Speech Acts as a Basis for U n d e r s t a n d i n g Dialogue Coherence

by

C. Raymond P e r r a u l t and J a m e s F. Allen

Dept. of Computer Science
U n i v e r s i t y of Toronto
T o r o n t o Canada

and

P h i l i p R. C o h e n
Bolt Beranek and N e w m a n
C a m b r i d g e Mass.

i. Introduction Each agent m a i n t a i n s a model of the

world, including a model of the m o d e l s of
Webster's dictionary defines other agents. L i n g u i s t i c utterances are
"coherence" as "the quality of being the result of the e x e c u t i o n of o p e r a t o r s
logically integrated, consistent, and whose effects are mainly on the models
intelligible". If one were asked whether that the speaker and hearer m a i n t a i n of
a sequence of physical acts being each other. These effects are intended by
performed by an agent was coherent, a the speaker to be produced p a r t l y by the
crucial factor in the d e c i s i o n would be hearer's recognition of the speaker's
whether the acts were perceived as plan.
contributing to the achievement of an
overall goal. In that case they can This view of the c o m m u n i c a t i o n process
frequently be described briefly, by naming is very close in spirit to the Austin-
the goal or the procedure executed to Grice-Strawson-Searle approach to
achieve it. Once the intended goal has illocutionary acts, and indeed was
been conjectured, the sequence can be s t r o n g l y influenced by it. We are working
d e s c r i b e d as a more or less correct, m o r e on a theory of speech acts based on the
or less optimal attempt at the a c h i e v e m e n t notions of plans, world models, plan
of the goal. c o n s t r u c t i o n and plan recognition. It is
intended that this theory should answer
One of the m a i n s t r e a m s of AI research q u e s t i o n s such as:
has been the study of problem solving
behaviour in humans and its simulation by (i) Under what circumstances can an
machines. This can be considered as the observer believe that a speaker has
task of transforming an initial state of s i n c e r e l y and n o n - d e f e c t i v e l y performed a
the world into a goal state by finding an particular i l l o c u t i o n a r y act in producing
appropriate sequence of applications of utterance for a hearer? The observer
o p e r a t o r s from a given set. Each operator could also be the hearer or speaker.
has two m o d e s of execution: in the first
it changes the "real world", and in the (2) What changes does the successful
second it changes a model of the real execution of a speech act make to the
world. S e q u e n c e s of these o p e r a t o r s we s p e a k e r ' s model of the hearer, and to the
call plans. They can be constructed, h e a t e r ' s model of the speaker?
simulated, executed, optimized and
debugged. Operators are usually thought (3) H o w is the meaning (sense/reference)
of as achieving certain effects and of of an u t t e r a n c e x related to the acts that
being applicable only when certain can be p e r f o r m e d in uttering x?
p r e c o n d i t i o n s hold.
A theory of speech acts based on plans
The effects of one agent executing his must specify at least the following:
plans m a y be o b s e r v a b l e by other agents,
who, assuming that these plans were (i) A P l a n n i n g System: a language for
produced by the first agent's plan d e s c r i b i n g states of the world, a l a n g u a g e
c o n s t r u c t i o n algorithms, may try to infer for describing operators and algorithms
the plan being executed from the observed for plan c o n s t r u c t i o n and plan inference.
changes to the world. The fact that this S e m a n t i c s for the l a n g u a g e s should also be
inferencing m a y be intended by the first given.
agent underlies human c o m m u n i c a t i o n .
(2) Definitions of speech acts as
operators in the planning system. What
are their effects? When are they
* This research was supported in part by applicable? How can they be realized in
the National Research Council of Canada. words?

