c o m p o n e n t . In example 5.3, " p e r m i t t e d " has m a y as a c o m p o n e n t in its definition, and in the knowledge base m a y is a specialization (i.e., subclass) of can.

T h e important point in these matches is that elements of internal level definitions of surface items are related to elements of ISA patterns via the hierarchical relationships of the k n o w l e d g e base, i.e., via p r e d e - fined classification links. Thus, proposing the request i n t e r p r e t a t i o n for examples 5.1 to 5.3 involves relatively well u n d e r s t o o d knowledge base manipulations.

Example 5.4 is a typical utterance that is not ac- c o u n t e d for by the ISA patterns given. The problem is that example 5.4, a question according to its interrogative form, contains " p l e a s e " , a construct reserved for request and related s p e e c h acts. U t t e r a n c e s of this form have b e e n much-discussed in the literature (e.g. Sadock [22], Searle [25], and M o r g a n [20]). T h e question is w h e t h e r this form has evolved to the point that it is "really" a request only, no longer also a question. The interest is fueled by questions of the nature of meaning that are involved. Because I am interested in focusing on generalizations possible a b o u t ISAs, these issues will be omitted from discussion here. It is w o r t h noting, h o w e v e r , that, w h a t e v e r the ultimate theoretical disposition of these forms m a y be, they will p r o b a b l y have to be handled in a computational system by specialized patterns, to represent their unique properties. One such representation scheme, closely related to M o r g a n ' s n o t i o n of s h o r t - c i r c u i t e d e n t a i l m e n t , is given in [4].

The rest of the examples above, 5.5 through 5.9, can be related to the " c a n " request p a t t e r n in a regular fashion, but they require a richer set of matching relationships. E x a m p l e 5.5 is typical of examples for which proposing a match m a y turn out to involve arbi- trarily c o m p l e x i n f e r e n c e . T o begin to a c c o u n t for example 5.5, we can posit some link b e t w e e n can in the ISA p a t t e r n and a r e p r e s e n t a t i o n for being in the appropriate spatial proximity to do an action. This link may be hierarchical, or part of a definition related to the internal node can, or both.

This t r e a t m e n t does not, however, fully solve the problem exemplified b y example 5.5. T h e r e is still a good distance b e t w e e n the relationship of being at h o m e with a dog and the idea of being in the fight range to p e r f o r m the action of taking it for a walk.

The level of detail in the u t t e r a n c e is so much more specific than the level of detail in the p a t t e r n that we c a n n o t e x p e c t a match b y merely traversing p r e c o m - p u t e d links in a knowledge base. A n o t h e r d i f f e r e n c e b e t w e e n 5.5 and the previous examples is that the knowledge n e e d e d to propose the m a t c h m a y go be- y o n d i n f o r m a t i o n c o n v e y e d b y the u t t e r a n c e to information from the surrounding context. Either of these two factors has the potential to turn the process of proposing interpretations for u t t e r a n c e s such as example 5.5 into a full-blown inference process, with all the a t t e n d a n t difficulties in controlling the inference.

T h e rest of the examples can be e x p e c t e d to be m o r e tractable, b e c a u s e we can take a d v a n t a g e of specialized links in the model of actions i n t r o d u c e d in Section 3. Example 5.6 makes a " c a n y o u " request b y asking w h e t h e r P2 has an assignment (the h a m m e r ) for the instrument semantic case of the action (putting up a h o o k ) . Several different types of semantic cases can be q u e r i e d in this way (see [4]); the structural model of actions supplies links b e t w e e n actions and their cases that can be traversed in this match.

F o r examples 5.7 to 5.9, we can again exploit the model of actions to propose matches. T h e three examples m a y have request i n t e r p r e t a t i o n s w h e r e the actions intended are, respectively, that P2 stop smok- ing, pass the salt, and close the window. N o t e that n o n e of the t h r e e examples describes these actions explicitly, and for that reason I have called u t t e r a n c e s of this class implicit-action I S A s . These three examples r e p r e s e n t t h r e e classes of implicit-action ISAs, which differ in the complexity of the search n e e d e d to p r o p o s e a match. F o r example 5.7 there is essentially no search; the implicit action is merely that P2 stop or avoid the action named. Example 5.8 names a prerequisite of the implicit salt passing action. Recall f r o m Section 3.1 that prerequisites are a m o n g the basic parts of methods. O t h e r c o m p o n e n t s of actions, including semantic input cases, steps, and principal re- suits, may also be used in implicit-action ISAs. All of these are related to the action by the explicit links of the m e t h o d representation. Finally, example 5.9 al- ludes to the i n t e n d e d action b y stating a basis for the action, i.e., a condition seen as sufficient to warrant the action. "Basis for action" can be related to the structural links of m e t h o d s (see [4]), but the relationship b e t w e e n the condition n a m e d in the utterance and the implicit action is relatively complex. Implicit- action ISAs are discussed in more detail in [5].

(11)

G r e t c h e n P. Brown Characterizing Indirect Speech Acts

unsolved, but the links identified at least specify the types of paths that we can expect to see in matches of ISA patterns.

