Ask a dog owner whether their dog understands them and the answer is an instant yes. Ask a scientist and the answer is more careful: it depends on what you mean by “understand.” Over the past two decades, researchers have tested dogs with word-learning experiments, brain scans and controlled behavior trials. The picture they have built is more impressive, and more limited, than most people expect.
Dogs can learn the names of hundreds of objects, and their brains handle the words you say separately from the tone you say them in. What nobody has shown so far is that dogs process grammar the way people do. One caution up front: much of this evidence comes from single dogs or from groups of a few dozen at most, so treat the findings as promising, not settled.
This post walks through what the research says, where the evidence is still thin, and what it means for the way you talk to your own dog. Each study is linked the first time it comes up in the text.
How Scientists Study Language in Dogs
A dog cannot tell a researcher what it understands, so scientists work from two things: what dogs do, and what their brains do.
Behavior experiments. The classic design asks a dog to fetch an object by name from a pile of toys. To rule out accidental cues, the owner often asks from another room, as in Dror and colleagues’ toy-name study. Researchers also test the same word in different tasks. In Pilley and Reid’s work with Chaser, a border collie, object names were randomly combined with different commands. Chaser responded correctly even on the first trial, with no special training, which suggests she tied each name to an object rather than to one trained action.
Brain imaging. Some labs train dogs to lie still in an fMRI scanner so they can watch which brain regions respond to different sounds. Andics and colleagues used this approach to learn that dogs handle word meaning and tone of voice in different parts of the brain.
The core problem is separating comprehension from conditioning. A dog that sits when it hears “sit” may simply have learned that a sound leads to an action and then a treat. A dog that understands a word as a label for something should be able to use it flexibly: find the right object in a new setting, pick out a never-before-heard word’s referent, or remember a name weeks later. Most of the studies below were built to test exactly that.
What Can Dogs Understand?
The research points to four things dogs do well: link words to objects, prefer their own names, read tone of voice, and weigh gestures against what they already know. Each has limits, covered below.
Words Can Refer to Things
The strongest evidence that dogs treat words as labels comes from two border collies. Rico knew the labels of more than 200 items. In Kaminski’s 2004 experiment, researchers placed a new object among familiar ones and asked for it using a word he had never heard. Rico inferred that the new word belonged to the new object, retrieving it right away and again four weeks later. The authors argued that this kind of “fast mapping,” which children use to pick up new words from a single exposure, may rely on general learning and memory mechanisms found in other animals rather than something special to humans. Later researchers questioned whether that reflects true fast mapping, arguing that other strategies could explain his success on the retention test.
Chaser went further. According to Pilley and Reid’s 2011 report, her owners spent about three years teaching her the names of 1,022 objects, and they stopped because of time, not because she hit a limit. She also learned category words such as “toy,” “ball” and “Frisbee,” and could pair up to three labels with one object. A separate study of a mixed-breed dog named Sofia, who was trained over about 22 months, found that she could handle two-part requests, such as “ball fetch” and “fetch ball,” in either order.
These dogs are exceptional, and they were trained intensively. They show what a dog’s mind can do under the right conditions, not what the average pet knows. Even so, Chaser’s results suggest that a dog can link a sound to a thing, not just to an action.
Knowing Names
Your dog’s response to its name is a good example of the gap between recognizing a sound and understanding a label. In a University of Maryland study of how dogs respond to their own names, dogs preferred their own name to a different word with the same stress pattern, even when a stranger said it. They kept preferring it with many people talking in the background, down to the point where the name was no louder than the chatter. That beat 1-year-old infants, who lose their preference for their own name at that level. Dogs that work with people, such as service, search-and-rescue and explosives-detection dogs, did better than companion dogs.
The study did not show that a dog understands “Max” as a word meaning “me.” What it shows is that dogs reliably prefer the sound of their own name, even in a stranger’s voice and in noise. The authors suggest the working dogs’ edge may reflect more training and better attention, not a special gift for names.
Listening to Tone and Emotion
Dogs process what you say and how you say it in different parts of the brain, and they combine the two. In an fMRI study of dogs, one side of the brain showed a bias for processing meaningful words regardless of tone, and a separate auditory region distinguished words said with praising intonation from words said flatly. The brain’s reward areas became more active only when both the words and the tone signaled praise. Praise words said flatly, or neutral words said warmly, did not do it.
That finding complicates a common assumption. It is not true that dogs respond only to tone and ignore words, and it is not true that they follow words and ignore tone. They use both, and “good dog” works best when it sounds like it means it.
The same pattern shows up in behavior. In an experiment using naturalistic speech, adult dogs paid more attention to a speaker using dog-directed speech than to one using adult-directed speech, and showed more affectionate behavior toward them. When the tone and the content were mismatched, the dogs showed no significant preference either way. The authors suggest it may be the combination of dog-directed tone and dog-related words that matters. An earlier study had found that puppies, but not adult dogs, preferred dog-directed speech. The “dog voice” many owners use naturally may be doing real work.