125
 To make possible a first attempt at a n a l y s i s of indirect speech acts (such as
such a theory we have imposed several "Can you pass the salt?") - utterances
r e s t r i c t i o n s on the system to be modelled. which appear to result from one
illocutionary act but can be used to
(I) Any agent Al's model of another agent p e r f o r m another.
A2 is defined in terms of "facts" that A1
b e l i e v e s A2 believes, and goals that A1 Section 2 of this paper o u t l i n e s some
b e l i e v e s A2 is attempting to achieve. We requirements on the models which the
are not attempting to model obligations, v a r i o u s agents m u s t have of each other.
feelings ~ etc. Section 3 d e s c r i b e s the planning o p e r a t o r s
for REQUEST and INFORM, and how they can
(2) The only speech acts we try to model be used to g e n e r a t e plans which include
are some that appear to be d e f i n a b l e in assertions, imperatives, and several types
terms of beliefs and goals, namely REQUEST of questions.
and INFORM. We have been taking these to
be prototypical members of Searle's Section 4 discusses the relation
"directive" and "representative" classes b e t w e e n the o p e r a t o r s of section 3 and the
(Searle (1976)). We represent questions linguistic sentences which can realize
as R E Q U E S T s to INFORM. These acts are them. We c o n c e n t r a t e on the p r o b l e m of
interesting for they have a wide range of identifying illocutionary force, in
syntactic realizations, and account for a particular on indirect speech acts. A
large p r o p o r t i o n of everyday utterances. useful c o n s e q u e n c e of the illocutionary
force identification process is that it
(3) We have limited o u r s e l v e s so far to p r o v i d e s a natural way to u n d e r s t a n d some
the study of so-called task-oriented elliptical utterances, and utterances
dialogues which we interpret to be whose purpose is to acknowledge, c o r r e c t
conversations between two agents or clarify interpretations of previous
c o o p e r a t i n g in the a c h i e v e m e n t of a single utterances.
high-level goal. These d i a l o g u e s do not
allow changes in the topic of discourse A critical part of c o m m u n i c a t i o n is
but still display a wide range of the process by which a speaker can
linguistic behaviour. c o n s t r u c t d e s c r i p t i o n s of o b j e c t s involved
in his plans such that the hearer can
Much of our work so far has dealt with identify the intended referent. Why can
the problem of g e n e r a t i n g plans containing someone asking "Where's the s c r e w d r i v e r ? "
REQUEST and INFORM, as well as non- be answered with "In the drawer with the
linguistic operators. Suppose that an hammer" if it is assumed he knows where
agent is a t t e m p t i n g to achieve some task, the hammer is, but maybe by "In the third
w i t h incomplete k n o w l e d g e of that task and drawer from the left" if he doesn't. How
of the m e t h o d s to complete it, but with accurate must descriptive phrases be?
some knowledge of the a b i l i t i e s of another Section 5 examines how the speaker and
agent. How can the first agent make use of hearer's m o d e l s of each other influence
the abilities of the second? Under what their references. Finally, section 6
circumstances can the first usefully c o n t a i n s some ideas on future research.
produce utterances to transmit or acquire
facts and goals? How can he initiate Most e x a m p l e s in the paper are drawn
action on the part of the second? from a s i t u a t i o n in which one p a r t i c i p a n t
is an information clerk at a train
We view the plan related aspects of station, whose objective is to assist
language generation and recognition as p a s s e n g e r s in boarding and m e e t i n g trains.
indissociable, and strongly related to the The d o m a i n is o b v i o u s l y limited, but still
process by which agents c o o p e r a t e in the provides a natural setting for a wide
achievement of goals. For example, for range of utterances, both in form and in
agent2 to reply "It's closed" to a g e n t l ' s intention.
query "Where's the nearest service
station?" seems to require him to infer
that agentl wants to m a k e use of t h e 2. On m o d e l s of others
service station which he could not do if
it were closed. The reply "Two blocks In this section we present criteria
east" would be seen as m i s l e a d i n g if given that one a g e n t ' s model of another ought to
alone, and u n n e c e s s a r y if given along with satisfy. For convenience we dub the
"It's closed". Thus p a r t of c o o p e r a t i v e agents SELF and OTHER. Our r e s e a r c h has
behaviour is the d e t e c t i o n by one a~ent of concentrated on modelling beliefs and
obstacles in the plans he b e l i e v e s the goals. We claim that a theory of language
other agent holds, p o s s i b l y f o l l o w e d by an need not be concerned with what is
attempt to overcome them. We claim that actually true in the real world: it
speakers expect (and intend) h e a r e r s to should describe language processing in
operate this way and therefore that any terms of a person's beliefs about the
hearer can a s s u m e that inferences that he world. Accordingly, SELF's model of O T H E R
can draw based on knowledge that is shared should be based on "believe" as described,
w i t h the speaker are in fact intended by for example, in Hintikka(1962) and not on
t h e speaker. These p r o c e s s e s u n d e r ! ~ e our "know" in its sense of "true belief".