5.2. Embedded ISAs

In the discussion of matching, the initial assumption was that matching of surface representations o c c u r r e d against p a t t e r n s p r o d u c e d b y single applications of rules to speech act methods. This assumption makes no provision f o r e m b e d d e d forms. Some e v i d e n c e does exist f o r this approach. F o r example, Sadock [22], in a n o t h e r context, observes that 5.10 is not a request for the hearer to move over, even though the similar form 5.11 is.

5.10 Tell me if you can move over. 5.11 Can y o u move over?

In terms of the rules p r e s e n t e d in Section 4, a request interpretation for 5.10 could only come from an em- b e d d e d rule application: the U N M A R K E D - A S K rule applied to request II, resulting in ask, t h e n the C O M P O S I T E - R E Q U E S T rule applied to four of the preconditions of ask (see the Appendix). Forbidding such a double application effectively blocks a request i n t e r p r e t a t i o n , leaving only the i n f o r m a t i o n - s e e k i n g alternative.

This s t r a i g h t f o r w a r d solution, a u g m e n t e d b y various classes of exceptions, was a d o p t e d in [4]. T h e e m b e d d e d examples that have accumulated since, however, are too numerous to be a c c o u n t e d for simply as exceptions. Consider the following indirect requests:

5.12 I w o n d e r if you can move over.

5.13 I believe it's your turn to do the dishes.

F o r 5.12, the internal level definition of wonder would include the following information:

P1 w o n d e r if < a c t i o n or s t a t e >

1. P1 wants to know if < a c t i o n or s t a t e > 2. P1 is speculating if < a c t i o n or s t a t e >

Example 5.12 is eharacterizable by applying rule P1- A S S E R T to ask I (P1 wants to know the answer to the question) a f t e r applying rule U N M A R K E D - A S K to request II. Example 5.13 is eharacterizable by applying rule P 1 - A S S E R T to state I (P1 believes X is a fact) a f t e r applying rule U N M A R K E D - A S S E R T to request IV.

These examples, and others like them, seem to be best a c c o u n t e d for by politeness conditions. 8 In particular, I suggest the following hypothesis: embedding of general ISA rules is p e r m i t t e d w h e n it f u r t h e r s the politeness intentions of P1, either to heighten polite-

8 "Politeness" is used here quite broadly to include not only observation of the conventions of etiquette, but also the expression of respect for the other participant and the expression only of emotions harmonious with the social expectations associated with the conversational environment.

ness or to lessen it. These processes are r e f e r r e d to here as mitigation and aggravation, respectively. ( T h e terms are b o r r o w e d from L a b o v and Fanshel [16] but apply to a s o m e w h a t b r o a d e r range of p h e n o m e n a here.) Embeddings within rules that are u n m a r k e d for politeness are forbidden, as are embeddings where the rules involved have conflicting politeness markings.

Evidence for this hypothesis is f o u n d in comparing example 5.12 to 5.14:

5.14 I want to k n o w if y o u can m o v e over.

Example 5.14 is derivable from the same set of rules as 5.12, but 5.12 conveys a request force while 5.14 does not. T h e r e a s o n for this, I suggest, is that the U N M A R K E D - A S K rule is a mitigator: questions, in general, p r o m o t e politeness b y giving P2 an o p p o r t u n i - ty to answer, allowing P2 to refuse to a c c e p t P l ' s goals in uttering the speech act. "I w o n d e r " is similarly undemanding: the emphasis is more on the specula- tion process P1 is involved in than the "wanting to k n o w " aspect. In contrast, the "I want to k n o w " in 5.14 works in the direction of aggravation. A goal stated explicitly leaves P2 v e r y little r o o m to refuse P1 without doing so explicitly. In narrowing P2's options, P1 has lowered the level of politeness. E x a m p l e 5.12, then, with b o t h rule applications working in the direction of mitigation, is a p e r m i t t e d embedding. E x a m p l e 5.14, with o n e rule application p r o d u c i n g mitigation and one aggravation, is blocked as an indirect request.

This a p p r o a c h can also be used to explain the block on embedding in example 5.10. T h e C O M P O S I T E - R E Q U E S T rule realized with an imperative is not a mitigator, while the U N M A R K E D - A S K rule realized with a question is. Thus, the indirect request interpretation is blocked.

T h e examples p r e s e n t e d make a case for the use of politeness c o n d i t i o n s to g o v e r n ISA rule e m b e d d i n g , but it must be emphasized that more work is needed. Despite the work on politeness conditions, m u c h of this area is not well u n d e r s t o o d . ( F o r three different perspectives on the implications of ISA choices, see L a k o f f [17], Davison [9], and E r v i n - T r i p p [10].) Con- elusive p r o o f or disproof of the hypothesis awaits an analysis of the implications of ISA choices at a level of detail and completeness that is not yet available.

6. Computational Implications

This p a p e r has c h a r a c t e r i z e d a significant n u m b e r of ISA forms, with a t t e n t i o n to r e p r e s e n t a t i o n a l issues. As n o t e d in the introduction, these are the only claims made, although the ultimate m o t i v a t i o n of the work was not only c o m p u t a t i o n a l but was d i r e c t e d b y a particular c o m p u t a t i o n a l philosophy of language recognition. This philosophy will be described briefly here, with emphasis on the way that the t h e o r y p r e s e n t e d fits into it. Due to the n u m b e r of issues involved, I