Some dogs can learn new words by eavesdropping on their owners
A new study reveals that dogs that have a unique talent for learning a vocabulary of toy names, can learn new names by overhearing their owners’ conversations
image:
Bryn, an 11-year-old male Border Collie from the UK, that knows the names of about 100 toys.
view moreCredit: Photo by: Helen Morgan
“Honey, will you take Luna to the P-A-R-K?” both parents and dog owners know that some words should not be spoken, but only spelled, to prevent small ears from eavesdropping on the conversation. At the age of 1.5 years toddlers can already learn new words by overhearing other people. Now, a groundbreaking study published in Science reveals that a special group of dogs are also able to learn names for objects by overhearing their owners’ interactions.
Similarly to 1.5 -year-old toddlers, that are equally good in learning from overheard speech and from direct interactions, these gifted dogs also excel in
learning from both situations
What makes this discovery remarkable
Although dogs excel at learning actions like “sit” or “down”, only a very small group of dogs have shown the ability to learn object names. These Gifted Word Learner (GWL) dogs can quickly learn hundreds of toy names through natural play sessions with their owners. Toddlers can easily learn new words through a variety of different processes. One of these processes is learning from overheard speech, where child learns new words by passively listening to interactions between adults. To do this, children must monitor the speakers’ gaze and attention, detect communicative cues, and extract the target words from a continuous stream of speech. Until now, it was unknown whether GWL dogs could also learn new object labels when not directly addressed.
“Our findings show that the socio-cognitive processes enabling word learning from overheard speech are not uniquely human,” says lead scientist Dr. Shany Dror, from ELTE and VetMedUni universities. “Under the right conditions, some dogs present behaviors strikingly similar to those of young children.”
How the researchers found that Gifted dogs learn toy names by eavesdropping
In the first experiment, the research team tested 10 Gifted dogs in two situations:
1. Addressed condition: Owners introduced two new toys and repeatedly labeled them while interacting directly with the dog.
2. Overheard condition: The dogs passively watched as their owners talked to another person about the toys, without addressing the dog at all.
Overall, in each condition, the dogs heard the name of each new toy for a total of only eight minutes, distributed across several brief exposure sessions. To test whether the dogs had learned the new labels, the toys were placed in a different room, and the owners asked the dogs to retrieve each toy by name (e.g., “Can you bring Teddy?”)
The result: In both conditions, seven out of ten dogs learned the new labels
The dog’s performance was very accurate already at the first trials of the test, with 80% correct choices in the addressed condition and 100% in the overhearing condition. Overall, the Gifted dogs performed just as well when learning from overheard speech, as when they were directly
taught, mirroring findings from infant studies.
But that’s not all: Gifted dogs overcome one of the key challenges in learning labels
In a second experiment, the researchers introduced a new challenge: owners first showed the dogs the toys and then placed them inside a bucket, naming the toys only when they were out of the dogs’ sight. This created a temporal separation between seeing the object and hearing its name. Despite this discontinuity, most of the Gifted dogs successfully learned the new labels.
“These findings suggest that GWL dogs can flexibly use a variety of different mechanisms to learn new object labels” says senior scientist Dr. Claudia Fugazza, from ELTE University in Budapest.
What we can learn from this study
The study suggests that the ability to learn from overheard speech may rely on general socio- cognitive mechanisms shared across species, rather than being uniquely tied to human language.
However, Gifted Word Learners are extremely rare, and their remarkable abilities likely reflect a combination of individual predispositions and unique life experiences.
“These dogs provide an exceptional model for exploring some of the cognitive abilities that enabled humans to develop language” says Dr. Shany Dror “But we do not suggest that all dogs learn in this way - far from it.”
Is your dog a Gifted Word Learner dog?
This research is part of the Genius Dog Challenge research project which aims to understand the unique talent that Gifted Word Learner dogs have. The researchers encourage dog owners who believe their dogs know multiple toy names, to contact them by email (geniusdogchallenge.offcial@gmail.com), Facebook or Instagram.
Journal
Science
Article Title
Dogs with a large vocabulary of object labels learn new labels by overhearing like 1.5-year-old infants
Article Publication Date
8-Jan-2026
Mental trauma succeeds 1 in 7 dog related injuries, claims data suggest
Time off work and/or loss of earnings reported in over half of cases; Explore mandatory lead use in certain public spaces to boost public safety, urge researchers
Mental trauma, including specific phobias and post traumatic stress disorder (PTSD), succeeds 1 in 7 dog related injuries, while over half of cases result in the need for time off work and/or loss of earnings, suggests a study of personal injury claims data for England and Wales, published online in the journal Injury Prevention.
As most of these claims involved unrestrained dogs in non-residential locations, mandating lead use on highways and in public spaces should now be explored to boost public safety, urge the researchers.
The latest estimates for 2024 indicate that there are 13.6 million dogs in the UK. Hospital admissions for dog related injuries have risen from 4.76/100,000 people in 1998 to 18.7 in 2023 in England. In Wales, they rose from 16.3/100,000 people in 2014 to 23.7 in 2022, note the researchers.