126
Henceforth, all uses of the words "know" Want
and "knowledge" are to be treated as
synonyms for "believe" and "beliefs". We Any r e p r e s e n t a t i o n of O T H E R ' s goals
have neglected other aspects of a model of (wants) m u s t d i s t i n g u i s h such information
another, such as focus of a t t e n t i o n (but from: O T H E R ' S beliefs, SELF's beliefs and
see G r o s z ( 1 9 7 7 ) ) . goals, and (recursively) from the o t h e r ' s
model of someone else's beliefs and goals.
Belief The representation for WANT must also
allow for d i f f e r e n t scopes of quantifiers.
Clearly, SELF ought to be able to For example, it should d i s t i n g u i s h between
distinguish his beliefs about the world the readings of "John wants to take a
from what he believes other believes. train" as "There is a specific train which
SELF ought to have the possibility of John wants to take" or as "John wants to
believing a proposition P, of believing take any train". Finally it should allow
not-P, or of being ignorant of P. a r b i t r a r y embeddings with BELIEVE. Wants
W h a t e v e r his stand on P, he should also be of beliefs (as in "SELF wants OTHER to
able to believe that O T H E R can hold any of believe P") become the reasons for telling
these p o s i t i o n s on P. N o t i c e that such P to OTHER, while beliefs of wants (e.g.,
d i s a g r e e m e n t s cannot be represented if the SELF B e l i e v e s SELF wants P) will be the
r e p r e s e n t a t i o n is based on "know" as in way to represent SELF's goals P.
Moore(1977).
Level____~s o f E m b e d d i n g
SELF's belief r e p r e s e n t a t i o n ought to
allow him to represent the fact that O T H E R A natural q u e s t i o n to ask is how many
knows whether some p r o p o s i t i o n P is true, levels of belief embedding are needed by
w i t h o u t SELFIs having to know which of P an agent capable of p a r t i c i p a t i n g in a
or -P he does believe. S u c h information dialogue. Obviously, to be able to deal
can be represented as a disjunction of with a d i s a g r e e m e n t , SELF needs two levels
beliefs (e.g., O R ( O T H E R BELIEVE P, O T H E R (SELF BELIEVE and SELF BELIEVE OTHER
BELIEVE ~P)). Such disjunctions are BELIEVE ). If SELF were to lie to OTHER,
essential to the planning of yes/no he would have to be able to believe some
questions. proposition P (i.e. SELF BELIEVE (P)),
while O T H E R believes that SELF believes
Finally, a belief r e p r e s e n t a t i o n must not P (i.e. SELF BELIEVE OTHER BELIEVE
distinguish between situations like the SELF BELIEVE (~P)), and hence he would
following: need at least three levels.

I. OTHER believes that the train leaves We show in Cohen (1978) how one can
from gate 8. represent, in a finite fashion, the
2. OTHER believes that the train has a u n b o u n d e d number of beliefs created by any
departure gate. communication act or by face-to-face
3. O T H E R knows what the d e p a r t u r e gate for situations. The finite representation,
the train is. which employs a circular data structure,
formalizes the concept of mutual belief
Case 1 can be represented by a p r o p o s i t i o n (cf. Schiffer (1972)). Typically, all
that c o n t a i n s no variables. Case 2 can be these levels of belief embedding can be
represented by a belief of a quantified represented in three levels, but
p r o p o s i t i o n -- i.e., theoretically, any finite number are
possible.
OTHER BELIEVE (
x (the y ~ GATE(TRAIN,y) = x))
3. U§in@ a Model of the Other to Decide
However, case 3 is represented by a What to Say
quantified belief namely,
As a n aid in evaluating speech act
x OTHER BELIEVE definitions, we have constructed a
(the y : GATE(TRAIN,y) = x) computer program, OSCAR, that plans a
range of speech acts. The goal of the
The formal semantics such beliefs have program is to c h a r a c t e r i z e a speaker's
been problematic for philosophers (cf. capacity to issue speech acts by
Quine (1956) and Hintikka (1962)). Our predicting, for specified situations, all
approach to them is d i s c u s s e d in Cohen and only those speech acts that would be
(1978). In Section 3, we discuss how a p p r o p r i a t e l y issued by a person under the
quantified beliefs are used during circumstances. In this section, we will
planning, and how they can be acquired make r e f e r e n c e to p r o t o t y p i c a l speakers by
during conversation. way of the OSCAR program, and to hearers
by way of the p r o g r a m ' s user.

