Verbal Behavior

Abstract

Verbal behavior is Skinner's term for language analyzed as operant behavior: speaking, writing, and signing shaped by their consequences in a community of listeners, not expressions of an inner grammar. In his 1957 book Skinner classified these actions by their controlling variables into verbal operants (the mand, tact, echoic, and intraverbal among them), each defined by what evokes and reinforces it rather than by its form. Noam Chomsky's 1959 critique argued that the generativity of language cannot be explained by reinforcement history, an exchange credited with catalyzing the cognitive revolution. MacCorquodale's reply and later naming research by Horne and Lowe kept the behavioral program active, and the taxonomy found durable application in language intervention for children with autism. Three demonstrations present the verbal operants, the combinatorial argument for generativity, and the functional independence of mand and tact.

Keywords: verbal behavior, operant conditioning, verbal operant

A child who says water while reaching toward a cup, a chemist who says water while pointing at a labeled flask, and a student who says water after the teacher says it are, to ordinary observation, all producing the same word. B. F. Skinner's wager in Verbal Behavior was that they are in fact producing three different behaviors, because the same sound is under the control of three different sets of variables — a state of deprivation, a visible object, and another speaker's utterance — each with its own history of reinforcement (Skinner, 1957). This is the central move of the behavioral analysis of language: to classify what a speaker does not by the shape of the words but by the conditions that evoke them and the consequences that maintain them. The sections that follow lay out the resulting taxonomy, the critique that challenged its reach, and the research tradition that carried it into modern practice.

What Verbal Behavior Is

Verbal behavior, in Skinner's sense, is behavior reinforced through the mediation of another person. The definition is deliberately narrow in one respect and broad in another. It is broad because it does not require speech: writing, signing, gesturing, and pressing symbols all qualify, since all are actions that act on the environment only by way of a listener trained to respond to them. It is narrow because it excludes behavior that affects the world directly — turning a key opens a lock whatever anyone thinks, but saying open sesame does nothing unless a listener has been conditioned to react. What makes an act verbal is thus not that it uses words but that its effect runs through a second organism whose responses have been shaped to reinforce the speaker in specific ways (Skinner, 1957).

This relocation of language from the head to the interaction is the program's defining commitment. Traditional accounts treat an utterance as the output of an inner meaning or intention; Skinner's analysis treats it as the terminal link in a chain of environmental causes — a motivating condition, a discriminative stimulus, a history of consequences — and asks what those causes are for each class of utterance. The unit of analysis is therefore not the word or the sentence but the verbal operant: a class of responses defined by the variables that control it. Because the same word can belong to different operants depending on what evokes it, and because a single operant can be expressed in many word forms, the classification cuts across the vocabulary of ordinary grammar entirely.

Types of Verbal Behavior

The Medical Subject Headings vocabulary files Verbal Behavior (D014705) directly beneath Communication, its parent descriptor, and lists one narrower descriptor beneath it, shown in Table 1. The single subtype should not be read as the whole of the field: a MeSH classification is an indexing classification, a controlled vocabulary built so that the published literature can be retrieved, not a theory of the kinds of verbal behavior or a claim that these categories carve language at its natural joints. It records how the literature is indexed, not what the true divisions are, and the functional operants Skinner defined — the mand, the tact, and the rest, treated in the next section — run orthogonally to it rather than appearing as MeSH children. Speech has no article on this site, so it is named in plain text and glossed from its MeSH sense.

The Verbal Operants

The heart of Verbal Behavior is a taxonomy of elementary operants, each defined by a pairing of antecedent and consequence rather than by anything in the utterance itself. The mand is evoked by a motivating operation — deprivation or aversive stimulation — and is reinforced by a consequence specific to it: a thirsty speaker says water and is reinforced by water, so the mand roughly maps onto requests, demands, and commands. The tact is evoked by a nonverbal object or event and is reinforced generally, by the listener's attention or approval rather than by any particular outcome: a speaker sees a dog and says dog, reinforced merely by being understood, so the tact is the operant of naming and describing. The echoic is evoked by another speaker's verbal behavior and reproduces its form — the child who repeats water after hearing it — while the intraverbal is evoked by verbal behavior whose form it does not copy, as in answering a question, completing capital of France with Paris, or any conversational exchange where one utterance occasions a different one. Beyond these primary classes Skinner added the autoclitic, a secondary operant that acts on the speaker's own verbal behavior — qualifying, quantifying, or ordering it — and carries the burden of what grammar and syntax accomplish in his account (Skinner, 1957).

