Verified by 
Not all dog growls are created equal. Most commonly growls are aggressive signals, however even here there are different meanings. One growl can mean, "Back off! This is mine," as in when a dog is guarding food or some other property. Another growl might mean "Trouble is coming. I don't like this." And then of course there are the growls which occur when a dog is playing with intensity, as in a tug-of-war or play fighting with another dog. Can the average person tell the difference?
There has been a lot of research on how well dogs can interpret human speech and communication (click here or click here for examples). Since communication is a two-way street it is strange to note that research on how well humans can understand the sounds that dogs make has been much less common, and that research has tended to focus more on the interpretation of barking patterns (click here for more on that). However a new study from the Department of Ethology at Eötvös Loránd University in Budapest, Hungary is beginning to shed some light on the question of whether people can understand specific growling sounds made by dogs.
The research team was headed by Tamás Faragó. For the purpose of this study, samples of dogs growling were obtained in three different natural situations. Play growls were recorded from dogs who were playing tug-of-war with their owner. Guarding growls were obtained in a situation where dogs were given some food in a bowl and then another dog approached. These growls are certainly telling the other dog to stay away. The third type of growl was in response to a threat. It was triggered when the dog was approached by a human stranger in a menacing manner. A large sample of these growls were collected, coming from dogs of various breeds and sizes.
The first thing that these investigators did was that they measured the acoustic nature of each of the growls to see if they could find differences. Obviously if all of the growls had identical sound patterns, regardless of the situation, then it would be senseless to ask if people could distinguish among them. They did find differences. For example, strings of playful growls are built up from short, quickly repeated growls, while the more aggressive growls were elongated and drawn out. In addition the food guarding growls differed from the threatening growls in what is called their formant dispersion. This feature affects the apparent pitch and loudness of the sound that we hear. To the human ear the guarding growl will sound as if it has a lower pitch than the growl from the threatened dog. A lower apparent pitch and a more intense sound gives the impression that the growl is coming from a larger dog.
Data was collected from 40 people who listened to various growl sound samples. They were asked to judge the emotional content of each growl using five different scales which measured the apparent amount of aggression, fear, despair, happiness, and playfulness that they perceived. They also had to decide if they could determine which of the three possible contexts (food guarding, threatening, or play) the sound came from.
Basically these researchers found that humans were pretty good at identifying the growl types. Their overall accuracy was around 63% (which is well above the chance level of 33%). The growls that were easiest to identify were the play growls, which were correctly identified 81% of the time. People were less accurate when it came to food guarding (60% correct) and threatening growls (50%). That makes a certain degree of sense since these last two obviously share an aggressive component and both were meant to ward off someone (a competitor for their food or a person who might be a threat to their safety). So it makes sense that they might share some of the same sound qualities and thus be a bit more confusable.
When we look at the emotional ratings for each of the growl types we find that the play growls rated highest in both playfulness and happiness. The food guarding growl rated highest in aggression (remember that these growls were recorded when a canine competitor was approaching their food). The threatening stranger growl produced high ratings for aggression but also a high rating for fear. This makes sense because the approaching stranger might pose a physical danger to the dog. It appears that the listeners were responding to the somewhat higher pitch in this growl and correctly reading it as a trace of fear in the dog.
The accuracy of judging these growls was affected by who the listener was. Dog owners were much better than other people at correctly identify a growl's meaning. This suggests that experience listening to dogs improves our recognition ability and also suggests that we can train people to better understand these communication signals from our canine companions.
Women were also better at distinguishing the growls than were men. This is actually a common pattern in many emotion recognition studies, and it has been suggested that women may be more empathic and sensitive to others' emotions than men.
So the results of this study seem to say that humans interpret long, low sounding, loud, growls as being the most aggressive. Any trace of a higher pitch entering into the growl will be interpreted as traces of fear. Shorter growls, especially if they occur in a sequence, and do not have a sustained low pitch are more likely to be interpreted as nonthreatening and even playful.
Stanley Coren is the author of many books including: Gods, Ghosts and Black Dogs; The Wisdom of Dogs; Do Dogs Dream? Born to Bark; The Modern Dog; Why Do Dogs Have Wet Noses? The Pawprints of History; How Dogs Think; How To Speak Dog; Why We Love the Dogs We Do; What Do Dogs Know? The Intelligence of Dogs; Why Does My Dog Act That Way? Understanding Dogs for Dummies; Sleep Thieves; The Left-hander Syndrome
Copyright SC Psychological Enterprises Ltd. May not be reprinted or reposted without permission
References
Faragó T, Takács N, Miklósi Á, Pongrácz P. (2017). Dog growls express various contextual and affective content for human listeners. Royal Society Open Science, 4: 170134. http://dx.doi.org/10.1098/rsos.170134