Specifially, the program is able to:

- Plan R E Q U E S T speech acts, for instance

a speech act that could be realized by

127
"Please open the door", w h e n its goal is Suppose, for example, that O S C A R is
to get the user to w a n t to p e r f o r m some outside a room w h o s e door is c l o s e d and
action. that it b e l i e v e s that the user is inside.
When planning to m o v e itself into the
- Plan I N F O R M speech acts, such as one room, it m i g h t R E Q U E S T that the user open
that could be realized by "The door is the door. However, it w o u l d only plan
locked", w h e n its goal is to g e t the user this speech act if it believed that the
to b e l i e v e some p r o p o s i t i o n . user did not a l r e a d y w a n t to open the door
and if it b e l i e v e d (and b e l i e v e d the user
- Combine the above to produce multiple believed) that the preconditions to
speech acts in one plan, w h e r e one speech o p e n i n g the door held. If that were not
act m a y e s t a b l i s h b e l i e f s of the user that so, O S C A R could plan a d d i t i o n a l I N F O R M or
can then be e m p l o y e d in the p l a n n i n g of R E Q U E S T speech acts. For example, a s s u m e
another speech act. that to open a door one needs to have the
key and OSCAR believes the user doesn't
- Plan q u e s t i o n s as r e q u e s t s that the know w h e r e it is. Then OSCAR could plan
user inform, w h e n its goal is to b e l i e v e "Please open the door. The key is in the
s o m e t h i n g and w h e n it b e l i e v e s that the closet". OSCAR thus employs its user
user knows the answer. m o d e l in telling him w h a t it b e l i e v e s he
needs to know.
- P l a n speech acts i n c o r p o r a t i n g third
parties, as in "Ask T o m to tell you w h e r e Mediating Acts and P e r l o c u t i o n a r y Effects
the key is and then tell me."
The e f f e c t s of INFORM (and REQUEST)
To i l l u s t r a t e the p l a n n i n g of speech are modelled so that the bearer's
acts, consider first the following b e l i e v i n g P (or w a n t i n g to do ACT) is not
simplified definitions of REQUEST and e s s e n t i a l to the s u c c e s s f u l c o m p l e t i o n of
INFORM as S T R I P S - l i k e o p e r a t o r s (cf. F i k e s the speech act. Speakers, we claim,
and N i l s s o n (1971)). Let SP d e n o t e the cannot i n f l u e n c e their hearers' beliefs
speaker, H the hearer, ACT some action, and goals directly. Thus, the
and PROP some p r o p o s i t i o n . Due to space p e r l o c u t i o n a r y e f f e c t s of a s p e e c h act are
limitations, the intuitive English not part of that act's d e f i n i t i o n . We
m e a n i n g s of the formal terms a p p e a r i n g in propose, then, as a principle of
these d e f i n i t i o n s will h a v e to s u f f i c e as c o m m u n i c a t i o n that a s p e a k e r ' s p u r p o s e in
explanation. s i n c e r e c o m m u n i c a t i o n is to p r o d u c e in the
hearer an a c c u r a t e model of his mental
REQUEST(SP,H,ACT) state.
preconditions:
SP B E L I E V E H C A N D O A C T To b r i d g e the g a p b e t w e e n the s p e e c h
SP B E L I E V E H B E L I E V E H C A N D O A C T acts and their intended perlocutionary
SP B E L I E V E SP W A N T TO R E Q U E S T effects, we p o s i t m e d i a t i n g acts, named
effects: C O N V I N C E and DECIDE, w h i c h model w h a t it
H B E L I E V E SP B E L I E V E SP W A N T H TO ACT takes to g e t s o m e o n e to b e l i e v e s o m e t h i n g
or want to do something. Our current
INFORM(SP,H,PROP) analysis of these mediating acts
preconditions: trivializes the p r o c e s s e s that they are
SP B E L I E V E PROP intended to m o d e l by p r o p o s i n g that to
SP B E L I E V E SP W A N T TO I N F O R M convince someone of something, for
effects: example, one need o n l y g e t that p e r s o n to
H B E L I E V E SP B E L I E V E PROP know that one b e l i e v e s it.

