Dog brains follow consonants to find words in unbroken speech, as human brains do

Set the two species side by side and the dogs' consonant-over-vowel margin, 42%, is larger than the humans' nearly 25% ^[5][6]. The difference comes from the vowels. If each percentage is a ratio on the same synchronisation measure, the dogs' response to vowel-patterned streams sat only about 9% above their response to random syllables. The humans' sat about 42% above random ^[1][2]. That matches Discover's account that human brains also tracked vowel-built words, less strongly than consonant-built ones, and dog brains did not ^[8]. Both species followed the syllable beat. Only the human brains also followed the faster rhythm of individual consonants and vowels ^[9].

On the narrow question of whether the consonant bias needs a language faculty, I think the evidence is strong. Dogs do not have language, though they live in a speech-rich environment, Andics told New Scientist ^[10]. The bias had not previously been found in primate species much more closely related to humans ^[11]. "This is the first evidence that a consonant bias can also emerge in the brain of a species other than humans," said Kinga G. Toth, the study's other co-lead author ^[25]. Andics said the findings suggest the bias "does not require a pre-wired brain and human-like language abilities" ^[12]. Nick Enfield of the University of Sydney told New Scientist the team convincingly shows the dogs are more sensitive to statistical patterns in consonants than in vowels ^[19].

Where the bias came from is a harder question, and the design is weaker there. All 20 dogs were companion animals with Hungarian owners ^[2]. The bias appeared even in dogs that seem to have spent their first three months on the street or in a shelter ^[14]. Neuroscience News takes that as a sign that general statistical learning is enough, with or without heavy early exposure ^[15]. Those dogs differed in puppyhood, but as adults every one of them lived with people who talk to them ^[2]. Andics goes further. "We argue that the only reason companion dogs have a consonant bias is that human speech, and specifically the characteristics of words, reshaped their brains," he told New Scientist ^[17]. The Neuroscience News account says the authors also allow a route through domestication ^[18]. The thing this doesn't tell you is whether a dog raised without speech would show the bias: the team expects it would not, but has not run the test, according to New Scientist ^[16].

The language setting is narrow as well. Enfield's caveat was that these are Hungarian dogs living with Hungarian speakers ^[20]. "There's 7000 different languages in the world and some of them rely a lot more on things like tonal distinctions," Enfield said ^[20]. Andics cited studies suggesting the human bias is weaker, or absent, in people who mostly listen to Mandarin or Danish ^[21]. Enfield also had a suggestion for owners: "The lesson here may be to give your dog a name with more consonants in it than one with more vowels" ^[26]. An earlier test using dogs' own names found no consonant bias. The authors say that searching for patterns in unfamiliar speech, as opposed to recognising a familiar name, may be why it showed up this time ^[22].