Linguist Geoff Pullum finds another example of someone finding an animal that can learn through conditioning to form associations between words in human language and actions or things in the world and then acting as if the animal has understood human language. It's a perfectly fine experiment to see what the cognitive capacities of dogs are, but Pullum says this is like teaching a goldfish to associate a Greek letter with a certain action and then saying it understands Greek. It's not quite as bad as that, but it's really close.
Chinese room!
I guess it's the same thing, but at least the Chinese room can approximate the whole of the outward signs of human language (by stipulation). Pullum's point is that these animals who are being claimed to have learned human language haven't even done anything close to what happens in Searle's case.