In the photo above: Jeff’s dogs Ozzie, Huck and Shiloh. Photo: Jeffrey Reed.
Ethologists have been naming dog sounds for a century, and the names are useful, but they don’t add up to a complete dictionary, because dogs don’t speak in neatly boxed words. It’s even harder for wild wolves. What we do have is something better than folklore and worse than a grammar: a repertoire. A set of sound families (howls, barks, growls, whines), each with variations that shift with distance, urgency, audience and mood. Scientists can measure those shifts. In some cases they can show that listeners (dogs, humans, mothers with pups) respond to them in reliable ways. That’s the ground truth: not poetry, not projection, but repeatable patterns in air.
What all of this adds up to is not a dictionary but a discipline: listen for families, not magic words. Start with the broad type (howl, bark, growl, whine), then attend to pitch, harshness, rhythm and the scene around the sound. Meaning, in dogs as in humans, is use. And use is a marriage of throat, body and circumstance.
An inventory of dog sound units
Evidence-based, grouped by function.
| Sound | Acoustics | Functions | Sources |
|---|---|---|---|
| Howl (solo howling) | Sustained tonal call; strong harmonic structure; long duration | Long-range contact and cohesion; can carry individual information; domestication and breed ancestry modulate responses to howls | Lehoczky et al., 2023; Cohen & Fox, 1976; Yeon, 2007 |
| Chorus howling (group howling) | Overlapping howls; temporally structured group bout | Group advertisement and cohesion; “joining” dynamics (in dogs, usually multi-dog households or kennels; experimentally much less mapped than in wolves) | Cohen & Fox, 1976; Yeon, 2007 |
| Bark-howl or “baying” (hound-type long bark) | Bark elements chained into longer, more tonal sequences; often rhythmic | Human-selected “working broadcast” during pursuit or tracking; can also alert or recruit attention | Feddersen-Petersen, 2000; Lord et al., 2009; Dorey et al., 2021 |
| Alarm or alert bark (“woof,” single bark) | Often lower F0, noisier; relatively isolated barks | Attention recruitment; alerts other dogs and humans; can work as a mobbing or harassment signal | Lord et al., 2009; Yin, 2002; Pongrácz et al., 2006 |
| Stranger or territorial bark (“guard bark,” “fence bark”) | Lower-pitched; shorter inter-bark intervals; lower HNR in aggressive contexts | Territorial defense and repulsion; communicates arousal and aggression; dogs and humans can identify this context above chance | Yin & McCowan, 2004; Pongrácz et al., 2005; Molnár et al., 2009 |
| Attack or aggressive bark (high-arousal aggression bark) | Low pitch plus harsh, noisy components; rapid cadence | Escalation within agonistic sequences; signals high arousal and aggression (human listeners can predict the context from acoustics) | Pongrácz et al., 2006; Morton, 1977 |
| Isolation or separation bark (“left alone” bark) | Typically higher F0 than the stranger bark; longer inter-bark intervals; more tonal and less harsh than the aggressive bark | Contact solicitation; listeners attribute distress or despair; dogs tell this bark apart from the stranger bark | Yin & McCowan, 2004; Molnár et al., 2009; Pongrácz et al., 2006 |
| Anticipation bark (“pre-walk” bark) | Often higher pitch and more tonal than threat barks | Positive arousal and anticipation; human listeners map acoustics to motivational state above chance | Pongrácz et al., 2006; Morton, 1977 |
| Solicitation bark (“ball bark,” “attention bark”) | Variable; often rhythmic bouts; moderate pitch | Requests interaction or an object; functionally closer to human-directed communication than to spacing or repulsion | Pongrácz et al., 2006; Yeon, 2007 |
| Play bark | Often higher pitch; can be short and repeated; mixed with play pants and growls | Maintains play; arousal without threat (anchored by play behavior) | Yeon, 2007; Yin & McCowan, 2004 |
| Threat or agonistic growl | Low F0; stable formants can encode body size; often harsh | Warning and repulsion; assessment signal; encodes affective and sometimes “referential” content (resource context) | Faragó et al., 2010; Faragó et al., 2017 |
| Food-guarding growl (“the bone is mine”) | Low F0; formant structure salient; can exaggerate apparent size | Resource defense; size projection and motivation; receivers integrate growl acoustics in cross-modal tasks | Faragó et al., 2010; Faragó et al., 2017 |
