Skip to main content

How to Teach Language to Dogs

AFTER A LONG DAY of being a dog, no dog in existence has ever curled up on a comfy couch to settle in with a good book. Dogs just don’t roll like that.

This article was published in Scientific American’s former blog network and reflects the views of the author, not necessarily those of Scientific American


AFTER A LONG DAY of being a dog, no dog in existence has ever curled up on a comfy couch to settle in with a good book. Dogs just don’t roll like that. But that shouldn’t imply that human words don’t or can’t have meaning for dogs.

Chaser, a Border Collie from South Carolina, first entered the news in 2011 when a Behavioral Processes paper reported she had learned and retained the distinct names of over 1,000 objects. But that’s not all. When tested on the ability to associate a novel word with an unfamiliar item, she could do that, too. She also learned that different objects fell into different categories: certain things are general “toys,” while others are the more specific “Frisbees” and, of course, there are many, many exciting “balls.” She differentiates between object labels and action commands, interpreting “fetch sock” as two separate words, not as the single phrase “fetchsock.”

Fast forward two years. Chaser and her owner and trainer Dr. John Pilley, an emeritus professor of psychology at Wofford College, appeared again in a scientific journal. This time, the study highlighted Chaser’s attention to the syntactical relationships between words, for example, differentiating “to ball take Frisbee” from “to Frisbee take ball.”


On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.


I’ve been keeping an eye on Chaser, and I’ve been keeping an eye on Rico, Sofia, Bailey, Paddy and Betsy, all companion dogs whose way with human language has been reported in scientific journals. Most media reports tend to focus on outcomes: what these dogs can -- or can’t -- do with our words. But I think these reports are missing the point. Learning the names of over 1,000 words doesn’t just happen overnight. What does the behind-the-scenes learning and training look like? How did Chaser develop this intimate relationship with human language?

Recently, I had the pleasure of making Chaser and John Pilley’s acquaintance when they visited New York City for their forthcoming book, Chaser: Unlocking the Genius of the Dog Who Knows a Thousand Words, out 10.29.13 (Facebook/Twitter). I sat down with Pilley to discuss Chaser’s initial training and learning and to pose questions offered up by readers of Dog Spies (Twitter /Facebook) and Do You Believe in Dog? (Twitter /Facebook). Pilley did most of the talking, and Chaser contributed too, periodically bringing me a Frisbee and sticks of various sizes to throw. As Pilley says, “Learning builds on learning,” so here’s what I learned from our conversation about the early building blocks of Chaser’s education:

Chaser is known for her extensive vocabulary. Can you walk us through how she came to learn the names of over 1,000 different objects?

Because learning depends upon learning, it helped to have her first learn certain simple behaviors, most of which were obedience behaviors like sit, stand, stay, and drop. She also learned herding commands like ‘way to me’, ‘come by,‘ ‘there,’ ‘freeze,’ ‘walk,’ ‘go out,’ and ‘crawl’ among others. I used positive reinforcement, and in the beginning, I used food when she came to me. I gradually phased out the food.

When I started teaching the name of objects, I said, ‘Watch Pop Pop’ and would show her the object and say, “This is Frisbee,” and I would hide the object just two feet away, so at that point, it wasn’t really hidden. When she went looking for it, I would keep saying, ‘Find Frisbee, find Frisbee,’ and when she mouthed it I would say, ‘Good dog! Good dog!’ Then we made it more difficult, and I would really hide it out of her sight.

That was the procedure. A couple of times a day, I would play with her with that object for five minutes or so. No other objects would be on the floor, and we would play with that object, and I would profusely repeat the name of that object between 20-40 times in each session. ‘Find Santa Claus, Find Santa Claus,’ and then, ‘Good dog! Good dog!’ when she mouthed it. That was the first step in teaching her the names of proper nouns.

After the first object was taught, it was set aside, and we worked on another one. She always learned one object at a time, a successive method of teaching. Before Chaser, we had tried a simultaneous method of teaching with other dogs [presenting two objects at a time], and that didn’t work.

As Chaser learned the names of different objects, how did you test what Chaser was learning? Did you have a way to assess her initial and long-term memory?

