THERE IS VERY LITTLE
MARGIN BETWEEN THE LEFT
AND RIGHT PAGES
158.8 S62v
Skinner
Verbal "behavior, .
61-05123
i.IAI JUL7 1975
OCT 1 1983
H'OV 8, 19*85" "
SEP 8 1986
THE CENTURY PSYCHOLOGY SERIES
Richard M. Elliott, Editor
Kenneth MacCorquodale, Assistant Editor
Verbal Behavior
BY THE SAME AUTHOR
The Behavior of Organisms
Walden Two
Science and Human Behavior
Schedules of Reinforcement
(with C. B. Ferster)
B. F. SKINNER
Verbal
Behavior
APPLETON-CENTURY-CROFTS, Inc.
New York
Copyright 1957 by
APPLETON-CENTURY-CROFTS, INC.
All rights reserved. This book, or parts
thereof, must not be reproduced in any
form without permission of the publisher.
597-1
LIBRARY OF CONGRESS CARD NUMBER:
57-11446
PRINTED IN THE UNITED STATES OF AMERICA
To
JULIE and DEBBIE,
my primary sources
PREFACE
IT HAS TAKEN a long time to write this book. A classification of verbal
responses in an early version of Part II was completed in the summer
of 1934. A few supporting experiments were then carried out with
the Verbal Summator, and statistical analyses were made of several
literary works, of data from word-association experiments, and of
guessing behavior. All this material was used in courses on Literary
and Verbal Behavior at the University of Minnesota in the late
thirties, at Harvard University in the summer of 1938, and at the
University of Chicago in the summer of 1939. A manuscript of the
present scope was to have been completed under a Guggenheim Fel-
lowship in 1941, but the war intervened. The Fellowship was resumed
in 1944-45 and a version nearly completed. It was the basis of a
course on Verbal Behavior at Columbia University in the summer of
1947, stenographic notes of which were circulated by Dr. Ralph
Heff erlein in mimeographed form the following year.
In the fall of 1947 material was extracted from the manuscript for
the William James Lectures at Harvard University, several hundred
mimeographed copies of which have since been circulated. In pre-
paring these lectures it was found that the manuscript had begun to
take on the character of a review of the literature and that the central
theme was becoming obscure. In completing the manuscript for pub-
lication, therefore, summaries of the literature were deleted. Com-
pletion of the final manuscript was postponed in favor of a general
book on human behavior (Science and Human Behavior) which
would provide a ready reference on matters not essentially verbal.
The present version is more than twice as long as the James Lectures
and contains many changes made to conform with recent progress in
the experimental analysis of behavior, human and otherwise. With
the exception of the last two chapters, it was written during the spring
term of 1955 at Putney, Vermont.
The work has been generously supported by the Society of Fellows
of Harvard University (a three-year fellowship), the University of
Minnesota (a one-half year sabbatical leave), the Guggenheim Foun-
vii
Vlll PREFACE
dation (a one-year fellowship), and Harvard University (the William
James Lectureship and a sabbatical leave). To all of these, thanks are
due. Unfortunately it is impossible to make an adequate acknowl-
edgement of the generous help received from students and colleagues
during these years and from criticisms of earlier versions, published
or unpublished. The final manuscript has profited greatly from criti-
cal and editorial help by Mrs. Susan R. Meyer and Dr. Dorothy
Cohen and from careful preparation by Mrs. Virginia N. MacLaury.
Cambridge, Mass. B. F, SKINNER
CONTENTS
PAGE
Preface vii
Part I: A Program
CHAPTER
1 . A Functional Analysis of Verbal Behavior i
2. General Problems 13
Part II: Controlling Variables
3. The Mand 35
4. Verbal Behavior Under the Control of Verbal Stimuli 52
5. The Tact 81
6. Special Conditions Affecting Stimulus Control 147
7. The Audience 172
8. The Verbal Operant as a Unit of Analysis 185
Part III: Multiple Variables
9. Multiple Causation 227
10. Supplementary Stimulation 253
1 1. New Combinations of Fragmentary Responses 293
Part IV: The Manipulation of Verbal Behavior
1 2 . The Autoclitic 3 1 1
13. Grammar and Syntax as Autoclitic Processes 331
14. Composition and Its Effects 344
X COM EMS
Part V: The Production of Verbal Behavior
CHAPTER PAGE
15. Self-Editing 369
16. Special Conditions of Self- Editing 384
17. Self-Strengthening of Verbal Behavior 403
18. Logical and Scientific Verbal Behavior 418
19. Thinking 432
Two Personal Epilogues 453
Appendix: The Verbal Community 461
Index
Verbal Behavior
Parti
A PROGRAM
Chapter i
A Functional Analysis
of Verbal Behavior
MEN ACT upon the world, and change it, and are changed in turn by
the consequences of their action. Certain processes, which the human
organism shares with other species, alter behavior so that it achieves
a safer and more useful interchange with a particular environment.
When appropriate behavior has been established, its consequences
work through similar processes to keep it in force. Jf by chance the
environment changes, old forms of behavior disappear, while new
consequences build new forms.
Behavior alters the environment through mechanical action, and
its properties or dimensions are often related in a simple way to the
effects produced. When a man walks toward an object, he usually finds
himself closer to it; if he reaches for it, physical contact is likely to fol-
low; and if he grasps and lifts it, or pushes or pulls it, the object fre-
quently changes position in appropriate directions. All this follows
from simple geometrical and mechanical principles.
Much of the time, however, a man acts only indirectly upon the en-
vironment from which the ultimate consequences of his behavior '
emerge. His first effect is upon other men. Instead of going to a drink-
ing fountain, a thirsty man may simply "ask for a glass of water" that
is, may engage in behavior which produces a certain pattern of sounds
which in turn induces someone to bring him a glass of water. The
sounds themselves are easy to describe in physical terms; but the glass
of water reaches the speaker only as the result of a complex series of
events including the behavior of a listener. The ultimate consequence,
the receipt of water, bears no useful geometrical or mechanical rela-
tion to the form of the behavior of "asking for water." Indeed, it is
VERBAL BEHAVIOR
characteristic of such behavior that it is impotent against the physical
world. Rarely do we shout down the walls of a Jericho or successfully
command the sun to stop or the waves to be still. Names do not
break bones. The consequences of such behavior are mediated by a
train of events no less physical or inevitable than direct mechanical
action, but clearly more difficult to describe.
Behavior which is effective only through the mediation of other
persons has so many distinguishing dynamic and topographical prop-
erties that a special treatment is justified and, indeed, demanded.
Problems raised by this special mode of action are usually assigned to
the field of speech or language. Unfortunately, the term "speech"
emphasizes vocal behavior and is only awkwardly applied to instances
in which the mediating person is affected visually, as in writing a
note. "Language" is now satisfactorily remote from its original com-
mitment to vocal behavior, but it has come to refer to the practices of
a linguistic community rather than the behavior of any one member.
The adjective "linguistic" suffers from the same disadvantage. The
term "verbal behavior" has much to recommend it. Its etymological
sanction is not too powerful, but it emphasizes the individual speaker
and, whether recognized by the user or not, specifies behavior shaped
and maintained by mediated consequences. It also has the advantage
of being relatively unfamiliar in traditional modes of explanation.
A definition of verbal behavior as behavior reinforced through the
mediation of other persons needs, as we shall see, certain refinements.
Moreover, it does not say much about the behavior of the listener,
even though there would be little verbal behavior to consider if
someone had not already acquired special responses to the patterns of
energy generated by the speaker. This omission can be justified, for the
behavior of the listener in mediating the 'consequences of the behavior
of the speaker is not necessarily verbal in any special sense. It cannot,
in fact, be distinguished from behavior, in general, and an adequate
account of verbal behavior need cover only as much of the behavior
of the listener as is needed to explain thfe behavior of the speaker. The
behaviors of speaker and listener taken together compose what may be
called a total speech episode. There is nothing in such an episode
which is more than the combined behravior of two or more individuals.
Nothing "emerges" in the social unit. The^speaker can be studied
while assuming a listener, and the listener while assuming a speaker.
The separate accounts which result exhaust the episode in which both
participate.
A FUNCTIONAL ANALYSIS
It would be foolish to underestimate the difficulty of this subject
matter, but recent advances in the analysis of behavior permit us to
approach it with a certain optimism. New experimental techniques
and fresh formulations have revealed a new level of order and pre-
cision. The basic processes and relations which give verbal behavior
its special characteristics are now fairly well understood. Much of the
experimental work responsible for this advance has been carried out
on other species, but the results have proved to be surprisingly free of
species restrictions. Recent work has shown that the methods can be
extended to human behavior without serious modification. Quite
apart from the possibility of extrapolating specific experimental find-
ings, the formulation provides a fruitful new approach to human
behavior in general, and enables us to deal more effectively, with that
subdivision called verbal.
The "understanding" of verbal behavior is something more than
the use of a consistent vocabulary with which specific instances may
be described. It is not to be confused with the confirmation of any set
of theoretical principles. The criteria are more demanding than that.
The extent to which we understand verbal behavior in a "causal"
analysis is to be assessed from the extent to which we can predict the
occurrence of specific instances and, eventually, from the extent to
which we can produce or control such behavior by altering the condi-
tions under which it occurs. In representing such a goal it is helpful
to keep certain specific engineering tasks in mind. How can the
teacher establish the specific verbal repertoires which are the princi-
pal end-products of education? How can the therapist uncover latent
verbal behavior in a therapeutic interview? How can the writer evoke
his own verbal behavior in the act of composition? How can the sci-
entist, mathematician, or logician manipulate his verbal behavior in
productive thinking? Practical problems of this sort are, of course,
endless. To solve them is not the immediate goal of a scientific anal-
ysis, but they underline the kinds of processes and relationships which
such an analysis must consider.
TRADITIONAL FORMULATIONS
A science of behavior does not arrive at this special field to find it
unoccupied. Elaborate systems of terms describing verbal behavior
have been developed. The lay vocabulary abounds with them. Clas-
sical rhetoric, grammar, logic, scientific methodology, linguistics,
4 VERBAL BEHAVIOR
literary criticism, speech pathology, semantics, and many other dis-
ciplines have contributed technical terms and principles. In general,
however, the subject here at issue has not been clearly identified, nor
have appropriate methods for studying it been devised. Linguistics,
for example, has recorded and analyzed speech sounds and semantic
and syntactical practices, but comparisons of different languages and
the tracing of historical changes have taken precedence over the study
of the individual speaker. Logic, mathematics, and scientific method-
ology have recognized the limitations which linguistic practices im-
pose on human thought, but have usually remained content with a
formal analysis; in any case, they have not developed the techniques
necessary for a causal analysis of the behavior of man thinking. Clas-
sical rhetoric was responsible for an elaborate system of terms describ-
ing the characteristics of literary works of art, applicable as well to
everyday speech. It *also gave some attention to effects upon the
listener. But the early promise of a science of verbal behavior was
never fulfilled. Modern literary criticism, except for some use of the
technical vocabulary of psychoanalysis, seldom goes beyond the terms
of the intelligent layman. An effective frontal attack, a formulation
appropriate to all special fields, has never emerged under the auspices
of any one of these disciplines.
Perhaps this fact is responsible for the rise of semantics as a general
account of verbal behavior. The technical study of meaning was al-
ready under way as a peripheral field of linguistics when, in 1923,
Ogden and Richards * demonstrated the need for a broader science of
symbolism. This was to be a general analysis of linguistic processes
applicable to any field and under the domination of no special inter-
est. Attempts have been made to carry out the recommendation, but
an adequate science of verbal behavior has not been achieved. There
are several current brands of semantics, and they represent the same
special interests and employ the same special techniques as heretofore.
The original method of Ogden and Richards was philosophical, with
psychological leanings. Some of the more rigorous systems are frankly
logical. In linguistics, semantics continues to be a question of how
meanings are expressed and how they change. Some semanticists deal
mainly with the verbal machinery of society, particularly propaganda.
