A Meaningful Hump

Epistolary speculations on the statistical analysis of text.
  November 9, 2003

Subject: Re: Hey.
From: A. Baylin
To: P.F.

Hi P.,

It was very nice to meet you, too; I was looking forward to it ever since J. mentioned that you attend these brunches.

The point I was making about Fish’s¹ theory and econometrics (or, more accurately, statistics, since econometrics is just a subset of that) was meant to illustrate how I think the methods of economics may be applied elsewhere.  Unfortunately, my example is moot because it’s utterly impracticable.  I don’t even have a rigorous theory to back it up at this point; mine is very much a raw conjecture in need of fleshing out.  The main problem is that following through all the ramifications eventually snowballs into an extremely unwieldy mental exercise that runs into the twin limits of my knowledge and patience.  Then I usually huff in frustration and go grab lunch.

The main idea is so simple it’s probably trivial: readers read text—readers interpret text—if someone (a telepath) could collect their interpretations and plot them they would fall into some sort of a distribution.  Leaving aside the issue of defining dimensions along which people’s interpretation of text can be measured, we know that as long as the final distribution can be formed, it will have the usual properties: mean-median-mode, skewness, kurtosis, etc.

Its median is a point that interests me.  This is the “hump” in the curve² marking the thickest concurrence in opinion; that is, for all practical purposes, this is what the majority of the readers consider the true meaning of the text.

I understand that the thrust of postmodern theories like Fish’s is to deny the possibility of an “authoritative” interpretation of text by refusing to grant one reader’s vision privilege over others’ because any grounds for such privilege are arbitrary.  My method is the least arbitrary way I can imagine to find such grounds.  The objection can be mounted that trusting in numerical descriptors is as arbitrary as anything else but I tend to dismiss such objections as unproductive because (a) the denial of objective reality is silly, whereas (b) the denial of the intelligibility of objective reality and therefore the attainability of objective truth is paradoxical since such denial itself aspires to the status of objective truth, and finally, (c) there are better things to mount.

A few more quick points, and then I’ll wrap up:

(1) Since I think semiotics, which I’ve never studied, extends the quality of “being a text” to all of reality, the above arguments may apply to our perception of the world in general.

(2) Note that we’re still relying on the readers’ (aggregate) vision to elucidate meaning, completely neglecting author’s intent (such neglect is the tenet of modern literary theory that annoys me the most).  I think an argument can be made that the reader’s median vision on average approaches the value at which author’s intent resides, although certain factors like cultural affinity and the passage of time can certainly interfere with that.  As such, statistics does not provide any normative guidance on who gets the cake: the author or his readers-in-the-aggregate.

(3) Another interesting property of the distribution is kurtosis.  Statisticians, humorless as they are, say that it describes the “fatness of tails.”  In other words, it’s a measure of how “bunched up” your distribution is at the hump.  The flatter is it, the fatter the tails, the less concentrated the agreement on one certain meaning of the text, ergo (a) the more obscure the text³; or (b) the dimmer the readers.  Some combination of (a) and (b) is usually the case, judging from life observations.

That’s where my thinking stands at the moment.  I don’t know if this is important or even intelligible, but feel free to comment or mock at your convenience.

<...>

Cheers,
A.

< | Thoughts Archive | >