At the McGill Office for Science and Society, science communicator Jonathan Jarry takes after a long overdue target — claims that animals are being taught (or even can be taught) to speak. He asks, “On social media, dogs press buttons that talk for them. But can they even make out the words coming out of these speakers?”

Buttons on the floor

Bunny, a sheepadoodle, is also a social media celeb with 8.5 million TikTok followers. Using the buttons on the floor, she can signal humans about various matters. “Dog parents” are buying these buttons at US$16 each, hoping to communicate with their pets at a deeper level than the ones we take for granted:

The way they work is simple. The human records themselves saying a single word into the button’s microphone—practical words like “OUTSIDE” and “PLAY”, but also more abstract ones like “STRANGER” and “SAD”—and train their dog to tap it when that is what they want. When pressed, the button plays back the owner’s recorded word on its built-in speaker, giving the dog the voice of the owner. Clever dogs are free to combine multiple button taps to form what their owners hope are meaningful sentences. “SAD, PLAY, OUTSIDE” becomes “I am sad and I want to go play outside.”

“The Dogs That Speak: Breakthrough or Illusion?” (July 23, 2026)

But Jarry asks, what’s really happening here? These sorts of experiments were conducted throughout the twentieth century, especially on chimpanzees and on gorillas like Koko or clever birds like Alex the parrot.

Even dolphins were studied in the hopes of learning “dolphinese.”

Jarry’s thesis is that we fool ourselves too readily:

When you spend years working with an animal on linguistic communication, you become personally invested in the outcome. You are likely to overinterpret what is being said. When your dog presses “IS, WENT” as smoke clears, your brain will look for it to make sense and ignore the possibility that this is just nonsense. Handlers are also likely to subconsciously cue their animal, much like with Hans’ owner. An animal can thus respond to a change in your body without understanding the words you are speaking. “Breakthrough or Illusion?”

Also, he notes, people tend to edit out the failures and present only the most striking results, making it difficult to determine what the dog routinely understands, as opposed to chance combinations that appear meaningful.

In any event, there is a genuine question whether the dogs even hear the sounds from the speakers clearly. According to one study:

There was a striking drop in understanding when the button were used, which suggested to the authors that “dogs do not recognize the words when they hear them from this device.” They measured that, while dogs can hear frequencies up to 45,000 Hertz, the sound coming out of a button’s speaker was much more limited and lost most of its power above 5,000 Hertz. In other words, the sound is muffled, compressed, and lacking in higher pitches, which can make understanding words coming out of it difficult. Even though some gifted dogs can recognize commands they have been trained on when they are spoken by someone with a different accent from their owner, and even though the Hungarian researchers also tested seven such gifted dogs, recognition of the words dropped by 70% when using a button compared to a live command. With a live command, the gifted dogs would almost always perform the required action; with the button, it was more like one out of three or one out of four times, depending on the dog. “Breakthrough or Illusion?”

The researchers whose work Jarry discusses are probably genuinely driven by a desire to find out what dogs can do with language. But, realistically, any results that could be interpreted as favorable would be immensely profitable too, given the amount of money that many people are willing to spend on their pet companions. There is certainly room for a commercially viable amount of exaggeration, even if it is not coming from the researchers themselves.

But there is a bigger issue here

None of these animals, however long they are trained, graduate to anywhere near the point of using language like a four-year-old child who has no training other than being raised by humans who speak:

Bobby: When Grampa comes, he’s gonna take me to the ZOO! He says they got lions and tigers there. I bet Grampa could fight a tiger! ROARRR!!! (runs around acting like a lion)

And Bobby has not gotten anywhere near the age of reason…

His seven-year-old sister Millie, who has reached the age of reason, sharply retorts, “Don’t be silly! Grampa could get killed that way!”

These are not conversations that it is at all likely dogs have.

The fundamental problem is that most human communication takes place at an intellectual level that dogs don’t — and can’t — access because they don’t do abstractions. Even simple questions like “What was the score last night?”, “What’s the weather forecast tomorrow?”, or “What did the doctor say about Mave’s arthritis?” are not under consideration here.

Imagining a treasure trove of insight from teaching dogs to use buttons — that is not available just from paying closer attention to the dogs — is surely a loss for both parties.