Two features of this scheme carry its real weight. The first is that the operants are defined functionally: the same topography, the word water, is a mand, a tact, an echoic, or an intraverbal depending entirely on the controlling variables, so the taxonomy classifies the behavior-environment relation and not the sound. The second is that the operants are, in principle, independent: acquiring water as a tact does not guarantee water as a mand, because the two are maintained by different consequences under different antecedents. Jack Michael, who did more than anyone to sharpen and extend Skinner's framework, pressed exactly this point, refining the analysis of the motivating conditions that evoke the mand — what he termed the establishing or motivating operation — and insisting on the sharp functional line between manding and tacting that later became central to applied work (Sundberg & Michael, 2001). The first demonstration lets the controlling variables be set independently and reports which operant the resulting contingency defines.

The utterance is held fixed at the word water. Change the controlling variables — what evokes it and what reinforces it — and watch the same sound become a different verbal operant. The taxonomy classifies the behavior–environment relation, not the topography.

Note. The mand is set apart by its specificreinforcement and its motivating antecedent; the tact by a nonverbal antecedent and generalized reinforcement; the echoic and intraverbal share a verbal antecedent and differ only in whether the response copies the model. This is why training one operant need not produce the others — the subject of the third demonstration.

The Chomsky Critique

No work in the behavioral analysis of language is remembered apart from the review that attacked it. In 1959 Noam Chomsky argued that Skinner's extrapolation from animal conditioning to human language was vacuous: terms like stimulus, response, and reinforcement, precisely defined in the laboratory, could be applied to speech only by stretching them until they meant nothing, so that to say a speaker emits the word beautiful under stimulus control amounts to no more than saying the speaker calls something beautiful (Chomsky, 1959). His deeper objection was about generativity. A competent speaker produces and understands an unbounded number of sentences never before encountered, and does so rapidly and on the basis of impoverished input, which Chomsky held to be impossible for any mechanism that acquires verbal responses one reinforced instance at a time. Language, on this view, requires an internal generative system — a grammar — of a kind the operant account has no room for. The review is routinely named as one of the sparks of the cognitive revolution, the shift that reinstated internal mental structure as a legitimate object of psychological science.

The combinatorial core of the generativity argument is worth making precise, because it is quantitative. If a speaker has even a modest number of word choices available at each position in a sentence, the number of possible sentences of ordinary length is astronomically larger than the number any person could produce or be reinforced for in a lifetime, so the overwhelming majority of the sentences a speaker will ever utter are necessarily novel. The worked example below carries this calculation through, and the second demonstration lets its parameters vary. The force of the point is not that reinforcement plays no role in language but that a repertoire of separately reinforced responses, however large, cannot be the whole story of a system that is productive by construction.

Give a speaker b grammatical word choices at each of Lpositions. The number of possible sentences is b^L. Compare it with the roughly 2.5 billion sentences a person could utter in an entire lifetime.

Note. At the defaults — b = 10,L = 20 — the space is 10²⁰ sentences and a whole lifetime of one sentence per second covers about 2.5 × 10⁻¹¹ of it: for every sentence a speaker ever produces, some forty billion of the same length go unused. Because the bars are on a logarithmic scale, each equal step rightward is a tenfold increase; the navy bar runs off the chart while the gold one barely leaves the origin. Almost every sentence a person utters is therefore one they were never specifically reinforced for.

Reply, Naming, and Modern Application

The behavioral program did not concede the field. In 1970 Kenneth MacCorquodale published a measured reply arguing that Chomsky had reviewed not Skinner's book but a composite behaviorism — a mixture of Hullian theory and reflexology — that Skinner had never held, and that the actual claims of Verbal Behavior, offered as an interpretive framework rather than an experimentally proven theory, were left largely untouched by the critique (MacCorquodale, 1970). Whatever one makes of the exchange, the research tradition continued. A notable later development was Charles Lowe and Pauline Horne's account of naming, which proposed that a bidirectional relation between hearing a word and saying it, once established, functions as a learned behavioral unit that could explain how children come to derive new stimulus relations without direct training — an attempt to give the behavioral account the generative reach its critics said it lacked (Horne & Lowe, 1996). That line of work rests on an older empirical discovery. Murray Sidman's demonstration of stimulus equivalence showed that after a few directly trained conditional relations among arbitrary stimuli, subjects spontaneously respond to many untrained relations as if the stimuli were interchangeable, so that a modest training history yields a far larger set of behaviors than was explicitly reinforced (Sidman & Tailby, 1982). Derived relations of this kind became the behavioral program's principal evidence that productivity need not imply an innate grammar, and the foundation on which later accounts of naming and relational responding were built.