Others are essentially therapists who hold that many of the troubles
of the world are linguistic error. The currency of the term "semantics"
shows the need for a science of verbal behavior which will be divorced
i Ogden, C. K., and Richards, I. A., The Meaning of Meaning (New York, 1923).
A FUNCTIONAL ANALYSIS 5
from special interests and helpful wherever language is used, but the
science itself has not emerged under this aegis.
The final responsibility must rest with the behaviorial sciences, and
particularly with psychology. What happens when a man speaks or
responds to speech is clearly a question about human behavior and
hence a question to be answered with the concepts and techniques of
psychology as an experimental science of behavior. At first blush, it
may not seem to be a particularly difficult question. Except on the
score of simplicity, verbal behavior has many favorable character-
istics as an object of study. It is usually easily observed (if it were not,
it would be ineffective as verbal behavior); there has never been any
shortage of material (men talk and listen a great deal); the facts are
substantial (careful observers will generally agree as to what is said
in any given instance); and the development of the practical art of
writing has provided a ready-made system of notation for reporting
verbal behavior which is more convenient and precise than any avail-
able in the nonverbal field. What is lacking is a satisfactory causal or
functional treatment. Together with other disciplines concerned with
verbal behavior, psychology has collected facts and sometimes put
them in convenient order, but in this welter of material it has failed
to demonstrate the significant relations which are the heart of a sci-
entific account. For reasons which, in retrospect, are not too difficult
to discover, it has been led to neglect some of the events needed in a
functional or causal analysis. It has done this because the place of such
events has been occupied by certain fictional causes which psychology
has been slow in disavowing? In examining some of these causes more
closely, we may find "an explanation of why a science of verbal be-
havior has been so long delayed.
It has generally been assumed that to explain behavior, or any
aspect of it, one must attribute it to events taking place inside the
organism. In the field of verbal behavior this practice was once repre-
sented by the doctrine of the expression of ideas. An utterance was
felt to be explained by setting forth the ideas which it expressed. If
the speaker had had a different idea, he would have uttered different
words or words in a different arrangement. If his utterance was
unusual, it was because of the novelty or originality of his ideas. If
it seemed empty, he must have lacked ideas or have been unable to
put them into words. If he could not keep silent, it was because of
the force of his ideas. If he spoke haltingly, it was because his ideas
6 VERBAL BEHAVIOR
came slowly or were badly organized. And so on. All properties of
verbal behavior seem to be thus accounted for.
Such a practice obviously has the same goal as a causal analysis,
but it has by no means the same results. The difficulty is that the
ideas for which sounds are said to stand as signs cannot be inde-
pendently observed. If we ask for evidence of their existence, we
are likely to be given a restatement in other words; but a restate-
ment is no closer to the idea than the original utterance. Restatement
merely shows that the idea is not identified with a single expression.
It is, in fact, often defined as something common to two or more
expressions. But we shall not arrive at this "something" even though
we express an idea in every conceivable way.
Another common answer is to appeal to images. The idea is said
to be what passes through the speaker's mind, \vhat the speaker sees
and hears and feels when he is "having" the idea. Explorations of
the thought processes underlying verbal behavior have been at-
tempted by asking thinkers to describe experiences of this nature.
But although selected examples are sometimes convincing, only a
small part of the ideas said to be expressed in words can be identified
with the kind of sensory event upon which the notion of image rests.
A book on physics is much more than a description of the images in
the minds of physicists.
There is obviously something suspicious in the ease with which we
discover in a set of ideas precisely those properties needed to account
for the behavior which expresses them. We evidently construct the
ideas at will from the behavior to be explained. There is, of course,
no real explanation. When we say that a remark is confusing because
the idea is unclear, we seem to be talking about two levels of observa-
tion although there is, in fact, only one. It is the remark which is un-
clear." The practice may have been defensible when inquiries into
verbal processes were philosophical rather than scientific, and when
a science of ideas could be imagined which would some day put the
matter in better order; but it stands in a different light today. It is the
function of an explanatory fiction to allay curiosity and to bring
inquiry to an end. The doctrine of ideas has had this effect by appear-
ing to assign important problems of verbal behavior to a psychology
of ideas. The problems have then seemed to pass beyond the range of
the techniques of the student of language, or to have become too ob-
scure to make further study profitable.
Perhaps no one today is deceived by an "idea" as an explanatory
A FUNCTIONAL ANALYSIS 7
fiction. Idioms and expressions which seem to explain verbal be-
havior in term of ideas are so common in our language that it is im-
possible to avoid them, but they may be little more than moribund
figures of speech. The basic formulation, however, has been pre-
served. The immediate successor to "idea" was "meaning," and the
place of the latter is in danger of being usurped by a newcomer, "in-
formation/* These terms all have the same effect of discouraging a
functional analysis and of supporting, instead, some of the practices
first associated with the doctrine of ideas.
One unfortunate consequence is the belief that speech has an in-
dependent existence apart from the behavior of the speaker. Words
are regarded as tools or instruments, analogous to the tokens,
counters, or signal flags sometimes employed for verbal purposes. It
is true that verbal behavior usually produces objective entities. The
sound-stream of vocal speech, the words on a page, the signals trans-
mitted on a telephone or telegraph wirethese are records left by
verbal behavior. As objective facts, they may all be studied, as they
have been from time to time in linguistics, communication engineer-
ing, literary criticism, and so on. But although the formal properties
of the records of utterances are interesting, we must preserve the dis-
tinction between an activity and its traces. In particular we must
avoid the unnatural formulation of verbal behavior as the "use of
words." We have no more reason to say that a man "uses the word
water" in asking for a drink than to say that he "uses a reach" in
taking the offered glass. In the arts, crafts, and sports, especially where
instruction is verbal, acts are sometimes named. We say that a tennis
player uses a drop stroke, or a swimmer a crawl. No one is likely to
be misled when drop strokes or crawls are referred to as things, but
words are a different matter. Misunderstanding has been common,
and often disastrous.
A complementary practice has been to assign an independent exist-
ence to meanings. "Meaning," like "idea," is said to be something
expressed or communicated by an utterance. A meaning explains the
occurrence of a particular set of words in the sense that if there had
been a different meaning to be expressed, a different set of words
would have been used. An utterance will be affected according to
whether a meaning is clear or vague, and so on. The concept has cer-
tain advantages. Where "ideas" (like "feelings" and "desires/ 1 which
are also said to be expressed by words) must be inside the organism,
8 VERBAL BEHAVIOR
there is a promising possibility that meanings may be kept outside the
skin. In this sense, they are as observable as any part of physics.
But can we identify the meaning of an utterance in an objective
way? A fair argument may be made in the case of proper nouns, and
some common nouns, verbs, adjectives, and adverbs roughly the
words with respect to which the doctrine of ideas could be supported
by the appeal to images. But what about words like atom or gene or
minus one or the spirit of the times where corresponding nonverbal
entities are not easily discovered? And for words like nevertheless,
although, and ouch! it has seemed necessary to look inside the or-
ganism for the speaker's intention, attitude, sentiment, or some other
psychological condition.
Even the words which seem to fit an externalized semantic frame-
work are not without their problems. It may be true that proper
nouns stand in a one-to-one correspondence with things, provided
everything has its own proper name, but what about common nouns?
What is the meaning of cat? Is it some one cat, or the physical totality
of all cats, or the class of all cats? Or must we fall back upon the idea
of cat? Even in the case of the proper noun, a difficulty remains. As-
suming that there is only one man named Doe, is Doe himself the
meaning of Doe? Certainly he is not conveyed or communicated when
the word, is used.
The existence of meaniiigs becomes even more doubtful when we
advance from single words to those collocations which "say some-
thing." What is said by a sentence is something more than what the
words in it mean. Sentences do not merely refer to trees and skies and
rain, they say something about them. This something is sometimes
called a "proposition" a somewhat more respectable precursor of
speech but very similar to the "idea" which would have been said to
be expressed by the same sentence under the older doctrine. To
define a proposition as "something which may be said in any lan-
guage" does not *ell us where propositions are, or of what stuff they
are made. Nor is the problem solved by defining a proposition as all
the sentences which have the same meaning as some one sentence,
since we cannot identify a sentence as a member of this class without
knowing its meaning at which point we find ourselves facing our
original problem.
It has been tempting to try to establish the separate existence of
words and meanings because a fairly elegant solution o certain
problems then becomes available. Theories of meaning usually deal
A FUNCTIONAL ANALYSIS g
with corresponding arrays of words and things. How do the linguistic
entities on one side correspond with the things or events which are
their meanings on the other side, and what is the nature of the rela-
tion between them called "reference"? Dictionaries seem, at first
blush, to support the notion of such arrays. But dictionaries do not
give meanings; at best they give words having the same meanings.
The semantic scheme, as usually conceived, has interesting properties.
Mathematicians, logicians, and information theorists have explored
possible modes of correspondence at length. For example, to w ? hat
extent can the dimensions of the thing communicated be represented
in the dimensions of the communicating medium? But it remains to
be shown that such constructions bear any close resemblances to the
products of genuine linguistic activities.
In any case the practice neglects many important properties of the
original behavior, and raises other problems. We cannot successfully
supplement a framework of semantic reference by appealing to the
"intention of the speaker" until a satisfactory psychological account
of intention can be given. If "connotative meaning" is to supplement
a deficient denotation, study of the associative process is required.
When some meanings are classed as "emotive," another difficult and
relatively undeveloped psychological field is invaded. These are all
efforts to preserve the logical representation by setting up additional
categories for exceptional words. They are a sort of patchwork which
succeeds mainly in showing how threadbare the basic notion is. When
we attempt to supply the additional material needed in this repre-
sentation of verbal behavior, we find that our task has been set in awk-
ward if not impossible terms. The observable data have been pre-
empted, and the student of behavior is left with vaguely identified
"thought processes."
The impulse to explicate a meaning is easily understood. We ask,
"What do you mean?" because the answer is frequently helpful.
Clarifications of meaning in this sense have an important place in
every sort of intellectual endeavor. For the purposes of effective dis-
course the method of paraphrase usually suffices; we may not need
extraverbal referents. But the explication of verbal behavior should
not be allowed to generate a sense of scientific achievement. One has
not accounted for a remark by paraphrasing "what it means."
We could no doubt define ideas, meanings, and so on, so that they
would be scientifically acceptable and even useful in describing
verbal behavior. But such an effort to retain traditional terms would
1O VERBAL BEHAVIOR
be costly. It is the general formulation which is wrong. We seek
"causes" of behavior which have an acceptable scientific status and
which, with luck, will be susceptible to measurement and manipula-
tion. To say that these are "all that is meant by" ideas or meanings is
to misrepresent the traditional practice. We must find the functional
relations which govern the verbal behavior to be explained; to call
such relations "expression" or "communication" is to run the danger
of introducing extraneous and misleading properties and events. The
only solution is to reject the traditional formulation of verbal be-
havior in terms of meaning.
A NEW FORMULATION
The direction to be taken in an alternative approach is Dictated
by the task itself. Our first responsibility is simple description: what
is the topography of this subdivision of human behavior? Once that
question has been answered in at least a preliminary fashion we may
advance to the stage called explanation: what conditions are rele-
vant to the occurrence of the behavior what are the variables of
which it is a function? Once these have been identified, we can ac-
count for the dynamic characteristics of verbal behavior within a
framework appropriate to human behavior as a whole. At the same
time, of course, we must consider the behavior of the listener. In re-
lating this to the behavior of the speaker, we complete our account of
the verbal episode.
But this is only the beginning. Once a repertoire of verbal behavior
has been set up, a host of new problems arise from the interaction of
its parts. Verbal behavior is usually the effect of multiple causes. Sepa-
rate variables combine to extend their functional control, and new
forms of behavior emerge from the recombination of old fragments.
All of this has appropriate effects upon the listener, whose behavior
then calls for analysis.
