Dogs Use Consonants to Recognise Words, and What That Actually Means for How You Talk to Your Dog

Published September 29, 2026

Researchers at Eotvos Lorand University wired up dogs' brains and found something that, honestly, nobody expected: dogs process consonants the way human babies do, using them as the anchors that separate one word from the next.

The study, published in Science on September 26, 2026, measured brain activity in dogs as they listened to speech. The finding wasn't just that dogs can distinguish words, we already had a rough sense of that. It was that the mechanism they use mirrors early human language acquisition. Consonants act as the boundary markers. Vowels carry emotional tone. This is apparently quite unusual in mammals outside of humans.

Which raises an obvious question for anyone who owns a dog: have you been talking to yours all wrong?

What the Study Actually Found

The Eotvos Lorand team recorded neural responses while dogs listened to familiar and unfamiliar words. When consonants were swapped out, dogs registered them as different words. When vowels changed, the response was less distinct. This matches what developmental linguists have observed in infants around six to eight months old, before they can produce any words themselves.

So dogs aren't processing "sit" as one undifferentiated sound blob. They're identifying the s and the t as the load-bearing parts of that word.

That's a meaningful distinction. It suggests dog cognition around language is more structured than a simple association between sounds and outcomes. They're doing something closer to parsing.

Why This Changes How You Should Speak to Your Dog

Most dog owners intuitively speak to their dogs in a higher pitch with exaggerated vowels. The sweet baby voice. And that does have a function, the vowel emphasis probably carries emotional cues that dogs respond to. But the Eotvos Lorand research suggests that if you want your dog to reliably recognise a word, the consonants are the part that needs to be clear.

A few things follow from this.

Short commands with strong consonants tend to work better. "Sit", "stay", "down", "back" all have sharp starts or ends. This isn't coincidence, trainers have landed on these through decades of trial and error. The science now gives you a reason.

Mumbled commands probably do fail for linguistic reasons, not just volume. If you're saying "siiiit" with the consonants swallowed, you may literally be speaking a different word as far as your dog's brain is concerned.

Consistency in how you say a word matters more than researchers previously assumed. If you alternate between "lie down", "down", and "lay" depending on your mood, you're not just being inconsistent about which command you use. You're potentially presenting what a dog's brain interprets as three different words.

What This Means for Training New Commands

If you're working on teaching your dog something new, the consonant finding is worth keeping in mind.

Pick a command word where the consonants are distinct and not easily confused with something else your dog already knows. "Fetch" and "fresh" are going to register as different words, which is fine. "Down" and "round" are trickier, the d and r are distinct but in fast speech you can lose the difference. Worth thinking through before you settle on a word.

Also, the study's framing around word boundaries is interesting. Dogs use consonants partly to detect where one word ends and another begins. So if you're training a two-word cue, run it together. "Leave it" said quickly as "leavit" may actually be cleaner for a dog to parse than the deliberate two-word version where you pause between them. The pause potentially signals a boundary where there isn't one.

The Behaviour Piece That Often Gets Missed

A dog that seems to be ignoring a command is almost always being blamed for a focus or motivation problem. Sometimes that's right. But sometimes the word has just degraded in pronunciation over months of use, or you've unconsciously started replacing it with a near-synonym, and the dog genuinely isn't recognising it anymore.

This is something Tailo's episode tracking can help surface, because if you're logging training sessions and noting which commands are landing and which aren't, you start to see patterns that are easy to miss in the moment. A command that worked reliably six months ago and suddenly seems to need retraining might not be a behaviour regression at all.

One Thing Worth Trying This Week

Say your three most-used commands out loud, slowly, to yourself. Notice which parts you're actually pronouncing and which parts you're eliding. Then say them quickly, the way you actually say them to your dog mid-walk.

If the consonants are still there, you're probably fine. If "wait" has become "aaait" and "come" has become "uuhm", those words may need to be re-introduced almost from scratch, said cleanly and consistently until the dog has a new clean version to attach to the behaviour.

The Eotvos Lorand team didn't set out to give dog owners training tips. They were studying cognition. But the practical implication is harder to ignore than most academic findings: your dog is listening more carefully to specific parts of your words than you are.