Behaviorism Could Count Words but Not the Structure of Language

B.F. Skinner arrives at language by way of a box with a lever in it.

The rat presses. A pellet appears. The rate of pressing changes.

The apparatus has a beautiful vulgarity. It makes a consequence visible, records it continuously, and leaves no room for a little inner homunculus to claim the credit.

Skinner had been building this machinery since the 1930s. The Behavior of Organisms gave it a programme. Verbal Behavior, in 1957, extended the programme to language.

The wager was not modest. Private events were real, on radical behaviorism’s account, but they were not privileged causes. Explanation lived in the individual’s history and current environment.

The preceding stimulus. The emitted response. What follows it.

That is a formidable way to make learning public.

Pavlov had already shown how a neutral stimulus can acquire the power to elicit a reflex. Operant conditioning concerns a different creature: behavior emitted rather than elicited, made more or less likely by what comes after.

The distinction is not pedantry. A bell makes a dog salivate. A dog sits for a treat. One has acquired a respondent reflex. The other has entered a contingency.

Skinner and Charles Ferster’s Schedules of Reinforcement appeared the same year as Verbal Behavior, largely out of work with pigeons, and it gave the contingency a personality.

Fixed ratio: a post-reinforcement pause, then a brisk run. Variable ratio: the highest sustained rate, and the greatest resistance to extinction. Fixed interval: the familiar scallop, quiet after a reinforcer and urgent as the interval closes. Variable interval: steadier work, because time has become untrustworthy.

The ledger can be exquisite.

It can distinguish a slot machine’s variable ratio from a paycheck’s loose fixed-interval analogue. It can show why intermittent reinforcement makes extinction harder: an omitted reinforcer might be the beginning of the end, or merely an ordinary gap. It can observe an extinction burst, then spontaneous recovery after a rest.

A rat, it turns out, can make a procedure look like a short story.

So the ledger begins to look as though it could go all the way.

If words are behavior, and behavior has consequences, perhaps meaning is only the name for a very complicated reinforcement history. Or perhaps the other party must insist that language floats free of learning entirely, pure structure above the grubby economy of pellets, approval, correction, and attention.

Neither escape route earns its rent.

Austin enters through a door Skinner’s box cannot quite instrument.

His Harvard lectures, published after his death as How to Do Things with Words, begin by objecting to the descriptive fallacy: the assumption that the serious work of a sentence is reporting facts and becoming true or false.

Sometimes an utterance does that. Sometimes it names a ship, marries a couple, bequeaths a watch, warns a friend, or makes a promise and leaves everyone suddenly in possession of an obligation.

The words have not ceased to have meaning. They have acquired a job description.

Austin’s final distinction is less tidy than the old performative and constative split, because he had the intellectual manners to notice when his own neat drawer no longer closed.

A locution is the words with their conventional grammatical and lexical meaning. An illocution is the act performed in saying them: asserting, warning, requesting, promising. A perlocution is the effect achieved by saying them: persuading, frightening, convincing.

The third category looks especially hospitable to behaviorism.

Effects can be observed. Perhaps measured. A warning that startles, a request that produces salt, a promise that secures cooperation. Surely these are all outcomes in a social environment, the sort of thing an ambitious ledger has always wanted.

But an effect is not an illocutionary force wearing a lab coat.

A promise made insincerely can still be a promise. It may be defective. It has not become a different species of noise because the promisor is a scoundrel.

An order may fail to move anyone. A warning may be ignored.

The action performed in speaking and the later behavior of a hearer can come apart, which is inconvenient for any account that wishes the second to exhaust the first.

John Searle’s Speech Acts makes the inconvenience almost administrative. A promise requires propositional content, preparatory conditions, sincerity, and an essential condition: the utterance counts as undertaking an obligation.

A failure in the conditions can produce a misfire, or an abuse. It does not turn every linguistic act into a successful button that releases its intended human pellet.

At dinner, someone asks whether you can pass the salt.

No anatomical examination has unexpectedly begun. You hear a request.

The literal question concerns ability. The request travels on that question’s felicity conditions, including the hearer’s ability and the fact that the salt has not already arrived. The sentence is doing double duty, and you performed the second job without noticing the first.

Grice gives the same table a second, quieter ghost. His Cooperative Principle and its maxims of Quantity, Quality, Relation, and Manner describe what a hearer is entitled to assume.

Asked about a party, a speaker says the food was good. The words can be perfectly true. Their conspicuous incompleteness communicates that the party was not.

Nothing here proves that conditioning is irrelevant to speaking, listening, correcting a child, or sustaining a conversational practice. Social consequences plainly matter.

Nor does speech-act theory supply a complete account of how semantic competence is acquired. It classifies what speakers do with meaningful words. It does not hand over a developmental mechanism in a velvet pouch.

That restraint matters.

Watson’s 1913 manifesto made psychology the prediction and control of behavior, and his Little Albert study made an enormous claim from one uncontrolled infant. It was a demonstration with poor manners and poorer aftercare, not a general theory’s coronation.

Broad claims about language learning deserve convergent developmental evidence, not a famous child, a classroom, or a very confident rat.

There may be a sharply separable innate module for some linguistic capacity. That is speculation, not a conclusion supplied by the ledger or by the dining table.

What survives the quarrel is not that behaviorism secretly contained speech-act theory, nor that meaning sits outside the material life of language like a duchess refusing the train.

Language is public action with structure.

It has histories, consequences, conventions, grammatical forms, obligations, and inferences.

A word can be counted. An utterance can be reinforced. But its structure lets it promise without persuading, request while asking, and mean more than it says.

The world supplies responses. Language arranges them into acts.