Reading Gestures
In daily life, speech rarely arrives alone. You say “walk,” pick up the leash and head for the door, and your dog gets the message from all three. Dogs are also not simply obeying gestures. In one set of pointing experiments, dogs ignored a person pointing at an empty cup when they already knew the cup was empty, and they ignored misleading pointing from both children and adults. The authors concluded that dogs do not treat pointing as a command, and it remains unclear whether they read it as helpful information or something else.
In practice, that suggests a word, a gesture and a familiar routine probably reinforce each other, and a mismatch between them may leave a dog unsure what you want. The pointing studies did not test that directly, so treat it as a reasonable inference.
Following Pointing and Gaze
Following a point is one of the better-studied dog skills, and where it comes from is still debated. In a 2002 Science study, dogs beat great apes at using human signals to find hidden food, and dog puppies only a few weeks old did too, including puppies with little human contact. Wolves raised by humans did not show the skill, which led the authors to argue that domestication shaped dogs’ social thinking.
Later work fills in the picture. Riedel and colleagues tested puppies aged 6 to 24 weeks and found that all ages used pointing to find hidden food, and that six-week-old puppies still managed when the cups were placed beside them, so they could not simply follow the experimenter’s hand. In one version of the task adult dogs did not succeed, which the authors could not fully explain. They also noted that none of this shows puppies understand the intent behind a point.
The domestication story is not settled. A 2021 study in Current Biology tested 44 dog puppies and 37 wolf puppies and found the dog puppies were more attracted to people, better at reading human gestures and made more eye contact. A 2022 reanalysis of that data argued that once rearing conditions and willingness to approach a stranger are taken into account, dogs and wolves perform very similarly on pointing. Researchers are still sorting out how much of the difference comes from genes and how much from how the animals were raised.
Gaze matters too. In an eye-tracking study by Téglás and colleagues, dogs watched videos of a person turning toward one of two objects and followed the head turn more often when the person had first made eye contact and spoken to them, a pattern also seen in human infants. That suggests a pointing hand or a glance works best when you already have your dog’s attention, though that is an inference, since the study used videos rather than owners.
Common Questions About How Dogs Understand Us
Can Dogs Understand Sentences or Grammar?
Not in any way that has been shown. Dogs can handle short combinations of known words. Sofia, the mixed-breed dog mentioned above, responded to two-part requests that paired an object with an action, and Chaser responded correctly when object names and commands were randomly combined. Sofia did about as well in either word order and struggled with brand-new combinations, so this is memory for two items together, not proof that a dog parses word order or sentence structure. Her researchers described it, at most, as a possible forerunner of syntax.
There is little direct research on how dogs treat negation, so this part is an inference. Because a dog has learned “sit” as a signal, “don’t sit” probably still contains the word it knows best. That is one reason trainers favor short, positive cues, such as telling a dog what to do rather than what to avoid.
Are Words Just Linked to Rewards?
Partly, but that is not the whole story. Dogs are excellent associative learners, and a word paired with a treat will get a response. Several findings go beyond simple reward conditioning, though.
- Chaser responded correctly to object names she had learned separately from the commands, pairings that had never been trained together.
- Dogs in toy-name studies were asked to retrieve by name from another room, with no cues from the owner.
- In a study published in Science in January 2026, gifted word learner dogs, a rare group that picks up object names unusually fast, learned toy names from conversations they overheard, about as well as from being taught directly. The researchers tested 10 of these dogs, and ordinary family dogs showed no clear evidence of learning that way.
None of this means every word a dog knows works this way. Many everyday cues, like “sit,” probably do rest largely on learned associations.
Are Some Dogs Better Word Learners?
Yes, and the differences are large. Researchers use the term Gifted Word Learner for dogs that learn object names far faster than typical dogs. One study followed six border collies that each already knew more than 26 toy names. Their owners taught them six new names in a week and, a few weeks later, 12 more. The dogs averaged about 85 percent correct on a test at the end of each week. They still remembered many of the names one and two months later, though accuracy fell to roughly 60 percent (chance was 25 percent), and two dogs did no better than chance at two months.
These dogs are rare. The team that studied them found only six worldwide in 2021 and later tested 10. In the 2026 overhearing study, a comparison group of typical family dogs showed no clear evidence of learning names that way, and the authors caution against generalizing from these dogs to most pets. Why some dogs learn like this is still unclear. Breed, genetics, early experience and how much a person talks to a dog are all possible factors, but the research has not yet separated them.
What Happens in a Dog’s Brain?
Brain scans give us a different angle. All of these studies used awake dogs trained to lie still in a scanner.
- Words versus tone. Word meaning and intonation are processed in different regions, and reward areas respond when they match.
