Stuff for the Ironing Pile

"This is an interesting avenue of thought. I should iron this out, and then write about it on that blog I allegedly maintain."

Who has time for ironing?

Can you computationally discriminate obscurantist nonsense from legitimately hard subject matter on a linguistic basis?

Kicked off by this post on Hegel. Are there linguistic (lexical, syntactic, whatever) differences between texts that are deliberately written to sound impenetrably profound, texts that are explaining something badly, and texts that are working with legitimately complex subject matter? Could you write a document classifier to discriminate between the different cases?

Related #1: is "bad communication" stylistically convergent (characterised by common amateurish mistakes), or stylistically divergent (the result of any number of deleterous mistakes, cf. the Anna Karenina Principle)?

Related #2: It's my observation that people emulate the written style of their influences. Is it possible to track this?

Why do people want to pathologically classify things?

I occasionally find myself in "multidisciplinary" study groups. Just recently I've been doing a "for fun" MOOC on political economy, and the discussion forums are particularly bad for a behaviour which I've noticed before but not thought about too much: obsessive classification. Is this an art or a science? Is this a positivist or an interpretivist approach? Is this a left-wing or right-wing idea? Is it holistic or reductionist? This seems to be universally pointless class of discussion. Why is it so pervasive?

One Thought on “Stuff for the Ironing Pile

  1. I suspect one could train a document classifier to do better than chance at discriminating deliberately obscure texts from unintentionally obscure texts from clear expositions of difficult ideas, and that "bad communication" is partly stylistically convergent. (Needlessly obscure diction and wordiness seem quite common.) But yeah, I think the AK principle does apply here too.

    "It's my observation that people emulate the written style of their influences. Is it possible to track this?"

    I'm optimistic about that one. People already use stylometry for automated authorship attribution, as far as I know successfully, and tracking writing style influences is quite close to that.

    (Kind of amused by the idea that this might be an example of the pathological urge to classify you mention near the end!)

Leave a Reply

Your email address will not be published. Required fields are marked *

Post Navigation