Logo and Natural Language

Logo is well suited to explorations of natural language. This is because Logo's data structures - words and lists - closely parallel the words, phrases, and sentences that make up spoken and written language.

For example, if we type

print sentence "strawberry [ice cream]

Logo displays

strawberry ice cream

on the screen, having made a sentence out of the word "strawberry and the list of two words [ice cream].

The instruction

print sentence [vanilla fudge] [ice cream]

prints

vanilla fudge ice cream

In Exploring Language with Logo, Paul Goldenberg and Wally Feurzeig begin with some gossip;

to gossip
output sentence who doeswhat
end

to who
output pick [Sandy Dale Dana Chris]
end

to doeswhat
output pick [cheats. [loves to walk.] [talks a mile a minute] yells.]
end

Pick reports an item randomly picked from the list that follows. Who reports a randomly chosen name of a person. Doeswhat reports a randomly selected verb phrase. Gossip puts the two together and reports a sentence.

The instruction print gossip might display

Chris loves to walk.

or

Dana cheats.

or

Dale talks a mile a minute.

or

Dale cheats.

or...

What if we change doeswhat to be:

to doeswhat
output pick [cheats. [loves to walk.] [talks a mile a minute] yells.[ Ford Mustang.]]
end

We could end up with a non-sentence like

Dale Ford Mustang.

So with this playful exercise we can learn something about nouns and verbs.

to pluralize :singular
output word :singular "s
end

print pluralize "dog
dogs

print pluralize "cat
cats

print pluralize "box
boxs

Oops. Time to debug...

to pluralize :singular
if (last :singular) = "x [output word :singular "es]
output word :singular "s
end

print pluralize "box
boxes

Good.

print pluralize "mouse
mouses

looks like we have some more debugging to do...

Instead of being taught the rules of pluralization and then applying them, Logo allows us to construct the rules, building them up from a series of concrete cases.

There's much much more. Check out Paul and Wally's book.