Still another set of problems arises from the fact, often pointed out,
that a speaker is normally also 3. listener. He reacts to his own be-
havior in several important ways. Part of what he says is under the
control of other parts of his verbal behavior. We refer to this inter-
action when we say that the speaker qualifies, orders, or elaborates his
behavior at the moment it is produced. The mere emission of re-
sponses is an incomplete characterization when behavior is composed.
As another consequence of the fact that the speaker is also a listener,
A FUNCTIONAL ANALYSIS 11
some of the behavior of listening resembles the behavior of speaking,
particularly when the listener "understands" what is said.
The speaker and listener within the same skin engage in activities
which are traditionally described as "thinking." The speaker manipu-
lates his behavior; he reviews it, and may reject it or emit it in modi-
fied form. The extent to which he does so varies over a wide range,
determined in part by the extent to which he serves as his own lis-
tener. The skillful speaker learns to tease out weak behavior and to
manipulate variables which will generate and strengthen new re-
sponses in his repertoire. Such behavior is commonly observed in the
verbal practices of literature as well as of science and logic. An anal-
ysis of these activities, together with their effects upon the listener,
leads us in the end to the role of verbal behavior in the problem of
knowledge.
The present book sets forth the principal features of an analysis
from this point of view. Part II sketches the topography of verbal
behavior in relation to its controlling variables and Part III some of
the consequences of the interaction of variables. Part IV describes
the manipulation of verbal behavior in the act of composition, while
Part V considers the activities involved in editing and in the creative
production of behavior which are usually called verbal thinking. No
assumption is made of any uniquely verbal characteristic, and the
principles and methods employed are adapted to the study of human
behavior as a whole. An extensive treatment of human behavior in
general from the same point of view may be found elsewhere. 2 The
present account is self-contained.
.One important feature of the analysis is that it is directed to the
behavior of the individual speaker and listener; no appeal is made
to statistical concepts based upon data derived from groups. Even
with respect to the individual speaker or listener, little use is made
of specific experimental results. The basic facts to be analyzed are
well known to every educated person and do not need to be sub-
stantiated statistically or experimentally at the level of rigor here
attempted. No effort has been made to survey the relevant "litera-
ture," The emphasis is upon an orderly arrangement of well-known
facts, in accordance with a formulation of behavior derived from an
experimental analysis of a more rigorous sort. The present extension
to verbal behavior is thus an exercise in interpretation rather than a
quantitative extrapolation of rigorous experimental results.
2 Skinner, B. F., Science and Human Behavior (New York, 1954).
12 VERBAL BEHAVIOR
The lack of quantitative rigor is to some extent offset by an insist-
ence that the conditions appealed to in the analysis be, so far as
possible, accessible and manipulable. The formulation is inherently
practical and suggests immediate technological applications at almost
every step. Although the emphasis is not upon experimental or sta-
tistical facts, the book is not theoretical in the usual sense. It makes
no appeal to hypothetical explanatory entities. The ultimate aim is
the prediction and control of verbal behavior.
Chapter 2
General Problems
VERBAL BEHAVIOR AS A DEPENDENT VARIABLE
OUR SUBJECT matter is verbal behavior, and we must accept this in
the crude form in which it is observed. In studying speech, we have
to account for a series of complex muscular activities which produce
noises. In studying writing or gesturing, we deal with other sorts of
muscular responses. It has long been recognized that this is the stuff
of which languages are made, but the acknowledgement has usu-
ally been qualified in such a way as to destroy the main point. As
Jespersen * said many years ago, "The only unimpeachable definition
of a word is that it is a human habit." Unfortunately, he felt it neces-
sary to add, "an habitual act on the part of one human individual
which has, or may have, the effect of evoking some idea in the mind
of another individual." Similarly, Bertrand Russell 2 asserts that "just
as jumping is one class of movement ... so the word 'dog' is [another]
class," but he adds that words differ from other classes of bodily move-
ments because they have "meaning." In both cases something has
been added to an objective description.
It is usually argued that the addition is necessary, even when behav-
ior is not verbal. Any effort to deal with behavior as a movement of
the parts of ai\ organism meets at once the objection that it cannot
be mere movement which is important but rather what the move-
ment means, either to the behaving organism or to the observer. It
is usually asserted that we can see meaning or purpose in behavior
and should not omit it from our account. But meaning is not a prop-
erty of behavior as such but of the conditions under which behavior
1 Jespersen, O., Language (New York, 1922).
2 Russell, B., Inquiry into Meaning and Truth (New York, 1940) .
13
14 VERBAL BEHAVIOR
occurs. Technically, meanings are to be found among the inde-
pendent variables in a functional account, rather than as properties
of the dependent variable. When someone says that he can see the
meaning of a response, he means that he can infer some of the vari-
ables of which the response is usually a function. The issue is particu-
larly important in the field of verbal behavior where the concept of
meaning enjoys unusual prestige.
In defining verbal behavior as behavior reinforced through the
mediation of other persons we do not, and cannot, specify any one
form, mode, or medium. Any movement capable of affecting another
organism may be verbal. We are likely to single out \ocal behavior,
not only because it is commonest, but because it has little effect upon
the physical environment and hence is almost necessarily verbal. But
there are extensive written languages, sign languages, and languages
in which the "speaker" stimulates the skin of the "listener/' Audible
behavior which is not vocal (for example, clapping the hands for a
servant, or blowing a bugle) and gestures are verbal, although they
may not compose an organized language. The skilled telegraphist
behaves verbally by moving his wrist. Some of these forms normally
arise only after vocal behavior has been established, but this is not
necessarily so. Writing and typing may be either primordially verbal
or transcriptions of a prior vocal form. Pointing to words is verbal
as, indeed, is all pointing, since it is effective only when it alters the
behavior of someone. The definition also covers manipulations of
physical objects which are undertaken because of the effect upon
people, as in the use of ceremonial trappings. In the case of any
medium, the behavior is both verbal and nonverbal at once non-
verbal in the effect upon the medium verbal in the ultimate effect
upon the observer. Ceremonial languages, and the languages of
flowers, gems, and so on, are of little interest, because they have small
vocabularies and little or no grammar, but they are nevertheless
verbal under the terms of the definition. Because vocal verbal be-
havior is the commonest form, we may deal with it as representative.
Where necessary or helpful, parallel problems in other forms may
be considered.
VOCAL BEHAVIOR
Vocal verbal behavior is executed by an extensive musculature
the diaphragm, the vocal cords, the false vocal cords, the epiglottis,
the soft palate, the tongue, the cheek, the lips, and the jaw. The most
GENERAL PROBLEMS 15
complete record of a single instance of an utterance would be an
electrical or mechanical report of the action of all the muscles in-
volved. At the moment this is o theoretical interest only, since noth-
ing like it has ever been made. Fortunately, a science of verbal
behavior need not wait. The complex muscular responses of vocal
behavior affect the verbal environment by producing audible
"speech." This is a much more accessible datum.
The acoustic product of vocal verbal behavior may be recorded
phonographically. The record may be converted into visible form
and analyzed for greater convenience ^into pitch-intensity spectra.
The acoustic report is less accurate than a report of muscular action
because different muscular patterns presumably produce the same
sounds, but it is at least feasible. It is also more convenient because
it uses fewer terms or dimensions. Probably nothing of importance
is lost, because the scientist stands in essentially the same position as
the listener and for many purposes may ignore any property of verbal
behavior which does not produce a difference in the sound-stream.
Even so, an acoustic report tells us more than we usually want to
know, except when acoustic details are to be specially emphasized,
and it soon becomes awkward.
Another kind of record was made possible by the discovery that
speech could be broken into constituent sounds and by the invention
of a phonetic alphabet to represent these sounds. (Both of these ad-
vances, of course, antedated scientific study.) A sample of verbal be-
havior can be recorded by placing appropriate symbols in a corres-
ponding order, as is done, however inexactly, in writing with the Eng-
lish alphabet. So far as we are concerned here, such a record simply
makes it possible to identify some of the acoustic properties of an ut-
terance. The transcription permits the reader to construct a facsimile
of the behavior which will have the same effect upon the verbal com-
munity as the original sample. It is a practical and economical record,
because an indefinite number of different acoustic events may be
represented with a few symbols.
This use of a "phonetic" alphabet makes no commitments about
the functional significance of the units identified. We may use Eng-
lish spelling to record bird calls (to-whit> to-whoo, or peewee), or the
noises of inanimate things (pop and boom), in the sense that in read-
ing such records aloud one constructs a reasonable facsimile of the
original songs or noises. But this does not mean that birds and drums
speak in English "phonemes." The analytical (rather than tran-
l6 VERBAL BEHAVIOR
scriptive) function of the phoneme in modern linguistics arises, on
the one hand, from an excursion into phonology which will not have
to be made here and, on the other, from the study and comparison of
the practices of whole verbal communities. The linguist is concerned
with such facts as these: (i) in one verbal community the responses
pin and bin have different effects or occur under different conditions,
while in another verbal community they have the same effect or occur
under the same conditions; (2) in one verbal community the re-
sponses pit and bit have different effects or occur under different cir-
cumstances, while in another verbal community they have the same
effect or occur under the same circumstances; (3) in that community
in which pin and bin have the same effect, pit and bit also have the
same effect; and in that community in which pin and bin have differ-
ent effects, pit and bit also have different effects. These facts present
problems which lie beyond the mere transcription of verbal behavior,
because they include references to the conditions of occurrence of
verbal behavior or to effects upon a listener. We shall deal with these
additional facts in another way here.
A record of an utterance in a phonetic alphabet provides, of "course,
less information about its properties than an acoustic report, but
there should be no objection if we can show that the properties which
have been preserved are the effective properties of verbal behavior.
This brings us to an important principle in the analysis of behavior.
We distinguish between an instance of a response and a class of re-
sponses. A single response, as an instance of the activity of an or-
ganism, may be described as fully as facilities will permit. But when
we are concerned with the prediction of future behavior it may be
either impossible to predict the great detail of the single instance or,
more likely, unimportant to do so. All we want to know is whether
or not a response of a given class will occur. By "of a given class" we
mean a response showing certain selected properties. We may want to
know whether a man will open a door although we do not care how
he turns the knob. We do not dismiss the details of turning the knob
as unlawful or undetermined; we simply deal with his opening the
door without accounting for them. The property of behavior by
virtue of which we classify a response as "opening a door" is our prin-
cipal interest. In the same way, we do not need to know all the details
of a vocal response so long as the sound-pattern which it produces
achieves a given effect upon a specified verbal community. There
are many practical and theoretical reasons for recording and analyz-
GENERAL PROBLEMS 17
ing given instances of vocal behavior in as great detail as possible,
but they do not coincide with our interests in the prediction and
control o verbal behavior, at least in the present state of the science.
The "phoneme" was an early recognition of the principle of the
defining property of a response. Unfortunately for our present pur-
poses the extension of the concept to historical and comparative
linguistics has obscured its relevance in defining a unit of verbal
behavior in the individual speaker.
The problem of the speech-sound becomes somewhat clearer, and
perhaps loses some of its importance, when we compare other modes
of behavior. If verbal behavior were never vocal, there would be no
sciences of phonology and phonetics. Yet most of the problems to be
considered in the study of verbal behavior would remain. In a com-
munity in which all verbal behavior was written, we should have to
identify "speech-marks," and discover their essential geometric prop-
erties. If such a language resembled modern script, we should have
to study a large number of marks which functioned as, say, the letter
a in order to identify their common features and to discover what
properties could for most purposes be ignored. If such a community
spoke only with typewriters, the range of properties would be narrow.
The advantage of a narrow range for the reader, as well as the scien-
tist, is suggested by the frequent instruction "Please print." Graph-
ology provides a rudimentary "phonetics" of written verbal behavior;
here again the "significances" require other techniques of analysis.
A "direct quotation" is a record of verbal behavior which depends
more explicitly upon a knowledge of the conditions under which the
behavior occurred. It is often, however, little more than an acoustic
or phonetic transcription which permits the reader to reconstruct
relevant properties of the original behavior. The spoken report that
someone said It is four o'clock actually reconstructs an instance of
verbal behavior. A written report permits the reader to reconstruct
it for himself.