The program uses a simplistic Using Quantified Beliefs -- Planning

backward-chaining algorithm that plans Questions
actions when their e f f e c t s are w a n t e d as
subgoals that are not believed to be Notice that the precondition to
true. It is the testing of p r e c o n d i t i o n s O S C A R ' s g e t t i n g the key -- knowing w h e r e
of the newly planned action before it is -- is of the form:
creating new subgoals that e x e r c i s e s the
program's model of its user. We shall x OSCAR BELIEVE
b r i e f l y sketch h o w to plan a REQUEST. (the y : LOC(KEY,y) = x)

E v e r y a c t i o n has "want p r e c o n d i t i o n s " , W h e n such a q u a n t i f i e d b e l i e f is a goal,

which specify that b e f o r e an agent does it leads O S C A R to plan the q u e s t i o n "Where
that action, he m u s t w a n t to do it. OSCAR is the key?" (i.e., R E Q U E S T ( O S C A R , USER,
plans REQUEST speech acts to achieve INFORM(USER, OSCAR, the y
precisely this precondition of actions LOC(KEY,y))). In c r e a t i n g this q u e s t i o n ,
that it wants the user to perform. OSCAR first plans a CONVINCE and then
Similarly, the goal of the user's p l a n s the u s e r ' s I N F O R M s p e e c h act, w h i c h
believing some p r o p o s i t i o n PROP becomes it then tries to get him to p e r f o r m by way
O S C A R ' S r e a s o n for p l a n n i n g to INFORM h i m of r e q u e s t i n g .
of PROP.

128
 The above definition of INFORM is the speaker intends to perform, but' as is
inadequate for dealing with the q u a n t i f i e d well known, utterances which taken
beliefs that arise in m o d e l l i n g someone l i t e r a l l y would indicate one i l l o c u t i o n a r y
else. This INFORM should be viewed as force can be used to indicate another.
that version of the speech act that the Thus "Can you close the door?" can be a
planning agent (e.g., OSCAR) plans for request as well as a question. These so-
itself to perform. A different view of called indirect speech acts are the acid
INFORM, say I N F O R M - B Y - O T H E R , is n e c e s s a r y test of a theory of speech acts. We claim
to represent acts of informing by agents that a plan-based theory gives some
other than the speaker. The difference insight into this phenomenon.
between the two INFORMs is that for the
first, the planner knows what he wants to Searle(1975) c o r r e c t l y suggests that
say, but he o b v i o u s l y does not have such "In cases where these sentences < i n d i r e c t
knowledge of the content of the second forms of requests> are uttered as
act. requests, they still have their literal
m e a n i n g and are uttered with and as having
The p r e c o n d i t i o n for this new act is a that literal meaning". How then can they
quantified speaker-belief: also have their indirect m e a n i n g ?

x USER BELIEVE Our answer relies in part on the fact

(the y : LOC(KEY,y) = x) that an agent participating in a
cooperative dialogue m u s t have processes
where the user is to be the speaker. For to:
the system to plan an I N F O R M - B Y - O T H E R act
for the user, it m u s t believe that the (I) Achieve goals based on what he
user knows where the key is, but it does believes.
not have to know that location! (2) Adopt goals of other agents as his
Similarly, the effects of the INFORM-BY- own.
O T H E R act is also a quantified belief, as (3) Infer goals of other agents.
in (4) Predict future behaviour of other
agents.
x O S C A R BELIEVE
USER BELIEVE These p r o c e s s e s would be n e c e s s a r y even if
(the y .~ LOC(KEY,y) = x) all speech acts were l i t e r a l to account
for e x c h a n g e s where the response indicates
Thus, O S C A R plans this I N F O R M - B Y - O T H E R act a knowledge of the speaker's plan. For
of the key's location in order to know example
where the user thinks the key is.
Passenger: "When does the next train to
Such information has been lacking M o n t r e a l leave?"
from all other formulations of ASK (or Clerk : "At 6:15 at Gate 7"
INFORM) that we have seen in the or
literature (e.g., Schank (1975), Mann et Clerk - "There won't be one until
al. (1976), Searle (1969)). Cohen (1978) tomorrow."
p r e s e n t s one approach to defining this new
view of INFORM, and its associated Speakers expect hearers to be
m e d i a t i n g act CONVINCE. e x e c u t i n g these p r o c e s s e s and they expect
hearers to know this. Inferences that a
hearer can draw by executing these
4. R e c o g n i z i n @ Speech Acts p r o c e s s e s based on information he thinks
the speaker b e l i e v e s can be taken by the
In the p r e v i o u s section we d i s c u s s e d hearer to be intended by the speaker.
the structure of plans that include This accounts for many of the standard
instances of the o p e r a t o r s REQUEST and e x a m p l e s of indirect speech acts such as
INFORM without explaining the relation "Can you close the door?" and "It's cold
between these speech acts and sentences here". For instance, even if "It's cold
used to perform them. This section here" is intended literally and is
sketches our first steps in exploring this r e c o g n i z e d as such, the helpful hearer may
relation. We have been particularly still close the window. When the sentence
c o n c e r n e d with the p r o b l e m of recognizing is uttered as a request, the speaker
illocutionary force and propositional intends the hearer to recognize the
content of the utterances of a speaker. s p e a k e r ' s intention that the hearer should
Detailed algorithms which handle the p e r f o r m the helpful behaviour.
examples given in this section have been
designed by J. Allen and are being If indirect speech acts are to be
implemented by him. Further d e t a i l s can e x p l a i n e d in terms of inferences speakers
be found in (Allen and P e r r a u l t 1978) and can expect of hearers, then a theory of
A l l e n ' s forthcoming Ph.D. d i s s e r t a t i o n . speech acts m u s t concern itself with how
such inferences are controlled. Some
Certain syntactic clues in an h e u r i s t i c s are p a r t i c u l a r l y helpful. If a
utterance s u c h as its mood and the use of chain of inference by the hearer has the
explicit performatives indicate what act speaker planning an action whose effects