| Play growl | Low-frequency like threat growls, but used within play | Play maintenance; “virtual size” effects: dogs can sound larger than they are without implying danger | Bálint et al., 2013; Faragó et al., 2017 |
| Puppy whine (distress or need call) | High F0; tonal; may include nonlinear phenomena (NLP) | Triggers maternal care; encodes litter and individual identity and condition cues (e.g., body weight through F0 within litter ranges) | Massenet et al., 2024; Massenet et al., 2022 |
| Adult whine (general) | High-pitched tonal call; shorter than howls; often repeated | Appeasement and solicitation; contact-seeking; flexible function depending on context and receiver | Yeon, 2007; Cohen & Fox, 1976 |
| Begging or solicitation whine (“food whine”) | Tonal; can be modulated (pitch contours) | Asks humans for resources; acoustic structure can bias perceived need or urgency (evidence strongest in controlled puppy data; adult contexts less standardized) | Sibiryakova et al., 2021; Yeon, 2007 |
| Whimper | Very soft, high-pitched, often brief | Submission, appeasement or low-grade distress; often treated as a low-intensity form of the whine | Cohen & Fox, 1976; Yeon, 2007 |
| Yelp (pain or distress yelp) | Sudden, high F0, short | Acute pain or startle; distress broadcast; usually diagnostic in context (but less standardized experimentally than bark, growl and whine work) | Cohen & Fox, 1976; Yeon, 2007 |
| Yip (sharp, high-pitched burst) | Very brief, high F0 | Startle or distress; can occur in play accidents or mild pain; classified as a distinct type in canid repertoires | Cohen & Fox, 1976 |
| Scream or squeal | Extended high pitch, often noisy or rough | Extreme distress or fear; classified in canid repertoires, but little dog-specific experimental mapping | Cohen & Fox, 1976; Yeon, 2007 |
| Play pant or “dog laugh” (breathy forced exhalation) | Breath-like, non-tonal exhalations; acoustically distinct from cooling pants in some datasets | Play initiation and maintenance; playback can reduce stress-related behavior in shelters; matches “distinctive play pant” findings | Volsche et al., 2023; Simonet et al., 2005 |
| Grunt | Low, short, often pulsed | Low-intensity contact signal; can accompany greeting, handling or frustration; often treated as a quiet close-range canid call | Cohen & Fox, 1976; Feddersen-Petersen, 2000 |
| Mew | Quiet, high-pitched, brief | Classified in canid repertoires; in dogs mostly reported from observation (often juveniles) rather than experiments | Cohen & Fox, 1976 |
| Coo | Soft, low-intensity tonal call | Classified in canid repertoires; function in domestic dogs not well isolated experimentally | Cohen & Fox, 1976 |
| Tooth snap (jaw snap) | Impulsive broadband “clack” | Threat display; distance-increasing signal; often paired with growl or snarl displays | Cohen & Fox, 1976; Yeon, 2007 |
| Clicks (teeth or tongue clicks) | Short transient clicks | Classified as a distinct canid sound type; communicative function in dogs under-specified | Cohen & Fox, 1976 |
| Cough-like vocalization | Cough-shaped burst (not an illness cough) | Listed in canid repertoires as a vocal type; communicative role in dogs not well standardized in experiments | Cohen & Fox, 1976; Yeon, 2007 |
| Mixed vocalizations (bark-growl, growl-bark, bark-howl sequences) | Concatenations or partial overlaps of the units above | Often graded escalation or context blending; domestication may increase flexible recombination of bark-based sequences (a hypothesis with mixed evidence) | Cohen & Fox, 1976; Lord et al., 2009 |
Long-range voices: the calls meant to travel
Howls are the old architecture of canid communication, built for distance, built for nights when bodies are scattered across miles and vision is useless. In dogs, howling is uneven (some breeds do it readily, others barely at all), but when it happens it keeps the shape of its ancestry: a sustained, tonal signal with harmonics stacked like rungs. The best-supported functions are still the classic ones (contact, cohesion, advertisement), though in domestic dogs the social ecology is warped by houses, fences and human schedules. Still, when researchers play wolf howls to dogs, response patterns track genetic proximity to wolves: the closer the dog is to wolves, the more “wolfish” the reaction tends to be. The howl remembers where the species came from. (Lehoczky et al., 2023; Cohen & Fox, 1976)
