Washoe, the signing chimpanzees, and the man who refuted his own experiment. · published 14 September 2026
Herbert Terrace set out to prove apes build sentences, found statistical structure in over nineteen thousand of Nim Chimpsky’s multi-sign utterances, and then discovered on the videotapes that most were prompted by what the teacher had signed seconds earlier — so he published the refutation of his own project in Science in 1979, while the underlying fact that chimps like Washoe genuinely learned hundreds of signs to name and request survived intact.
These are the working notes, unedited. Everything the video
states as fact comes from here, and nothing else may be stated as fact.
The sections on limits and on what the finding does not show are
part of the brief, not an afterthought — see
the method.
The work: Allen and Beatrix Gardner began teaching American Sign Language to a young chimpanzee named Washoe at the University of Nevada, Reno, in the late 1960s. She was cross-fostered — raised in a human household rather than a cage — and by the end she was credited with roughly 350 signs. She later passed signs to her adopted son, Loulis, without human teaching.
The Gardners were not casual about counting. A sign was only added to Washoe’s vocabulary after she had produced it spontaneously rather than in response to a prompt, appropriately, and in the right context, on every day for at least fourteen consecutive days, with double-blind scoring. That standard is stricter than most of what was later done in the field and it is the reason the work was taken seriously.
The claim that travelled. What reached the public was larger than the vocabulary: that a chimpanzee had acquired language the way a child does, including combining signs into something like sentences. Washoe signing “water bird” at a swan is the example everyone repeats.
The refutation, from inside. Herbert Terrace raised a chimpanzee he named Nim Chimpsky — a joke at Noam Chomsky’s expense — expecting to demonstrate exactly that. He analysed more than nineteen thousand of Nim’s multi-sign utterances and found apparent structure: particular signs tending to occupy particular positions, in a way that looked too regular to be habit. Then he went back to the videotapes. Most of Nim’s utterances turned out to be prompted by what the teacher had just signed a moment earlier. The structure was in the conversation, not in the chimpanzee. Terrace and his colleagues published this against their own project in Science in 1979, under the title “Can an ape create a sentence?” Their answer was no. Laura-Ann Petitto, who had worked most closely with Nim, was one of the authors.
Limits
Terrace studied Nim, not Washoe, and the Gardners and Roger Fouts disputed that the finding carried across — Nim was handled by a large rotating cast of teachers in conditions nothing like Washoe’s household, which is a fair objection. “Water bird” is a single anecdote and cannot settle anything either way: it may be a coinage, or simply the sign for water followed by the sign for bird. And the dispute is about grammar, not about whether the animals learned anything.
What this does not show
Not that the apes were faking, not that the researchers were dishonest, and not that chimpanzees learned nothing. They demonstrably learned to use symbols to ask for things and to label them, which was not obvious beforehand and is still interesting. What did not survive is the specific claim that they combined those symbols into sentences the way a child does. And Terrace did not expose someone else: he set out to prove the claim, found he had produced an artefact, and said so in print.