After a few weeks, it was time to test Chaser. The criteria for learning was that she had to successfully select a particular object out of 8 other objects. My criteria was not just that one test. We replicated it at least 8 times, and the items surrounding the object of interest would be different from what she had seen in earlier presentations. She had to be successful in each of those independent sessions 8 times in a row. If she was successful on the first 6 but missed on the 7th, I stopped the testing and put the object aside and worked with it some more. Then I would start the testing some more.

In addition to testing her initial learning of objects, each month we tested her memory on all of the objects that she had learned up to that point. We did it in sets of 20. I put 20 objects on the floor, using only objects which she had shown initial learning, and we asked Chaser to retrieve each of the 20 objects. Over the course of 3 years of teaching, Chaser was successful in retrieving 90% of the objects correctly. Most of the time she retrieved 100% of the objects and other times 95%, but it never got below 90%. If she was not successful, I put that object aside for additional training. The reason we tested her each month was we wanted to make sure that she was not learning new objects to replace an old object. We wanted to be sure that there was complete retention in the long term.

How about grammar? What were the general training and testing procedures?

Chaser initially learned actions independent of specific objects. I might be holding a piece of food, and I would say, ‘Take.’ Over time, she learned the three commands, 'take,' 'paw,' and 'nose' and they were later applied to different objects. The first step in grammar is two elements like, 'Take Lamb,’ and Chaser demonstrated that each word had an independent meaning. I like to say that learning builds on learning. Not until she learned the two elements should she learn three elements. When she heard the phrase, ‘Take ball to Frisbee,’ she was successful three fourths of the time. But she seemed to act on the last thing she heard, so we changed the expression to ‘To ball take Frisbee.’ We also tested the inverse, ‘To Frisbee take ball.’ This work tests her syntax, her understanding of the rule. When we invert it, we’re demonstrating semantics, that she actually understands each of those components of the sentence and the different meanings of the words.

Can you talk about the use of play in Chaser’s learning and training?

Tension, fear and anxiety inhibit creative learning, and play and tension are incompatible. During play, creative learning can take place. If you are going for complex learning, you are not going to get any learning with tension and anxiety; I’ve learned that in over 30 years of teaching. One of the things that runs throughout the entire book, Chaser: Unlocking the Genius of the Dog Who Knows a Thousand Words, is that play is one of the secrets of Chaser’s learning.

Most people don’t explicitly teach their dog words, yet they think that the dog understands many words. Do you think this assumption could ever be problematic?

As long as this doesn’t result in some kind of negative response like criticism or punishing the dog, it could be innocuous. It could have a negative impact on the dog if someone thinks a dog should know a word or phrase and holds it against the dog if he doesn’t seem to be performing. The best and simplest assumption is that if a dog is not emitting a behavior, then the dog might not understand.

Does knowing words and syntax help Chaser have a "happier" life with humans?

I think when you have more understanding, there is better communication and thereby more happiness.

What do you hope people take away from Chaser’s story?

The big thing is that Chaser is not special, and that anybody’s dog is smarter than he or she thinks. I’m hoping people will work with their dogs, of course teaching them words, but also getting to know their dogs. Find out what makes your dog happy, and give your dog opportunities to explore its interests.

Image: Julie Hecht and Chaser via Julie Hecht. Chaser and John Pilley via © Mark Olencki. Interview from July 17, 2013 shortened for publication.

Related Reading

Hecht, J. 2012. Say What? Do Dogs Understand Our Words? The Bark Magazine

Goldman, J. Monday Pets: How Do Dogs Learn New Words? Scientific American. May 10, 2010.

Pilley, J.W. with Hilary Hinzmann. Chaser: Unlocking the Genius of the Dog Who Knows a Thousand Words. Houghton Mifflin Harcourt. October 2013 Release

References

Pilley, J.W., and A.K. Reid. 2011. Border collie comprehends object names as verbal referents. Behavioural Processes 86: 184–195.

Pilley, J.W. 2013. Border collie comprehends sentences containing a prepositional object, verb and direct object. Learning and Motivation. Published online May 13, 2013.