Overview
- A peer-reviewed paper published in Science in late September 2026 reports that pet dogs and adult humans both show a consonant-based bias when extracting word-like units from continuous, meaningless speech.
- Researchers measured brain activity with EEG in 20 adult humans and 20 pet dogs who listened to continuous streams of three-syllable pseudowords arranged by consonant patterns, vowel patterns, or random order.
- Both species showed stronger neural tracking of word rhythms when the sequences were organized by consonants, with differences emerging about 0.5 seconds after word onset and humans showing stronger tracking of individual syllables.
- The consonant preference appeared even in dogs without documented rich early speech exposure, but the authors stress that group sizes and incomplete life-history data make those subgroup findings tentative.
- Authors caution that the result supports the idea that exposure-driven statistical learning can build some speech-segmentation tools, yet questions remain about causation, the role of domestication, and how the effect varies by breed, age, and language environment.