The information on these injuries isn’t as comprehensive as it needs to be, suggest the researchers. Currently, hospital records don’t differentiate between dog bites and dog strikes (any dog-related injury not caused by a bite); not everyone who needs hospital treatment will be admitted to hospital; and there’s little or no information on the longer term effects.
While civil claims data are more informative, these claims are only pursued when there are sufficient assets to cover damages and legal costs, point out the researchers.
Most solicitors’ firms, however, collect initial contextual and impact data before making a judgement as to whether the case can be taken on. If these initial data were routinely collated and analysed, this would remove the inherent socioeconomic bias of analysing court records, they explain.
To find out if this might be a viable option, and how informative these data would be, the researchers analysed anonymised civil claims enquiry data from 1 January 2017 to 31 March 2024, provided by a specialist law firm operating throughout England and Wales.
The enquiry data contained information about the injured person, including their age and sex; date, location/land use, and context of the incident; breed and level of restraint of the dog; and consequences including physical injuries, mental trauma, medical treatment, time off work and loss of earnings.
Data were categorised into incidents involving dog bites and those involving strikes or other types of behaviour.
Analysis of the data showed that 816 dog-related incidents, comprising 842 individual claims, were recorded between January 2017 and March 2024.
Most (94%) incidents occurred in England and primarily involved dog bites (just over 91%); 7% were dog strikes.
Half of dog bite victims were male (53%), while dog strike victims were mainly female (70%). Women were more than twice as likely to be involved in a non-bite incident as men. Most of the injured didn’t know the dog involved (80%).
The three most common locations for dog bites were in front of a private residential property (just over 34%), on a highway or pavement (18%), and inside a private residential property (11%).
Almost half of non-bite incidents occurred in public spaces (49%), the most frequent of which were outdoor recreational areas, such as parks and nature reserves (34%), highways or pavements (23%), and ‘forestry, open land and water’ (11.5%).
Delivery workers made up over 1 in 4 (28%) of those bitten, most often during a delivery to a private residential property, when an unrestrained dog came out of the front door (12%). Other situations involved walking, exercising, playing in public without a dog (11.5%); and walking with one’s own dog (11%).
The most common circumstances for non-bite incidents were either with a dog when out walking, exercising, or playing in public (34%) or without (27%); and when a dog escaped from a private property (10%).
In both types of incident, most of the dogs weren’t restrained at the time: 79% of biting dogs; and 86% of non-bite incidents. Most dogs involved in a bite (69%) or non-bite (77.5%) incident were reported to be with their owner.
Almost all (98%) of bite, and 78% of non-bite, incidents ended up with a physical injury. Fractures comprised nearly 4% of injuries, while tissue loss or amputations made up 3%. Non-bite incidents were primarily described as fractures (73%), muscle/tendon/ligament damage (9%), and soft tissue damage (9%). The head was injured in 1 in 7 bite cases.
Most of those injured reported psychological consequences: (90%) of those who were bitten; and 76% of those who were the victims of other types of injuries.
In all, 15% (1 in 7) of all those injured were formally diagnosed with a mental illness as a result of the incident, while 6.5% were diagnosed with a specific phobia and 4% with PTSD. Other mental health consequences included anxiety, disturbed sleep, and avoidance.
Most of the physical injuries sustained resulted in a hospital visit. A quarter of those bitten, and nearly a third of those who weren’t bitten, required surgery.
Of the claimants still working when injured, 59.5% of those bitten and 56% of those who weren’t bitten took time off work, up to a maximum of 5 years. Over half of those bitten (54%) and 41.5% of those who weren’t reported loss of earnings as a result of their injuries.
There is no centralised registry of these cases across all law firms. And the researchers acknowledge that their study was based on data from one law firm and therefore may not be fully representative.
But they suggest: “These data implicate unrestrained dogs in non-residential locations as a major inciting factor for dog-related injuries, and injury prevention strategies need to explore how lead use can be effectively legislated.”
They add: “Most claimants reported that dogs were with their owners and off lead. Almost half of bite and more than 80% of non-bite incidents occurred in non-residential locations and the majority involved unrestrained dogs… These findings raise concerns over owner control.”
Current national legislation on lead control doesn’t affect public highways or urban green spaces, where most injuries occur, they argue.
The Highway Code advises that dogs should be ‘kept on a short lead when walking on the pavement, road or path shared with cyclists or horse riders’, but this is only guidance, not law. And while local authorities can introduce Public Space Protection Orders (PSPOs), it’s unclear how often these are applied or how much of a deterrent they are, they highlight.
“We recommend that national legislation is updated so that all dogs should be on a fixed- length short lead (less than 2 metres) on public highways and in urban green spaces (unless a local authority provides provisions for off-lead areas, or make areas exempt).
“This exemption provision is to ensure that the important balance between public safety and dog welfare can be achieved. This should be partnered with a nationally coordinated public communication campaign,” they conclude.
Journal
Injury Prevention
Method of Research
Data/statistical analysis
Subject of Research
People
Article Title
Using civil claim enquiry data to understand the context and impact of dog- related injuries in England and Wales between 2017 and 2024
Article Publication Date
13-Jan-2026
COI Statement
JM and JMB are current employees of Slee Blackwell Solicitors LLP.