A technique which permits the reconstruction of a datum is un-
usual. Science does not generally resort to models or mimicry; its
descriptions of events do not resemble those events. In the field of
nonverbal behavior we usually do not report behavior by imitating
it. Yet in speaking a language under study the scientist uses mimicry
in lieu of the more usual method of description which bears no point-
to-point correspondence with the thing described. (This distinction
l8 VERBAL BEHAVIOR
is discussed further in Chapter 5.) Russell 3 has pointed out that some
rare instances of verbal behavior, such as the Coronation Oath or the
Lord's Prayer, have proper names. He also mentions the method, due
to Godel, of assigning numbers to words and hence to all possible
sentences. The indexing system in a library assigns proper names
(identifying numbers) to the large samples of verbal behavior known
as books. It is not probable, however, that these foreshadow a de-
scriptive system in which all verbal responses will be given names
which bear no greater resemblances to the things named than the
resemblances between events and descriptions in science elsewhere.
No matter how tempting it may be to utilize the special possibility
of phonetic transcription or direct quotation to reconstruct the be-
havior being analyzed, it must be emphasized that from the point of
view of scientific method an expression such as It is four o'clock is
the name of a response. It is obviously not the response being studied,
because that was made by someone else at some other time. It simply
resembles that response in point of form. The conditions responsible
for the original response may not share anything in common with
the conditions responsible for the response on the part of the de-
scribing scientist. This practice, called hypostasis, is an anomaly in
scientific method. The field of verbal behavior is distinguished by the
fact that the names of the things with which it deals are acoustically
similar to the things themselves. As Quine 4 has said, "A quotation is
not a description, but a hieroglyph; it designates its object not by
describing it in terms of other objects, but by picturing it." Quine is
speaking here of the written report of written verbal behavior. In
no other science is this possible, because in no other science do names
and the things named have similar structures.
A quotation is usually something more than an acoustic or pho-
netic transcription, hieroglyph, or name. In the first place, it usually,
though not inevitably, breaks a fairly continuous sample of behavior
into parts. Such breaks need not reflect actual pauses or other prop-
erties of the temporal or stress pattern of the behavior. In quoting a
speech episode, we separate it not only into speech-sounds, repre-
sented by letters, but into larger units called words or sentences,
represented by spatial breaks or punctuation. The difference between
a phonetic report and a direct quotation is seen in the training
needed in the two cases. A small phonetic repertoire will suffice to
3 Russell, B., Inquiry into Meaning and Truth (New York, 1940).
4 Quine, W. V., Mathematical Logic (New York, 1940), p. 26.
GENERAL PROBLEMS 19
transcribe English speech for purposes of reconstruction. But thou-
sands of different * 'words" must ( be learned before direct quotations
can effectively be written down. The process includes, of course,
"learning to spell" and, in particular, to distinguish between homo-
phones. The ability is generally acquired in the process of learning
to write and, once acquired, is often taken for granted. We are likely
to overlook the fact that a process of analysis is actually taking place.
We are also likely to overlook the fact that in a direct quotation
we are inferring something about the conditions under which a
response was emitted, or about characteristic effects on a listener. A
fairly good phonetic transcription may be made of a language one
does not speak, or, as the stenographer often shows, of a familiar
language without otherwise reacting as a listener. But the units of
direct quotation specify verbal responses as units under functional
control! In making a distinction between through and threw, or
between Send me two and Send me, too we are specifying either the
normal conditions under which the responses are made or their
normal effects upon a listener. In the indirect quotation greater
emphasis is placed upon these additional variables. He said that he
would go permits only a very rough reconstruction of an actual verbal
response; only "go" has survived from the possible original I will go.,
and we cannot even be sure that another response characteristic of
the same situation was not actually made. But we know with some
certainty what kind of situation it was and what kind of effect the
remark could have had.
A UNIT OF VERBAL BEHAVIOR
From the muscular or acoustic record of verbal behavior we pass
through phonetic transcription to direct and indirect quotation. As
we do so, we retain less and less information about the specific in-
stance. This loss of detail can be tolerated if properties essential for
prediction continue to be described. At the same time we begin to
add inferences or facts about the conditions under which the response
was made. In undertaking to predict or control verbal behavior, we
rnust^ of course, take such additional variables into account, but their
status must be clarified. Traditional units of verbal behavior never
njake a sharp distinction between observed and inferred. Consider,
for example, the concept of "word." As used by the layman and by
many linguists, a word may be nothing more than an utterance ("I
want a word with you" or "The last word"), or a conventional sub-
2O VERBAL BEHAVIOR
division of an utterance ("What would be two or three words in
English is often only one in German"), or a supposed or real objective
counter or token ("to choose a word" or "to string words together"),
or something common to two or more modes of behavior ("a word may
be either spoken or written"). With less justification we even speak of
the same word in two languages ("French and English use the same
word for 'accord' "), or in two historical stages of the same language,
or in two cognate forms (" 'adamant' is the same word as 'dia^
mond* "). Sometimes "word" seems to mean merely a standard lexical
design ("the word 'fast' ").
What is needed for present purposes and what the traditional
"word" occasionally approximates is a unit of behavior composed
of a response of identifiable form functionally related to one or more
independent variables. In traditional terms we might say that we
need a unit of behavior defined in terms of both "form and meaning."
The analysis of nonverbal behavior has clarified the nature of such
a unit under laboratory conditions in which the expediency of the
unit may be submitted to rigorous checks. An extrapolation of this
concept to the verbal field is central to the analysis represented by
the rest of this book. The kinds of behavior in which we are usually
interested have, as we have seen, an effect upon the environment
which has a return effect upon the organism. Such behavior may be
distinguished from activities which are primarily concerned with the
internal economy of the organism by calling activities which operate
upon the environment "operant behavior." Any unit of such be-
havior is conveniently called "an operant." For most purposes
"operant" is interchangeable with the traditional "response," but
the terms permit us to make the distinction between an instance of
behavior ("So-and-so smoked a cigarette between 2:00 and 2:10 P.M.
yesterday") and a kind of behavior ("cigarette smoking"). The term
"response" is often used for both of these although it does not carry
the second meaning easily. The description of an instance of behavior
does not require a description of related variables or of a functional
relation. The term operant, on the other hand, is concerned with
the prediction and control of a kind of behavior. Although we ob-
serve only instances, we are concerned with laws which specify kinds.
The distinction raises the issue of formalism. A response, as an
instance, can be completely described as a form of behavior. An
operant specifies at least one relation to a variable the effect which
the behavior characteristically, though perhaps not inevitably, has
GENERAL PROBLEMS 21
upon the environment and is therefore not a purely formal unit. A
formal specification cannot be avoided, since a response can be said
to be an instance of an operant only through objective identification.
But identification is not enough. As an instance of a verbal operant,
the response must occur as a function of a certain variable. In this
way we may distinguish between the operant fast in which the con-
trolling variable is shared by the operant speedy and the operant fast
in which the controlling variable is similar to that in the operant
fixed.
A long-standing problem in the analysis of verbal behavior is the
size of the unit. Standard linguistic units are of various sizes. Below
the level of the word lie roots and affixes or, more rigorously, the
small "meaningful" units called morphemes. Above the word come
phrases, idioms, clauses, sentences, and so on. Any one of these may
have functional unity as a verbal operant. A bit of behavior as small
as a single speech-sound, or even a pitch or stress pattern, may be
under independent control of a manipulable variable (we shall see
evidence of such "atomic" verbal operants later). On the other hand,
a large segment of behavior perhaps a phrase like vast majority or
when all is said and done or the truth, the whole truthj and nothing
but the truth or a whole sentence such as Haste makes waste may be
shown to vary under a similarly unitary functional control. Although
parts of these larger operants have the same form as parts of other
operants or even of whole units, there may be no functional inter-
action. If this seems at odds with traditional linguistic analysis, it
must be remembered that the verbal operant is exclusively a unit
of behavior in the individual speaker. The functional unity of a large
operant and the extent to which the presence of that operant in the
repertoire of the speaker may affect operants of similar form must be
decided by a study of the behavior of that speaker. In the practices
characteristic of a verbal community, it may not be possible to estab-
lish the functional unity of a similar large sample of behavior.
We observe that a speaker possesses a verbal repertoire in the sense
that responses of various forms appear in his behavior from time to
time in relation to identifiable conditions. A repertoire, as a collec-
tion of verbal operants, describes the potential behavior of a speaker.
To ask where a verbal operant is when a response is not in the course
of being emitted is like asking where one's knee-jerk is when the
physician is not tapping the patellar tendon. A repertoire of verbal
22 VERBAL BEHAVIOR
behavior is a convenient construct. The distinction between "verbal
operant" and "word" is matched by that between "verbal repertoire"
and "vocabulary." A person is said to possess a vocabulary of so many
thousands of words if these words are observed in his verbal behavior
during a period of time. But a vocabulary is usually regarded as a
warehouse of inanimate tools from which the speaker makes appropri-
ate selections as he speaks. We are concerned here not only with the
fact that certain specific forms of verbal behavior are observed but
that they are observed under specific circumstances. These control-
ling circumstances add a dynamic character to "repertoire" which is
lacking in "vocabulary."
PROBABILITY OF RESPONSE
Some parts of a verbal repertoire are more likely to occur than
others. This likelihood is an extremely important, though difficult,
conception. Our basic datum is not the occurrence of a given re-
sponse as such, but the probability that it will occur at a given time.
Every verbal operant may be conceived of as having under specified
circumstances an assignable probability of emission conveniently
called its "strength." We base the notion of strength upon several
kinds of evidence.
EMISSION OF RESPONSE
If a response is emitted at all, the operant is probaBly strong. Emis-
sion is a better sign of strength, however, if the circumstances are unu-
sual. In one type of verbal slip, for example, the response which in-
trudes upon or distorts behavior (see Chapter 1 1) is not appropriate to
the immediate situation and therefore appears to be especially strong.
A response which appears under inappropriate, difficult, or ambiguous
circumstances but is not a slip is probably strong for the same reason.
The scientist who continues to talk shop during a thrilling football
game or in a noisy subway and the steamrolling conversationalist who
will brook no interruption give evidence of especially strong reper-
toires. Other forms of verbal behavior for example, writing present
evidence of the same sort.
Among the unusual circumstances which give evidence of strength
we may include inadequate verbal stimuli; from the fact that one sees
his name in unclear or briefly exposed printed material or hears his
name in a noisy conversation in a room we infer the strength of his
name in his own repertoire.
GENERAL PROBLEMS %%
ENERGY-LEVEL
Emission of a response is an all-or-none measure. It enables us to
infer strength only in terms of the adequacy of the conditions under
which emission occurs. A second sort of evidence suggests that
strength lies along a continuum from zero to a very high value. A re-
sponse may be executed with a certain energy, which is not to be
confused with "strength" as a synonym for "probability." Energy
seems to vary with probability, and is frequently accepted as a meas-
ure of strength. 5 An energetic and prolonged NO! is not only a strong
response, it suggests a strong tendency to respond which would not
easily be overcome by competing forces. On the other hand, a timid
brief No is accepted as an instance of a weak operant from which we
infer some inadequacy in the independent variables. Relative energy
permits a similar inference. From the response a RED kite we con-
clude that the redness was of special importance to the speaker, while
from a red KITE we infer the special effectiveness of the kite itself
as a variable. Under certain circumstances, a change in energy level
may take place rapidly, as in the case of Mr. Winkle in the Pickwick
Papers, who, just before falling into an alcoholic sleep, cried,
"Let's have 'nother bottle," commencing in a very loud key, and end-
ing in a very faint one.