Sometimes dogs howl together, what you could call chorus howling: a pile-up of voices that becomes, briefly, a social object in its own right. In wolves, chorus howling is a big deal; in dogs it’s less mapped, more anecdotal, more variable. It’s fair to describe it as the domestic echo of a wild system: the same instrument, played in a different room. But we should be honest that the rigorous experimental work here is thinner than it is for barks and growls. (Cohen & Fox, 1976; Yeon, 2007)
Then there is baying, the hound’s long bark that stretches toward a howl. It’s the sound of pursuit turned into broadcast: not simply “I see something,” but “I’m still on it; come find me.” Baying is where domestication shows its heavy hand. We didn’t just inherit a canid voice; we tuned it. In some working lines, humans selected for a call that can ride wind and terrain, a vocal beacon you can follow when the dog has vanished into timber. (Feddersen-Petersen, 2000)
Barks: short-range punctuation and social management
If the howl is architecture, the bark is punctuation: fast, flexible and often aimed at a nearby audience. One of the surprises of modern research is that barks are not random noise. They carry information about who is calling and about what kind of situation is unfolding. Humans can classify bark contexts above chance. Dogs can discriminate too. Not perfectly, not magically, but reliably enough that “bark” stops being a single word and starts looking like a family of related signals. (Yin & McCowan, 2004; Molnár et al., 2009)
A single alert bark, the old “woof,” sometimes low and blunt, often works like a tripwire. It recruits attention. It says: look. It doesn’t always specify what to look at, but it can prime the group (dogs, humans, whoever is in earshot) to shift from loose living to watchfulness. (Yin, 2002; Lord et al., 2009)
A stranger or territorial bark tends to be different: often lower, harsher, delivered in bouts that feel less like inquiry and more like pressure. The function is distance-increasing: it pushes. And it’s one of the cleanest cases where acoustics and context line up. The same dog can produce barks that listeners classify as “stranger at the door” versus “left alone” at rates above chance, which tells you the sound is not just arousal; it is structured. (Yin & McCowan, 2004; Pongrácz et al., 2006)
An isolation or separation bark, the one that erupts when the car leaves the driveway and the house becomes too quiet, often carries a different emotional signature. It’s less about repelling and more about pulling the world back. Contact seeking. Distress. A social tether thrown out into the air. Again, not a perfect “meaning,” but a reliable communicative orientation: come back, don’t leave me, where are you. (Yin & McCowan, 2004)
There are also anticipation barks, the pre-walk, pre-dinner, pre-ball explosion. They sit on the brighter edge of arousal. In experiments, humans do surprisingly well at mapping bark acoustics to the emotional content of the situation, as if the dog’s body is already speaking through the throat: higher pitch, different rhythm, less harshness when the world promises something good. (Pongrácz et al., 2006; Morton, 1977)
And there are play barks, which live inside the play frame, signals that help keep roughness from turning into a fight. Here we have to be careful: play is not a single sound, it’s a choreography of bodies and cues. But barks do show up as part of that choreography, and their job is to maintain the contract: this is still play. (Yeon, 2007)
Growls: negotiated boundaries, with body size hidden in the physics
A growl is a low-frequency truth: the vocal cords slow down, the tract resonates, and suddenly the dog sounds like a larger animal. That’s not metaphor; it’s acoustics. Formants, the resonant frequencies shaped by the vocal tract, can provide cues to body size, and listeners use them. That is why growls are so often part of assessment: a warning that tests whether the other party will back away without making the situation expensive. (Faragó et al., 2010)
The best-studied growl contexts are clean and almost painfully familiar: food-guarding growls over the bone, the bowl, the prized object. Here the function is not vague. It is resource defense. In controlled studies, the acoustic structure of these growls carries both affective content and contextual information. They are not simply “angry noises”; they are structured signals in a recurring ecological problem: who gets to keep what. (Faragó et al., 2010)
