Taxonomy for Dogs and Children
When my niece Susan (not her real name) was a baby, she adored my big dog Mina. One day, Susan and Mina were sitting with me and playing a game. I would point to my eye, and Susan would point to her eye. Then I would point to the dog, and Susan would point to Mina's eye. Then I would say "Susan's eye" and Susan would point to her eye.
We played "eye," and "head," and "ear." Susan hesitated, but then lifted one of Mina's floppy ears and laughed. An ear is still an ear, even when it's furry and floppy. Then we switched, and I pointed to Mina's mouth, and then to mine. Susan said, "mouth" and pointed to her mouth. I said, "Auntie Lynn's mouth," and Susan pointed to my mouth. Then I pointed to the dog, said, "Mina," and pointed to Susan. Susan pointed to her chest and said proudly, "Susan!"
Susan was learning both the names for parts of the body, and that those names belonged to individuals. She also knew that there was a unique name for each whole. At two years old, she had a basic grasp of taxonomy and metadata. She knew simple terms, and could use them as labels on an object. She could do this visually, well before she could write, and even before she could speak full sentences.
Susan was on a roll. I pointed to my foot, and Susan said "foot." Then I tickled Susan's feet, and she said, "feet." She knew a plural term. She pointed to Mina's hind feet and said, "feet." Then to Mina's front feet, and looked puzzled.
I waited. She looked at her own hands. She knew "hands" without saying the word. Then she looked at Mina's front paws. Her eyes snapped from hind to front, and the penny dropped. "Feet. Feet." She pointed, and said it again, eyes wide with amazement. "Feet, FEET!" It was an epiphany. Dogs had all feet, and no hands. Suddenly, Mina wasn't just Susan's beloved friend. She was an individual representative of a different species. Susan said, "Dog!" She knew the species name.
It would be a while before Susan would grasp the abstract concept of "person," let alone "human," as an equivalent to "dog." It is hard at two years old to understand abstract terms that apply to yourself. However, Susan had a grasp of that most abstract thing, "self," and could apply it. "Who are you?" "Susan!"
She knew her own name.
Human beings are natural taxonomists. Language is a term set that we apply to the world to understand abstract concepts and organize them into relationships. This is a core difference between humans and dogs.
Or is it? We had been playing with nouns, which are abstract names for categories. Elementary school children learn this as "person, place or thing." But Mina knew her own name, and perked up when either of us said, "Mina." Dogs, on the whole, are better with verbs. Most dogs can even understand simple sentences. "Want to go for a walk?" Mina jumped to her feet (all four of them). So did Susan and Sarah. Clearly, the game was over.
Dogs, and children, hear words that they know and understand more clearly than words they don't. Children, unlike dogs, learn to talk as well as listen, and to write as well as \speak. Vocabulary grows as we develop. We can write Susan's simple vocabulary as either a list of words (a folksonomy) or as a structured hierarchical grammar (taxonomy). It is easier to learn words as a folksonomy.
FOLKSONOMY
Hands eye Susan feet
nose Auntie Lynn head want mouth dog Sarah go foot sister walk
As our vocabulary grows, we learn to organize words into complete sentences. "This is my sister Susan. Want to go for a walk?"
We learn grammar first in spoken language,. Only later do we learn formal grammar as a structured set of abstract rules for organizing sentences. This ability to learn "natural language" is a key difference between humans and computers, which are not as smart as dogs. AI and machine learning are redefining "smart" here, but they are still largely dependent on controlled vocabularies and structured taxonomies.
TAXONOMY
(Noun)
(Name)
Alice
Rose
Mina
Lynn
(Me)
(Person)
(Body)
Hands
Hand
Eyes
Eye
Nose
Mouth
Feet
Foot
Dog
(Body)
Eyes
Eye
Nose
Mouth
Feet
Foot
(Family)
Aunt
Sister
(Verb)
Want
Go
We can use both natural language and visualization to learn and construct taxonomies. Indeed, it is much easier to learn this way. Picture books associate images with words, making learning so simple that even a child can do it. But individuals have different learning styles and capabilities. Sarah is in art college now, and Susan, as a high school senior, is already an accomplished writer. Some people are visual learners, some learn with words.
However, words are more compact than pictures. Look at the picture of Susan and Mina, with and without the labels. There is only room for three or four labels on the picture (Susan, Mina, hands, feet). The picture also has things that aren't in the vocabulary (rug, blocks, bookcase). And where would we place the labels for "sister" and "walk"?
A visual model of a taxonomy, then would have a number of features. It would be able to:
- Illustrate concepts with images
- Associate images with words
- Show associations between words
- Be compact for easy viewing
- Expand to multiple levels
- Show terms as folksonomies
- Show structured terms as taxonomies
- Show relationships among terms, both hierarchically and flat
- Illustrate and distinguish between categories, names, and individuals
- Show nouns, verbs, and adjectives visually
- Be understandable by a preschooler
Tall order. But a new class of visual modeling tools can help us learn as simply as Susan learns. Check out Whole Elephant Modeling, a 3D visualization of natural language ontologies for architectural frameworks.
Comments