129
are true before the action is executed, achieve a goal which would allow plan
then the chain is likely to be wrong, or deduction to continue. Consider the
else must be continued further. This following example.
accounts for "Can you pass the salt?" as a
request for the salt, not a q u e s t i o n about Passenger : When is the Windsor train?
salt-passing prowess. As Searle(1975) Clerk : The train to W i n d s o r ?
points out, a crucial part of Passenger : Yes.
understanding indirect speech acts is Clerk : 3:15.
being able to recognize that they are not
to be interpreted literally. After the first sentence the clerk
cannot distinguish between the
A second h e u r i s t i c is that a chain of e x p e c t a t i o n s "Passenger travel by train to
inference that leads to an action whose Windsor" and "Passenger meets train from
p r e c o n d i t i o n s are known to be not easily W i n d s o r " , so he sets up a goal : (clerk
a c h i e v a b l e is likely to be wrong. believes passenger wants to travel) or
(clerk believes passenger wants to meet
Inferencing can also be controlled train). The planning for this goal
through the use of e x p e c t a t i o n s about the p r o d u c e s a plan that involves asking the
s p e a k e r ' s goals. P r i o r i t y can be given to passenger if he wants one of the
inferences which relate an o b s e r v e d speech alternatives, and receiving back the
act to an expected goal. Expectations answer. The execution of this plan
enable inferencing to work top-down as p r o d u c e s the clerk response "The train to
well as bottom-up. Windsor?" and recognizes the response
"Yes". Once the passenger's goal is
The use of expected goals to guide the known, the clerk can c o n t i n u e the o r i g i n a l
inferencing has another advantage: it deduction process with the "travel to
allows for the recognition of Windsor" a l t e r n a t i v e favoured. This plan
illocutionary force in elliptical is accepted and the clerk p r o d u c e s the
utterances such as "The 3:15 train to response "3:15" to overcome the o b s t a c l e
Windsor?", without requiring that the "passenger knows d e p a r t u r e time".
syntactic and semantic analysis
"reconstitute" a complete semantic
representation such as "Where does the
3:15 train to W i n d s o r leave?". For 5. Reference and the Model of the Other
example, let the clerk assume that
passengers want to either m e e t incoming We have shown that q u a n t i f i e d beliefs
trains or board d e p a r t i n g ones. Then the are needed in d e c i d i n g to ask someone a
utterance "The 3:15 train to W i n d s o r ? " is question. They are also involved, we
first interpreted as a REQUEST about a claim, in the r e p r e s e n t a t i o n of singular
train to Windsor with 3:15 as either definite noun phrases and hence any
arrival or d e p a r t u r e time. Only d e p a r t i n g natural language system will need them.
trains have d e s t i n a t i o n s different from According to our analysis, a hearer should
T o r o n t o and this leads to believing that represent the referring phrase in a
the passenger wants to board a 3:15 train s p e a k e r ' s s t a t e m e n t "The pilot of TWA 510
to Windsor. Attempting to identify is drunk" by:
o b s t a c l e s in the p a s s e n g e r ' s plan leads to
finding that the p a s s e n g e r knows the time x S P E A K E R BELIEVE
but p r o b a b l y not the place of departure. (the y : PILOT(y,TWA510) = x &
Finally, overcoming the obstacle then DRUNK (x))
leads to an INFORM like "Gate 8".
This is the reading w h e r e b y the speaker is
Our analysis of elliptical u t t e r a n c e s b e l i e v e d to "know who the pilot of TW~ 510
raises two questions. First, what is" (at least p a r t i a l l y accounting for
information does the i l l o c u t i o n a r y force Donnellan's (1966) referential reading).
r e c o g n i t i o n m o d u l e expect from the syntax This is to be c o n t r a s t e d with the reading
and semantics? Our approach here has been of whoever is p i l o t i n g that plane is drunk
to require from the syntax and semantics a (Donnellan's attributive noun phrases).
h y p o t h e s i s about the literal i l l o c u t i o n a r y In this latter case, the existential
force and a predicate calculus-like q u a n t i f i e r would be inside the scope of
representation of the propositional the belief.
content, but where u n d e t e r m i n e d p r e d i c a t e s
and objects could be replaced by p a t t e r n s These existential presuppositions of
on which certain restrictions can be d e f i n i t e referential noun p h r a s e s give one
imposed. As part of the plan inferencing important way for hearers to acquire
process these patterns become further quantified s p e a k e r - b e l i e f s . Such beliefs,
specified. we have seen, can be used as the basis for
p l a n n i n g further c l a r i f i c a t i o n questions.
The second q u e s t i o n is: w h a t should
the hearer do if more than one path We agree with S t r a w s o n (1950) (and
between the observed u t t e r a n c e and the many others) that hearers understand
e x p e c t a t i o n s is p o s s i b l e ? He may suspend referring phrases based on what they
plan deduction and start planning to believe speakers intend to refer to.

