Dogs, like humans, may favor consonants over vowels when parsing continuous speech, according to a new study comparing brain activity in the two species. The findings suggest that similar ways of processing speech can emerge in humans and dogs without requiring human-like language abilities. Across many human languages, consonants generally provide more useful information for identifying words than vowel sounds. Human infants increasingly rely on consonants to segment continuous speech by around one year of age, shifting away from their initial greater reliance on acoustically prominent vowels. However, it is unknown if this consonant bias is specific to humans and develops as a result of language acquisition, or if it could also emerge in other species exposed to human speech. According to Boglárka Morvai and colleagues, dogs provide a useful model to investigate these questions because, unlike laboratory animals, companion dogs are spoken to daily. They can detect changes in individual sounds and extract words from continuous human speech. Moreover, dogs differ in how much human speech they encounter early in life, making it possible to examine whether early exposure to speech shapes the development of a consonant bias.
In this study, Morvai et al. directly compared electroencephalogram (EEG) responses in the brains of 20 adult humans and 20 companion dogs as they listened to continuous streams of three-syllable nonsense words. The words followed patterns based either on consonants or vowels, while a third set had no consistent word patterns. None of the sounds clearly marked where one word ended and another began. Although the vowels carried more acoustic energy and were therefore more perceptually prominent, the design allowed the authors to test whether humans and dogs preferentially tracked consonant-based word patterns. The authors found that dogs, like humans, tend to rely more on consonants than vowels when identifying word boundaries in speech. Because the vowels in the experiment were acoustically stronger, the preference for consonants cannot simply be attributed to hearing the louder sounds. Instead, consonants may be more useful for detecting patterns that help break continuous speech into words. The bias was also found in dogs with both early and later exposure to human speech, suggesting that extensive early exposure is not necessary and that statistical learning – the ability to recognize recurring patterns – may be enough to produce it. Morvai et al. caution that dogs' consonant bias may reflect either their history of domestication and exposure to human speech or a broader auditory or cognitive ability that predates domestication.