The taxonomy's most consequential afterlife, however, has been applied. Mark Sundberg and Jack Michael argued that Skinner's analysis offered concrete advantages for teaching language to children with autism, precisely because its functional operants can be trained as separate targets: a child may be taught to tact an object without thereby being able to mand for it, so an effective curriculum must program both (Sundberg & Michael, 2001). Empirical reviews have since assessed how far the laboratory operants transfer to intervention. Sautter and LeBlanc catalogued the growing body of applied studies using Skinner's analysis with human participants and found the mand and tact the most thoroughly researched, with the derived and conversational operants far less so (Sautter & LeBlanc, 2006), and DeSouza, Akers, and Fisher later reviewed interventions for children with autism specifically, confirming substantial support for mand and tact training while noting that the evidence thins markedly for intraverbal and more complex verbal behavior (DeSouza et al., 2017). A bibliometric analysis by Petursdottir and Devine traced how the book's influence on the scholarly literature grew rather than faded across the 2005 to 2016 decade, concentrated heavily in applied behavior analysis (Petursdottir & Devine, 2017). Table 2 summarizes the principal strands.

Table 2

Principal Lines of Work on Verbal Behavior

Note. The founding analysis and its critique frame the theoretical dispute; the later strands show where the program delivered — in derived relations and, above all, in applied language training.

Worked Example

Chomsky's generativity argument can be made numerical, and doing so shows why a repertoire of separately reinforced utterances cannot be the whole account of language. Suppose, conservatively, that at each position in a sentence a competent speaker could grammatically choose among about 10 words — a gross underestimate, since real vocabularies run to tens of thousands, but enough to make the point. For a sentence of 20 words, the number of distinct sentences the speaker could construct is 10 raised to the 20th power, that is 10^20, or one hundred quintillion. This is the size of the space the speaker commands, not a count of sentences anyone would bother to utter.

Now ask how many sentences a person could possibly produce in a lifetime. A life of 79 years contains about 2.5 x 10^9 seconds (79 x 365.25 x 24 x 3600 = 2,493,050,400). Even on the absurd assumption that someone produced one brand-new sentence every second, awake and asleep, for an entire life, the total would be roughly 2.5 x 10^9 sentences — about two and a half billion. Dividing, the fraction of the 20-word sentence space that this heroic talker could ever cover is 2.5 x 10^9 divided by 10^20, which is 2.5 x 10^-11: about one hundred-billionth of the space, a proportion so small that for every sentence the speaker ever produces there remain some forty billion unused sentences of the same length. A more realistic estimate — one sentence every five seconds across sixteen waking hours for 79 years — gives about 3.3 x 10^8 sentences, roughly 3 x 10^-12 of the space, smaller still.

The conclusion is unavoidable: almost every sentence a person utters is one they were never specifically reinforced for producing, because the overwhelming majority of possible sentences will never be uttered by anyone at all. A theory in which verbal responses are acquired as individually reinforced units therefore cannot, by itself, explain the ordinary fact that speakers routinely produce and understand sentences that are new to them. This is the quantitative heart of Chomsky's objection, and it is why the debate pushed psychology toward positing an internal generative system rather than a stored repertoire. It does not show that reinforcement is irrelevant to how language is learned or used; it shows that productivity is a structural property that a list, however long, does not have. The third demonstration turns to a different lesson of the taxonomy — the functional independence of the operants — which is where the behavioral analysis proved most useful in practice.

A child is taught the word cookie. Train it as one operant and then probe the other. Because the mand and the tact are maintained by different consequences under different antecedents, acquiring one does not confer the other — each must be trained in its own right.

Note. The functional independence of the operants is the taxonomy’s most consequential applied lesson: a child taught to tact an object is not thereby able to mand for it, so interventions after Sundberg and Michael (2001) program the mand and the tact as separate targets rather than assuming one will generalize to the other.