130
Undoubtedly, a hearer will understand a handle p r o m i s e s w i t h o u t first d e a l i n g with
s p e a k e r ' s (reference) intentions by using obligations, or warnings without the
a model of that speaker's beliefs. n o t i o n s of danger and u n d e s i r a b i l i t y ? We
Speakers, of course, know of these are c u r r e n t l y c o n s i d e r i n g an e x t e n s i o n of
interpretation s t r a t e g i e s and thus plan the approach to understanding stories
their referring phrases to take the which report simple dialogue.
a p p r o p r i a t e referent within the hearer's
model of them. A speaker cannot use Much remains to be done on the
private descriptions, nor descriptions r e p r e s e n t a t i o n of the a b i l i t i e s of angther
that he thinks the hearer thinks are agent. A simple setting suggests a number
private, for c o m m u n i c a t i o n . of problems. Let one agent H be seated in
a room in front of a table with a
For instance, consider the following c o l l e c t i o n of blocks. Let another agent
variant of an example of Donnellan's S be outside the room but c o m m u n i c a t i n g by
(1966): At a party, a woman is holding a telephone. If S b e l i e v e s that there is a
martini glass which Jones believes green block on the table and wants it
contains water, but of which he is c e r t a i n cleared, but knows nothing about any other
everyone else believes (and b e l i e v e s he blocks except that H can see them, then
believes) contains a martini. Jones would how can S ask H to clear the g r e e n block?
understand that Smith, via q u e s t i o n (I), The blocks S wants removed are those which
but not via q u e s t i o n (2) is referring to are in fact there, p e r h a p s those which he
this woman. could p e r c e i v e to be there if he were in
the room. The goal seems to be of the
(i) Who is the woman holding the m a r t i n i ? form
(2) Who is the woman holding the water?
S BELIEVE
since Jones does not believe Smith knows x (x on the green block => S WANT
about the water in her glass. (x removed from green block))

