3.269 evidence (39)

Wed, 19 Jul 89 19:17:07 EDT

Humanist Discussion Group, Vol. 3, No. 269. Wednesday, 19 Jul 1989.

Date: Tue, 18 Jul 89 21:53:17 CDT
From: Steven J. DeRose <D106GFS@UTARLVM1>
Subject: Evidence

In general I agree with Donald Spaeth's note, pointing out that one
cannot prove a literary point by heaping up examples; but I would
suggest a small adjustment. While it is sometimes the case that
100 examples prove no more than 10, it is also *only* sometimes the case.
It all depends on just what claim the author is making. For example,
in the realms of linguistics and philology, one can examine (say) the
TLG or GramCord or what-have-you and say with substantial authority that
such an event does or does not occur, or always occurs when another does,
**in the extant literature**.

The proviso is important, of course, since as DS rightly pointed out the
surviving works may not be representative. But such a claim, for example
that a particular Greek verb occurs *only* in funeral narrations in
the surviving literature, may be a significant and interesting one.

Another use for the computer is in finding not *positive* but *negative*
truths. If one is planning to state a universal claim, one is well
advised to check as thoroughly as possible for counterexample. Writing
a paper on certain syntactic constructions in Greek, I found a source
which made a universal claim about syntax; so I scanned the TLG for the
verbs involved, and found that while the claim was true, common elements
in all the syntactic contexts made it predictable on other grounds, and
therefore less significant. Without checking a large number of examples,
the more comprehensive generalization would have been hard to see.

Steve DeRose