Other properties of verbal behavior vary with the energy level. At
low levels the part of the response which produces "voicing" drops
out to leave the familiar whisper. At the other end of the continuum
other topographical properties are affected. Probably because of the
mechanism of the speech apparatus, the pitch level of a response tends
to vary with the energy. Other things being equal, the louder the
response the higher the pitch. Pitch level may therefore sometimes be
taken as an indicator of strength. In the behavior of young children
the low and scarcely audible "proper remark" upon a social occasion
and high-pitched playground shouting suggest the range of possible
values. Other forms of verbal behavior generally have a more limited
range. In written verbal behavior some indication of strength may be
found in the size of letters, pressure of the pen, underlining, and so
on. Some allowance for comparable characteristics is made in the de-
s It is possible that energy and probability co-vary only after the energy of the
response has been differentially reinforced (see Science and Human Behavior, p. 95) .
24 VERBAL BEHAVIOR
sign of type. These are now mainly conventional devices, but they
retain some trace of an original variation with operant strength.
SPEED
Another property of emitted verbal behavior is the speed with
which successive parts of a sample follow one another or the speed
with which a response appears after the occasion for it has arisen. In
general we accept the implication that strong verbal behavior is rapid
and that hesitant speech indicates little strength. A ready answer is
one which the speaker is "strongly inclined to make"; a delay in an-
swering leads us to suspect that something is possibly amiss in the
controlling circumstances. The weakness may be due to competitive
behavior. A man deeply engrossed in a book may respond to a call
or a question with delays of the order of several seconds. In young
children, when verbal behavior is weak because it is still in the
process of being acquired, delays of the order of minutes are some-
times observed. A child thirteen months old had acquired the re-
sponse Light. Upon one occasion he was shown a light and asked,
"What is it? What is that?" He made no response for at least a full
minute, and the attempt to get him to respond was given up. He had
turned to play with a toy when the response came out clearly. In path-
ological behavior delays may be still greater. An early report of an
example is due to Head, 6 who asked one of his aphasic patients to
count. The patient did not reply until ten minutes had passed, when
he suddenly began One, two, three, four, We sometimes infer
the strength of the verbal behavior of a correspondent from the speed
with which a letter is answered, and traces of speed in handwriting
supply similar evidence. The frantic gesture exemplifies speed of re-
sponding in still another mode of verbal behavior.
REPETITION
A third possible indication of relative strength is the immediate rep-
etition of a response. Instead of saying NO! with great energy one
may say No! No! No! A sort of wholesale repetition is implied in A
thousand times no! Energy and repetition may be combined. Occa-
sionally it is possible to observe a decline in strength as successive re-
sponses drop off in energy, pitch, and speed: NO! NO/ No! no. Repeti-
tion is apparently responsible for a class of expressions which imply
special emphasis for example, Gome, come, come and Now, now. Ex-
e Head, Henry, Aphasia (New York, 1926).
GENERAL PROBLEMS 25
pressions such as again and again, round and round, and miles and
miles are complicated by an additional principle but probably also
show the effect of strength. A very,, very sad mistake serves in place of
A VERY sad mistake. Repetition may be diluted by intervening be-
havior. In the response No, it's not. Not at all. It's not a question of
what I think the exceptional strength of the form not is evident in its
repetition.
LIMITATIONS ON EVIDENCE OF STRENGTH
It is easy to overestimate the significance of these indicators. If two
or more properties of behavior indicate the same thing, they must
vary together; but energy, speed, and repetitiveness do not always
satisfy this test. We classify people according to the general strength
of their verbal behavior in a way which suggests that our measures
are closely associated. For example, the garrulous person (when he is
garrulous) talks loudly, rapidly, and repeats himself, while the taci-
turn man speaks slowly, quietly, and seldom repeats. But in single
instances these measures are altered through other circumstances, and
the exceptions must be explained. For example, a poorly memorized
answer may be delayed because of its weakness, but during the delay
the aversive character of the situation increases, and when the re-
sponse is finally emitted the energy level may be high. The apparent
discrepancy between delay and force of response requires a special
account.
Another complication is that our measures energy level, speed of
response, and even repetitionenter into the construction of different
forms of response. In English this presents no great difficulty. Abso-
lute levels of pitch and intensity are not "distinctive," nor are relative
pitch levels important. Changes in pitch, however, distinguish differ-
ent types of utterance. Energy of response cannot be taken as an
inevitable indicator of strength so long as it serves to make DE-sert a
different response from de-SERT. The prolonging of a sound does not
necessarily mean strength when it serves as "quantity," nor is redu-
plication always a useful instance of repetition of form.
Energy, speed, and repetitiveness are all affected by special condi-
tions of reinforcement. We speak more energetically to the deaf and
more slowly to anyone who has difficulty in following us; and we re-
peat in both cases. Repetition may be needed against a noisy back-
ground (Hear ye! Hear ye!). To someone at a distance we raise the
2 6 VERBAL BEHAVIOR
energy and pitch of our voice and prolong each sound when possible.
A quick loud response is more likely to get results in a competitive
situation, for example, in reciting in a classroom. We can allow for
special conditions of this sort in evaluating any given measure only
by inferring operant strength, not from the fact that one speaks
loudly, but from the fact that he speaks at an energy level above that
which would ordinarily prevail under the same circumstances. There
is some consolation in the fact that changes in strength due to these
special conditions usually exaggerate "natural" strength. They may
lead us to mistake the relative importance of an indicator but not its
direction or sign.
Unfortunately other kinds of consequences oppose normal evi-
dences of strength. Extreme values of any of these properties interfere
with the effect upon the listener. The verbal community, as a collec-
tion of listeners, forces speech toward a standard level of speed,
energy, and repetitiveness. If a child speaks loudly, he is told not to
shout. If he mumbles, he is told to speak up. If he hesitates, he is told
to hurry. If his words come tumbling out, he is told to be deliberate.
To repeat oneself is bad form, and the double negative, which is
merely the innocent result of a strong No, is called ungrammatical
and illogical.
But if the indicators are somewhat obscured by these conflicting
interests, evidence of strength still survives. We still make practical
inferences about a speaker's behavior from his energy, speed, and re-
petitiveness. A complete levelling to a monotone is not achieved and
is in fact also opposed by the community. In some kinds of verbal
behavior for example, in reading aloud the controlling variable
generates behavior at a tairly constant level of strength. Except for
unfamiliar or poorly learned responses, a text ordinarily does not
strengthen one response above another. But a series of responses of
uniform energy and speed is not effective upon the listener. The
reader is therefore encouraged to introduce spurious signs of
strength. He reads as if his behavior were determined, not by a text,
but by an assortment of variables similar to those in "real" speech.
Now it is significant that he does this by modulating pitch, energy,
and speed. From these indicators of strength the listener infers a
plausible set of determining conditions. The reader has shown good
* interpretation."
We also supply indicators for other reasons. If we are shown a
GENERAL PROBLEMS 07
prized work of art and exclaim Beautiful!, the speed and energy o
the response will not be lost on the owner. We may accentuate the
effect by using repetition: Beautiful, beautiful, simply beautiful!
This is so fully understood by everyone that it becomes part of a
culture to simulate characteristics of strength whether appropriate
independent variables are present or not whether the picture is an
occasion upon which such verbal behavior would naturally be strong.
This would scarcely be the case if the significance of our indicators
had been entirely obscured by other considerations.
OVER-ALL FREQUENCY
A third type of evidence is the over-all frequency with which a
response appears in a large sample of verbal behavior. For example,
the number of times a speaker emits I, me, my, and mine is sometimes
taken to indicate the strength of his behavior with respect to himself
as a controlling variable his "egocentricity" or "conceit." Other re-
sponses have been used to indicate other themes. With such a meas-
ure it can be shown that a writer's interests change from year to year
that he becomes more or less preoccupied with sex, death, or any
other subject. The practice recognizes the general notion of a varying
probability of response and the relevance of an over-all frequency in
measuring it, but such interpretations depend upon certain assump-
tions which are not always justified.
Word counts are often attempts to develop a purely formal analysis
of the dependent variable alone. Verbal behavior is studied without
regard to the circumstances under which it is emitted. But although
it may be useful to know that a response of a given form is frequently
emitted, it is also important to know the prevailing conditions. Since
our unit of analysis is not purely formal, we cannot be sure that all
instances of a response are instances of the same operant. Nor can we
be sure that frequency is not primarily attributable to the frequency
of occurrence of controlling variables. In the case of egocentricity,
the speaker himself is always present and his changing inclination to
talk about that subject may be significant; but a response such as snow
presumably varies with the seasons. A change in frequency may not
reflect a changing tendency to "talk about snow when snow is pres-
ent" but merely certain changing circumstances. Even the frequency
of responses such as /, me, my, and mine may vary as a function of the
listener to whom the verbal behavior is addressed. Unless we know
that such a listener remains present or absent, a change in frequency
2 8 VERBAL BEHAVIOR
cannot be used to infer a change in an underlying tendency to emit
such forms.
Although over-all frequencies are interesting and often satisfactory
data, they depart from our program of dealing with the individual
speaker upon a given occasion. The data are more often relevant to
studies of characteristic practices of a given verbal community, and
hence to the commoner preoccupations of linguistics. Nevertheless,
use may sometimes be made of such data in inferring characteristic
processes in the individual speaker.
PROBABILITY AND THE SINGLE INSTANCE
Although the English language contains many expressions which
suggest that the concept of probability of response is a familiar and
useful one, certain problems remain to be solved in using it in the
analysis of behavior. Under laboratory conditions probability of re-
sponse is easily studied in an individual organism as frequency of
responding. Under these conditions simple changes in frequency can
be shown to be precise functions of specific variables, and such studies
supply some of the most reliable facts about behavior now available.
But we need to move on from the study of frequencies to a considera-
tion of the probability of a single event. The problem is by no means
peculiar to the field of behavior. It is a basic one wherever the data of
a science are probabilistic, and this means the physical sciences in
general. Although the data upon which both the layman and the
scientist base their concepts of probability are in the form of frequen-
cies, both want to talk about the probability of a single forthcoming
event. In later chapters in this book we shall want to consider the way
in which several variables, combining at a given time, contribute
strength to a given response. In doing so we may appear to be going
well beyond a frequency interpretation of probability, yet our evi-
dence for the contribution of each variable is based upon observa-
tions of frequencies alone.
INDEPENDENT VARIABLES AND RELATED PROCESSES
The probability that a verbal response of given form will occur at
a given time is the basic datum to be predicted and controlled. It is
the "dependent variable" in a functional analysis. The conditions
and events to which we turn in order to achieve prediction or control
the "independent variables" must now be considered.
GENERAL PROBLEMS 29
CONDITIONING AND EXTINCTION
Any operant, verbal or otherwise, acquires strength and continues
to be maintained in strength when responses are frequently followed
by the event called "reinforcement." The process of "operant con-
ditioning" is most conspicuous when verbal behavior is first acquired.
The parent sets up a repertoire of responses in the child by reinforc-
ing many instances of a response. Obviously, a response must appear
at least once before it is strengthened by reinforcement. It does not
follow, however, that all the complex forms of adult behavior are
in the child's unconditioned vocal repertoire. The parent need not
wait for the emergence of the final form. Responses of great intricacy
can be constructed in the behavior of an organism through a proce-
dure illustrated by the following demonstration experiment. We
undertake to condition a pigeon to pace the floor of its cage in the
pattern of a figure-8. Let us assume that the pigeon is hungry and that
we can present food quickly and conveniently as a reinforcer. We
need not wait until a figure-8 emerges in its entirety in order to rein-
force the behavior. We begin by reinforcing any behavior which is
part of the final pattern. In case the pigeon remains relatively im-
mobile, we may have to begin by reinforcing any slight movement.
The bird will soon become active, though as yet in no particular pat-
tern. We then withhold reinforcement until the bird begins turning
in one specific direction, let us say clockwise. The slightest movement
in this direction is immediately reinforced. Later, reinforcement is
withheld until an extensive movement is made. Complete circular
movements soon appear. This is half the desired result. The operant
is then partially extinguished as reinforcements are withheld until
the bird turns in a counterclockwise direction. It may be necessary
to reinforce an occasional clockwise movement. Eventually the bird
makes complete turns in both directions. The two parts of the pattern
are now available but not yet in the required order. It is now possible
to wait for a single figure-8 pattern before reinforcing. Under suitable
conditions, the final relatively complex performance can be achieved
in a short period of time.