Discussion

The lasting significance of Verbal Behavior is double, and the two halves pull in opposite directions. As a general theory of language it is widely judged to have failed: the generativity that Chomsky foregrounded is a genuine and central property of human language, and no account that treats sentences as separately acquired operants can explain it, as the worked example makes concrete. The book's place in the history of psychology is in large part the place of a foil, the target whose perceived inadequacy helped license the return of internal mental structure to the center of the field. To read Skinner only this way, however, is to miss what survived. The functional taxonomy — the insight that the same word form can serve different behavioral functions depending on what evokes and maintains it, and that these functions can be trained and must sometimes be trained separately — turned out to be a durable and practically powerful tool, independent of whether operant conditioning is the right theory of grammar.

That is why the two literatures diverged. The theoretical dispute was largely settled against Skinner and absorbed into the history of the cognitive revolution, while the applied program grew, finding in autism intervention a setting where classifying language by function rather than form yields concrete teaching targets. MacCorquodale's reply has force here: Skinner offered his analysis explicitly as an interpretation, a way of seeing, rather than as a set of experimentally established laws, and an interpretation can remain useful for some purposes long after it has been rejected for others. The honest summary is that verbal behavior is a failed theory of language that bequeathed a successful technology of language training, and that its chief service to cognitive psychology was to provoke the argument that defined the discipline's modern shape.

Current Directions

Contemporary work on verbal behavior is concentrated in applied behavior analysis rather than in theoretical disputes about grammar, and its central question has shifted from whether the operants exist to how far they transfer. The best-supported targets remain the mand and the tact, trained as functionally independent repertoires, while the derived and conversational operants — the intraverbal above all — remain comparatively under-researched and harder to establish reliably, a gap that recent reviews of autism intervention identify as the field's main unfinished business (DeSouza et al., 2017). A parallel strand pursues the generativity problem on behavioral terms, asking whether phenomena such as naming and derived stimulus relations can give the account the productive reach its critics denied it (Horne & Lowe, 1996). Bibliometric evidence indicates that the book's influence on the published literature continued to rise rather than settle across the most recent decade examined, which suggests the applied program is still expanding rather than consolidating (Petursdottir & Devine, 2017).

Common Misconceptions

Chomsky's review disproved and ended Skinner's program.

The review was influential but not a conclusive refutation. MacCorquodale argued it attacked a general behaviorism Skinner had not advanced, and the research tradition continued — most productively in applied language intervention — long after the review appeared (MacCorquodale, 1970; DeSouza et al., 2017).

A verbal operant is a kind of word.

It is a behavior-environment relation, not a vocabulary item. The same word is a mand, a tact, an echoic, or an intraverbal depending on what evokes it and what reinforces it, so the operants cut across ordinary grammatical categories rather than aligning with them (Skinner, 1957).

Teaching a child to name an object teaches them to ask for it.

The tact and the mand are functionally independent: they are maintained by different consequences under different antecedents, so acquiring one does not guarantee the other. Effective language programs train them as separate targets (Sundberg & Michael, 2001).

Glossary

Autoclitic.

A secondary verbal operant that acts on the speaker's own other verbal behavior, qualifying or ordering it, as in the words that mark assertion, negation, or grammatical relation; Skinner's device for word order and syntax.

Behaviorism.

The school of psychology that explains behavior by its environmental causes and consequences rather than by inner mental states; the framework within which verbal behavior is analyzed.

Discriminative Stimulus.

An antecedent stimulus in whose presence a response has been reinforced and which therefore comes to evoke it; the object or event that occasions a tact.

Echoic.

A verbal operant evoked by another speaker's verbal behavior and reproducing its form, as when a child repeats a word just heard; the basis of vocal imitation.

Functional Independence.

The property that one verbal operant can be acquired without the others, because each is maintained by its own antecedent and consequence; the basis for training the mand and the tact as separate targets.

Generativity.

The capacity of a competent speaker to produce and understand an unbounded number of novel sentences; the central property Chomsky argued a reinforcement account cannot explain.

Intraverbal.

A verbal operant evoked by verbal behavior whose form it does not copy — answering a question or completing a phrase — and generally reinforced; the operant of conversation.

Mand.

