It’s been a while since I’ve shared a post on DR just for the sheer coolness factor, but there’s no way I couldn’t pass along the Washington Post’s article on Google’s natural language parser: Parsey McParseface. What the parser does is fairly easy to explain:
McParseface does what most students learn to do in elementary school. It takes a sentence and breaks it down, identifying nouns, verbs and so forth — and how all of these parts relate to one another. It can tell you, for instance, what the root verb of a sentence is; what is being done to whom and who is doing it.
Along the way, however, the article introduced me to the notion of garden path sentences. These are sentences which, although they are perfectly correct, are incredibly hard for human brains (and AI language parsers) to figure out. Here are two of my favorite examples (from the article):
- The horse raced past the barn fell.
- The old man the boat.
- The coach smiled at the player tossed the frisbee.
Yes, those are all really real sentence. The funny thing is that, now that I’ve wrapped my head around them, it’s hard for me not to understand them effortlessly when I read them. But I remember that the first time I tried to figure them out, it was like a gear in my brain was stuck. They’re called garden path sentences because they are deliberately constructed to make your brain think a word is being used in a certain way when, to understand the sentence, you have to let go of the initial impression and reinterpret a word to make sense of the sentence. So, for example, “the old man” is a phrase that naturally makes you think of the word “man” being used a noun. But, in order to understand that sentence, you have to discard your first impression and realize that it’s being used as a verb in this case: The old [people are the ones who man] the boat.
While I was reading, I couldn’t help but remembering a book I read a couple of years ago called Lexicon.1 The central conceit of the book is that if you can master language at a truly fundamental level–beneath the level of conscious human understanding–you can use the constituent sounds and their implications to effectively mind-control people. It’s an incredibly cool blend of fantasy magic and technology in a way I’ve never quite seen anywhere else, and feeling your brain struggle to wrap itself around those garden path words is definitely the closest real-life experience you can get to the fantastic mind-powers discussed in the book.
2 thoughts on “Linguistics, AI, and Garden Path Sentences”
“The central conceit of the book is that if you can master language at a truly fundamental level–beneath the level of conscious human understanding–you can use the constituent sounds and their implications to effectively mind-control people. ”
Sounds like the ancient Greek Sophist, Gorgias:
“Speech is a powerful lord, which by means of the finest and most invisible body effects the divinest works: it can stop fear and banish grief and create joy and nurture pity . . .
Fearful shuddering and tearful pity and grievous longing come upon its hearers, and at the actions and physical sufferings of others in good fortunes and in evil fortunes, through the agency of words, the soul is wont to experience a suffering of its own . . .
Sacred incantations sung with words are bearers of pleasure and banishers of pain, for, merging with opinion in the soul, the power of the incantation is wont to beguile it and persuade it and alter it by witchcraft. There have been discovered two arts of witchcraft and magic: one consists of errors of soul and the other of deceptions of opinion . . .
The effect of speech upon the condition of the soul is comparable to the power of drugs over the nature of bodies. For just as different drugs dispel different secretions form the body, and some bring an end to disease and others to life, so also in the case of speeches, some distress, others delight, some cause fear, others make the hearers bold, and some drug and bewitch the soul with a kind of evil persuasion.”
That is a really cool reference, Ivan. Thanks.
Comments are closed.