In teaching the young child to talk, the formal specifications upon
which reinforcement is contingent are at first greatly relaxed. Any
response which vaguely resembles the standard behavior of the com-
munity is reinforced. When these begin to appear frequently, a closer
gO VERBAL BEHAVIOR
approximation is insisted upon. In this manner very complex verbal
forms may be reached. (We shall see in Chapter 4 that there are other
ways of evoking a complex response in order to reinforce it. The
present method of "progressive approximation" is usually relevant
only in the early stages of setting up a verbal repertoire.)
If the contingencies of reinforcement are for any reason ever
relaxed, the properties of the verbal response undergo a change in
the other direction. The degeneration of the forms of military com-
mands is an example. Consider a sergeant with a new squad to be
conditioned to follow his commands. The sergeant begins with a
verbal response borrowed from the larger verbal community, for
example, the response March! At first this may need to be clearly
enunciated, but the squad soon executes the appropriate response
regardless of many specifications of the command, partly because other
aspects of the situation begin to control the behavior. The form of
the response then characteristically degenerates, and may eventually
reach the stage of a mere forceful expulsion of air with some voicing
but little or no shaping. It is only because the appropriate behavior
of the squad survives the deterioration in the behavior of the sergeant
that the final form is effective. The squad, as a group of listeners, has
been progressively reconditioned. A new squad, however, may bring
back the more specific form of response in the behavior of the ser-
geant.
Reinforcing consequences continue to be important after verbal
behavior has been acquired. Their principal function is then to
maintain the response in strength. How often the speaker will emit a
response depends, other things being equal, upon the over-all fre-
quency of reinforcement in a given verbal community. If reinforce-
ments cease altogether through some change of circumstance, an
operant grows weak and may effectively disappear in "extinction."
Operant reinforcement, then, is simply a way of controlling the
probability of occurrence of a certain class of verbal responses. If we
wish to make a response of given form highly probable, we arrange
for the effective reinforcement of many instances. If we wish to elim-
inate it from a verbal repertoire, we arrange that reinforcement shall
no longer follow. Any information regarding the relative frequency
of reinforcement characteristic of a given verbal community is obvi-
ously valuable in predicting such behavior.
GENERAL PROBLEMS gl
STIMULUS CONTROL
A child acquires verbal behavior when relatively unpatterned
vocalizations, selectively reinforced, gradually assume forms which
produce appropriate consequences in a given verbal community. In
formulating this process we do not need to mention stimuli occurring
prior to the behavior to be reinforced. It is difficult, if not impossible,
to discover stimuli which evoke specific vocal responses in the young
child. There is no stimulus which makes a child say b or a or e, as one
may make him salivate by placing a lemon drop in his mouth or make
his pupils contract by shining a light into his eyes. The raw responses
from which verbal behavior is constructed are not "elicited." In order
to reinforce a given response we simply wait until it occurs.
Prior stimuli are, however, important in the control of verbal be-
havior. They are important because they enter into a three-term con-
tingency of reinforcement which may be stated in this way: in the
presence of a given stimulus, a given response is characteristically
followed by a given reinforcement. Such a contingency is a property
of the environment. When it prevails, the organism not only acquires
the response which achieves reinforcement, it becomes more likely
to emit that response in the presence of the prior stimulus. The process
through which this comes about, called "stimulus discrimination,"
has been extensively studied in nonverbal behavior. Numerous ex-
amples will be described in later chapters.
MOTIVATION AND EMOTION
Although reinforcement provides for the control of a response, we
do not use reinforcement as such when we later exercise control. By
reinforcing with candy we strengthen the response Candy! but the
response will be emitted only when the child is, as we say, hungry for
candy. Subsequently we control the response, not by further rein-
forcement, but by depriving or satiating the child with candy. Non-
verbal responses are controlled in the same way. Whether a door is
opened with a "twist-and-push" or with an Outl, we make the re-
sponse more or less likely by altering the deprivation associated with
the reinforcement of getting through the door. If the response has
been reinforced in several different ways, we may control it by chang-
ing, not the deprivation, but the impending reinforcement. We in-
crease the probability that a man will cross a room by placing a
currently reinforcing object on the other side. By removing such an
32 VERBAL BEHAVIOR
object or, better still, placing it near the man, we reduce the prob-
ability of his crossing the room.
When an operant is acquired it becomes a member of a group of
responses which vary together with the relevant deprivation. A man
gets a drink of water in many ways by reaching for a glass of water,
by opening a faucet, by pouring water from a pitcher, and so on. The
verbal operant Water! becomes a member of this group when it is
reinforced with water. The probabilities of all operants so reinforced
vary together. Responses in all classes are made more likely to occur
when we deprive the man of water or cause him to lose water for
example, by inducing violent exercise, by feeding him salt which
must be excreted, or by raising the temperature of his surroundings
so that he sweats. On the other hand, we make all such responses less
likely to occur by causing the man to drink large amounts of water.
Such operations are said by the layman to create or allay a "state
of thirst." Such a concept is only as valid or useful in prediction and
control as the observations upon which it rests. The important events
are the operations which are said to change the state of thirst. In
predicting and controlling the verbal response Water! we do not
change thirst directly; we engage in certain operations which are said
to change it. It is simpler to omit any reference to a "drive" and say
that the probability of the response Water! can be changed through
these operations.
Suppose, however, that in addition to drinking water our speaker
has also used water to extinguish fires. Until we have tested the point,
we cannot be sure that a response acquired when he has been rein-
forced with water while thirsty will be emitted when the wastebasket
catches fire. If there is any functional connection, it must be found in
certain events common to drinking water and extinguishing a fire.
If the response Water! has been reinforced with the visual stimula-
tion supplied by water prior to water in the mouth, and if this stimu-
lation plays a role in controlling the behavior of extinguishing a fire,
then the response acquired only under water deprivation may occur
in the case of a conflagration. The group of operations which affect
the strength of Water! suggests, in common parlance, some general
"need for water" rather than "thirst." But we should have to ex-
amine all behavior in which water plays an essential role in order to
define this need. We may say that we increase the strength of any
response which has been reinforced with water, including the verbal
response Water!, by strengthening any behavior which "requires
GENERAL PROBLEMS 33
water for its execution." (In more technical terms, the latter would
be described as any behavior under the control of water as a dis-
criminative stimulus.)
AVERSIVE CONTROL
There are other types of consequences which alter the strength of
a verbal response. Behavior may be reinforced by the reduction of
aversive stimulation. When an aversive stimulus itself is reduced, we
call the behavior escape. When some condition which characteristi-
cally precedes an aversive stimulus is reduced, we speak of avoidance.
Thus, if the verbal response Stop it! is reinforced when it brings
about the cessation of physical injury, the response is an example of
escape. But Don't touch me! may be reinforced when it brings about
the cessation of the threat of such injury of events which have pre-
viously been followed by such injury and which are therefore con-
ditioned aversive stimuli and the behavior is then called avoidance.
When a speaker has had a history of such reinforcement, we control
his verbal behavior by creating appropriate circumstances. We make
him say Stop it! by pummeling him, or Don't touch me! by threaten-
ing to do so.
A complete account of the verbal behavior of the individual
speaker would lead us to survey other variables in the fields of motiva-
tion and emotion, but the processes here are seldom, if ever, uniquely
related to verbal behavior. Some relevant points are discussed in
Chapter 8.
THE LISTENER AND THE TOTAL VERBAL EPISODE
Our definition of verbal behavior applies only to the speaker, but
the listener cannot be omitted from our account. The traditional
conception of verbal behavior discussed in Chapter i has generally
implied that certain basic linguistic processes were common to both
speaker and listener. Common processes are suggested when language
is said to arouse in the mind of the listener "ideas present in the mind
of the speaker/' or when communication is regarded as successful
only if an expression has "the same meaning for both speaker and
listener." Theories of meaning are usually applied to both speaker
and listener as if the meaning process were the same for both.
Much of the behavior of the listener has no resemblance to the
34 VERBAL BEHAVIOR
behavior of the speaker and is not verbal according to our definition. 7
But the listener (and the reader as well) is reacting to verbal stimuli
the end-products of the behavior here analyzed and we are natu-
rally interested in the fate of such stimuli. On the one hand they
evoke responses of glands and smooth muscles, mediated by the auto-
nomic nervous system, especially emotional reactions. These ex-
emplify classical conditioned reflexes. On the other hand verbal
stimuli control much of the complex skeletal behavior with which
the individual operates upon his environment. The relevant proc-
esses in both these broad areas will be taken up as needed in what
follows. In neither case do the verbal stimuli differ in any particular
from other kinds of stimulation. The behavior of a man as listener
is not to be distinguished from other forms of his behavior.
Our interest in the listener is not, however, merely an interest in
what happens to the verbal stimuli created by the speaker. In a com-
plete account of a speech episode we need to show that the behavior
of the listener does in fact provide the conditions we have assumed
in explaining the behavior of the speaker. We need separate but in-
terlocking accounts of the behaviors of both speaker and listener if
our explanation of verbal behavior is to be complete. In explaining
the behavior of the speaker we assume a listener who will reinforce
his behavior in certain ways. In accounting for the behavior of the
listener we assume a speaker whose behavior bears a certain relation
to environmental conditions. The interchanges between them must
explain all the conditions thus assumed. The account of the whole
episode is then complete.
7 We shall see later that in many important instances the listener is also behaving
at the same time as a speaker.
Part II
CONTROLLING VARIABLES
Chapter 3
The Mand
IN A GIVEN verbal community, certain responses are characteristically
followed by certain consequences. Wait! is followed by someone's
waiting and Sh-h! by silence. Much of the verbal behavior of young
children is of this sort. Candy! is characteristically followed by the
receipt of candy and Out! by the opening of a door. These effects are
not inevitable, but we can usually find one consequence of each
response which is commoner than any other. There are nonverbal
parallels. Out!, as we have seen, has the same ultimate effect as turn-
ing a knob and pushing against a door. Both forms of behavior be-
come part of the repertoire of the organism through operant condi-
tioning. When a response is characteristically reinforced in a given
way, its likelihood of appearing in the behavior of the speaker is a
function of the deprivation associated with that reinforcement. The
response Candy! will be more likely to occur after a period of candy
deprivation, and least likely after candy satiation. The response
Quiet! is reinforced through the reduction of an aversive condition,
and we can increase the probability of its occurrence by creating such
a condition that is, by making a noise.
It will be convenient to have a name for the type of verbal operant
in which a response of given form is characteristically followed by a
given consequense in a verbal community. The basic relationship
has been recognized in syntactic and grammatical analyses (expres-
sions such as the "imperative mood" and "commands and entreaties"
suggest themselves), but no traditional term can safely be used here.
The term "mand" has a certain mnemonic value derived from "com-
mand," "demand," "countermand," and so on, and is conveniently
brief. A ".mand," then, may be defined as a verbal operant in which
the response^ is reinforced by a characteristic consequence and is
35
36 VERBAL BEHAVIOR
therefore under the functional control of relevant conditions of
deprivation or aversive stimulation. Adjectival and verbal uses of the
term are self-explanatory. In particular, and in contrast with other
types of verbal operants to be discussed later, the response has no
specified relation to a prior stimulus.
A mand is characterized by the unique relationship between the
form of the response and the reinforcement characteristically re-
ceived in a given verbal community. It is sometimes convenient to
refer to this relation by saying that a mand "specifies" its reinforce-
ment. Listen!, Look!, Run!, Stop!, and Say yes! specify the behavior
of a listener; but when a hungry diner calls Bread!, or More soup!,
he is specifying the ultimate reinforcement. Frequently both the be-
havior of the listener and the ultimate reinforcement are specified.
The mand Pass the salt! specifies an action (pass) and an ultimate re-
inforcement (the salt).
