February 07, 2005

Metapragmatic apologia

Never mind fact checkers and theory checkers, out here in the blogosphere we've got joke checkers.

Let's start with the backstory. Yesterday I referred to an overheard snippet of conversation this way:

Over at In Passing, eve documented a new metapragmatic variable, quoting "a girl to a guy walking down Fulton st" who said "Sure, for values of 'neat' that involve you not getting your security deposit back."

The link on metapragmatic goes to the Jargon File entry for metasyntactic variable, glossed as "[a] name used in examples and understood to stand for whatever thing is under discussion, or any random member of a class of things under discussion". The Jargon File goes on to explain:

Metasyntactic variables are so called because (1) they are variables in the metalanguage used to talk about programs etc; (2) they are variables whose values are often variables (as in usages like “the value of f(foo,bar) is the sum of foo and bar”). However, it has been plausibly suggested that the real reason for the term “metasyntactic variable” is that it sounds good. To some extent, the list of one's preferred metasyntactic variables is a cultural signature.

The "sounds good" part also describes my invention-in-passing of the variant "metapragmatic variable". My reasoning, such as it was, involved several steps, or at least the interaction of several vague ideas. First, the phrase " for values of * that" indexes a subculture that includes programmers as well as mathematicians and others. Second, Michael Silverstein has already used the word metapragmatic to refer to folk reasoning about (reasoning about) meaning in context (in the phrase "metapragmatic ideology", which also sounds good, though I couldn't figure out how to use it in the space and time available). Finally, the originally cited remark was based on treating the interpretation of the word neat as if it were the instantiation of a variable. (I presume, FWIW, that it was in response to some remark like "That was a neat party last night!").

Anyhow, within two hours of posting the item in question, I got a note from Jim Apple:

I don't believe "Sure, for values of 'neat' that involve you not getting your security deposit back," is a use of "neat" as a variable. I think there are two possibilities here:

1. The speaker means "not neat at all", like the old engineers' joke that 2+2=5 for large values of 2.

2. The speaker wants the listener to understand that "neat" occupies a large enough semantic space so that some of the values are not neat enough to get the deposit back.

In either case, the reference to neat depends on its use earlier and on its value as a word. Variables have values that are not so bound. Compare two metasyntactic uses here:

A: I love parties of Canadians.
B: Sure, for large enough n.

A: He uses "adequate" every time he complains to me. "Your foo is not adequate! I am not adequately quxed with your bar! The adequacy of your baz is in question!"

Here, the use of n, foo, bar, qux, or baz is not limited by variables in previous scope. Computer people would call them "implicitly declared."

This is like when the guy on the next barstool taps you on the shoulder and says "I don't believe..." -- sometimes it's best just to say "You're probably right" and move on to another watering hole. Then again, the blogosphere would be a sere and lonely place without intense discussion among strangers about obscure points of interpretation. So...

Look, Jim, here's the first thing. A party is not necessarily as any less "neat" just because it puts the host's security deposit in jeopardy. Security deposits are good, but so is conviviality, and sometimes good things conflict. Engineers deal with this sort of thing all the time under the heading of "joint optimization", right? You and I, as sober adults, may inhabit a world in which apartment-wrecking parties are ipso facto not neat, but this may not be the world of random girls and guys walking down Fulton St.

And in the second place, it's normal in logic and in mathematics for variables to be specified inititially as taking on only certain types of values, or only values in a certain range. The same thing is true for "variables" in most programming languages, which may be restricted as to type or have other conditions placed on their values. I admit that it's not normal to limit the instantiations of a variable to the contextual interpretations of a word -- but the young woman overheard on Fulton St. was using the language of mathematics to express an insight about an aspect of her life not defined by any prior formalism. Our colleagues in the humanities call this a "metaphor", or sometimes a "joke".

And anyhow, you're probably right.

[Update: Jim wrote back with some further thoughts about words and variables, and ended by linking to an older blog post of his own on the "pedantic" setting of some compilers, and remarking that

Oh, bother. My pedantry has surpassed that of the professional pedants. I had a feeling this day would come . . .

Ipse dixit. ]


Posted by Mark Liberman at February 7, 2005 07:51 AM