|
Edited on Sat Oct-03-09 08:35 PM by Igel
Here's my take on it:
Science has a set of heuristics: observe data, go through abductive reasoning for formulate a hypothesis that, ideally, can make a prediction about further data; collect and observe more data, whether from what's predicted or not; compare the additional data to the hypothesis. Along the way you use logic to make sure the hypothesis is consistent, that nothing's left out and that nothing's unnecessarily put in. Logic is a constraint on arguments and a way of saying whether an argument is true or false, and a danged nifty tool for finding error.
But logic isn't a science, except in the etymological sense (it's certainly a kind of scientia). I can't imagine a set of observations that could cause me to reconsider some bit of logic, formulate a hypothesis to "correct" the logical system *as such*, and then test the revised system. If I were to revise it, it would fail to work as it has. It would be something different, not simply logic amended.
It's like mathematics. It's a set of axioms and operations. While I'm fairly sure I couldn't come up with alternative logics, I have a suspicion that we have the logic that we do because it is internally consistent and it works with science. I don't see why it needs to be the only possible set of axioms and operators, but that's not saying that I think there are alternative sets of axioms and operators that are internally consistent, much less find some application (as though that were important). It's the same with mathematics and geometry: For a long time the standard four arithmetic operators and the set of real numbers was sufficient for science, for dealing with reality. Not only was it adequate, it was necessary. It was only later that alternative mathematics and geometries started to be advanced, and a bit longer before they were found useful in areas where the formerly sole mathematics failed.
It may well be that alternative logics would fail to be internally consistent and are impossible. I'm a linguist by training, not a mathematician. I've had basic logic, I've had a semester of formal semantics, I've had a bit of math, but I'm still a linguist. That's not irrelevant info.
One problem I've run across is that formal semantics, based in traditional kinds of first order logic, fails to describe language at all adequately. Yeah, you can capture some nifty generalizations, but there are honking big ones that the logic not only doesn't capture, but hints are impossible. Yet every native speaker of pretty much every natural language performs those "logical" operations daily, and consistently. Moreover, there are operations logic says should be attested, but which no language known in the literature as of 1996 displayed. Second order semantics apparently also fails to handle at least tense and aspect adequately. I killed a couple of afternoons one term watching a visiting German logician explain to a group of semanticians why his approach using higher order logic to account for tense and aspect data was promising--the local guys argued that second order logic handled a lot of the data, but the logician squashed them with a litany of unresolved problems, then proceded to show how first and second order logic, in turn, each failed to account for the data and in some cases inevitably denied the data was possible. This is bad, even in linguistics, and even though we all say language has its own odd "logic" semanticists keep on trying. Since I failed to understand pretty much everything he said about second order logic, well, I didn't attend the lectures in which he developed his ideas.
At the end the German still hadn't cracked the problem--no surprise there--and the local guys were unconvinced he was even on the right track. It may be that we lacked the right semantic types for the operators to work on, and he simply failed to produce them. It might be that the set of operators was impoverished. It might be that traditional logic, devised to deal with the physical world, is a poor fit to human language, where we make all kinds of odd assumptions and connections that simply aren't quite "real". Dunno. I'm not a semanticist, BTW, and haven't seriously thought about semantics since, oh, 2003 (and that wasn't really formal stuff).
Rather than say that logic as a system is broken and needs to be fixed to account for notions of relevance and coherence in discourse pragmatics, not to mention tense and aspect--which is precisely the kind of thing I'd say in a chem lab if some "theory" suddenly failed to account for my data--it strikes me that perhaps it's like trying to apply real numbers to the problems in electrical engineering and fluid dynamics that formed some of the complex analysis problems I had to solve. Then again, what do I know?
Gotta do some work. Later.
|