A mand is a type of verbal operant singled out by its controlling
variables. It is not a formal unit of analysis. No response can be said
to be a mand from its form alone. As a general rule, in order to
identify any type of verbal operant we need to know the kind of
variables of which the response is a function. In a given verbal com-
munity, however, certain formal properties may be so closely asso-
ciated with specific kinds of variables that the latter may often be
safely inferred. In the present case, we may say that some responses,
simply because of formal properties, are very probably mands.
The pattern of response which characteristically achieves the given
reinforcement depends, of course, upon the 'language" that is, upon
the reinforcing practices of the verbal community (see Appendix).
But we have to explain not only the relationships between patterns
of response and reinforcements, but the maintenance of the behavior
of the listener. When we come to consider other types of verbal
operants, we shall find that the behavior functions mainly for the
benefit of the listener, and in that case his behavior is not difficult to
explain. The mand, however, works primarily for the benefit of the
speaker; why should the listener perform the necessary mediation of
reinforcement?
What needs to be explained, in other words, is the total speech
episode. This can be done by listing all relevant events in the be-
havior of both speaker and listener in their proper temporal order.
The deprivation or aversive stimulation responsible for the strength
of each must be specified, and the reinforcing contingencies must
THE MAND ' 37
explain the origin and continued maintenance o the behavior.
Several interchanges between the two organisms frequently occur.
Figure i represents an episode in which one person asks another
for bread. The problem of motivation is disposed of by assuming a
hungry speaker and a listener already predisposed to reinforce him
with bread. The first physical interchange takes place when the mere
presence of the listener provides the occasion (S D ) X for the speaker's
mand Bread, please! The speaker does not ordinarily emit the re-
sponse when no one is present, but when a listener appears, the proba-
bility of response is increased (Chapter 7). The visual and other
stimulation supplied by the listener is indicated by the first f in the
diagram. The speaker's response (Bread, please) produces a verbal
stimulus for the listener. The interchange here (the first J4) is in the
form of auditory stimulation which supplies the occasion (S DT ) for
the nonverbal response of passing the bread. Though we have as-
sumed a listener predisposed to give bread to the speaker, the behav-
ior does not appear indiscriminately. The speaker's mand (Breadj
^please) establishes an occasion upon which the listener can, so to
speak, successfully give bread. The interchange of the bread is indi-
cated by the second f- The effect upon the speaker is to reinforce the
mand by the presentation of bread, and this completes the account so
far as the speaker is concerned. It is characteristic of many cultures,
however, that the successful reinforcement of a mand is followed by
another verbal response, designed to assure similar behavior of the
listener in the future. In the diagram, this is indicated by the verbal
response Thank you. This response is under the control of the stimu-
lation provided by the preceding parts of the episode indicated
in the diagram as the second S D . The auditory stimulation (the
second J4,) supplies a reinforcing stimulus for the listener, which
accounts to some extent for the behavior of passing the bread. This
verbal stimulus may also contribute to the occasion for a verbal re-
sponse on the part of the listener (You're welcome) which, when
heard by the speaker, reinforces the response Thank you. These last
two interchanges are not an integral part of the speech episode con-
taining a mand; they supplement our assumptions respecting the
motivation of the two individuals. (The effect of a verbal response
in serving as a reinforcement is further discussed in Chapter 6.)
i S = stimulus, R = response. The superscript V identifies verbal terms. S*> is tech-
nically a discriminative stimulus, i.e., not an eliciting stimulus.
38 VERBAL BEHAVIOR
KINDS OF MANDS
The mand represented In Figure i, in which the listener is inde-
pendently motivated to reinforce the speaker, is commonly called
a request. The response serves merely to indicate that the speaker will
accept what the listener is already disposed to give. It is, to repeat, an
occasion for successful giving. Often, however, the speaker's response,
in addition to specifying a reinforcement, may need to establish an
aversive situation from which the listener can escape only by pro-
viding the appropriate mediation. When the listener's behavior is
(S P E A
K E R)
(Audience)
Bread, please
bread
Thank you
You're welcome
S D
R v *
s re!n +s D
R v
^ s rein Y
t
||
t
||
ft
1
u
1
u
II
S DV
R
* s"" 1 v + s
D * RV ~ *-
Bread, please
passes bread
Thank you
You're welcome
(L 1 S T E
N E R)
FIGURE i
thus reinforced by reducing a threat, the speaker's response is called
f ja command. Hands up! not only specifies a form of action, it consti-
tutes a threat from which the victim can escape only by holding up
his hands. The threat may be carried by a characteristic intonation
or may be made explicit, as in Your money or your life!, where the
first two words specify the reinforcement and the last two the aversive
consequences with which the listener is threatened. Military com-
mands are obeyed because of a sort of standing threat.
A paradigm showing the interaction of speaker and listener in a
command is shown in Figure 2. Here again the first interchange is
from listener to speaker. The preseitce-of -tfee4istenep -constitutes the
occasion for verbal behavior (S D ) and also in this instance an aversive
stimulus (S**) ^ffom which the speaker's response will bring escape.
Let us say that the listener is in the speaker's way. The response Step
aside! specifies an action on the part of the listener and its intonation
constitutes a threat. Heard by the listener (at 44)* these evoke the ap-
propriate response of stepping aside which, in clearing the way for
the speaker, reinforces his mand. The reinforcement is also the oc-
THE MANB
39
casion for a change in his behavior, possibly quite conspicuous, by
virtue of which the threat is withdrawn. This change reinforces the
listener for stepping aside (at 4).
(SPEAKER)
(Audience)
Step aside -I" threat way cleared
threat withdrawn
S D -f S av
-* R V4. R avY ^ S n (=-S*
*) . (_ R .v)
f
11 t
1
j
U 1
1
S DV +S- V * R
_ ^ $ r*in (== _ s avYj
Step aside (threatening) steps aside
threat withdrawn
(LISTENER)
FIGURE 2
There are other ways in which the speaker may alter the prob-
ability that the listener will respond in an appropriate fashion. A
mand which promotes reinforcement by generating an emotional
disposition is commonly called a prayer or entreaty. A question is a
mand which specifies verbal action, and the behavior of the listener
permits us to classify it as a request, a command, or a prayer, as the
case may be. In Figure 3 we assume that the listener not only provides
an audience for the speaker but creates a situation in which the
(Audience)
S D
t
(S P E A
What's your name?
K E R)
Lester
Thank you
D V
> S rein -l-S DV
R v
1
I 1
ft
1
11
S DV +?
What's your name?
(L I S T E
II
1
Lester
Thank you
N E R)
FIGURE 3
speaker will be reinforced by being told the listener's name. The
speaker's mand What's your name? becomes (at the first 4,|) a verbal
stimulus for the listener who replies either because of a standing
4O VERBAL BEHAVIOR
tendency to respond to the speaker or an implied threat in the
speaker's response, or because the speaker has emotionally predis-
posed him to reply. His reply at ft completes the paradigm for the
speaker, but it also serves as the occasion for the response Thank you,
which completes the paradigm for the listener if that is necessary. If
Jthe speaker has controlled the listener mainly through aversive stimu-
lation, Thank you may be replaced by some visible relaxation of a
threat.
(An analysis of this sort seems to do violence to the temporal
dimensions of behavior. All of the events represented in one of these
paradigms might take place in two or three seconds. The events
described, however, can occur within a brief period, and we can
demonstrate the reality of such a linkage by interrupting the chain
at any point. The function of the interlocking paradigm is to check
the completeness of our account of verbal behavior. Have the be-
haviors of both speaker and listener been fully accounted for? Have
we identified appropriate states of deprivation or aversive stimulation
in all cases? Have we correctly represented the actual physical inter-
change between the two organisms? In this account of the speech
episode, it should be noted that nothing is appealed to beyond the
separate behaviors of speaker and listener. By assuming the condi-
tions supplied by a listener, we analyze the behavior of a speaker,
and vice versa. By putting the two cases together we construct the
total episode and show how it naturally arises and completes itself.)
Several other classes of mands may be distinguished in terms of
the behavior of the listener. In mediating the reinforcement of the
speaker, the listener will occasionally enjoy consequences in which
the speaker does not otherwise participate but which are neverthe-
less reinforcing. When these consist of positive reinforcement, we call
the mand advice (Go west!). When by carrying out the behavior
specified by the speaker the listener escapes from aversive stimulation,
we call the mand a warning (Look out!). When the listener is already
inclined to act in a given way but is restrained by, for example, a
threat, the mand which cancels the threat is commonly called per-
mission (Go ahead!). When gratuitous reinforcement of the behavior
of the listener is extended by the speaker, the mand is called an offer
(Take one free!). When the speaker characteristically goes on to emit
other behavior which may serve as reinforcement for the listener, the
mand is a call either a call to attention or the "vocative" call-by-
name.
Classifying the behavior of the speaker in terms of the character-
THE MAND 41
istics of the mediating behavior of the listener may be distinguished
from the traditional practice of defining requests, commands, prayers,
advice, warnings, permission, offers, and calls in terms of "the inten-
tion" of the speaker. In general, intention may be reduced to con-
tingencies of reinforcement. In the present case the conspicuous dif-
ferences lie in the behavior of the listener and the conditions which
control it. But these result in different contingencies of reinforce-
ment for the speaker, which yield different dynamic properties, dif-
ferent interrelationships among responses, different intonations, and
so on.
Since verbal behavior in the form of the mand operates primarily
for the benefit of the speaker, repeated mands are likely to move the
listener to revolt. It is customary to soften or conceal the mand char-
acter. The response Water! is not so likely to be successful as I'm
thirsty , the form of which is characteristic of a type of verbal operant
to be described in Chapter 5, or May I have some waterf, which ap-
pears to specify only the less burdensome act of saying Yes. (The pre-
tense is exposed if the listener simply says Yes.) Would you mind
getting me a drink? also specifies merely a verbal response (No, not
at all), but the implied mand may be effective because of the sug-
gested deference to the inclination of the listener. Explicit deference
appears in tags such as if you don't mind, if you please, or simply
please. When emphasized, these may convert a mere request into the
stronger entreaty.
The inclination of the listener to respond may be heightened by
flattery or praise, as in Get me a drink, my good fellow. The Lord's
Prayer is a mixture of mands and praise following this pattern. The
praise may be made conditional upon the execution of the reinforce-
ment, as in Be a good fellow and get me a drink, which may be trans-
lated Only if you get me a drink will I call you a good fellow. Grati-
tude may be withheld until the listener responds, as in I'll thank you
to get me a drink. Open bargaining is sometimes resorted to, as in
Give me a drink and I'll tell you all about it. The abundance of such
supplementary techniques merely emphasizes the precariousness of
the reinforcement of the mand.
Any response used in conjunction with different mands specifying
different reinforcements comes under the control of different depriva-
tions and acquires certain general properties. Please is the best known
example. It is strengthened by almost any state of deprivation, and
is often emitted without further specification of the behavior of the
reinforcer. Mands of lesser generality include the emphatic forms
42 VERBAL BEHAVIOR
Soij Now!, Now,, then!., and Here! where the common consequence
is the response of the listener in paying attention. Since the listener's
subsequent behavior may be relevant to many states of deprivation,
these responses come under a rather broad control. Generalized
mands reinforced by the attention of the listener are often used in
conjunction with other types of verbal behavior to be considered
later.
The mand relation is clearest when it is in exclusive control of a
response, but it is also effective in combination with other kinds of
variables. A hungry man may show a high frequency of responses
which, if they were mands, would be said to specify food, even though
they appear under circumstances which more clearly suggest other
types of verbal operants to be described below. Such "multiple
causation" of a single response is treated in Chapter 9.
DYNAMIC PROPERTIES OF THE MAND
The energy level of the mand may vary from very faint to very
loud, and the speed with which it is emitted when the occasion arises
may vary from very fast to very slow. If the pattern is of substantial
length, it may be executed slowly or rapidly. If the reinforcement is
not immediately forthcoming, the response may be emitted only once
or may be repeated. These properties vary as the result of many condi-
tions in the past and present history of the speaker. Particularly rele-
vant are level of deprivation and intensity of aversive stimulation and
the extent to which a given listener or someone like him has rein-
forced similar responses in the past (or has refused to do so). Such
conditions have a relatively greater effect upon the mand than upon
the other types of verbal behavior to be discussed in later chapters.
