A child says “water”.
He says it when he is thirsty and wants some. He says it when he walks past the sink and notices it. He says it right after you say “water”. And he says it when you ask “what do you drink?”.
To anyone listening, that is one word four times. Behaviorally it is four different things, learned at different times, sometimes months apart, and one of them being present is no promise about the others.
That is the whole point of Skinner’s classification, and it is why “how many words does he have?” is a question that cannot be answered usefully.
The four you will use most
Skinner sorted verbal behavior by what sets the occasion for it and what follows it — not by meaning, not by grammar. Four of his categories carry most of the practical weight:
Mand — a request. Occasioned by wanting something; followed by getting it. The child says “water” and water arrives. This is the one where the consequence is specific to what was asked, which is exactly why it is usually the first one worth teaching: the reinforcer is the thing the child already wanted.
Tact — naming what is there. Occasioned by the thing itself, under the control of what is physically present rather than a want; maintained less by any single consequence than by a general history of social acknowledgement — attention, a smile, “yes, that’s water” — which is also why a response can be under more than one kind of control at once. The child sees water and says “water”, and gets no water for it.
Echoic — repeating what was just said. Occasioned by someone else’s words, with a technical requirement: point-to-point correspondence — the response reproduces the same sounds, in the same order, as the stimulus. It is often a teaching tool rather than a target, and a child who echoes fluently may have none of the other three.
Intraverbal — a verbal response controlled by someone else’s words, without that point-to-point correspondence — the response does not repeat the stimulus’s form back, even when it names something related to it. “What do you drink?” → “water” qualifies because the answer does not echo the question’s sounds; the thing itself does not have to be present at all. Conversation lives here, and it is usually the last of the four to arrive.
The consequence that changes practice
These are separate skills, and teaching one does not guarantee the others appear.
A child taught to name a glass — reliably, across many glasses — may still have no way to ask for one when thirsty. A child who asks for water twenty times a day may not name it when it is simply there. Both are common, and neither is a contradiction. Whether teaching one operant carries over to some of the others is itself something researchers have tested directly rather than assumed — sometimes it does, often only partly — so the safer reading is to check each one rather than assume either way (Gamba, Goyos & Petursdottir, 2015; Finn, Miguel & Ahearn, 2012).
A vocabulary count still says something — more words is more to work with — but it says little about which of the four operants a given word has, so alongside it the practical move is: for the words that matter to this child, which of the four are present, and which have never been taught? That question has a different answer for almost every word.
It is also why the same word turns up more than once in a well-built program, and why that is not redundancy: asking for water and naming water are two teaching targets that happen to share a sound.
What the classification does not do
It does not tell you what a child understands. It sorts responses by function, which is a claim about the conditions under which a response occurs — not a claim about comprehension.
It does not measure progress on its own either. A child can have a large tact repertoire and almost no way to ask for anything, and a count of words hides that completely. What tells you something is the breakdown: which operants, for which words, and which of them work with other people.
And listener behavior — following an instruction, touching a picture when you hear its name — is not one of the verbal operants. It matters enormously and it is taught deliberately, but it is a different kind of behavior and filing it under this table to make the table complete misrepresents what the table classifies.
Sources
- Skinner, B. F. (1957). Verbal Behavior. Appleton-Century-Crofts.
- Gamba, J., Goyos, C., & Petursdottir, A. I. (2015). Emergence of untrained mands or tacts of novel adjectives — reviews the empirical evidence on whether teaching one verbal operant produces another.
- Finn, H. E., Miguel, C. F., & Ahearn, W. H. (2012). The emergence of untrained mands and tacts in children with autism. Journal of Applied Behavior Analysis, 45(2), 265–280.
Further reading
- Two simple tasks, two different worlds — why matching a photo and responding to a word are not the same kind of easy.
- How to teach vocabulary with visual supports — choosing which words, and in what order.