Using NLP to Defeat NLP

Γραμματο-Χαοτικον

We are the Γραμματο-Χαοτικον, an underground alliance of linguists, philologists, and polyglots. Our self-appointed role is to encourage arbitrary and capricious change both in Language and among languages, world-wide. Our exploits are legion, and now reasonably well documented (see “The Γραμματο-Χαοτικον Manifesto”, SpecGram CL.4).

We seek to enrich the languages of the world through arbitrary and capricious change, but the statistical servants of stupidity, those mathematical minions of moronitude commonly referred to as “computational linguists” are trying to ruin our fun. Practitioners of so-called “Natural Language Processing” engage in the unnatural; like Icarus, they fly too high—their goals are both too lofty and too pedestrian: they seek a muddled ersatz understanding of language.

Know the enemy!

We briefly considered the delicious irony of using neurolinguistic programming to defeat natural language processing, but neurolinguistic programming represents a level of crap beyond even natural language processing. At least natural language processing is something, rather than a mixed bag of nothing.

So we settled on the fairly tasty irony of using natural language processing against itself. A common tool of computational linguists is the humble pattern, used, for example, to extract the so-called “relationships” between “words” or “concepts” or “entities”.

A well-known set of such patterns from “Automatic Acquisition of Hyponyms from Large Text Corpora” (Hearst, 1992) that purport to indicate that X “is a” Y include:

• Y such as X
• such Y as X
• X, or other Y
• X, and other Y
• Y, including X
• Y, especially X

We can use the techniques of NLP to identify patterns such as these, then go out of our way to construct locutions with atypical, nay pathological, instances of these patterns, rendering them useless:

• Patterns such as Hearst discusses have become useless.
• Felons such as Noam Chomsky talks about are often hardened criminals.
  
  
• Such filthy dogs as your mom rescues and cleans up deserve a good home.
• Such inbred cretins as sociolinguists have to deal with would be too much for me.
  
  
• Those who reject lexicalists, and other upstanding people, are welcome.
• We will not discuss creatures hypothesized by evolutionary psychologists, or other imaginary beings.
  
  
• The entire university community should be aware of the infestation of bedbugs, including grad students and undergrads.
• Syntacticians are less likely than other more anthropologically-inclined linguists to need to eat cockroaches, especially documentary linguists.

So what if some of these constructions seem a little off the first hundred times you say them? You’ll get used to them in time—everyone who has sat through a syntax class knows it’s true!

Conclusion

We do not welcome our new robot overlords. Arigatōmasen Mr. Robotōmasen! Fight the Power!

—Γραμματο-Χαοτικον

	Rethinking Social Sciences Under the Influence of N-Bar Theory—Nikita Barbarosa, Nigel Barclay, Norberto Barrowman, & Vino Heineken
	English Writing System Reform—An Immodest Proposal—Wekkan Shanj LeMonde, B.F.A., M.Eng., Ph.D. and V. Villma Kind-Ifferenz, A.Z., α.Ω., Q.E.D.
	SpecGram Vol CLXV, No 4 Contents