Overview
- In experiments with 10 Gifted Word Learner dogs, seven learned new toy names in both direct-address and overheard settings.
- Accuracy differed by context, with about 80% correct choices when taught directly and 100% when learning by overhearing.
- Each label was heard for roughly eight minutes across short sessions, then learning was tested by fetching named toys from a separate room to prevent owner cueing.
- A temporal-discontinuity test, with names spoken only after toys were hidden, still produced learning, and the new mappings were retained two weeks later.
- Comparable tests with typical dogs, including non-gifted Border Collies, showed no evidence of label learning, and the authors frame the effect as shared socio-cognitive learning capacities rather than language equivalence.