- Familiar versus unfamiliar words. In 12 dogs trained to retrieve two objects by name, never-heard pseudowords produced more activity in the parietotemporal cortex, on both sides of the brain, than trained words did. The authors link this to novelty detection.
- Familiar versus unfamiliar languages. In a study of 18 dogs, the secondary auditory cortex and another region called the precruciate gyrus responded differently to natural speech in a familiar versus an unfamiliar language. The difference was greater in older dogs, which suggests that language exposure plays a role. No region told the two languages apart once the speech was scrambled, which points to language-specific patterns in the sound rather than voice cues.
A caution applies to all of it. A brain scan shows that the brain is telling sounds apart. It does not prove the dog understands what they mean, so scans need to be paired with behavioral tests.
What About Talking Buttons?
Soundboards, which let a dog press a button that plays a recorded word, have become popular with owners. A 2024 study in PLOS ONE asked a narrower question than whether dogs can talk: do the words on the buttons mean anything to the dogs? In one experiment, 30 owner-trained pet dogs were tested with an unfamiliar researcher pressing the buttons. In a second, citizen-science experiment, 29 more dogs were tested at home with owners pressing buttons or speaking the words. The dogs behaved in ways that fit play-related and outside-related words, whether the owner or a stranger pressed the button and whether the word was pressed or spoken. A control nonsense word was included, and there was no conclusive evidence for food-related words.
The authors list several caveats. The dogs’ accuracy was modest, only three word concepts were tested, and owners could not be blinded in the home study, so unintentional cueing remains a concern. The button’s location was also not fully separated from its sound, so dogs may have learned the location. Several of the authors consult for or work for FluentPet, which makes soundboards. And the study did not test whether dogs use the buttons to express ideas, only whether they respond to the words.
Do Dogs Truly Understand What Humans Are Saying?
Dogs understand more than a trained reflex and far less than a person does. The studies above show real word learning in some dogs, attention to names, and a brain that separates meaning from tone. They do not show that dogs follow sentences the way we do.
A lot of what looks like language understanding in everyday life is probably context. Dogs notice that shoes mean you are leaving, that the leash means a walk, and that your voice is warm when something good is coming. Their noses pick up information we can’t, too, as we cover in things dogs can smell that humans cannot. That is intelligent, and it is also why it is hard to know how much of a specific word a specific dog has learned. If you are curious, try an informal test: say “walk” while you are sitting on the couch with no leash in sight, and see whether your dog reacts. It will not be a controlled experiment, but it is the same logic researchers use when they ask for a toy by name from another room. For a more structured look at how your own dog thinks, see how we got on when we put Harley and Charlie through the Dognition cognitive assessment.
Reducing everything a dog does to routine and cues would underrate them, though. Chaser learned category labels such as “toy,” and dogs in brain scans treated familiar words differently from nonsense ones. Both context and language seem to be at work, and the balance differs from dog to dog.
How to Talk to Your Dog, Based on the Research
Research cannot tell you how to train a particular dog, but it does point to some habits that line up with how dogs seem to process speech.
- Match your tone to your words. Dogs’ reward areas respond when praise words and a praising voice go together. Say “good dog” like you mean it.
- Use dog-directed speech for connection. Adult dogs paid more attention to a speaker who combined a dog-directed tone with dog-related words, and showed more affectionate behavior toward that speaker.
- Keep cues short and consistent. Dogs show no sign of parsing grammar, so one clear word for one action works better than a long phrase. Pick a word and stick with it, and tell your dog what to do instead of what to avoid. If you are starting from scratch, our obedience training guide and our lesson on teaching the place command cover the basics.
- Use your dog’s name to get attention. Dogs reliably prefer their own name, even in a stranger’s voice and with noise in the background. Save it for getting your dog’s attention rather than using it as a catch-all.
- Teach words through play and objects. Studies that taught toy names did it in everyday play sessions. Say the name each time you hand over a toy, then ask for it by name later. Our post on teaching your dog to pick up or take an object is a good first step.
- Practice in different places. Because dogs lean on context, a word may be solid in the kitchen and shaky at the park. Repeat it in new rooms and with new distractions before assuming a dog knows it.
- Pair a word with a gesture, and keep them in agreement. Dogs weigh gestures against what they already know, so a hand cue that contradicts your words may confuse them. That second part is a practical inference, since the pointing studies did not test mixed signals.
Conclusion
Dogs understand some of what we say, though how much depends on the dog and the word. Some dogs can link hundreds of words to objects, dogs pay attention to their names and respond to how you sound, and they are good at reading the room. No study to date has shown that dogs follow grammar or can take in a complex sentence.
The question worth asking is less “how much does my dog understand?” and more “which of my signals is my dog actually using?” Dogs have lived alongside people for thousands of years, and they have become very good at reading us. The best way to build on that is to be clear, consistent and warm, and to let your dog show you what it knows.
