Summary:
A comparative electroencephalogram (EEG) study reveals that companion dogs, like humans, rely primarily on consonants rather than vowels to segment continuous speech into distinct words. The discovery demonstrates that sophisticated auditory parsing and statistical learning can emerge independently of human language acquisition or formal linguistic competence.
Key Facts:
- Consonant Priority in Canine Brains: Despite vowel sounds carrying higher acoustic energy and loudness, EEG recordings showed that dogs track recurring patterns based on consonants to identify word boundaries in unbroken speech.
- Independent of Early Language Exposure: The consonant bias manifested in dogs regardless of whether they had extensive early exposure to human speech as puppies, suggesting general statistical learning mechanisms are sufficient to drive the preference.
- Shared Auditory Computation: The findings demonstrate that speech segmentation strategies previously thought to require dedicated human linguistic machinery can operate across non-human companion animals.
Source: AAAS
When humans speak, sounds flow together in a continuous acoustic stream without clean physical gaps between words. To decipher meaning, the listener’s brain must execute speech segmentation—identifying where one word ends and the next begins.
In human languages, consonants typically carry the heaviest informational load for identifying words, while vowels convey acoustic richness, tone, and emotional prosody. By approximately 12 months of age, human infants undergo a developmental transition, shifting away from their initial reliance on acoustically prominent vowels toward tracking consonant patterns.
Whether this “consonant bias” is unique to humans, developing as a specialized byproduct of language acquisition, or reflects a broader mammalian capacity remained unresolved.
Now, an EEG study led by researcher Boglárka Morvai and colleagues provides evidence that companion dogs process continuous speech using the very same consonant-driven strategy.
Measuring Brain Waves in Dogs and Humans
Because companion dogs live alongside humans, are spoken to daily, and can recognize spoken words, they represent a model system for probing how non-human brains navigate human language.
To compare neural auditory processing directly, Morvai and her team recorded EEG activity from 20 adult humans and 20 companion dogs. While hooked up to non-invasive scalp electrodes, participants of both species listened to continuous audio streams composed of three-syllable nonsense words.
The experiment was calibrated across three distinct auditory conditions:
- Consonant-Structured Streams: Word boundaries were defined by statistical regularities among consonants.
- Vowel-Structured Streams: Word boundaries were defined by statistical regularities among vowels.
- Random Streams: Syllables occurred without predictive transitional patterns.
Crucially, vowels within the synthesized streams were louder and packed with greater acoustic energy than the consonants. If either species relied purely on acoustic salience, their brains would naturally track the vowel transitions.
Statistical Learning Over Acoustic Loudness
The neural data showed that both humans and dogs preferentially tracked word patterns organized by consonants rather than vowels.
Because the vowels were louder, the canine brain’s consonant preference cannot be dismissed as a simple auditory volume reflex. Instead, dog brains actively leverage consonants as reliable markers to detect transitional probabilities and segment fluid speech into discrete lexical units.
Furthermore, the researchers compared dogs with varying levels of early-life exposure to human speech. The consonant bias was equally robust in dogs with limited early exposure as in those raised with heavy verbal interaction from puppyhood. This suggests that extensive early exposure to language is not a strict prerequisite; rather, general statistical learning, the fundamental ability to track recurring statistical transitions across sounds, is sufficient to generate the bias.
Domestication or Ancient Auditory Heritage?
The authors emphasize that dogs’ reliance on consonants could stem from two evolutionary routes: tens of thousands of years of selective pressure during domestication alongside human vocal communication, or an ancient, general mammalian auditory mechanism that predates domestication entirely.
Regardless of its evolutionary origin, the study confirms that word-segmentation biases long viewed as hallmarks of human linguistic development operate naturally in man’s best friend.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About This Auditory Neuroscience & Canine Cognition Research News
- Media Contact: Science Press Package Team
- Source: AAAS
- Image Credit: Image credited to Neuroscience News
- Original Research is Open Access: Science (September 24, 2026). “Neural evidence that dogs segment the speech they hear with a humanlike consonant bias.” Authors: Boglárka Morvai, Kinga G. Tóth, Marianna Boros, Dorottya S. Rácz, Ivaylo Iotchev, Kitti Szabó, and Attila Andics.
- DOI: 10.1126/science.adw7709
Abstract
Neural evidence that dogs segment the speech they hear with a humanlike consonant bias
Across many human languages, consonants carry more lexical information than vowels. During speech segmentation, humans, unlike nonhuman primates, rely more on consonant than vowel patterns, despite vowels’ greater acoustic saliency.
To investigate whether this consonant bias is specific to humans or could also emerge in other species exposed to human speech, we performed noninvasive electroencephalography in humans and dogs. Intertrial coherence measures of neural entrainment and event-related potentials revealed enhanced word-level tracking for consonant- compared with vowel-structured streams in both species.
This suggests that dogs, similar to humans, exhibit consonant bias when segmenting the speech they hear. Linguistic regularity–based processing biases can thus also emerge in a nonspeaking species exposed to speech.