Then there are threat growls more broadly, agonistic warning signals that sit near the front edge of violence but often prevent it. Faragó and colleagues showed that humans can infer context from growls at above-chance levels, and that growls differ across situations in ways listeners are sensitive to. Again, the important point isn’t that people can translate dog speech. It’s that the signal contains recoverable information. (Faragó et al., 2017)
And then, almost comically, there are play growls, where dogs can sound large and fierce while remaining harmless. It’s a kind of vocal cosplay. In play, a growl can exaggerate size without implying actual threat, and other dogs seem to understand the frame. That fact alone is a warning against simplistic decoding: the same “word” can be used under different rules depending on the social game being played. (Bálint et al., 2013)
Whines and whimpers: the high road to care
If growls live low in the spectrum, whines live high: thin, tonal, often urgent. Their central function is solicitation: help me, come closer, attend to me. In adult dogs, whines are flexible and can mean many things, which is why they’re less satisfying to classify. But in puppies, the science gets sharp.
Puppy whines don’t just correlate with care; they can mediate maternal behavior. In recent work, researchers showed that specific features of puppy whines can influence how mothers respond. That’s not interpretation; that’s a causal relationship inside a social system that evolved under strong selection: keep the helpless alive. (Massenet et al., 2024)
Even more intriguing, puppy whines can include nonlinear phenomena (subharmonics, chaos), acoustic roughness that tends to increase perceived distress and urgency. Across mammals, nonlinearities often work as attention-grabbers in high-arousal calls, and puppy whines appear to follow that rule. The physics becomes emotion in the listener’s ear. (Massenet et al., 2022; Morton, 1977)
At the softer end, whimpers are the low-amplitude cousins of whines, small sounds that often accompany appeasement or low-grade distress. They’re common in real life and oddly under-measured, perhaps because science prefers signals that declare themselves loudly. (Cohen & Fox, 1976; Yeon, 2007)
Yelps, yips and screams: the involuntary edge
A yelp is a flare: sudden, high, sharp. It’s the sound you hear when a dog steps wrong, gets pinched in play, collides with a boundary. The primary function is broadcast distress, pain or surprise, often so diagnostic in context that experiments have not needed to do as much work to prove it. (Cohen & Fox, 1976)
A yip is briefer, still high, still sharp: sometimes startle, sometimes minor distress. And at the extreme there is the scream or squeal, longer and more chaotic, the kind of sound you hear rarely and remember too well. These are the signals at the cliff edge of welfare, where the body’s alarm system commandeers the voice. (Cohen & Fox, 1976; Yeon, 2007)
Breath, laughter and the sound of play as a social technology
Then there’s the odd one: the play pant, sometimes called “dog laughter.” It isn’t voiced like a bark or whine; it’s breath shaped into a social cue. In shelter settings, recordings of this sound have been used in ways that suggest it can influence behavior, reducing stress-related actions and nudging dogs toward play. More recent work supports the idea that play pants can be acoustically distinctive and tied to play contexts. It’s one of those moments where the line between physiology and communication gets thin: the dog’s breathing becomes part of the message. (Simonet et al., 2005; Volsche et al., 2023)
The small percussive signals: teeth and clicks
Finally, there are the non-laryngeal sounds, the tooth snap and the click, brief percussive events that often accompany threat displays or tight social negotiation. They’re reminders that not all communication needs the vocal cords. Sometimes a mouth makes its point with mechanics. The research here is older and thinner, but the categories persist across canid repertoires. (Cohen & Fox, 1976)
Curious how your own dog reacts to its wild relatives? Play it a few calls from the gray wolf dictionary.