Conversely, if Jones wanted to refer but our planning m a c h i n e r y and d e f i n i t i o n

to the woman in an u t t e r a n c e intended for of R E Q U E S T are inadequate for g e n e r a t i n g
Smith, he could do so using (i) but not "I request you to clear the green block".
(2) since in the latter case he would not
think the hearer could pick out his We have not yet spent much time
intended referent. investigating the process of giving
answers to How and Why questions, or to WH
q u e s t i o n s requiring an event description
Thus it appears that for a speaker to as an answer. We c o n j e c t u r e that because
plan a successful singular definite of the speech act approach answers to
referential e x p r e s s i o n requires that the "What did he say?" should be found in much
speaker believe the e x p r e s s i o n he finally the same way as answers to "What did he
chooses have the right referent in the do?" and that this parallelism should
hearer's model of the speaker. Our extend to other question types. The
c o n c e p t of mutual belief can be used (as natural e x t e n s i o n of our a n a l y s i s would
in Cohen (1978)) to ensure that the suggest r e p r e s e n t i n g "How did AGT achieve
expression denotes appropriately in all goal G?" as a R E Q U E S T by the speaker that
further embedded belief models. This the hearer inform him of a plan by which
example is p r o b l e m a t i c for any approach to AGT achieved G. We have not yet
reference where a communicating party investigated the repercussions of this
assumes that its reality is the only e x t e n s i o n on the r e p r e s e n t a t i o n language.
reality. Speakers and hearers can be
"wrong" or "ignorant" and yet Finally consider the following
c o m m u n i c a t i o n can still be m e a n i n g f u l and dialogue. Assume that S is a shady
successful. businessman, A his secretary.

A : IRS is on the phone.

6. Further Research
S : I'm not here.
We believe that speech acts p r o v i d e an
e x c e l l e n t way of e x p l a i n i n g the relations How is A to understand S's utterance?
between utterances in a dialogue, as well Although its propositional content is
as relating linguistic to non-linguistic l i t e r a l l y false, maybe even nonsensical,
activity. Until we better understand the the u t t e r a n c e ' s intention is unmistakable.
m e c h a n i s m s by which c o n v e r s a n t s change the How tolerant does the u n d e r s t a n d i n g system
topic and goals of the c o n v e r s a t i o n it have to be to infer its way to a c o r r e c t
will be d i f f i c u l t to extend this analysis interpretation? Must "I'm not here" be
beyond e x c h a n g e s of a few utterances, in treated i d i o m a t i c a l l y ?
particular to non-task oriented dialogues.
Fuller j u s t i f i c a t i o n of our approach also
requires its a p p l i c a t i o n to a much broader
range of speech acts. Here the p r o b l e m is
mainly representational: how can we

131
 B ibl log r aphy

Allen, J.F. and Perrault, C.R.,

"Participating in Dialogue:
Understanding via Plan Deduction", 2nd
National Conference of the Canadian
Society for Studies in Computational
Intelligence, Toronto, July, 1978.

Cohen, P.R., "On Knowing What to Say:

Planning Speech Acts", TRII8 Dept. of
Computer Science, University of
Toronto, 1978.

Donnellan, K., "Reference and Definite

Description", The Philosophical
Review, vol. 75, 1960, pp280-304.
Reprinted in Semantics, Steinberg and
Jacobovits, eds., Cambridge University
Press, 1970.

Fikes, R. E. and Nilsson, N. J., 1970,

"STRIPS: A new approach to the
application of theorem proving",
Artificial Intelligence 2, 1970.

Grosz, B. J., "The Representation and Use

of Focus in Natural Language
Dialogues", 5IJCAI, 1977.

Hintikka, K.J., Knowled~[e and Belief,

Cornell University Press, 1962.

Mann, W.C., Moore, J.A., Levin, J.A.; "A

Comprehension Model for Human
Dialogue", 5IJCAI, 1977.

Moore, R.C.; "Reasoning about Knowledge

and Action", 5IJCAI, 1977.

Quine, w.v., "Quantifiers and

Propositional Attitudes", The Journal
of Philosophy 53, (1956), 177-187.

Schiffer, S., Meaning, Oxford University

Press, 1972.

Schank, R. and Abelson, R., "Scripts,

Plans and Knowledge", 4IJCAI, 1975.

Searle, J. R., Speech Acts, Cambridge

University Press, 1969.

Searle, J. R.; "Indirect Speech Acts" in

Syntax and Semantics, Vol. 3: Speech
Acts, Cole and Morgan (eds), Academic
Press, 1975.

Searle, J. R., "A Taxonomy of

Illocutionary Acts", Language, Mind
and Knowledge, K. Gunderson (ed.),
University of Minnesota Press, 1976.

Strawson, P. F., "On Referring", Mind,

1950.

132