The wide range of dynamic properties which result makes the mand
a very expressive type of operant.
The probability and intensity of the listener's behavior may also
vary over a wide range. If the listener is not already predisposed to
act, the probability of his mediating a reinforcement may depend
upon the effectiveness of the aversive stimulation supplied by the
speaker. Some listeners are accustomed to taking orders they have
felt the unconditioned aversive consequences of not doing so and
respond appropriately to simple mands. Others are more likely to
react to softened forms. The intonation, loudness, or other indication
that the speaker will supply aversive consequences has an appropriate
effect. A hesitant or weak request or command is least likely to be
THE MAND 43
reinforced. A loud and threatening response is likely to be rein-
forced subject only to the relative strength of listener and speaker.
It is to be noted that mands are characteristic of most hypnotic in-
structions, and the extent to which the subject co-operates or obliges
the hypnotist will depend upon the kinds of variables here being
considered. These variables enter into what is called the authority
or prestige of the speaker.
The net result of a long history of responding to mands is a general
tendency no longer easily traced to any form of deprivation or aver-
sive stimulation. The listener obliges and may not even be aware
(see Chapter 5) that he is doing so. A classroom experiment designed
by F. S. Keller illustrates this point. The instructor says, "Before
summing up these influences, there is an additional one that should
be mentioned. I can illustrate this best with an example." At this
point he turns to the blackboard and writes
DO IT ON PAPER
The instructor then continues, "What you did was the result of the
'set' or 'attitude* that you had at the moment you were presented
with this stimulus situation. Examples of this are multiple and you
could supply them from your own experience by the hour. Usually
no one is aware of the times when they occur in everyday life, but
our generalization is the product of laboratory experimentation and
can readily be checked." He then puts on the board
DO IT ON PAPER
When the number of those who multiplied in the first instance is
compared with the number who multiplied in the second, there is
almost always more multiplying in the second case. The underlined
words, which of course are not emphasized in the instructions, exert
some control over the listener's behavior.
TRADITIONAL TREATMENT
In the traditional treatment of verbal behavior, the "meaning" of
a mand is presumably the reinforcement which characteristically
44 VERBAL BEHAVIOR
follows it. The meaning of Candy! is the kind of object frequently
produced by that response. But "what is communicated" would ap-
pear to be "the speaker's need for candy," which refers to the con-
trolling state of deprivation. The concept of the mand, or of the
verbal operant in general, explicitly recognizes both contingency
of reinforcement and deprivation or aversive stimulation and is free
to deal with these variables in appropriate fashion without trying to
identify a relation of reference or a process of communication.
Apart from these questions of semantics, the formulation carries
some of the burden of grammar and syntax in dealing with the
dynamic properties of verbal behavior. The mand obviously suggests
the imperative mood, but interrogatives are also mands, as are most
interjections and vocatives, and some subjunctives and optatives. The
traditional classifications suffer from a mixture of levels of analysis.
In particular they show the influence of formal descriptive systems
in which sentences are classified with little or no reference to the be-
havior of the speaker. It is here that the shortcomings of grammar
and syntax in a causal analysis are most obvious. Appropriate tech-
niques are lacking. As Epictetus said, "When you are to write to your
friend, grammar will tell you how to write; but whether you are to
write to your friend at all, grammar will not tell you." The use of
the mand as a unit of analysis does not mean that the work of lin-
guistic analysis can be avoided, but it simplifies our task by isolating
the behavior of the individual speaker as an object of study and by
making appropriate techniques available.
In choosing between descriptive systems on the basis of simplicity
and effectiveness, the greater familiarity of the classical approach
should not be put into the balance. Consider, for example, the fol-
lowing quotation:
In many countries it has been observed that very early a child uses a
long m (without a vowel) as a sign that it wants something, but we can
hardly be right in supposing that the sound is originally meant by
children in this sense. They do not use it consciously until they see that
grown-up people, on hearing the sound, come up and find out what the
child wants. 2
Although this passage may be said to make an intelligible point in
connection with an episode which is intelligibly reported, much is
left to be done. It is not the most advantageous account for all con-
cernedj for the psychological terms it contains raise many problems.
2 Jesperson, O v Language (New York, 1922), p. 157.
THE MAND 45
How would the point be made in the present terms? The expres-
sion "uses a long m as a sign that it wants something" becomes "emits
the sound m in a given state of deprivation or aversive stimulation/'
The expression "the sound is not originally meant in this sense"
becomes "the relation between the sound and the state of depriva-
tion or aversive stimulation is innate, or at least of some earlier
origin, and the response is not verbal according to our definition."
"They do not use it consciously . . ." becomes "It is not conditioned
as a verbal response " And ". . . until they see that grown-up
people, on hearing the sound, come up and find out what the child
wants" becomes ". . . until the emission of the sound leads listeners
to supply reinforcements appropriate to a particular deprivation."
The whole passage might be translated:
It has been observed that very early a child emits the sound m in certain
states of deprivation or aversive stimulation, but we can hardly be right
in calling the response verbal at this stage. It is conditioned as a verbal
operant only when people, upon hearing the sound, come up and supply
appropriate reinforcement.
The distinction between learned and unlearned response is much
easier to make in terms of a history of reinforcement than in terms
of meaning and conscious use. An important example is crying. Vocal
behavior of this sort is clearly an unconditioned response in the new-
born infant. For some time it is a function of various states of depri-
vation and aversive stimulation. But when crying is characteristically
followed by parental attentions which are reinforcing, it may become
verbal according to our definition. It has become a different behav-
ioral unit because it is now under the control of different variables.
It has also probably acquired different properties, for parents are
likely to react differently to different intonations or intensities of
crying.
The simplicity of such a translation is very different from the sim-
plicity of the original account. The translation is simple because its
terms can be defined with respect to experimental operations and
because it is consistent with other statements about verbal and non-
verbal behavior. The original account is simple because it is familiar
and appropriate for casual discourse. It is the difference between the
systematic simplicity of science and the ready comprehensibility of
the layman's account. Newton's Principia was not simple to the man
in the street, but in one sense it was simpler than everything which
the man in the street had to say about the same subject.
46 VERBAL BEHAVIOR
THE EXTENDED MAND
A mand assumes a given form because of contingencies of reinforce-
ment maintained by the listener or by the verbal community as a
whole. The stimulating conditions which prevail when such a response
is emitted and reinforced do not enter into the definition of the unit.
When a mand is reinforced by a reduction in unconditioned or
conditioned aversive stimuli, stimuli occurring prior to the response
must, of course, be taken into account, but these serve a different
function from the stimuli being considered here. Stimuli affecting
the speaker prior to the emission of verbal behavior are often im-
portant and are never wholly irrelevant, as we shall see in the follow-
ing chapters. The probability of emission of a response is greatest
when the stimulating conditions closely resemble those which have
previously prevailed before reinforcement. But past and present cir-
cumstances need not be identical; indeed, any aspect or feature of the
present situation which resembles the situation at the time of re-
inforcement may be supposed to make some contribution to the
probability of response.
An example of extended stimulus control is seen when people
mand the behavior of dolls, small babies, and untrained animals.
These "listeners" cannot possibly reinforce the behavior in character-
istic fashion. Nevertheless, they have enough in common with
listeners who have previously provided reinforcement to control the
response, at least when it shows appreciable strength. The fact that
reinforcement is unlikely or impossible may affect the dynamic prop-
erties. The response may be weak, or emitted in a whimsical fashion,
or accompanied by suitable comment (Chapter 12). On the other
hand, such behavior often occurs when its "irrational" aspects are
not seen by the speaker. We acquire and retain the response Stop!
because many listeners stop whatever they are doing when we emit it,
but as a result we may say Stop! to a car with faulty brakes or to a
cue ball which threatens to drop into a pocket of the pool table.
The same process leads in the extreme case to the emission of
mands in the absence of any listener whatsoever. The lone man dying
of thirst gasps Water! An unattended king calls A horse, a horse, my
kingdom for a horse! These responses are "unreasonable" in the sense
that they can have no possible effect upon the momentary environ-
ment, but the underlying process is lawful. Through a process of
stimulus induction situations which are similar to earlier situations
THE MAND 4/7
come to control the behavior, and in the extreme case a very strong
response is emitted when no comparable stimulus can be detected.
There are many familiar nonverbal instances of stimulus induction.
It may be true that one cannot open a door without a door or eat a
meal without a meal, but in a state of great strength parts of even
the most practical behavior occur in the absence of the stimulation
required for proper execution. A baseball player who has dropped
the ball at a crucial moment may pantomime the correct throw with
an empty hand. A thirsty person may "pretend" to drink from an
empty glass. Many gestures appear to have originated as "irrational"
extension of practical responses. The traffic officer extends his hand,
palm outward, toward an oncoming car, as if to bring the car to a
stop by physical means. The gesture functions as a verbal response,
but it exemplifies the extension of a practical response through stimu-
lus induction to a situation in which normal reinforcement is im-
possible. Verbal behavior may more easily break free from stimulus
control, because by its very nature it does not require environmental
support that is, no stimuli need be present to direct it or to form
important links in chaining responses.
SUPERSTITIOUS MANDS
There are mands which cannot be explained by arguing that
responses of the same form have been reinforced under similar cir-
cumstances. The dice player exclaims Come seven!, for example,
even though he has not asked for and got sevens anywhere. Accidental
reinforcement of the response appears to be the explanation. The
experimental study of nonverbal behavior has shown that merely
intermittent reinforcement, such as that provided by chance throws
of seven, is sufficient to maintain a response in strength. The player
may readily admit that there is no mechanical connection between
his response and the behavior of the dice, but he retains the response
in some strength and continues to utter it, either whimsically or
seriously under sufficient stress, because of its occasional "conse-
quences." Mands which specify the behavior of inanimate objects
often receive some reinforcement in this sense. The response Blow,
blow, thou winter wind, for example, is usually uttered when the
wind is already blowing, and the correlation between behavior and
effect, though spurious, may work a change in operant strength.
Other "unreasonable" mands owe their strength to collateral effects
not strictly specified in the form of the response. Many responses
48 VERBAL BEHAVIOR
mand emotional behavior even though, because of the special ways
in which such behavior is conditioned, true emotional responses on
the part of the listener cannot be carried out to order. The mand
O dry your tears has no effect upon lacrimal secretion. We cannot
write a paradigm similar to that of Figure i in which the mand has
the form Weep, please! because we cannot complete the account
of the listener. A verbal response may be part of a larger pattern, how-
ever, which produces tears in the sensitive listener or reader for other
reasons. Intonation and other properties are important in eliciting
emotional behavior, and an emotional speaker will supplement his
responses with very generous sound effects. We do not say Cheer up!
in a dull tone, for we cannot leave the effect upon the listener to the
mand alone. Properly pronounced, however, such a response may
have an effect. The general process is not characteristic of the mand,
and the same result is frequently (and probably more easily) obtained
without the mand form.
THE MAGICAL MAND
There are mands which cannot be accounted for by showing that
they have ever had the effect specified or any similar effect upon
similar occasions. The speaker appears to create new mands on the
analogy of old ones. Having effectively manded bread and butter, he
goes on to mand the jam, even though he has never obtained jam
before in this way. The poet exclaims Milton, thou shouldst be living
in this hour!; although he has never successfully addressed Milton
before nor brought anyone to life with a similar response. The special
relation between response and consequence exemplified by the mand
establishes a general pattern of control over the environment. In
moments of sufficient stress, the speaker simply describes the re-
inforcement appropriate to a given state of deprivation or aversive
stimulation. The response must, of course, already be part of his
verbal repertoire as some other type of verbal operant (Chapters 4
and 5).
This sort of extended o 