A verbal operant evoked by a motivating operation such as deprivation and reinforced by a consequence specific to it; roughly, a request, demand, or command.

Motivating Operation.

An environmental condition, such as deprivation or aversive stimulation, that alters the momentary value of a consequence and so evokes the behavior it reinforces; Michael's refinement of Skinner's antecedent for the mand.

Naming.

In Horne and Lowe's account, a learned bidirectional relation between hearing a word and saying it that functions as a unit and lets a child derive untrained stimulus relations.

Operant Conditioning.

Learning in which behavior is shaped and maintained by its consequences; the form of learning Skinner extended to language.

Reinforcement.

A consequence that increases the future probability of the behavior it follows; specific for the mand, generalized for the tact.

Tact.

A verbal operant evoked by a nonverbal object or event and maintained by generalized reinforcement such as attention; the operant of naming and describing.

Verbal Operant.

A class of verbal responses defined by the variables that control it — its antecedent and its consequence — rather than by the form of the words; the unit of Skinner's analysis.

Key Researchers

Noam Chomsky

. Linguist and cognitive scientist whose 1959 review argued that the generativity of language defeats a reinforcement account, a critique widely credited with helping catalyze the cognitive revolution. Wikipedia - Wikidata

Jack Michael

. Behavior analyst who sharpened Skinner's taxonomy, drawing the functional distinction between the mand and the tact and developing the concept of the motivating operation. Wikipedia - Wikidata

Anna I. Petursdottir

. Contemporary researcher on verbal behavior, stimulus equivalence, and language acquisition at the University of Nevada, Reno; co-author of the bibliometric review mapping the modern verbal-behavior literature. Google Scholar - Faculty Page

B. F. Skinner

. Founder of radical behaviorism and the experimental analysis of behavior; his 1957 Verbal Behavior recast language as operant behavior under environmental control, defining the verbal operants the field still uses. Wikipedia - Wikidata

Mark L. Sundberg

. Behavior analyst specializing in the applied analysis of verbal behavior; author of the VB-MAPP language assessment and founding editor of The Analysis of Verbal Behavior. Faculty Page

Frequently Asked Questions

What is verbal behavior?

Verbal behavior is B. F. Skinner's analysis of language as operant behavior: speaking, writing, or signing that is reinforced through the mediation of a listener, classified by the variables that control it rather than by the form of the words (Skinner, 1957).

What are the verbal operants?

They are the elementary classes Skinner defined by antecedent and consequence: the mand (evoked by deprivation, specifically reinforced), the tact (evoked by an object, generally reinforced), the echoic (copying a verbal model), and the intraverbal (answering other verbal behavior), among others (Skinner, 1957).

What is the difference between a mand and a tact?

A mand is a request driven by a motivating operation and reinforced by a specific outcome, as in saying water to get water. A tact is a naming response evoked by an object and reinforced by general approval, as in saying water on seeing it. They are functionally independent, so one can be present without the other (Sundberg & Michael, 2001).

What was Chomsky's criticism of Verbal Behavior?

Chomsky argued that behavioral terms lose their meaning when stretched to cover speech, and that the generativity of language (the endless production of novel sentences from limited input) cannot be explained by a history of reinforced responses, implying an internal generative system instead (Chomsky, 1959).

Did Chomsky's review end behavioral approaches to language?

No. MacCorquodale replied that the review attacked a composite behaviorism Skinner had not proposed, and the research program continued, most visibly in applied behavior analysis and autism language intervention, where it remains active today (MacCorquodale, 1970; DeSouza et al., 2017).

How is verbal behavior used in autism intervention?

Because the operants are functionally independent, a language curriculum built on Skinner's analysis trains the mand, the tact, and other operants as separate targets. Reviews find strong support for mand and tact training and weaker evidence for the more complex operants (Sundberg & Michael, 2001; DeSouza et al., 2017).

Does verbal behavior require speech?

No. Any behavior reinforced through a listener qualifies, including writing, signing, gesturing, or exchanging picture symbols, because what makes an act verbal is that its effect runs through a second person rather than acting on the world directly (Skinner, 1957).

What is naming in the behavioral analysis of language?

Naming, in Horne and Lowe's account, is a learned bidirectional relation between hearing a word and saying it that functions as a single behavioral unit, proposed to explain how children derive new word-object relations without direct training (Horne & Lowe, 1996).

References